From a5c24a786e46f7bff73eb741d0c0d1878c43e39d Mon Sep 17 00:00:00 2001
From: Miguel Alonso Jr <miguel.alonso@unity3d.com>
Date: Tue, 2 Apr 2024 05:48:48 -0400
Subject: [PATCH] Deployed fb2af761a with MkDocs version: 1.5.3

---
 404.html                                      |  70 +++++++----
 API-Reference/index.html                      |  77 ++++++++----
 Background-Machine-Learning/index.html        |  77 ++++++++----
 Background-PyTorch/index.html                 |  79 ++++++++----
 Background-Unity/index.html                   |  77 ++++++++----
 CODE_OF_CONDUCT/index.html                    |  77 ++++++++----
 CONTRIBUTING/index.html                       |  77 ++++++++----
 Custom-SideChannels/index.html                |  77 ++++++++----
 ELO-Rating-System/index.html                  |  77 ++++++++----
 FAQ/index.html                                |  77 ++++++++----
 Getting-Started/index.html                    |  77 ++++++++----
 Glossary/index.html                           |  77 ++++++++----
 Hugging-Face-Integration/index.html           |  77 ++++++++----
 Installation-Anaconda-Windows/index.html      |  77 ++++++++----
 Installation/index.html                       | 117 +++++++++++++-----
 Integrations-Match3/index.html                |  77 ++++++++----
 Integrations/index.html                       |  77 ++++++++----
 LICENSE/index.html                            |  77 ++++++++----
 Learning-Environment-Create-New/index.html    |  77 ++++++++----
 Learning-Environment-Design-Agents/index.html |  89 ++++++++-----
 Learning-Environment-Design/index.html        |  77 ++++++++----
 Learning-Environment-Examples/index.html      |  79 ++++++++----
 Learning-Environment-Executable/index.html    |  77 ++++++++----
 Limitations/index.html                        |  77 ++++++++----
 ML-Agents-Envs-README/index.html              |  77 ++++++++----
 ML-Agents-Overview/index.html                 |  97 +++++++++------
 ML-Agents-README/index.html                   |  77 ++++++++----
 ML-Agents-Toolkit-Documentation/index.html    |  77 ++++++++----
 Migrating/index.html                          | 107 ++++++++++------
 Package-Settings/index.html                   |  81 +++++++-----
 Profiling-Python/index.html                   |  79 ++++++++----
 Python-Custom-Trainer-Plugin/index.html       |  79 ++++++++----
 Python-Gym-API-Documentation/index.html       |  77 ++++++++----
 Python-Gym-API/index.html                     |  79 ++++++++----
 Python-LLAPI-Documentation/index.html         |  81 +++++++-----
 Python-LLAPI/index.html                       |  77 ++++++++----
 .../index.html                                |  79 ++++++++----
 Python-Optimizer-Documentation/index.html     |  81 +++++++-----
 .../index.html                                |  77 ++++++++----
 Python-PettingZoo-API/index.html              |  87 ++++++++-----
 Readme/index.html                             | 103 +++++++++------
 Sentis/index.html                             |  77 ++++++++----
 Training-Configuration-File/index.html        |  79 ++++++++----
 Training-ML-Agents/index.html                 |  79 ++++++++----
 Training-Plugins/index.html                   |  77 ++++++++----
 Training-on-Amazon-Web-Service/index.html     |  77 ++++++++----
 Training-on-Microsoft-Azure/index.html        |  83 ++++++++-----
 Tutorial-Custom-Trainer-Plugin/index.html     |  81 +++++++-----
 Unity-Environment-Registry/index.html         |  93 +++++++++-----
 Using-Docker/index.html                       |  77 ++++++++----
 Using-Tensorboard/index.html                  |  79 ++++++++----
 Using-Virtual-Environment/index.html          |  79 ++++++++----
 Versioning/index.html                         |  77 ++++++++----
 assets/javascripts/bundle.1e8ae164.min.js     |  29 +++++
 assets/javascripts/bundle.1e8ae164.min.js.map |   7 ++
 assets/javascripts/bundle.6c14ae12.min.js     |  29 -----
 assets/javascripts/bundle.6c14ae12.min.js.map |   7 --
 ...f886a092.min.js => search.b8dbb3d2.min.js} |   2 +-
 ....min.js.map => search.b8dbb3d2.min.js.map} |   2 +-
 assets/stylesheets/main.bcfcd587.min.css      |   1 +
 assets/stylesheets/main.bcfcd587.min.css.map  |   1 +
 assets/stylesheets/main.fad675c6.min.css      |   1 -
 assets/stylesheets/main.fad675c6.min.css.map  |   1 -
 assets/stylesheets/palette.06af60db.min.css   |   1 +
 .../stylesheets/palette.06af60db.min.css.map  |   1 +
 assets/stylesheets/palette.356b1318.min.css   |   1 -
 .../stylesheets/palette.356b1318.min.css.map  |   1 -
 com.unity.ml-agents.extensions/index.html     |  77 ++++++++----
 com.unity.ml-agents/index.html                |  77 ++++++++----
 doxygen/Readme/index.html                     |  77 ++++++++----
 index.html                                    | 103 +++++++++------
 ml-agents-envs/index.html                     |  77 ++++++++----
 ml-agents/index.html                          |  77 ++++++++----
 search/search_index.json                      |   2 +-
 sitemap.xml                                   | 116 ++++++++---------
 sitemap.xml.gz                                | Bin 801 -> 800 bytes
 76 files changed, 3233 insertions(+), 1727 deletions(-)
 create mode 100644 assets/javascripts/bundle.1e8ae164.min.js
 create mode 100644 assets/javascripts/bundle.1e8ae164.min.js.map
 delete mode 100644 assets/javascripts/bundle.6c14ae12.min.js
 delete mode 100644 assets/javascripts/bundle.6c14ae12.min.js.map
 rename assets/javascripts/workers/{search.f886a092.min.js => search.b8dbb3d2.min.js} (99%)
 rename assets/javascripts/workers/{search.f886a092.min.js.map => search.b8dbb3d2.min.js.map} (79%)
 create mode 100644 assets/stylesheets/main.bcfcd587.min.css
 create mode 100644 assets/stylesheets/main.bcfcd587.min.css.map
 delete mode 100644 assets/stylesheets/main.fad675c6.min.css
 delete mode 100644 assets/stylesheets/main.fad675c6.min.css.map
 create mode 100644 assets/stylesheets/palette.06af60db.min.css
 create mode 100644 assets/stylesheets/palette.06af60db.min.css.map
 delete mode 100644 assets/stylesheets/palette.356b1318.min.css
 delete mode 100644 assets/stylesheets/palette.356b1318.min.css.map

diff --git a/404.html b/404.html
index 4697012553..deb9499b82 100644
--- a/404.html
+++ b/404.html
@@ -16,7 +16,7 @@
       
       
       <link rel="icon" href="/ml-agents/images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -24,10 +24,10 @@
       
     
     
-      <link rel="stylesheet" href="/ml-agents/assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="/ml-agents/assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="/ml-agents/assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="/ml-agents/assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -68,9 +68,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -120,21 +117,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -142,9 +139,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,6 +150,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -196,7 +196,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -241,7 +241,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -256,6 +256,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/." class="md-nav__link">
         
@@ -275,6 +276,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/ML-Agents-Overview/" class="md-nav__link">
         
@@ -294,6 +296,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Installation/" class="md-nav__link">
         
@@ -313,6 +316,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -332,7 +336,7 @@
   
   
   
-    
+  
     
     
     
@@ -341,7 +345,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -368,6 +371,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Background-Machine-Learning/" class="md-nav__link">
         
@@ -388,6 +392,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Background-PyTorch/" class="md-nav__link">
         
@@ -408,6 +413,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Background-Unity/" class="md-nav__link">
         
@@ -428,6 +434,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/ELO-Rating-System/" class="md-nav__link">
         
@@ -455,7 +462,7 @@
   
   
   
-    
+  
     
     
     
@@ -464,7 +471,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -491,6 +497,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Python-Gym-API/" class="md-nav__link">
         
@@ -511,6 +518,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Python-PettingZoo-API/" class="md-nav__link">
         
@@ -531,6 +539,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Python-LLAPI/" class="md-nav__link">
         
@@ -558,7 +567,7 @@
   
   
   
-    
+  
     
     
     
@@ -567,7 +576,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -594,6 +602,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -614,6 +623,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -634,6 +644,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -654,6 +665,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -681,7 +693,7 @@
   
   
   
-    
+  
     
     
     
@@ -690,7 +702,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -717,6 +728,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Training-Plugins/" class="md-nav__link">
         
@@ -737,6 +749,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -757,6 +770,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Hugging-Face-Integration/" class="md-nav__link">
         
@@ -784,7 +798,7 @@
   
   
   
-    
+  
     
     
     
@@ -793,7 +807,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -820,6 +833,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/FAQ/" class="md-nav__link">
         
@@ -840,6 +854,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Limitations/" class="md-nav__link">
         
@@ -860,6 +875,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Migrating/" class="md-nav__link">
         
@@ -880,6 +896,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="/ml-agents/Versioning/" class="md-nav__link">
         
@@ -936,6 +953,7 @@ <h1>404 - Not found</h1>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -968,10 +986,10 @@ <h1>404 - Not found</h1>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "/ml-agents/", "features": [], "search": "/ml-agents/assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "/ml-agents/", "features": [], "search": "/ml-agents/assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="/ml-agents/assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="/ml-agents/assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/API-Reference/index.html b/API-Reference/index.html
index bf58042d5b..d838422f61 100644
--- a/API-Reference/index.html
+++ b/API-Reference/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -964,11 +981,19 @@ <h1 id="api-reference">API Reference</h1>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1001,10 +1026,10 @@ <h1 id="api-reference">API Reference</h1>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Background-Machine-Learning/index.html b/Background-Machine-Learning/index.html
index 1feb71f706..2e6d15c63f 100644
--- a/Background-Machine-Learning/index.html
+++ b/Background-Machine-Learning/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -345,7 +349,7 @@
     
   
   
-    
+  
     
     
     
@@ -354,7 +358,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" checked>
         
           
@@ -383,6 +386,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -486,6 +490,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -506,6 +511,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -526,6 +532,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -553,7 +560,7 @@
   
   
   
-    
+  
     
     
     
@@ -562,7 +569,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -589,6 +595,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -609,6 +616,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -629,6 +637,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -656,7 +665,7 @@
   
   
   
-    
+  
     
     
     
@@ -665,7 +674,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -692,6 +700,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -712,6 +721,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -732,6 +742,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -752,6 +763,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -779,7 +791,7 @@
   
   
   
-    
+  
     
     
     
@@ -788,7 +800,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -815,6 +826,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -835,6 +847,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -855,6 +868,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -882,7 +896,7 @@
   
   
   
-    
+  
     
     
     
@@ -891,7 +905,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -918,6 +931,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -938,6 +952,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -958,6 +973,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -978,6 +994,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1269,11 +1286,19 @@ <h2 id="deep-learning">Deep Learning</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1306,10 +1331,10 @@ <h2 id="deep-learning">Deep Learning</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Background-PyTorch/index.html b/Background-PyTorch/index.html
index d5645db548..de10c41766 100644
--- a/Background-PyTorch/index.html
+++ b/Background-PyTorch/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -345,7 +349,7 @@
     
   
   
-    
+  
     
     
     
@@ -354,7 +358,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" checked>
         
           
@@ -381,6 +384,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -403,6 +407,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -479,6 +484,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -499,6 +505,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -526,7 +533,7 @@
   
   
   
-    
+  
     
     
     
@@ -535,7 +542,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -562,6 +568,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -582,6 +589,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -602,6 +610,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -629,7 +638,7 @@
   
   
   
-    
+  
     
     
     
@@ -638,7 +647,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -665,6 +673,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -685,6 +694,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -705,6 +715,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -725,6 +736,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -752,7 +764,7 @@
   
   
   
-    
+  
     
     
     
@@ -761,7 +773,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -788,6 +799,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -808,6 +820,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -828,6 +841,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -855,7 +869,7 @@
   
   
   
-    
+  
     
     
     
@@ -864,7 +878,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -891,6 +904,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -911,6 +925,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -931,6 +946,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -951,6 +967,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1056,7 +1073,7 @@ <h2 id="tensorboard">TensorBoard</h2>
 certain model attributes (called <em>hyperparameters</em>). Finding the right values of
 these hyperparameters can require a few iterations. Consequently, we leverage a
 visualization tool called
-<a href="https://www.tensorflow.org/programmers_guide/summaries_and_tensorboard">TensorBoard</a>.
+<a href="https://www.tensorflow.org/tensorboard">TensorBoard</a>.
 It allows the visualization of certain agent attributes (e.g. reward) throughout
 training which can be helpful in both building intuitions for the different
 hyperparameters and setting the optimal values for your Unity environment. We
@@ -1070,11 +1087,19 @@ <h2 id="tensorboard">TensorBoard</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1107,10 +1132,10 @@ <h2 id="tensorboard">TensorBoard</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Background-Unity/index.html b/Background-Unity/index.html
index e05f618e3d..8446682200 100644
--- a/Background-Unity/index.html
+++ b/Background-Unity/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -345,7 +349,7 @@
     
   
   
-    
+  
     
     
     
@@ -354,7 +358,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" checked>
         
           
@@ -381,6 +384,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -401,6 +405,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -423,6 +428,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -451,6 +457,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -478,7 +485,7 @@
   
   
   
-    
+  
     
     
     
@@ -487,7 +494,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -514,6 +520,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -534,6 +541,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -554,6 +562,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -581,7 +590,7 @@
   
   
   
-    
+  
     
     
     
@@ -590,7 +599,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -617,6 +625,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -637,6 +646,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -657,6 +667,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -677,6 +688,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -704,7 +716,7 @@
   
   
   
-    
+  
     
     
     
@@ -713,7 +725,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -740,6 +751,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -760,6 +772,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -780,6 +793,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -807,7 +821,7 @@
   
   
   
-    
+  
     
     
     
@@ -816,7 +830,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -843,6 +856,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -863,6 +877,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -883,6 +898,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -903,6 +919,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -985,11 +1002,19 @@ <h1 id="background-unity">Background: Unity</h1>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1022,10 +1047,10 @@ <h1 id="background-unity">Background: Unity</h1>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/CODE_OF_CONDUCT/index.html b/CODE_OF_CONDUCT/index.html
index 2e790cdecd..1f71ae4d8e 100644
--- a/CODE_OF_CONDUCT/index.html
+++ b/CODE_OF_CONDUCT/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1071,11 +1088,19 @@ <h2 id="attribution">Attribution</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1108,10 +1133,10 @@ <h2 id="attribution">Attribution</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/CONTRIBUTING/index.html b/CONTRIBUTING/index.html
index 332d53d1df..8082f84466 100644
--- a/CONTRIBUTING/index.html
+++ b/CONTRIBUTING/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1068,11 +1085,19 @@ <h2 id="9-merge-your-changes">9. Merge your changes</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1105,10 +1130,10 @@ <h2 id="9-merge-your-changes">9. Merge your changes</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Custom-SideChannels/index.html b/Custom-SideChannels/index.html
index f45267966a..9cffe8fb3b 100644
--- a/Custom-SideChannels/index.html
+++ b/Custom-SideChannels/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1222,11 +1239,19 @@ <h3 id="example-python-code">Example Python code</h3>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1259,10 +1284,10 @@ <h3 id="example-python-code">Example Python code</h3>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/ELO-Rating-System/index.html b/ELO-Rating-System/index.html
index 1b86b2c07d..405743b7d8 100644
--- a/ELO-Rating-System/index.html
+++ b/ELO-Rating-System/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -345,7 +349,7 @@
     
   
   
-    
+  
     
     
     
@@ -354,7 +358,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" checked>
         
           
@@ -381,6 +384,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -401,6 +405,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -421,6 +426,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -443,6 +449,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -541,7 +548,7 @@
   
   
   
-    
+  
     
     
     
@@ -550,7 +557,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -577,6 +583,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -597,6 +604,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -617,6 +625,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -644,7 +653,7 @@
   
   
   
-    
+  
     
     
     
@@ -653,7 +662,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -680,6 +688,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -700,6 +709,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -720,6 +730,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -740,6 +751,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -767,7 +779,7 @@
   
   
   
-    
+  
     
     
     
@@ -776,7 +788,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -803,6 +814,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -823,6 +835,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -843,6 +856,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -870,7 +884,7 @@
   
   
   
-    
+  
     
     
     
@@ -879,7 +893,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -906,6 +919,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -926,6 +940,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -946,6 +961,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -966,6 +982,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1123,11 +1140,19 @@ <h3 id="the-tennis-example">The Tennis example</h3>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1160,10 +1185,10 @@ <h3 id="the-tennis-example">The Tennis example</h3>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/FAQ/index.html b/FAQ/index.html
index adbc927f52..29efbd6db6 100644
--- a/FAQ/index.html
+++ b/FAQ/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -343,7 +347,7 @@
   
   
   
-    
+  
     
     
     
@@ -352,7 +356,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -379,6 +382,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -399,6 +403,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -419,6 +424,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -439,6 +445,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -466,7 +473,7 @@
   
   
   
-    
+  
     
     
     
@@ -475,7 +482,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -502,6 +508,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -522,6 +529,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -542,6 +550,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -569,7 +578,7 @@
   
   
   
-    
+  
     
     
     
@@ -578,7 +587,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -605,6 +613,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -625,6 +634,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -645,6 +655,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -665,6 +676,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -692,7 +704,7 @@
   
   
   
-    
+  
     
     
     
@@ -701,7 +713,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -728,6 +739,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -748,6 +760,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -768,6 +781,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -797,7 +811,7 @@
     
   
   
-    
+  
     
     
     
@@ -806,7 +820,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" checked>
         
           
@@ -835,6 +848,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -947,6 +961,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -967,6 +982,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -987,6 +1003,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1169,11 +1186,19 @@ <h2 id="file-name-cannot-be-opened-because-the-developer-cannot-be-verified">"Fi
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1206,10 +1231,10 @@ <h2 id="file-name-cannot-be-opened-because-the-developer-cannot-be-verified">"Fi
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Getting-Started/index.html b/Getting-Started/index.html
index b33282dbeb..1ff5c1c472 100644
--- a/Getting-Started/index.html
+++ b/Getting-Started/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1304,11 +1321,19 @@ <h2 id="next-steps">Next Steps</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1341,10 +1366,10 @@ <h2 id="next-steps">Next Steps</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Glossary/index.html b/Glossary/index.html
index fc5a7c5d2c..f6d0aab43a 100644
--- a/Glossary/index.html
+++ b/Glossary/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -987,11 +1004,19 @@ <h1 id="ml-agents-toolkit-glossary">ML-Agents Toolkit Glossary</h1>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1024,10 +1049,10 @@ <h1 id="ml-agents-toolkit-glossary">ML-Agents Toolkit Glossary</h1>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Hugging-Face-Integration/index.html b/Hugging-Face-Integration/index.html
index 235eb0e191..d7252d20cf 100644
--- a/Hugging-Face-Integration/index.html
+++ b/Hugging-Face-Integration/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -343,7 +347,7 @@
   
   
   
-    
+  
     
     
     
@@ -352,7 +356,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -379,6 +382,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -399,6 +403,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -419,6 +424,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -439,6 +445,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -466,7 +473,7 @@
   
   
   
-    
+  
     
     
     
@@ -475,7 +482,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -502,6 +508,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -522,6 +529,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -542,6 +550,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -569,7 +578,7 @@
   
   
   
-    
+  
     
     
     
@@ -578,7 +587,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -605,6 +613,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -625,6 +634,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -645,6 +655,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -665,6 +676,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -694,7 +706,7 @@
     
   
   
-    
+  
     
     
     
@@ -703,7 +715,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" checked>
         
           
@@ -730,6 +741,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -750,6 +762,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -772,6 +785,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -864,7 +878,7 @@
   
   
   
-    
+  
     
     
     
@@ -873,7 +887,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -900,6 +913,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -920,6 +934,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -940,6 +955,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -960,6 +976,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1100,11 +1117,19 @@ <h2 id="visualize-an-agent-playing">Visualize an agent playing</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1137,10 +1162,10 @@ <h2 id="visualize-an-agent-playing">Visualize an agent playing</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Installation-Anaconda-Windows/index.html b/Installation-Anaconda-Windows/index.html
index d3cce1faf1..8964a13446 100644
--- a/Installation-Anaconda-Windows/index.html
+++ b/Installation-Anaconda-Windows/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1333,11 +1350,19 @@ <h2 id="acknowledgments">Acknowledgments</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1370,10 +1395,10 @@ <h2 id="acknowledgments">Acknowledgments</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Installation/index.html b/Installation/index.html
index 84fdb01018..339c88ef54 100644
--- a/Installation/index.html
+++ b/Installation/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -307,6 +309,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -352,9 +355,9 @@
     <ul class="md-nav__list" data-md-component="toc" data-md-scrollfix>
       
         <li class="md-nav__item">
-  <a href="#install-unity-20223-or-later" class="md-nav__link">
+  <a href="#install-unity-20232-or-later" class="md-nav__link">
     <span class="md-ellipsis">
-      Install Unity 2022.3 or Later
+      Install Unity 2023.2 or Later
     </span>
   </a>
   
@@ -458,6 +461,15 @@
     </span>
   </a>
   
+</li>
+        
+          <li class="md-nav__item">
+  <a href="#os-x-installing-grpc-libraries" class="md-nav__link">
+    <span class="md-ellipsis">
+      (OS X) Installing GRPC libraries
+    </span>
+  </a>
+  
 </li>
         
           <li class="md-nav__item">
@@ -514,6 +526,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -533,7 +546,7 @@
   
   
   
-    
+  
     
     
     
@@ -542,7 +555,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -569,6 +581,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -589,6 +602,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -609,6 +623,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -629,6 +644,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -656,7 +672,7 @@
   
   
   
-    
+  
     
     
     
@@ -665,7 +681,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -692,6 +707,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -712,6 +728,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -732,6 +749,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -759,7 +777,7 @@
   
   
   
-    
+  
     
     
     
@@ -768,7 +786,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -795,6 +812,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -815,6 +833,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -835,6 +854,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -855,6 +875,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -882,7 +903,7 @@
   
   
   
-    
+  
     
     
     
@@ -891,7 +912,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -918,6 +938,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -938,6 +959,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -958,6 +980,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -985,7 +1008,7 @@
   
   
   
-    
+  
     
     
     
@@ -994,7 +1017,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -1021,6 +1043,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -1041,6 +1064,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -1061,6 +1085,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -1081,6 +1106,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1130,9 +1156,9 @@
     <ul class="md-nav__list" data-md-component="toc" data-md-scrollfix>
       
         <li class="md-nav__item">
-  <a href="#install-unity-20223-or-later" class="md-nav__link">
+  <a href="#install-unity-20232-or-later" class="md-nav__link">
     <span class="md-ellipsis">
-      Install Unity 2022.3 or Later
+      Install Unity 2023.2 or Later
     </span>
   </a>
   
@@ -1236,6 +1262,15 @@
     </span>
   </a>
   
+</li>
+        
+          <li class="md-nav__item">
+  <a href="#os-x-installing-grpc-libraries" class="md-nav__link">
+    <span class="md-ellipsis">
+      (OS X) Installing GRPC libraries
+    </span>
+  </a>
+  
 </li>
         
           <li class="md-nav__item">
@@ -1322,7 +1357,7 @@ <h1 id="installation">Installation</h1>
 </ul>
 <p>Consequently, to install and use the ML-Agents Toolkit you will need to:</p>
 <ul>
-<li>Install Unity (2022.3 or later)</li>
+<li>Install Unity (2023.2 or later)</li>
 <li>Install Python (3.10.12 or higher)</li>
 <li>Clone this repository (Recommended for the latest version and bug fixes)</li>
 <li><strong>Note:</strong> If you do not clone the repository, then you will not be
@@ -1335,7 +1370,7 @@ <h1 id="installation">Installation</h1>
 <li>Install the <code>mlagents-envs</code></li>
 <li>Install the <code>mlagents</code> Python package</li>
 </ul>
-<h3 id="install-unity-20223-or-later">Install <strong>Unity 2022.3</strong> or Later</h3>
+<h3 id="install-unity-20232-or-later">Install <strong>Unity 2023.2</strong> or Later</h3>
 <p><a href="https://unity3d.com/get-unity/download">Download</a> and install Unity. We
 strongly recommend that you install Unity through the Unity Hub as it will
 enable you to manage multiple Unity versions.</p>
@@ -1418,13 +1453,19 @@ <h3 id="install-the-mlagents-python-package">Install the <code>mlagents</code> P
 offer a dedicated <a href="../Using-Virtual-Environment/">guide on Virtual Environments</a>.</p>
 <h4 id="windows-installing-pytorch">(Windows) Installing PyTorch</h4>
 <p>On Windows, you'll have to install the PyTorch package separately prior to
-installing ML-Agents. Activate your virtual environment and run from the command line:</p>
-<pre><code class="language-sh">pip3 install torch~=1.13.1 -f https://download.pytorch.org/whl/torch_stable.html
+installing ML-Agents in order to make sure the cuda-enabled version is used,
+rather than the CPU-only version. Activate your virtual environment and run from
+the command line:</p>
+<pre><code class="language-sh">pip3 install torch~=2.2.1 --index-url https://download.pytorch.org/whl/cu121
 </code></pre>
 <p>Note that on Windows, you may also need Microsoft's
 <a href="https://support.microsoft.com/en-us/help/2977003/the-latest-supported-visual-c-downloads">Visual C++ Redistributable</a>
 if you don't have it already. See the <a href="https://pytorch.org/get-started/locally/">PyTorch installation guide</a>
 for more installation options and versions.</p>
+<h4 id="os-x-installing-grpc-libraries">(OS X) Installing GRPC libraries</h4>
+<p>On OS X, you may need to explicitly install the GRPC runtime libraries to avoid hitting errors when training like <code>dlopen(/Users/alex.mccarthy/miniconda3/envs/mlagents/lib/python3.10/site-packages/grpc/_cython/cygrpc.cpython-310-darwin.so, 0x0002): symbol not found in flat namespace '_CFRelease'</code>.</p>
+<pre><code class="language-sh">pip3 install grpcio
+</code></pre>
 <h4 id="installing-mlagents">Installing <code>mlagents</code></h4>
 <p>To install the <code>mlagents</code> Python package, activate your virtual environment and
 run from the command line:</p>
@@ -1474,11 +1515,19 @@ <h2 id="help">Help</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1511,10 +1560,10 @@ <h2 id="help">Help</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Integrations-Match3/index.html b/Integrations-Match3/index.html
index fcfe45b7d2..4085972523 100644
--- a/Integrations-Match3/index.html
+++ b/Integrations-Match3/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1209,11 +1226,19 @@ <h3 id="action-space">Action Space</h3>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1246,10 +1271,10 @@ <h3 id="action-space">Action Space</h3>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Integrations/index.html b/Integrations/index.html
index c6717ef9f3..d8efe7573a 100644
--- a/Integrations/index.html
+++ b/Integrations/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -984,11 +1001,19 @@ <h2 id="interested-in-more-game-templates">Interested in more game templates?</h
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1021,10 +1046,10 @@ <h2 id="interested-in-more-game-templates">Interested in more game templates?</h
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/LICENSE/index.html b/LICENSE/index.html
index 1af4b3db3e..c3085b245d 100644
--- a/LICENSE/index.html
+++ b/LICENSE/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -122,21 +119,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -144,9 +141,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -155,6 +152,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -198,7 +198,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -243,7 +243,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -258,6 +258,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -277,6 +278,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -296,6 +298,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -315,6 +318,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -334,7 +338,7 @@
   
   
   
-    
+  
     
     
     
@@ -343,7 +347,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -370,6 +373,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -390,6 +394,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -410,6 +415,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -430,6 +436,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -457,7 +464,7 @@
   
   
   
-    
+  
     
     
     
@@ -466,7 +473,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -493,6 +499,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -513,6 +520,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -533,6 +541,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -560,7 +569,7 @@
   
   
   
-    
+  
     
     
     
@@ -569,7 +578,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -596,6 +604,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -616,6 +625,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -636,6 +646,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -656,6 +667,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -683,7 +695,7 @@
   
   
   
-    
+  
     
     
     
@@ -692,7 +704,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -719,6 +730,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -739,6 +751,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -759,6 +772,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -786,7 +800,7 @@
   
   
   
-    
+  
     
     
     
@@ -795,7 +809,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -822,6 +835,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -842,6 +856,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -862,6 +877,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -882,6 +898,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1137,11 +1154,19 @@ <h1>LICENSE</h1>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1174,10 +1199,10 @@ <h1>LICENSE</h1>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Learning-Environment-Create-New/index.html b/Learning-Environment-Create-New/index.html
index f87e490b3a..f99b591de8 100644
--- a/Learning-Environment-Create-New/index.html
+++ b/Learning-Environment-Create-New/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1546,11 +1563,19 @@ <h2 id="optional-training-using-concurrent-unity-instances">Optional: Training U
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1583,10 +1608,10 @@ <h2 id="optional-training-using-concurrent-unity-instances">Optional: Training U
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Learning-Environment-Design-Agents/index.html b/Learning-Environment-Design-Agents/index.html
index 02b009fd40..081f9aa1e5 100644
--- a/Learning-Environment-Design-Agents/index.html
+++ b/Learning-Environment-Design-Agents/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1912,14 +1929,16 @@ <h3 id="variable-length-observations">Variable Length Observations</h3>
 the padded observations. Note that attention layers are invariant to
 the order of the entities, so there is no need to properly "order" the
 entities before feeding them into the <code>BufferSensor</code>.</p>
-<p>The <code>BufferSensorComponent</code> Editor inspector has two arguments:
- - <code>Observation Size</code> : This is how many floats each entities will be
+<p>The <code>BufferSensorComponent</code> Editor inspector has two arguments:</p>
+<ul>
+<li><code>Observation Size</code> : This is how many floats each entities will be
  represented with. This number is fixed and all entities must
  have the same representation. For example, if the entities you want to
  put into the <code>BufferSensor</code> have for relevant information position and
- speed, then the <code>Observation Size</code> should be 6 floats.
- - <code>Maximum Number of Entities</code> : This is the maximum number of entities
- the <code>BufferSensor</code> will be able to collect.</p>
+ speed, then the <code>Observation Size</code> should be 6 floats.</li>
+<li><code>Maximum Number of Entities</code> : This is the maximum number of entities
+ the <code>BufferSensor</code> will be able to collect.</li>
+</ul>
 <p>To add an entity's observations to a <code>BufferSensorComponent</code>, you need
 to call <code>BufferSensorComponent.AppendObservation()</code> in the
 Agent.CollectObservations() method
@@ -2419,11 +2438,19 @@ <h2 id="recording-demonstrations">Recording Demonstrations</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -2456,10 +2483,10 @@ <h2 id="recording-demonstrations">Recording Demonstrations</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Learning-Environment-Design/index.html b/Learning-Environment-Design/index.html
index 8e524891c4..548b443809 100644
--- a/Learning-Environment-Design/index.html
+++ b/Learning-Environment-Design/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1187,11 +1204,19 @@ <h2 id="recording-statistics">Recording Statistics</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1224,10 +1249,10 @@ <h2 id="recording-statistics">Recording Statistics</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Learning-Environment-Examples/index.html b/Learning-Environment-Examples/index.html
index f5af330a38..b19736e6ba 100644
--- a/Learning-Environment-Examples/index.html
+++ b/Learning-Environment-Examples/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1342,7 +1359,7 @@ <h2 id="food-collector">Food Collector</h2>
     objects around agent's forward direction (40 by 40 with 6 different categories).</li>
 <li>Actions:<ul>
 <li>3 continuous actions correspond to Forward Motion, Side Motion and Rotation</li>
-<li>1 discrete acion branch for Laser with 2 possible actions corresponding to
+<li>1 discrete action branch for Laser with 2 possible actions corresponding to
   Shoot Laser or No Action</li>
 </ul>
 </li>
@@ -1633,11 +1650,19 @@ <h2 id="dungeon-escape">Dungeon Escape</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1670,10 +1695,10 @@ <h2 id="dungeon-escape">Dungeon Escape</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Learning-Environment-Executable/index.html b/Learning-Environment-Executable/index.html
index 163a29de75..a6100ebbd5 100644
--- a/Learning-Environment-Executable/index.html
+++ b/Learning-Environment-Executable/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1174,11 +1191,19 @@ <h2 id="training-on-headless-server">Training on Headless Server</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1211,10 +1236,10 @@ <h2 id="training-on-headless-server">Training on Headless Server</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Limitations/index.html b/Limitations/index.html
index 72530e2f5b..e5bfa68506 100644
--- a/Limitations/index.html
+++ b/Limitations/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -343,7 +347,7 @@
   
   
   
-    
+  
     
     
     
@@ -352,7 +356,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -379,6 +382,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -399,6 +403,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -419,6 +424,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -439,6 +445,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -466,7 +473,7 @@
   
   
   
-    
+  
     
     
     
@@ -475,7 +482,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -502,6 +508,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -522,6 +529,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -542,6 +550,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -569,7 +578,7 @@
   
   
   
-    
+  
     
     
     
@@ -578,7 +587,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -605,6 +613,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -625,6 +634,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -645,6 +655,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -665,6 +676,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -692,7 +704,7 @@
   
   
   
-    
+  
     
     
     
@@ -701,7 +713,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -728,6 +739,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -748,6 +760,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -768,6 +781,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -797,7 +811,7 @@
     
   
   
-    
+  
     
     
     
@@ -806,7 +820,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" checked>
         
           
@@ -833,6 +846,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -855,6 +869,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -883,6 +898,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -903,6 +919,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -973,11 +990,19 @@ <h1 id="limitations">Limitations</h1>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1010,10 +1035,10 @@ <h1 id="limitations">Limitations</h1>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/ML-Agents-Envs-README/index.html b/ML-Agents-Envs-README/index.html
index 4a3a93fae1..f83f1b1257 100644
--- a/ML-Agents-Envs-README/index.html
+++ b/ML-Agents-Envs-README/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1019,11 +1036,19 @@ <h2 id="limitations">Limitations</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1056,10 +1081,10 @@ <h2 id="limitations">Limitations</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/ML-Agents-Overview/index.html b/ML-Agents-Overview/index.html
index d4784ffd56..6818ba17bb 100644
--- a/ML-Agents-Overview/index.html
+++ b/ML-Agents-Overview/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -288,6 +289,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -648,6 +650,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -667,6 +670,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -686,7 +690,7 @@
   
   
   
-    
+  
     
     
     
@@ -695,7 +699,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -722,6 +725,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -742,6 +746,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -762,6 +767,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -782,6 +788,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -809,7 +816,7 @@
   
   
   
-    
+  
     
     
     
@@ -818,7 +825,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -845,6 +851,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -865,6 +872,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -885,6 +893,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -912,7 +921,7 @@
   
   
   
-    
+  
     
     
     
@@ -921,7 +930,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -948,6 +956,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -968,6 +977,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -988,6 +998,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -1008,6 +1019,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -1035,7 +1047,7 @@
   
   
   
-    
+  
     
     
     
@@ -1044,7 +1056,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -1071,6 +1082,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -1091,6 +1103,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -1111,6 +1124,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -1138,7 +1152,7 @@
   
   
   
-    
+  
     
     
     
@@ -1147,7 +1161,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -1174,6 +1187,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -1194,6 +1208,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -1214,6 +1229,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -1234,6 +1250,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1962,7 +1979,7 @@ <h4 id="a-quick-note-on-reward-signals">A Quick Note on Reward Signals</h4>
 <h3 id="deep-reinforcement-learning">Deep Reinforcement Learning</h3>
 <p>ML-Agents provide an implementation of two reinforcement learning algorithms:</p>
 <ul>
-<li><a href="https://blog.openai.com/openai-baselines-ppo/">Proximal Policy Optimization (PPO)</a></li>
+<li><a href="https://openai.com/research/openai-baselines-ppo">Proximal Policy Optimization (PPO)</a></li>
 <li><a href="https://bair.berkeley.edu/blog/2018/12/14/sac/">Soft Actor-Critic (SAC)</a></li>
 </ul>
 <p>The default algorithm is PPO. This is a method that has been shown to be more
@@ -2003,11 +2020,13 @@ <h4 id="rnd-for-sparse-reward-environments">RND for Sparse-reward Environments</
 <p>Similarly to Curiosity, Random Network Distillation (RND) is useful in sparse or rare
 reward environments as it helps the Agent explore. The RND Module is implemented following
 the paper <a href="https://arxiv.org/abs/1810.12894">Exploration by Random Network Distillation</a>.
-RND uses two networks:
- - The first is a network with fixed random weights that takes observations as inputs and
- generates an encoding
- - The second is a network with similar architecture that is trained to predict the
- outputs of the first network and uses the observations the Agent collects as training data.</p>
+RND uses two networks:</p>
+<ul>
+<li>The first is a network with fixed random weights that takes observations as inputs and
+ generates an encoding</li>
+<li>The second is a network with similar architecture that is trained to predict the
+ outputs of the first network and uses the observations the Agent collects as training data.</li>
+</ul>
 <p>The loss (the squared difference between the predicted and actual encoded observations)
 of the trained model is used as intrinsic reward. The more an Agent visits a state, the
 more accurate the predictions and the lower the rewards which encourages the Agent to
@@ -2056,9 +2075,9 @@ <h3 id="imitation-learning">Imitation Learning</h3>
 This can directly conflict with goal-oriented tasks like our PushBlock or Pyramids
 example environments where an agent must reach a goal state thus ending the
 episode as quickly as possible. In these cases, we strongly recommend that you
-use a low strength GAIL reward signal and a sparse extrinisic signal when
+use a low strength GAIL reward signal and a sparse extrinsic signal when
 the agent achieves the task. This way, the GAIL reward signal will guide the
-agent until it discovers the extrnisic signal and will not overpower it. If the
+agent until it discovers the extrinsic signal and will not overpower it. If the
 agent appears to be ignoring the extrinsic reward signal, you should reduce
 the strength of GAIL.</p>
 <h4 id="gail-generative-adversarial-imitation-learning">GAIL (Generative Adversarial Imitation Learning)</h4>
@@ -2114,7 +2133,7 @@ <h2 id="training-methods-environment-specific">Training Methods: Environment-spe
 <h3 id="training-in-competitive-multi-agent-environments-with-self-play">Training in Competitive Multi-Agent Environments with Self-Play</h3>
 <p>ML-Agents provides the functionality to train both symmetric and asymmetric
 adversarial games with
-<a href="https://openai.com/blog/competitive-self-play/">Self-Play</a>. A symmetric game is
+<a href="https://openai.com/research/competitive-self-play">Self-Play</a>. A symmetric game is
 one in which opposing agents are equal in form, function and objective. Examples
 of symmetric games are our Tennis and Soccer example environments. In
 reinforcement learning, this means both agents have the same observation and
@@ -2346,11 +2365,19 @@ <h2 id="summary-and-next-steps">Summary and Next Steps</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -2383,10 +2410,10 @@ <h2 id="summary-and-next-steps">Summary and Next Steps</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/ML-Agents-README/index.html b/ML-Agents-README/index.html
index 16c1f891cc..9fe27e70c6 100644
--- a/ML-Agents-README/index.html
+++ b/ML-Agents-README/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1008,11 +1025,19 @@ <h2 id="limitations">Limitations</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1045,10 +1070,10 @@ <h2 id="limitations">Limitations</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/ML-Agents-Toolkit-Documentation/index.html b/ML-Agents-Toolkit-Documentation/index.html
index 0912bfd0cf..961b6b31e1 100644
--- a/ML-Agents-Toolkit-Documentation/index.html
+++ b/ML-Agents-Toolkit-Documentation/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -326,6 +329,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -482,7 +486,7 @@
   
   
   
-    
+  
     
     
     
@@ -491,7 +495,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -518,6 +521,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -538,6 +542,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -558,6 +563,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -578,6 +584,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -605,7 +612,7 @@
   
   
   
-    
+  
     
     
     
@@ -614,7 +621,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -641,6 +647,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -661,6 +668,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -681,6 +689,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -708,7 +717,7 @@
   
   
   
-    
+  
     
     
     
@@ -717,7 +726,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -744,6 +752,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -764,6 +773,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -784,6 +794,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -804,6 +815,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -831,7 +843,7 @@
   
   
   
-    
+  
     
     
     
@@ -840,7 +852,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -867,6 +878,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -887,6 +899,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -907,6 +920,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -934,7 +948,7 @@
   
   
   
-    
+  
     
     
     
@@ -943,7 +957,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -970,6 +983,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -990,6 +1004,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -1010,6 +1025,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -1030,6 +1046,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1282,11 +1299,19 @@ <h2 id="deprecated-docs">Deprecated Docs</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1319,10 +1344,10 @@ <h2 id="deprecated-docs">Deprecated Docs</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Migrating/index.html b/Migrating/index.html
index c562d7255e..de584d60a7 100644
--- a/Migrating/index.html
+++ b/Migrating/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -343,7 +347,7 @@
   
   
   
-    
+  
     
     
     
@@ -352,7 +356,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -379,6 +382,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -399,6 +403,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -419,6 +424,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -439,6 +445,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -466,7 +473,7 @@
   
   
   
-    
+  
     
     
     
@@ -475,7 +482,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -502,6 +508,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -522,6 +529,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -542,6 +550,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -569,7 +578,7 @@
   
   
   
-    
+  
     
     
     
@@ -578,7 +587,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -605,6 +613,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -625,6 +634,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -645,6 +655,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -665,6 +676,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -692,7 +704,7 @@
   
   
   
-    
+  
     
     
     
@@ -701,7 +713,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -728,6 +739,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -748,6 +760,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -768,6 +781,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -797,7 +811,7 @@
     
   
   
-    
+  
     
     
     
@@ -806,7 +820,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" checked>
         
           
@@ -833,6 +846,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -853,6 +867,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -875,6 +890,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -903,6 +919,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -972,19 +989,23 @@ <h2 id="migrating-to-the-ml-agents-envs-0300-package">Migrating to the ml-agents
   Please update your python installation to 3.10.12 or higher.</li>
 <li>The <code>gym-unity</code> package has been refactored into the <code>ml-agents-envs</code> package. Please update your imports accordingly.</li>
 <li>Example:</li>
-<li>Before</li>
+<li>Before
+    <code>python
+    from gym_unity.unity_gym_env import UnityToGymWrapper</code></li>
+<li>After:
+    <code>python
+    from mlagents_envs.envs.unity_gym_env import UnityToGymWrapper</code></li>
 </ul>
-<pre><code class="language-python">from gym_unity.unity_gym_env import UnityToGymWrapper
-</code></pre>
+<h2 id="migrating-the-package-to-version-3x">Migrating the package to version 3.x</h2>
 <ul>
-<li>After:</li>
+<li>The official version of Unity ML-Agents supports is now 2023.2. If you run
+  into issues, please consider deleting your project's Library folder and reopening your
+  project.</li>
 </ul>
-<pre><code class="language-python">from mlagents_envs.envs.unity_gym_env import UnityToGymWrapper
-</code></pre>
 <h2 id="migrating-the-package-to-version-2x">Migrating the package to version 2.x</h2>
 <ul>
 <li>The official version of Unity ML-Agents supports is now 2022.3 LTS. If you run
-  into issues, please consider deleting your project's Library folder and reponening your
+  into issues, please consider deleting your project's Library folder and reopening your
   project.</li>
 <li>If you used any of the APIs that were deprecated before version 2.0, you need to use their replacement. These
 deprecated APIs have been removed. See the migration steps bellow for specific API replacements.</li>
@@ -1126,7 +1147,7 @@ <h3 id="match3-integration-changes">Match3 integration changes</h3>
 values from <code>GetMaxBoardSize()</code>.</p>
 <h3 id="gridsensor-changes">GridSensor changes</h3>
 <p>The sensor configuration has changed:
-* The sensor implementation has been refactored and exsisting GridSensor created from extension package
+* The sensor implementation has been refactored and existing GridSensor created from extension package
 will not work in newer version. Some errors might show up when loading the old sensor in the scene.
 You'll need to remove the old sensor and create a new GridSensor.
 * These parameters names have changed but still refer to the same concept in the sensor: <code>GridNumSide</code> -&gt; <code>GridSize</code>,
@@ -1146,8 +1167,8 @@ <h3 id="gridsensor-changes">GridSensor changes</h3>
 * The observation data should be written to the input <code>dataBuffer</code> instead of creating and returning a new array.
 * Removed the constraint of all data required to be normalized. You should specify it in <code>IsDataNormalized()</code>.
 Sensors with non-normalized data cannot use PNG compression type.
-* The sensor will not further encode the data recieved from <code>GetObjectData()</code> anymore. The values
-recieved from <code>GetObjectData()</code> will be the observation sent to the trainer.</p>
+* The sensor will not further encode the data received from <code>GetObjectData()</code> anymore. The values
+received from <code>GetObjectData()</code> will be the observation sent to the trainer.</p>
 <h3 id="lstm-models-from-previous-releases-no-longer-supported">LSTM models from previous releases no longer supported</h3>
 <p>The way that Sentis processes LSTM (recurrent neural networks) has changed. As a result, models
 trained with previous versions of ML-Agents will not be usable at inference if they were trained with a <code>memory</code>
@@ -1162,7 +1183,7 @@ <h3 id="implementing-iheuristic-in-your-iactuator-implementations">Implementing
 <li><code>VectorSensor.AddObservation(IEnumerable&lt;float&gt;)</code> is deprecated. Use <code>VectorSensor.AddObservation(IList&lt;float&gt;)</code>
   instead.</li>
 <li><code>ObservationWriter.AddRange()</code> is deprecated. Use <code>ObservationWriter.AddList()</code> instead.</li>
-<li><code>ActuatorComponent.CreateAcuator()</code> is deprecated.  Please use override <code>ActuatorComponent.CreateActuators</code>
+<li><code>ActuatorComponent.CreateActuator()</code> is deprecated.  Please use override <code>ActuatorComponent.CreateActuators</code>
   instead.  Since <code>ActuatorComponent.CreateActuator()</code> is abstract, you will still need to override it in your
   class until it is removed.  It is only ever called if you don't override <code>ActuatorComponent.CreateActuators</code>.
   You can suppress the warnings by surrounding the method with the following pragma:
@@ -1367,7 +1388,7 @@ <h3 id="important-changes_4">Important changes</h3>
   method names will be removed in a later release:</li>
 <li><code>InitializeAgent()</code> was renamed to <code>Initialize()</code></li>
 <li><code>AgentAction()</code> was renamed to <code>OnActionReceived()</code></li>
-<li><code>AgentReset()</code> was renamed to <code>OnEpsiodeBegin()</code></li>
+<li><code>AgentReset()</code> was renamed to <code>OnEpisodeBegin()</code></li>
 <li><code>Done()</code> was renamed to <code>EndEpisode()</code></li>
 <li><code>GiveModel()</code> was renamed to <code>SetModel()</code></li>
 <li>The <code>IFloatProperties</code> interface has been removed.</li>
@@ -1523,7 +1544,7 @@ <h3 id="important-changes_6">Important changes</h3>
   depended on <a href="https://www.python.org/dev/peps/pep-0420/">PEP420</a>, which caused
   problems with some of our tooling such as mypy and pylint.</li>
 <li>The official version of Unity ML-Agents supports is now 2022.3 LTS. If you run
-  into issues, please consider deleting your library folder and reponening your
+  into issues, please consider deleting your library folder and reopening your
   projects. You will need to install the Sentis package into your project in
   order to ML-Agents to compile correctly.</li>
 </ul>
@@ -1883,11 +1904,19 @@ <h3 id="semantics">Semantics</h3>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1920,10 +1949,10 @@ <h3 id="semantics">Semantics</h3>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Package-Settings/index.html b/Package-Settings/index.html
index ccd412a91e..d03ab542bb 100644
--- a/Package-Settings/index.html
+++ b/Package-Settings/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -977,14 +994,14 @@ <h1 id="ml-agents-package-settings">ML-Agents Package Settings</h1>
 <p>You can find them at <code>Edit</code> &gt; <code>Project Settings...</code> &gt; <code>ML-Agents</code>. It lists out all the available settings and their default values.</p>
 <h2 id="create-custom-settings">Create Custom Settings</h2>
 <p>In order to to use your own settings for your project, you'll need to create a settings asset.</p>
-<p>You can do this by clicking the <code>Create Settings Asset</code> buttom or clicking the gear on the top right and select <code>New Settings Asset...</code>.
+<p>You can do this by clicking the <code>Create Settings Asset</code> button or clicking the gear on the top right and select <code>New Settings Asset...</code>.
 The asset file can be placed anywhere in the <code>Asset/</code> folder in your project.
 After Creating the settings asset, you'll be able to modify the settings for your project and your settings will be saved in the asset.</p>
 <p><img alt="Package Settings" src="../images/package-settings.png" /></p>
 <h2 id="multiple-custom-settings-for-different-scenarios">Multiple Custom Settings for Different Scenarios</h2>
 <p>You can create multiple settings assets in one project.</p>
 <p>By clicking the gear on the top right you'll see all available settings listed in the drop-down menu to choose from.</p>
-<p>This allows you to create different settings for different scenatios. For example, you can create two
+<p>This allows you to create different settings for different scenarios. For example, you can create two
 separate settings for training and inference, and specify which one you want to use according to what you're currently running.</p>
 <p><img alt="Multiple Settings" src="../images/multiple-settings.png" /></p>
 
@@ -992,11 +1009,19 @@ <h2 id="multiple-custom-settings-for-different-scenarios">Multiple Custom Settin
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1029,10 +1054,10 @@ <h2 id="multiple-custom-settings-for-different-scenarios">Multiple Custom Settin
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Profiling-Python/index.html b/Profiling-Python/index.html
index 4717e6ff66..0ea13d7200 100644
--- a/Profiling-Python/index.html
+++ b/Profiling-Python/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1011,7 +1028,7 @@
 
 
 <h1 id="profiling-in-python">Profiling in Python</h1>
-<p>As part of the ML-Agents Tookit, we provide a lightweight profiling system, in
+<p>As part of the ML-Agents Toolkit, we provide a lightweight profiling system, in
 order to identity hotspots in the training process and help spot regressions
 from changes.</p>
 <p>Timers are hierarchical, meaning that the time tracked in a block of code can be
@@ -1070,11 +1087,19 @@ <h4 id="threads">Threads</h4>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1107,10 +1132,10 @@ <h4 id="threads">Threads</h4>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Python-Custom-Trainer-Plugin/index.html b/Python-Custom-Trainer-Plugin/index.html
index d2c0961e0b..8c7826b5c6 100644
--- a/Python-Custom-Trainer-Plugin/index.html
+++ b/Python-Custom-Trainer-Plugin/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -986,7 +1003,7 @@ <h1 id="unity-ml-agents-custom-trainers-plugin">Unity Ml-Agents Custom trainers
 in <code>Ml-agents</code> Package. This will allow rerouting <code>mlagents-learn</code> CLI to custom trainers and extending the config files
 with hyper-parameters specific to your new trainers. We will expose a high-level extensible trainer (both on-policy,
 and off-policy trainers) optimizer and hyperparameter classes with documentation for the use of this plugin. For more
-infromation on how python plugin system works see <a href="../Training-Plugins/">Plugin interfaces</a>.</p>
+information on how python plugin system works see <a href="../Training-Plugins/">Plugin interfaces</a>.</p>
 <h2 id="overview">Overview</h2>
 <p>Model-free RL algorithms generally fall into two broad categories: on-policy and off-policy. On-policy algorithms perform updates based on data gathered from the current policy. Off-policy algorithms learn a Q function from a buffer of previous data, then use this Q function to make decisions. Off-policy algorithms have three key benefits in the context of ML-Agents: They tend to use fewer samples than on-policy as they can pull and re-use data from the buffer many times. They allow player demonstrations to be inserted in-line with RL data into the buffer, enabling new ways of doing imitation learning by streaming player data.</p>
 <p>To add new custom trainers to ML-agents, you would need to create a new python package.
@@ -1026,11 +1043,19 @@ <h2 id="tutorial">Tutorial</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1063,10 +1088,10 @@ <h2 id="tutorial">Tutorial</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Python-Gym-API-Documentation/index.html b/Python-Gym-API-Documentation/index.html
index 0909565fd8..af4750bc61 100644
--- a/Python-Gym-API-Documentation/index.html
+++ b/Python-Gym-API-Documentation/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -343,7 +347,7 @@
   
   
   
-    
+  
     
     
     
@@ -352,7 +356,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -379,6 +382,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -399,6 +403,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -419,6 +424,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -439,6 +445,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -466,7 +473,7 @@
   
   
   
-    
+  
     
     
     
@@ -475,7 +482,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -502,6 +508,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -522,6 +529,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -542,6 +550,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -571,7 +580,7 @@
     
   
   
-    
+  
     
     
     
@@ -580,7 +589,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" checked>
         
           
@@ -609,6 +617,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -637,6 +646,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -657,6 +667,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -677,6 +688,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -704,7 +716,7 @@
   
   
   
-    
+  
     
     
     
@@ -713,7 +725,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -740,6 +751,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -760,6 +772,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -780,6 +793,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -807,7 +821,7 @@
   
   
   
-    
+  
     
     
     
@@ -816,7 +830,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -843,6 +856,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -863,6 +877,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -883,6 +898,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -903,6 +919,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1085,11 +1102,19 @@ <h4 id="lookup_action">lookup_action</h4>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1122,10 +1147,10 @@ <h4 id="lookup_action">lookup_action</h4>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Python-Gym-API/index.html b/Python-Gym-API/index.html
index e72029523a..5e6cecba33 100644
--- a/Python-Gym-API/index.html
+++ b/Python-Gym-API/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -343,7 +347,7 @@
   
   
   
-    
+  
     
     
     
@@ -352,7 +356,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -379,6 +382,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -399,6 +403,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -419,6 +424,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -439,6 +445,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -468,7 +475,7 @@
     
   
   
-    
+  
     
     
     
@@ -477,7 +484,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" checked>
         
           
@@ -506,6 +512,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -684,6 +691,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -704,6 +712,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -731,7 +740,7 @@
   
   
   
-    
+  
     
     
     
@@ -740,7 +749,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -767,6 +775,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -787,6 +796,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -807,6 +817,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -827,6 +838,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -854,7 +866,7 @@
   
   
   
-    
+  
     
     
     
@@ -863,7 +875,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -890,6 +901,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -910,6 +922,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -930,6 +943,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -957,7 +971,7 @@
   
   
   
-    
+  
     
     
     
@@ -966,7 +980,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -993,6 +1006,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -1013,6 +1027,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -1033,6 +1048,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -1053,6 +1069,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1248,7 +1265,7 @@ <h1 id="unity-ml-agents-gym-wrapper">Unity ML-Agents Gym Wrapper</h1>
 our <code>UnityEnvironment</code> class, which is the default way of interfacing with a
 Unity environment via Python.</p>
 <h2 id="installation">Installation</h2>
-<p>The gym wrapper is part of the <code>mlgents_envs</code> package. Please refer to the
+<p>The gym wrapper is part of the <code>mlagents_envs</code> package. Please refer to the
 <a href="../ML-Agents-Envs-README/">mlagents_envs installation instructions</a>.</p>
 <h2 id="using-the-gym-wrapper">Using the Gym Wrapper</h2>
 <p>The gym interface is available from <code>gym_unity.envs</code>. To launch an environment
@@ -1543,11 +1560,19 @@ <h3 id="example-gridworld">Example: GridWorld</h3>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1580,10 +1605,10 @@ <h3 id="example-gridworld">Example: GridWorld</h3>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Python-LLAPI-Documentation/index.html b/Python-LLAPI-Documentation/index.html
index ec984b8942..52ed7e2806 100644
--- a/Python-LLAPI-Documentation/index.html
+++ b/Python-LLAPI-Documentation/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -343,7 +347,7 @@
   
   
   
-    
+  
     
     
     
@@ -352,7 +356,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -379,6 +382,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -399,6 +403,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -419,6 +424,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -439,6 +445,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -466,7 +473,7 @@
   
   
   
-    
+  
     
     
     
@@ -475,7 +482,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -502,6 +508,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -522,6 +529,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -542,6 +550,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -571,7 +580,7 @@
     
   
   
-    
+  
     
     
     
@@ -580,7 +589,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" checked>
         
           
@@ -607,6 +615,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -627,6 +636,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -649,6 +659,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -677,6 +688,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -704,7 +716,7 @@
   
   
   
-    
+  
     
     
     
@@ -713,7 +725,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -740,6 +751,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -760,6 +772,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -780,6 +793,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -807,7 +821,7 @@
   
   
   
-    
+  
     
     
     
@@ -816,7 +830,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -843,6 +856,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -863,6 +877,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -883,6 +898,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -903,6 +919,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1505,7 +1522,7 @@ <h3 id="unityenvregistry">UnityEnvRegistry</h3>
 of downloading the Unity Editor.
 The UnityEnvRegistry implements a Map, to access an entry of the Registry, use:</p>
 <pre><code class="language-python">registry = UnityEnvRegistry()
-entry = registry[&lt;environment_identifyier&gt;]
+entry = registry[&lt;environment_identifier&gt;]
 </code></pre>
 <p>An entry has the following properties :
  * <code>identifier</code> : Uniquely identifies this environment
@@ -1514,7 +1531,7 @@ <h3 id="unityenvregistry">UnityEnvRegistry</h3>
  * <code>description</code> : A human readable description of the environment.</p>
 <p>To launch a Unity environment from a registry entry, use the <code>make</code> method:</p>
 <pre><code class="language-python">registry = UnityEnvRegistry()
-env = registry[&lt;environment_identifyier&gt;].make()
+env = registry[&lt;environment_identifier&gt;].make()
 </code></pre>
 <p><a name="mlagents_envs.registry.unity_env_registry.UnityEnvRegistry.register"></a></p>
 <h4 id="register">register</h4>
@@ -1957,11 +1974,19 @@ <h4 id="channel_id">channel_id</h4>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1994,10 +2019,10 @@ <h4 id="channel_id">channel_id</h4>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Python-LLAPI/index.html b/Python-LLAPI/index.html
index b55be99161..79ee71048a 100644
--- a/Python-LLAPI/index.html
+++ b/Python-LLAPI/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -343,7 +347,7 @@
   
   
   
-    
+  
     
     
     
@@ -352,7 +356,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -379,6 +382,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -399,6 +403,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -419,6 +424,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -439,6 +445,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -468,7 +475,7 @@
     
   
   
-    
+  
     
     
     
@@ -477,7 +484,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" checked>
         
           
@@ -504,6 +510,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -524,6 +531,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -546,6 +554,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -728,7 +737,7 @@
   
   
   
-    
+  
     
     
     
@@ -737,7 +746,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -764,6 +772,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -784,6 +793,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -804,6 +814,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -824,6 +835,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -851,7 +863,7 @@
   
   
   
-    
+  
     
     
     
@@ -860,7 +872,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -887,6 +898,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -907,6 +919,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -927,6 +940,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -954,7 +968,7 @@
   
   
   
-    
+  
     
     
     
@@ -963,7 +977,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -990,6 +1003,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -1010,6 +1024,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -1030,6 +1045,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -1050,6 +1066,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1551,11 +1568,19 @@ <h4 id="custom-side-channels">Custom side channels</h4>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1588,10 +1613,10 @@ <h4 id="custom-side-channels">Custom side channels</h4>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Python-On-Off-Policy-Trainer-Documentation/index.html b/Python-On-Off-Policy-Trainer-Documentation/index.html
index c06eb73083..f252927035 100644
--- a/Python-On-Off-Policy-Trainer-Documentation/index.html
+++ b/Python-On-Off-Policy-Trainer-Documentation/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -343,7 +347,7 @@
   
   
   
-    
+  
     
     
     
@@ -352,7 +356,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -379,6 +382,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -399,6 +403,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -419,6 +424,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -439,6 +445,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -466,7 +473,7 @@
   
   
   
-    
+  
     
     
     
@@ -475,7 +482,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -502,6 +508,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -522,6 +529,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -542,6 +550,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -571,7 +580,7 @@
     
   
   
-    
+  
     
     
     
@@ -580,7 +589,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" checked>
         
           
@@ -607,6 +615,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -627,6 +636,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -647,6 +657,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -669,6 +680,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -704,7 +716,7 @@
   
   
   
-    
+  
     
     
     
@@ -713,7 +725,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -740,6 +751,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -760,6 +772,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -780,6 +793,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -807,7 +821,7 @@
   
   
   
-    
+  
     
     
     
@@ -816,7 +830,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -843,6 +856,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -863,6 +877,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -883,6 +898,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -903,6 +919,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1465,7 +1482,7 @@ <h2 id="lesson-objects">Lesson Objects</h2>
 class Lesson()
 </code></pre>
 <p>Gathers the data of one lesson for one environment parameter including its name,
-the condition that must be fullfiled for the lesson to be completed and a sampler
+the condition that must be fulfilled for the lesson to be completed and a sampler
 for the environment parameter. If the completion_criteria is None, then this is
 the last lesson in the curriculum.</p>
 <p><a name="mlagents.trainers.settings.EnvironmentParameterSettings"></a></p>
@@ -1530,11 +1547,19 @@ <h4 id="from_argparse">from_argparse</h4>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1567,10 +1592,10 @@ <h4 id="from_argparse">from_argparse</h4>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Python-Optimizer-Documentation/index.html b/Python-Optimizer-Documentation/index.html
index 36ec3c96e7..739dc4c271 100644
--- a/Python-Optimizer-Documentation/index.html
+++ b/Python-Optimizer-Documentation/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -982,8 +999,8 @@ <h4 id="get_trajectory_value_estimates">get_trajectory_value_estimates</h4>
 <p><strong>Arguments</strong>:</p>
 <ul>
 <li><code>batch</code>: An AgentBuffer that consists of a trajectory.</li>
-<li><code>next_obs</code>: the next observation (after the trajectory). Used for boostrapping
-    if this is not a termiinal trajectory.</li>
+<li><code>next_obs</code>: the next observation (after the trajectory). Used for bootstrapping
+    if this is not a terminal trajectory.</li>
 <li><code>done</code>: Set true if this is a terminal trajectory.</li>
 <li><code>agent_id</code>: Agent ID of the agent that this trajectory belongs to.</li>
 </ul>
@@ -1017,11 +1034,19 @@ <h4 id="update">update</h4>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1054,10 +1079,10 @@ <h4 id="update">update</h4>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Python-PettingZoo-API-Documentation/index.html b/Python-PettingZoo-API-Documentation/index.html
index f3faa30a5f..ac98d084d7 100644
--- a/Python-PettingZoo-API-Documentation/index.html
+++ b/Python-PettingZoo-API-Documentation/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -343,7 +347,7 @@
   
   
   
-    
+  
     
     
     
@@ -352,7 +356,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -379,6 +382,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -399,6 +403,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -419,6 +424,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -439,6 +445,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -466,7 +473,7 @@
   
   
   
-    
+  
     
     
     
@@ -475,7 +482,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -502,6 +508,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -522,6 +529,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -542,6 +550,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -571,7 +580,7 @@
     
   
   
-    
+  
     
     
     
@@ -580,7 +589,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" checked>
         
           
@@ -607,6 +615,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -629,6 +638,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -657,6 +667,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -677,6 +688,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -704,7 +716,7 @@
   
   
   
-    
+  
     
     
     
@@ -713,7 +725,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -740,6 +751,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -760,6 +772,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -780,6 +793,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -807,7 +821,7 @@
   
   
   
-    
+  
     
     
     
@@ -816,7 +830,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -843,6 +856,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -863,6 +877,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -883,6 +898,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -903,6 +919,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1137,11 +1154,19 @@ <h4 id="close">close</h4>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1174,10 +1199,10 @@ <h4 id="close">close</h4>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Python-PettingZoo-API/index.html b/Python-PettingZoo-API/index.html
index 48bd8695b3..49df06459b 100644
--- a/Python-PettingZoo-API/index.html
+++ b/Python-PettingZoo-API/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -343,7 +347,7 @@
   
   
   
-    
+  
     
     
     
@@ -352,7 +356,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -379,6 +382,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -399,6 +403,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -419,6 +424,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -439,6 +445,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -468,7 +475,7 @@
     
   
   
-    
+  
     
     
     
@@ -477,7 +484,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" checked>
         
           
@@ -504,6 +510,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -526,6 +533,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -611,6 +619,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -638,7 +647,7 @@
   
   
   
-    
+  
     
     
     
@@ -647,7 +656,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -674,6 +682,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -694,6 +703,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -714,6 +724,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -734,6 +745,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -761,7 +773,7 @@
   
   
   
-    
+  
     
     
     
@@ -770,7 +782,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -797,6 +808,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -817,6 +829,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -837,6 +850,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -864,7 +878,7 @@
   
   
   
-    
+  
     
     
     
@@ -873,7 +887,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -900,6 +913,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -920,6 +934,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -940,6 +955,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -960,6 +976,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1055,11 +1072,11 @@
 
 <h1 id="unity-ml-agents-pettingzoo-wrapper">Unity ML-Agents PettingZoo Wrapper</h1>
 <p>With the increasing interest in multi-agent training with a gym-like API, we provide a
-PettingZoo Wrapper around the <a href="https://www.pettingzoo.ml/">Petting Zoo API</a>. Our wrapper
+PettingZoo Wrapper around the <a href="https://pettingzoo.farama.org/">Petting Zoo API</a>. Our wrapper
 provides interfaces on top of our <code>UnityEnvironment</code> class, which is the default way of
 interfacing with a Unity environment via Python.</p>
 <h2 id="installation-and-examples">Installation and Examples</h2>
-<p>The PettingZoo wrapper is part of the <code>mlgents_envs</code> package. Please refer to the
+<p>The PettingZoo wrapper is part of the <code>mlagents_envs</code> package. Please refer to the
 <a href="../ML-Agents-Envs-README/">mlagents_envs installation instructions</a>.</p>
 <p><a href="https://colab.research.google.com/github/Unity-Technologies/ml-agents/blob/develop-python-api-ga/ml-agents-envs/colabs/Colab_PettingZoo.ipynb">[Colab] PettingZoo Wrapper Example</a></p>
 <p>This colab notebook demonstrates the example usage of the wrapper, including installation,
@@ -1068,7 +1085,7 @@ <h2 id="installation-and-examples">Installation and Examples</h2>
 which is a multi-agents environment with multiple different behavior names.</p>
 <h2 id="api-interface">API interface</h2>
 <p>This wrapper is compatible with PettingZoo API. Please check out
-<a href="https://www.pettingzoo.ml/api">PettingZoo API page</a> for more details.
+<a href="https://pettingzoo.farama.org/api/aec/">PettingZoo API page</a> for more details.
 Here's an example of interacting with wrapped environment:</p>
 <pre><code class="language-python">from mlagents_envs.environment import UnityEnvironment
 from mlagents_envs.envs import UnityToPettingZooWrapper
@@ -1083,8 +1100,8 @@ <h2 id="api-interface">API interface</h2>
 </code></pre>
 <h2 id="notes">Notes</h2>
 <ul>
-<li>There is support for both <a href="https://www.pettingzoo.ml/api#interacting-with-environments">AEC</a>
-  and <a href="https://www.pettingzoo.ml/api#parallel-api">Parallel</a> PettingZoo APIs.</li>
+<li>There is support for both <a href="https://pettingzoo.farama.org/api/aec/">AEC</a>
+  and <a href="https://pettingzoo.farama.org/api/parallel/">Parallel</a> PettingZoo APIs.</li>
 <li>The AEC wrapper is compatible with PettingZoo (PZ) API interface but works in a slightly
   different way under the hood. For the AEC API, Instead of stepping the environment in every <code>env.step(action)</code>,
   the PZ wrapper will store the action, and will only perform environment stepping when all the
@@ -1106,11 +1123,19 @@ <h2 id="notes">Notes</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1143,10 +1168,10 @@ <h2 id="notes">Notes</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Readme/index.html b/Readme/index.html
index 99c716f727..8330352eae 100644
--- a/Readme/index.html
+++ b/Readme/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1059,18 +1076,20 @@ <h2 id="releases-documentation">Releases &amp; Documentation</h2>
 to get started with the latest release of ML-Agents.</strong></p>
 <p><strong>You can also check out our new <a href="https://unity-technologies.github.io/ml-agents/">web docs</a>!</strong></p>
 <p>The table below lists all our releases, including our <code>main</code> branch which is
-under active development and may be unstable. A few helpful guidelines:
-- The <a href="../Versioning/">Versioning page</a> overviews how we manage our GitHub
-  releases and the versioning process for each of the ML-Agents components.
-- The <a href="https://github.com/Unity-Technologies/ml-agents/releases">Releases page</a>
-  contains details of the changes between releases.
-- The <a href="../Migrating/">Migration page</a> contains details on how to upgrade
-  from earlier releases of the ML-Agents Toolkit.
-- The <strong>Documentation</strong> links in the table below include installation and usage
+under active development and may be unstable. A few helpful guidelines:</p>
+<ul>
+<li>The <a href="../Versioning/">Versioning page</a> overviews how we manage our GitHub
+  releases and the versioning process for each of the ML-Agents components.</li>
+<li>The <a href="https://github.com/Unity-Technologies/ml-agents/releases">Releases page</a>
+  contains details of the changes between releases.</li>
+<li>The <a href="../Migrating/">Migration page</a> contains details on how to upgrade
+  from earlier releases of the ML-Agents Toolkit.</li>
+<li>The <strong>Documentation</strong> links in the table below include installation and usage
   instructions specific to each release. Remember to always use the
-  documentation that corresponds to the release version you're using.
-- The <code>com.unity.ml-agents</code> package is <a href="https://docs.unity3d.com/2020.1/Documentation/Manual/pack-safe.html">verified</a>
-  for Unity 2020.1 and later. Verified packages releases are numbered 1.0.x.</p>
+  documentation that corresponds to the release version you're using.</li>
+<li>The <code>com.unity.ml-agents</code> package is <a href="https://docs.unity3d.com/2020.1/Documentation/Manual/pack-safe.html">verified</a>
+  for Unity 2020.1 and later. Verified packages releases are numbered 1.0.x.</li>
+</ul>
 <table>
 <thead>
 <tr>
@@ -1187,7 +1206,7 @@ <h2 id="additional-resources">Additional Resources</h2>
 <h3 id="more-from-unity">More from Unity</h3>
 <ul>
 <li><a href="https://unity.com/products/sentis">Unity Sentis</a></li>
-<li><a href="https://blog.unity.com/engine-platform/introducing-unity-muse-and-unity-sentis-ai">Introductin Unity Muse and Sentis</a></li>
+<li><a href="https://blog.unity.com/engine-platform/introducing-unity-muse-and-unity-sentis-ai">Introducing Unity Muse and Sentis</a></li>
 </ul>
 <h2 id="community-and-feedback">Community and Feedback</h2>
 <p>The ML-Agents Toolkit is an open-source project and we encourage and welcome
@@ -1219,11 +1238,19 @@ <h2 id="privacy">Privacy</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1256,10 +1283,10 @@ <h2 id="privacy">Privacy</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Sentis/index.html b/Sentis/index.html
index 3f7342be9b..717d3b2ce9 100644
--- a/Sentis/index.html
+++ b/Sentis/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1013,11 +1030,19 @@ <h2 id="model-inference-outside-of-unity">Model inference outside of Unity</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1050,10 +1075,10 @@ <h2 id="model-inference-outside-of-unity">Model inference outside of Unity</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Training-Configuration-File/index.html b/Training-Configuration-File/index.html
index 47adac6c92..685ccf12ea 100644
--- a/Training-Configuration-File/index.html
+++ b/Training-Configuration-File/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1523,7 +1540,7 @@ <h2 id="self-play">Self-Play</h2>
 </tr>
 <tr>
 <td style="text-align: left;"><code>play_against_latest_model_ratio</code></td>
-<td style="text-align: left;">(default = <code>0.5</code>) Probability an agent will play against the latest opponent policy. With probability 1 - <code>play_against_latest_model_ratio</code>, the agent will play against a snapshot of its opponent from a past iteration. <br><br> A larger value of <code>play_against_latest_model_ratio</code> indicates that an agent will be playing against the current opponent more often. Since the agent is updating it's policy, the opponent will be different from iteration to iteration. This can lead to an unstable learning environment, but poses the agent with an <a href="https://openai.com/blog/emergent-tool-use/">auto-curricula</a> of more increasingly challenging situations which may lead to a stronger final policy. <br><br> Typical range: <code>0.0</code> - <code>1.0</code></td>
+<td style="text-align: left;">(default = <code>0.5</code>) Probability an agent will play against the latest opponent policy. With probability 1 - <code>play_against_latest_model_ratio</code>, the agent will play against a snapshot of its opponent from a past iteration. <br><br> A larger value of <code>play_against_latest_model_ratio</code> indicates that an agent will be playing against the current opponent more often. Since the agent is updating it's policy, the opponent will be different from iteration to iteration. This can lead to an unstable learning environment, but poses the agent with an <a href="https://openai.com/research/emergent-tool-use">auto-curricula</a> of more increasingly challenging situations which may lead to a stronger final policy. <br><br> Typical range: <code>0.0</code> - <code>1.0</code></td>
 </tr>
 <tr>
 <td style="text-align: left;"><code>window</code></td>
@@ -1560,11 +1577,19 @@ <h3 id="note-on-swap-steps">Note on Swap Steps</h3>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1597,10 +1622,10 @@ <h3 id="note-on-swap-steps">Note on Swap Steps</h3>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Training-ML-Agents/index.html b/Training-ML-Agents/index.html
index 41ed05e251..8b7b30e394 100644
--- a/Training-ML-Agents/index.html
+++ b/Training-ML-Agents/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1528,7 +1545,7 @@ <h3 id="default-behavior-settings">Default Behavior Settings</h3>
 <p>In some cases, you may want to specify a set of default configurations for your Behaviors.
 This may be useful, for instance, if your Behavior names are generated procedurally by
 the environment and not known before runtime, or if you have many Behaviors with very similar
-settings. To specify a default configuraton, insert a <code>default_settings</code> section in your YAML.
+settings. To specify a default configuration, insert a <code>default_settings</code> section in your YAML.
 This section should be formatted exactly like a configuration for a Behavior.</p>
 <pre><code class="language-yaml">default_settings:
   # &lt; Same as Behavior configuration &gt;
@@ -1748,11 +1765,19 @@ <h3 id="training-using-concurrent-unity-instances">Training Using Concurrent Uni
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1785,10 +1810,10 @@ <h3 id="training-using-concurrent-unity-instances">Training Using Concurrent Uni
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Training-Plugins/index.html b/Training-Plugins/index.html
index a81e117741..c756dda0b9 100644
--- a/Training-Plugins/index.html
+++ b/Training-Plugins/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -343,7 +347,7 @@
   
   
   
-    
+  
     
     
     
@@ -352,7 +356,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -379,6 +382,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -399,6 +403,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -419,6 +424,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -439,6 +445,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -466,7 +473,7 @@
   
   
   
-    
+  
     
     
     
@@ -475,7 +482,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -502,6 +508,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -522,6 +529,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -542,6 +550,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -569,7 +578,7 @@
   
   
   
-    
+  
     
     
     
@@ -578,7 +587,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -605,6 +613,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -625,6 +634,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -645,6 +655,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -665,6 +676,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -694,7 +706,7 @@
     
   
   
-    
+  
     
     
     
@@ -703,7 +715,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" checked>
         
           
@@ -732,6 +743,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -871,6 +883,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -891,6 +904,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -918,7 +932,7 @@
   
   
   
-    
+  
     
     
     
@@ -927,7 +941,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -954,6 +967,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -974,6 +988,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -994,6 +1009,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -1014,6 +1030,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1214,11 +1231,19 @@ <h4 id="registration">Registration</h4>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1251,10 +1276,10 @@ <h4 id="registration">Registration</h4>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Training-on-Amazon-Web-Service/index.html b/Training-on-Amazon-Web-Service/index.html
index 4b0aa19725..60e6764b8f 100644
--- a/Training-on-Amazon-Web-Service/index.html
+++ b/Training-on-Amazon-Web-Service/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1385,11 +1402,19 @@ <h3 id="could-not-launch-x-server">Could not launch X Server</h3>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1422,10 +1447,10 @@ <h3 id="could-not-launch-x-server">Could not launch X Server</h3>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Training-on-Microsoft-Azure/index.html b/Training-on-Microsoft-Azure/index.html
index 6b00f68a97..543bb29cd7 100644
--- a/Training-on-Microsoft-Azure/index.html
+++ b/Training-on-Microsoft-Azure/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1030,13 +1047,13 @@ <h1 id="training-on-microsoft-azure-works-with-ml-agents-toolkit-v03">Training o
 correctly. We've decided to keep it up just in case it is helpful to you.</p>
 <p>This page contains instructions for setting up training on Microsoft Azure
 through either
-<a href="https://azure.microsoft.com/services/container-instances/">Azure Container Instances</a>
+<a href="https://azure.microsoft.com/en-us/products/container-instances/">Azure Container Instances</a>
 or Virtual Machines. Non "headless" training has not yet been tested to verify
 support.</p>
 <h2 id="pre-configured-azure-virtual-machine">Pre-Configured Azure Virtual Machine</h2>
 <p>A pre-configured virtual machine image is available in the Azure Marketplace and
 is nearly completely ready for training. You can start by deploying the
-<a href="https://azuremarketplace.microsoft.com/en-us/marketplace/apps/microsoft-dsvm.ubuntu-1804">Data Science Virtual Machine for Linux (Ubuntu)</a>
+<a href="https://learn.microsoft.com/en-us/azure/machine-learning/data-science-virtual-machine/dsvm-ubuntu-intro?view=azureml-api-2">Data Science Virtual Machine for Linux (Ubuntu)</a>
 into your Azure subscription.</p>
 <p>Note that, if you choose to deploy the image to an
 <a href="https://docs.microsoft.com/azure/virtual-machines/linux/sizes-gpu">N-Series GPU optimized VM</a>,
@@ -1126,7 +1143,7 @@ <h2 id="monitoring-your-training-run-with-tensorboard">Monitoring your Training
    <code>&lt;Your_VM_IP_Address&gt;:6060</code> to view the TensorBoard report.</li>
 </ol>
 <h2 id="running-on-azure-container-instances">Running on Azure Container Instances</h2>
-<p><a href="https://azure.microsoft.com/services/container-instances/">Azure Container Instances</a>
+<p><a href="https://azure.microsoft.com/en-us/products/container-instances/">Azure Container Instances</a>
 allow you to spin up a container, on demand, that will run your training and
 then be shut down. This ensures you aren't leaving a billable VM running when it
 isn't needed. Using ACI enables you to offload training of your models without
@@ -1216,11 +1233,19 @@ <h2 id="custom-instances">Custom Instances</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1253,10 +1278,10 @@ <h2 id="custom-instances">Custom Instances</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Tutorial-Custom-Trainer-Plugin/index.html b/Tutorial-Custom-Trainer-Plugin/index.html
index e2ed32c07c..bf34ada4dc 100644
--- a/Tutorial-Custom-Trainer-Plugin/index.html
+++ b/Tutorial-Custom-Trainer-Plugin/index.html
@@ -22,7 +22,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -30,10 +30,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -74,9 +74,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -131,21 +128,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -153,9 +150,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -164,6 +161,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -207,7 +207,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -252,7 +252,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -286,6 +287,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -305,6 +307,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -324,6 +327,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -343,7 +347,7 @@
   
   
   
-    
+  
     
     
     
@@ -352,7 +356,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -379,6 +382,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -399,6 +403,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -419,6 +424,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -439,6 +445,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -466,7 +473,7 @@
   
   
   
-    
+  
     
     
     
@@ -475,7 +482,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -502,6 +508,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -522,6 +529,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -542,6 +550,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -569,7 +578,7 @@
   
   
   
-    
+  
     
     
     
@@ -578,7 +587,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -605,6 +613,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -625,6 +634,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -645,6 +655,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -665,6 +676,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -694,7 +706,7 @@
     
   
   
-    
+  
     
     
     
@@ -703,7 +715,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" checked>
         
           
@@ -730,6 +741,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -752,6 +764,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -870,6 +883,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -897,7 +911,7 @@
   
   
   
-    
+  
     
     
     
@@ -906,7 +920,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -933,6 +946,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -953,6 +967,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -973,6 +988,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -993,6 +1009,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1128,7 +1145,7 @@ <h3 id="step-1-write-your-custom-trainer-class">Step 1: Write your custom traine
 </code></pre>
 <p>Users of the plug-in system are responsible for implementing the trainer class subject to the API standard. Let us follow an example by implementing a custom trainer named "YourCustomTrainer". You can either extend <code>OnPolicyTrainer</code> or <code>OffPolicyTrainer</code> classes depending on the training strategies you choose.</p>
 <p>Please refer to the internal <a href="../ml-agents/mlagents/trainers/ppo/trainer.py">PPO implementation</a> for a complete code example. We will not provide a workable code in the document. The purpose of the tutorial is to introduce you to the core components and interfaces of our plugin framework. We use code snippets and patterns to demonstrate the control and data flow.</p>
-<p>Your custom trainers are responsible for collecting experiences and training the models. Your custom trainer class acts like a co-ordinator to the policy and optimizer. To start implementing methods in the class, create a policy class objects from method <code>create_policy</code>:</p>
+<p>Your custom trainers are responsible for collecting experiences and training the models. Your custom trainer class acts like a coordinator to the policy and optimizer. To start implementing methods in the class, create a policy class objects from method <code>create_policy</code>:</p>
 <pre><code class="language-python">def create_policy(
     self, parsed_behavior_id: BehaviorIdentifiers, behavior_spec: BehaviorSpec
 ) -&gt; TorchPolicy:
@@ -1304,7 +1321,7 @@ <h3 id="step-4-install-your-custom-trainer-and-run-training">Step 4: Install you
 <p>Before installing your custom trainer package, make sure you have <code>ml-agents-env</code> and <code>ml-agents</code> installed</p>
 <pre><code class="language-shell">pip3 install -e ./ml-agents-envs &amp;&amp; pip3 install -e ./ml-agents
 </code></pre>
-<p>Install your cutom trainer package(if your package is pip installable):</p>
+<p>Install your custom trainer package(if your package is pip installable):</p>
 <pre><code class="language-shell">pip3 install your_custom_package
 </code></pre>
 <p>Or follow our internal implementations:</p>
@@ -1350,11 +1367,19 @@ <h3 id="validate-your-implementations">Validate your implementations:</h3>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1387,10 +1412,10 @@ <h3 id="validate-your-implementations">Validate your implementations:</h3>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Unity-Environment-Registry/index.html b/Unity-Environment-Registry/index.html
index d92c41afc2..740c472871 100644
--- a/Unity-Environment-Registry/index.html
+++ b/Unity-Environment-Registry/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -991,13 +1008,15 @@ <h2 id="loading-an-environment-from-the-registry">Loading an Environment from th
 env.close()
 </code></pre>
 <h2 id="create-and-share-your-own-registry">Create and share your own registry</h2>
-<p>In order to share the <code>UnityEnvironemnt</code> you created, you must :
- - <a href="../Learning-Environment-Executable/">Create a Unity executable</a> of your environment for each platform (Linux, OSX and/or Windows)
- - Place each executable in a <code>zip</code> compressed folder
- - Upload each zip file online to your preferred hosting platform
- - Create a <code>yaml</code> file that will contain the description and path to your environment
- - Upload the <code>yaml</code> file online
-The <code>yaml</code> file must have the following format :</p>
+<p>In order to share the <code>UnityEnvironment</code> you created, you must:</p>
+<ul>
+<li><a href="../Learning-Environment-Executable/">Create a Unity executable</a> of your environment for each platform (Linux, OSX and/or Windows)</li>
+<li>Place each executable in a <code>zip</code> compressed folder</li>
+<li>Upload each zip file online to your preferred hosting platform</li>
+<li>Create a <code>yaml</code> file that will contain the description and path to your environment</li>
+<li>Upload the <code>yaml</code> file online
+The <code>yaml</code> file must have the following format :</li>
+</ul>
 <pre><code class="language-yaml">environments:
   - &lt;environment-identifier&gt;:
      expected_reward: &lt;expected-reward-float&gt;
@@ -1021,11 +1040,19 @@ <h2 id="create-and-share-your-own-registry">Create and share your own registry</
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1058,10 +1085,10 @@ <h2 id="create-and-share-your-own-registry">Create and share your own registry</
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Using-Docker/index.html b/Using-Docker/index.html
index 17ed39104f..db53dcd246 100644
--- a/Using-Docker/index.html
+++ b/Using-Docker/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1182,11 +1199,19 @@ <h3 id="stopping-container-and-saving-state">Stopping Container and Saving State
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1219,10 +1244,10 @@ <h3 id="stopping-container-and-saving-state">Stopping Container and Saving State
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Using-Tensorboard/index.html b/Using-Tensorboard/index.html
index 698ab4413d..5cce08414f 100644
--- a/Using-Tensorboard/index.html
+++ b/Using-Tensorboard/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1034,7 +1051,7 @@
 <h1 id="using-tensorboard-to-observe-training">Using TensorBoard to Observe Training</h1>
 <p>The ML-Agents Toolkit saves statistics during learning session that you can view
 with a TensorFlow utility named,
-<a href="https://www.tensorflow.org/programmers_guide/summaries_and_tensorboard">TensorBoard</a>.</p>
+<a href="https://www.tensorflow.org/tensorboard">TensorBoard</a>.</p>
 <p>The <code>mlagents-learn</code> command saves training statistics to a folder named
 <code>results</code>, organized by the <code>run-id</code> value you assign to a training session.</p>
 <p>In order to observe the training process, either during training or afterward,
@@ -1181,11 +1198,19 @@ <h2 id="custom-metrics-from-unity">Custom Metrics from Unity</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1218,10 +1243,10 @@ <h2 id="custom-metrics-from-unity">Custom Metrics from Unity</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Using-Virtual-Environment/index.html b/Using-Virtual-Environment/index.html
index 841a23fc3b..a1177911a7 100644
--- a/Using-Virtual-Environment/index.html
+++ b/Using-Virtual-Environment/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1044,7 +1061,7 @@ <h2 id="python-version-requirement-required">Python Version Requirement (Require
 <p>This guide has been tested with Python 3.10.12. Newer versions might not
 have support for the dependent libraries, so are not recommended.</p>
 <h2 id="use-conda-or-mamba">Use Conda (or Mamba)</h2>
-<p>While there are many options for setting up virtual environments for python, by far the most</p>
+<p>While there are many options for setting up virtual environments for python, by far the most common and simpler approach is by using Anaconda (aka Conda). You can read the documentation on how to get started with Conda <a href="https://learning.anaconda.cloud/get-started-with-anaconda">here</a>.</p>
 <h2 id="installing-pip-required">Installing Pip (Required)</h2>
 <ol>
 <li>Download the <code>get-pip.py</code> file using the command
@@ -1095,11 +1112,19 @@ <h2 id="windows-setup">Windows Setup</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1132,10 +1157,10 @@ <h2 id="windows-setup">Windows Setup</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/Versioning/index.html b/Versioning/index.html
index 4b150c2aeb..63ac6576d2 100644
--- a/Versioning/index.html
+++ b/Versioning/index.html
@@ -20,7 +20,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -28,10 +28,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -72,9 +72,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -129,21 +126,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -151,9 +148,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -162,6 +159,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -205,7 +205,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -250,7 +250,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -265,6 +265,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -284,6 +285,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -303,6 +305,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -322,6 +325,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -341,7 +345,7 @@
   
   
   
-    
+  
     
     
     
@@ -350,7 +354,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -377,6 +380,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -397,6 +401,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -417,6 +422,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -437,6 +443,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -464,7 +471,7 @@
   
   
   
-    
+  
     
     
     
@@ -473,7 +480,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -500,6 +506,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -520,6 +527,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -540,6 +548,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -567,7 +576,7 @@
   
   
   
-    
+  
     
     
     
@@ -576,7 +585,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -603,6 +611,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -623,6 +632,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -643,6 +653,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -663,6 +674,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -690,7 +702,7 @@
   
   
   
-    
+  
     
     
     
@@ -699,7 +711,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -726,6 +737,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -746,6 +758,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -766,6 +779,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -795,7 +809,7 @@
     
   
   
-    
+  
     
     
     
@@ -804,7 +818,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" checked>
         
           
@@ -831,6 +844,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -851,6 +865,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -871,6 +886,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -893,6 +909,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -1219,11 +1236,19 @@ <h2 id="side-channels">Side Channels</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1256,10 +1281,10 @@ <h2 id="side-channels">Side Channels</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/assets/javascripts/bundle.1e8ae164.min.js b/assets/javascripts/bundle.1e8ae164.min.js
new file mode 100644
index 0000000000..212979889b
--- /dev/null
+++ b/assets/javascripts/bundle.1e8ae164.min.js
@@ -0,0 +1,29 @@
+"use strict";(()=>{var _i=Object.create;var br=Object.defineProperty;var Ai=Object.getOwnPropertyDescriptor;var Ci=Object.getOwnPropertyNames,Ft=Object.getOwnPropertySymbols,ki=Object.getPrototypeOf,vr=Object.prototype.hasOwnProperty,eo=Object.prototype.propertyIsEnumerable;var Zr=(e,t,r)=>t in e?br(e,t,{enumerable:!0,configurable:!0,writable:!0,value:r}):e[t]=r,F=(e,t)=>{for(var r in t||(t={}))vr.call(t,r)&&Zr(e,r,t[r]);if(Ft)for(var r of Ft(t))eo.call(t,r)&&Zr(e,r,t[r]);return e};var to=(e,t)=>{var r={};for(var o in e)vr.call(e,o)&&t.indexOf(o)<0&&(r[o]=e[o]);if(e!=null&&Ft)for(var o of Ft(e))t.indexOf(o)<0&&eo.call(e,o)&&(r[o]=e[o]);return r};var gr=(e,t)=>()=>(t||e((t={exports:{}}).exports,t),t.exports);var Hi=(e,t,r,o)=>{if(t&&typeof t=="object"||typeof t=="function")for(let n of Ci(t))!vr.call(e,n)&&n!==r&&br(e,n,{get:()=>t[n],enumerable:!(o=Ai(t,n))||o.enumerable});return e};var jt=(e,t,r)=>(r=e!=null?_i(ki(e)):{},Hi(t||!e||!e.__esModule?br(r,"default",{value:e,enumerable:!0}):r,e));var ro=(e,t,r)=>new Promise((o,n)=>{var i=c=>{try{s(r.next(c))}catch(p){n(p)}},a=c=>{try{s(r.throw(c))}catch(p){n(p)}},s=c=>c.done?o(c.value):Promise.resolve(c.value).then(i,a);s((r=r.apply(e,t)).next())});var no=gr((xr,oo)=>{(function(e,t){typeof xr=="object"&&typeof oo!="undefined"?t():typeof define=="function"&&define.amd?define(t):t()})(xr,function(){"use strict";function e(r){var o=!0,n=!1,i=null,a={text:!0,search:!0,url:!0,tel:!0,email:!0,password:!0,number:!0,date:!0,month:!0,week:!0,time:!0,datetime:!0,"datetime-local":!0};function s(C){return!!(C&&C!==document&&C.nodeName!=="HTML"&&C.nodeName!=="BODY"&&"classList"in C&&"contains"in C.classList)}function c(C){var ct=C.type,Ne=C.tagName;return!!(Ne==="INPUT"&&a[ct]&&!C.readOnly||Ne==="TEXTAREA"&&!C.readOnly||C.isContentEditable)}function p(C){C.classList.contains("focus-visible")||(C.classList.add("focus-visible"),C.setAttribute("data-focus-visible-added",""))}function l(C){C.hasAttribute("data-focus-visible-added")&&(C.classList.remove("focus-visible"),C.removeAttribute("data-focus-visible-added"))}function f(C){C.metaKey||C.altKey||C.ctrlKey||(s(r.activeElement)&&p(r.activeElement),o=!0)}function u(C){o=!1}function h(C){s(C.target)&&(o||c(C.target))&&p(C.target)}function w(C){s(C.target)&&(C.target.classList.contains("focus-visible")||C.target.hasAttribute("data-focus-visible-added"))&&(n=!0,window.clearTimeout(i),i=window.setTimeout(function(){n=!1},100),l(C.target))}function A(C){document.visibilityState==="hidden"&&(n&&(o=!0),Z())}function Z(){document.addEventListener("mousemove",J),document.addEventListener("mousedown",J),document.addEventListener("mouseup",J),document.addEventListener("pointermove",J),document.addEventListener("pointerdown",J),document.addEventListener("pointerup",J),document.addEventListener("touchmove",J),document.addEventListener("touchstart",J),document.addEventListener("touchend",J)}function te(){document.removeEventListener("mousemove",J),document.removeEventListener("mousedown",J),document.removeEventListener("mouseup",J),document.removeEventListener("pointermove",J),document.removeEventListener("pointerdown",J),document.removeEventListener("pointerup",J),document.removeEventListener("touchmove",J),document.removeEventListener("touchstart",J),document.removeEventListener("touchend",J)}function J(C){C.target.nodeName&&C.target.nodeName.toLowerCase()==="html"||(o=!1,te())}document.addEventListener("keydown",f,!0),document.addEventListener("mousedown",u,!0),document.addEventListener("pointerdown",u,!0),document.addEventListener("touchstart",u,!0),document.addEventListener("visibilitychange",A,!0),Z(),r.addEventListener("focus",h,!0),r.addEventListener("blur",w,!0),r.nodeType===Node.DOCUMENT_FRAGMENT_NODE&&r.host?r.host.setAttribute("data-js-focus-visible",""):r.nodeType===Node.DOCUMENT_NODE&&(document.documentElement.classList.add("js-focus-visible"),document.documentElement.setAttribute("data-js-focus-visible",""))}if(typeof window!="undefined"&&typeof document!="undefined"){window.applyFocusVisiblePolyfill=e;var t;try{t=new CustomEvent("focus-visible-polyfill-ready")}catch(r){t=document.createEvent("CustomEvent"),t.initCustomEvent("focus-visible-polyfill-ready",!1,!1,{})}window.dispatchEvent(t)}typeof document!="undefined"&&e(document)})});var zr=gr((kt,Vr)=>{/*!
+ * clipboard.js v2.0.11
+ * https://clipboardjs.com/
+ *
+ * Licensed MIT © Zeno Rocha
+ */(function(t,r){typeof kt=="object"&&typeof Vr=="object"?Vr.exports=r():typeof define=="function"&&define.amd?define([],r):typeof kt=="object"?kt.ClipboardJS=r():t.ClipboardJS=r()})(kt,function(){return function(){var e={686:function(o,n,i){"use strict";i.d(n,{default:function(){return Li}});var a=i(279),s=i.n(a),c=i(370),p=i.n(c),l=i(817),f=i.n(l);function u(D){try{return document.execCommand(D)}catch(M){return!1}}var h=function(M){var O=f()(M);return u("cut"),O},w=h;function A(D){var M=document.documentElement.getAttribute("dir")==="rtl",O=document.createElement("textarea");O.style.fontSize="12pt",O.style.border="0",O.style.padding="0",O.style.margin="0",O.style.position="absolute",O.style[M?"right":"left"]="-9999px";var I=window.pageYOffset||document.documentElement.scrollTop;return O.style.top="".concat(I,"px"),O.setAttribute("readonly",""),O.value=D,O}var Z=function(M,O){var I=A(M);O.container.appendChild(I);var W=f()(I);return u("copy"),I.remove(),W},te=function(M){var O=arguments.length>1&&arguments[1]!==void 0?arguments[1]:{container:document.body},I="";return typeof M=="string"?I=Z(M,O):M instanceof HTMLInputElement&&!["text","search","url","tel","password"].includes(M==null?void 0:M.type)?I=Z(M.value,O):(I=f()(M),u("copy")),I},J=te;function C(D){"@babel/helpers - typeof";return typeof Symbol=="function"&&typeof Symbol.iterator=="symbol"?C=function(O){return typeof O}:C=function(O){return O&&typeof Symbol=="function"&&O.constructor===Symbol&&O!==Symbol.prototype?"symbol":typeof O},C(D)}var ct=function(){var M=arguments.length>0&&arguments[0]!==void 0?arguments[0]:{},O=M.action,I=O===void 0?"copy":O,W=M.container,K=M.target,Ce=M.text;if(I!=="copy"&&I!=="cut")throw new Error('Invalid "action" value, use either "copy" or "cut"');if(K!==void 0)if(K&&C(K)==="object"&&K.nodeType===1){if(I==="copy"&&K.hasAttribute("disabled"))throw new Error('Invalid "target" attribute. Please use "readonly" instead of "disabled" attribute');if(I==="cut"&&(K.hasAttribute("readonly")||K.hasAttribute("disabled")))throw new Error(`Invalid "target" attribute. You can't cut text from elements with "readonly" or "disabled" attributes`)}else throw new Error('Invalid "target" value, use a valid Element');if(Ce)return J(Ce,{container:W});if(K)return I==="cut"?w(K):J(K,{container:W})},Ne=ct;function Pe(D){"@babel/helpers - typeof";return typeof Symbol=="function"&&typeof Symbol.iterator=="symbol"?Pe=function(O){return typeof O}:Pe=function(O){return O&&typeof Symbol=="function"&&O.constructor===Symbol&&O!==Symbol.prototype?"symbol":typeof O},Pe(D)}function xi(D,M){if(!(D instanceof M))throw new TypeError("Cannot call a class as a function")}function Xr(D,M){for(var O=0;O<M.length;O++){var I=M[O];I.enumerable=I.enumerable||!1,I.configurable=!0,"value"in I&&(I.writable=!0),Object.defineProperty(D,I.key,I)}}function yi(D,M,O){return M&&Xr(D.prototype,M),O&&Xr(D,O),D}function Ei(D,M){if(typeof M!="function"&&M!==null)throw new TypeError("Super expression must either be null or a function");D.prototype=Object.create(M&&M.prototype,{constructor:{value:D,writable:!0,configurable:!0}}),M&&dr(D,M)}function dr(D,M){return dr=Object.setPrototypeOf||function(I,W){return I.__proto__=W,I},dr(D,M)}function wi(D){var M=Oi();return function(){var I=Pt(D),W;if(M){var K=Pt(this).constructor;W=Reflect.construct(I,arguments,K)}else W=I.apply(this,arguments);return Ti(this,W)}}function Ti(D,M){return M&&(Pe(M)==="object"||typeof M=="function")?M:Si(D)}function Si(D){if(D===void 0)throw new ReferenceError("this hasn't been initialised - super() hasn't been called");return D}function Oi(){if(typeof Reflect=="undefined"||!Reflect.construct||Reflect.construct.sham)return!1;if(typeof Proxy=="function")return!0;try{return Date.prototype.toString.call(Reflect.construct(Date,[],function(){})),!0}catch(D){return!1}}function Pt(D){return Pt=Object.setPrototypeOf?Object.getPrototypeOf:function(O){return O.__proto__||Object.getPrototypeOf(O)},Pt(D)}function hr(D,M){var O="data-clipboard-".concat(D);if(M.hasAttribute(O))return M.getAttribute(O)}var Mi=function(D){Ei(O,D);var M=wi(O);function O(I,W){var K;return xi(this,O),K=M.call(this),K.resolveOptions(W),K.listenClick(I),K}return yi(O,[{key:"resolveOptions",value:function(){var W=arguments.length>0&&arguments[0]!==void 0?arguments[0]:{};this.action=typeof W.action=="function"?W.action:this.defaultAction,this.target=typeof W.target=="function"?W.target:this.defaultTarget,this.text=typeof W.text=="function"?W.text:this.defaultText,this.container=Pe(W.container)==="object"?W.container:document.body}},{key:"listenClick",value:function(W){var K=this;this.listener=p()(W,"click",function(Ce){return K.onClick(Ce)})}},{key:"onClick",value:function(W){var K=W.delegateTarget||W.currentTarget,Ce=this.action(K)||"copy",It=Ne({action:Ce,container:this.container,target:this.target(K),text:this.text(K)});this.emit(It?"success":"error",{action:Ce,text:It,trigger:K,clearSelection:function(){K&&K.focus(),window.getSelection().removeAllRanges()}})}},{key:"defaultAction",value:function(W){return hr("action",W)}},{key:"defaultTarget",value:function(W){var K=hr("target",W);if(K)return document.querySelector(K)}},{key:"defaultText",value:function(W){return hr("text",W)}},{key:"destroy",value:function(){this.listener.destroy()}}],[{key:"copy",value:function(W){var K=arguments.length>1&&arguments[1]!==void 0?arguments[1]:{container:document.body};return J(W,K)}},{key:"cut",value:function(W){return w(W)}},{key:"isSupported",value:function(){var W=arguments.length>0&&arguments[0]!==void 0?arguments[0]:["copy","cut"],K=typeof W=="string"?[W]:W,Ce=!!document.queryCommandSupported;return K.forEach(function(It){Ce=Ce&&!!document.queryCommandSupported(It)}),Ce}}]),O}(s()),Li=Mi},828:function(o){var n=9;if(typeof Element!="undefined"&&!Element.prototype.matches){var i=Element.prototype;i.matches=i.matchesSelector||i.mozMatchesSelector||i.msMatchesSelector||i.oMatchesSelector||i.webkitMatchesSelector}function a(s,c){for(;s&&s.nodeType!==n;){if(typeof s.matches=="function"&&s.matches(c))return s;s=s.parentNode}}o.exports=a},438:function(o,n,i){var a=i(828);function s(l,f,u,h,w){var A=p.apply(this,arguments);return l.addEventListener(u,A,w),{destroy:function(){l.removeEventListener(u,A,w)}}}function c(l,f,u,h,w){return typeof l.addEventListener=="function"?s.apply(null,arguments):typeof u=="function"?s.bind(null,document).apply(null,arguments):(typeof l=="string"&&(l=document.querySelectorAll(l)),Array.prototype.map.call(l,function(A){return s(A,f,u,h,w)}))}function p(l,f,u,h){return function(w){w.delegateTarget=a(w.target,f),w.delegateTarget&&h.call(l,w)}}o.exports=c},879:function(o,n){n.node=function(i){return i!==void 0&&i instanceof HTMLElement&&i.nodeType===1},n.nodeList=function(i){var a=Object.prototype.toString.call(i);return i!==void 0&&(a==="[object NodeList]"||a==="[object HTMLCollection]")&&"length"in i&&(i.length===0||n.node(i[0]))},n.string=function(i){return typeof i=="string"||i instanceof String},n.fn=function(i){var a=Object.prototype.toString.call(i);return a==="[object Function]"}},370:function(o,n,i){var a=i(879),s=i(438);function c(u,h,w){if(!u&&!h&&!w)throw new Error("Missing required arguments");if(!a.string(h))throw new TypeError("Second argument must be a String");if(!a.fn(w))throw new TypeError("Third argument must be a Function");if(a.node(u))return p(u,h,w);if(a.nodeList(u))return l(u,h,w);if(a.string(u))return f(u,h,w);throw new TypeError("First argument must be a String, HTMLElement, HTMLCollection, or NodeList")}function p(u,h,w){return u.addEventListener(h,w),{destroy:function(){u.removeEventListener(h,w)}}}function l(u,h,w){return Array.prototype.forEach.call(u,function(A){A.addEventListener(h,w)}),{destroy:function(){Array.prototype.forEach.call(u,function(A){A.removeEventListener(h,w)})}}}function f(u,h,w){return s(document.body,u,h,w)}o.exports=c},817:function(o){function n(i){var a;if(i.nodeName==="SELECT")i.focus(),a=i.value;else if(i.nodeName==="INPUT"||i.nodeName==="TEXTAREA"){var s=i.hasAttribute("readonly");s||i.setAttribute("readonly",""),i.select(),i.setSelectionRange(0,i.value.length),s||i.removeAttribute("readonly"),a=i.value}else{i.hasAttribute("contenteditable")&&i.focus();var c=window.getSelection(),p=document.createRange();p.selectNodeContents(i),c.removeAllRanges(),c.addRange(p),a=c.toString()}return a}o.exports=n},279:function(o){function n(){}n.prototype={on:function(i,a,s){var c=this.e||(this.e={});return(c[i]||(c[i]=[])).push({fn:a,ctx:s}),this},once:function(i,a,s){var c=this;function p(){c.off(i,p),a.apply(s,arguments)}return p._=a,this.on(i,p,s)},emit:function(i){var a=[].slice.call(arguments,1),s=((this.e||(this.e={}))[i]||[]).slice(),c=0,p=s.length;for(c;c<p;c++)s[c].fn.apply(s[c].ctx,a);return this},off:function(i,a){var s=this.e||(this.e={}),c=s[i],p=[];if(c&&a)for(var l=0,f=c.length;l<f;l++)c[l].fn!==a&&c[l].fn._!==a&&p.push(c[l]);return p.length?s[i]=p:delete s[i],this}},o.exports=n,o.exports.TinyEmitter=n}},t={};function r(o){if(t[o])return t[o].exports;var n=t[o]={exports:{}};return e[o](n,n.exports,r),n.exports}return function(){r.n=function(o){var n=o&&o.__esModule?function(){return o.default}:function(){return o};return r.d(n,{a:n}),n}}(),function(){r.d=function(o,n){for(var i in n)r.o(n,i)&&!r.o(o,i)&&Object.defineProperty(o,i,{enumerable:!0,get:n[i]})}}(),function(){r.o=function(o,n){return Object.prototype.hasOwnProperty.call(o,n)}}(),r(686)}().default})});var Kn=gr((Cw,qn)=>{"use strict";/*!
+ * escape-html
+ * Copyright(c) 2012-2013 TJ Holowaychuk
+ * Copyright(c) 2015 Andreas Lubbe
+ * Copyright(c) 2015 Tiancheng "Timothy" Gu
+ * MIT Licensed
+ */var Va=/["'&<>]/;qn.exports=za;function za(e){var t=""+e,r=Va.exec(t);if(!r)return t;var o,n="",i=0,a=0;for(i=r.index;i<t.length;i++){switch(t.charCodeAt(i)){case 34:o="&quot;";break;case 38:o="&amp;";break;case 39:o="&#39;";break;case 60:o="&lt;";break;case 62:o="&gt;";break;default:continue}a!==i&&(n+=t.substring(a,i)),a=i+1,n+=o}return a!==i?n+t.substring(a,i):n}});var mM=jt(no());/*! *****************************************************************************
+Copyright (c) Microsoft Corporation.
+
+Permission to use, copy, modify, and/or distribute this software for any
+purpose with or without fee is hereby granted.
+
+THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES WITH
+REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY
+AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT,
+INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM
+LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR
+OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
+PERFORMANCE OF THIS SOFTWARE.
+***************************************************************************** */var yr=function(e,t){return yr=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(r,o){r.__proto__=o}||function(r,o){for(var n in o)Object.prototype.hasOwnProperty.call(o,n)&&(r[n]=o[n])},yr(e,t)};function se(e,t){if(typeof t!="function"&&t!==null)throw new TypeError("Class extends value "+String(t)+" is not a constructor or null");yr(e,t);function r(){this.constructor=e}e.prototype=t===null?Object.create(t):(r.prototype=t.prototype,new r)}function io(e,t,r,o){function n(i){return i instanceof r?i:new r(function(a){a(i)})}return new(r||(r=Promise))(function(i,a){function s(l){try{p(o.next(l))}catch(f){a(f)}}function c(l){try{p(o.throw(l))}catch(f){a(f)}}function p(l){l.done?i(l.value):n(l.value).then(s,c)}p((o=o.apply(e,t||[])).next())})}function Ut(e,t){var r={label:0,sent:function(){if(i[0]&1)throw i[1];return i[1]},trys:[],ops:[]},o,n,i,a;return a={next:s(0),throw:s(1),return:s(2)},typeof Symbol=="function"&&(a[Symbol.iterator]=function(){return this}),a;function s(p){return function(l){return c([p,l])}}function c(p){if(o)throw new TypeError("Generator is already executing.");for(;r;)try{if(o=1,n&&(i=p[0]&2?n.return:p[0]?n.throw||((i=n.return)&&i.call(n),0):n.next)&&!(i=i.call(n,p[1])).done)return i;switch(n=0,i&&(p=[p[0]&2,i.value]),p[0]){case 0:case 1:i=p;break;case 4:return r.label++,{value:p[1],done:!1};case 5:r.label++,n=p[1],p=[0];continue;case 7:p=r.ops.pop(),r.trys.pop();continue;default:if(i=r.trys,!(i=i.length>0&&i[i.length-1])&&(p[0]===6||p[0]===2)){r=0;continue}if(p[0]===3&&(!i||p[1]>i[0]&&p[1]<i[3])){r.label=p[1];break}if(p[0]===6&&r.label<i[1]){r.label=i[1],i=p;break}if(i&&r.label<i[2]){r.label=i[2],r.ops.push(p);break}i[2]&&r.ops.pop(),r.trys.pop();continue}p=t.call(e,r)}catch(l){p=[6,l],n=0}finally{o=i=0}if(p[0]&5)throw p[1];return{value:p[0]?p[1]:void 0,done:!0}}}function ue(e){var t=typeof Symbol=="function"&&Symbol.iterator,r=t&&e[t],o=0;if(r)return r.call(e);if(e&&typeof e.length=="number")return{next:function(){return e&&o>=e.length&&(e=void 0),{value:e&&e[o++],done:!e}}};throw new TypeError(t?"Object is not iterable.":"Symbol.iterator is not defined.")}function V(e,t){var r=typeof Symbol=="function"&&e[Symbol.iterator];if(!r)return e;var o=r.call(e),n,i=[],a;try{for(;(t===void 0||t-- >0)&&!(n=o.next()).done;)i.push(n.value)}catch(s){a={error:s}}finally{try{n&&!n.done&&(r=o.return)&&r.call(o)}finally{if(a)throw a.error}}return i}function z(e,t,r){if(r||arguments.length===2)for(var o=0,n=t.length,i;o<n;o++)(i||!(o in t))&&(i||(i=Array.prototype.slice.call(t,0,o)),i[o]=t[o]);return e.concat(i||Array.prototype.slice.call(t))}function ot(e){return this instanceof ot?(this.v=e,this):new ot(e)}function ao(e,t,r){if(!Symbol.asyncIterator)throw new TypeError("Symbol.asyncIterator is not defined.");var o=r.apply(e,t||[]),n,i=[];return n={},a("next"),a("throw"),a("return"),n[Symbol.asyncIterator]=function(){return this},n;function a(u){o[u]&&(n[u]=function(h){return new Promise(function(w,A){i.push([u,h,w,A])>1||s(u,h)})})}function s(u,h){try{c(o[u](h))}catch(w){f(i[0][3],w)}}function c(u){u.value instanceof ot?Promise.resolve(u.value.v).then(p,l):f(i[0][2],u)}function p(u){s("next",u)}function l(u){s("throw",u)}function f(u,h){u(h),i.shift(),i.length&&s(i[0][0],i[0][1])}}function so(e){if(!Symbol.asyncIterator)throw new TypeError("Symbol.asyncIterator is not defined.");var t=e[Symbol.asyncIterator],r;return t?t.call(e):(e=typeof ue=="function"?ue(e):e[Symbol.iterator](),r={},o("next"),o("throw"),o("return"),r[Symbol.asyncIterator]=function(){return this},r);function o(i){r[i]=e[i]&&function(a){return new Promise(function(s,c){a=e[i](a),n(s,c,a.done,a.value)})}}function n(i,a,s,c){Promise.resolve(c).then(function(p){i({value:p,done:s})},a)}}function k(e){return typeof e=="function"}function pt(e){var t=function(o){Error.call(o),o.stack=new Error().stack},r=e(t);return r.prototype=Object.create(Error.prototype),r.prototype.constructor=r,r}var Wt=pt(function(e){return function(r){e(this),this.message=r?r.length+` errors occurred during unsubscription:
+`+r.map(function(o,n){return n+1+") "+o.toString()}).join(`
+  `):"",this.name="UnsubscriptionError",this.errors=r}});function Ve(e,t){if(e){var r=e.indexOf(t);0<=r&&e.splice(r,1)}}var Ie=function(){function e(t){this.initialTeardown=t,this.closed=!1,this._parentage=null,this._finalizers=null}return e.prototype.unsubscribe=function(){var t,r,o,n,i;if(!this.closed){this.closed=!0;var a=this._parentage;if(a)if(this._parentage=null,Array.isArray(a))try{for(var s=ue(a),c=s.next();!c.done;c=s.next()){var p=c.value;p.remove(this)}}catch(A){t={error:A}}finally{try{c&&!c.done&&(r=s.return)&&r.call(s)}finally{if(t)throw t.error}}else a.remove(this);var l=this.initialTeardown;if(k(l))try{l()}catch(A){i=A instanceof Wt?A.errors:[A]}var f=this._finalizers;if(f){this._finalizers=null;try{for(var u=ue(f),h=u.next();!h.done;h=u.next()){var w=h.value;try{co(w)}catch(A){i=i!=null?i:[],A instanceof Wt?i=z(z([],V(i)),V(A.errors)):i.push(A)}}}catch(A){o={error:A}}finally{try{h&&!h.done&&(n=u.return)&&n.call(u)}finally{if(o)throw o.error}}}if(i)throw new Wt(i)}},e.prototype.add=function(t){var r;if(t&&t!==this)if(this.closed)co(t);else{if(t instanceof e){if(t.closed||t._hasParent(this))return;t._addParent(this)}(this._finalizers=(r=this._finalizers)!==null&&r!==void 0?r:[]).push(t)}},e.prototype._hasParent=function(t){var r=this._parentage;return r===t||Array.isArray(r)&&r.includes(t)},e.prototype._addParent=function(t){var r=this._parentage;this._parentage=Array.isArray(r)?(r.push(t),r):r?[r,t]:t},e.prototype._removeParent=function(t){var r=this._parentage;r===t?this._parentage=null:Array.isArray(r)&&Ve(r,t)},e.prototype.remove=function(t){var r=this._finalizers;r&&Ve(r,t),t instanceof e&&t._removeParent(this)},e.EMPTY=function(){var t=new e;return t.closed=!0,t}(),e}();var Er=Ie.EMPTY;function Dt(e){return e instanceof Ie||e&&"closed"in e&&k(e.remove)&&k(e.add)&&k(e.unsubscribe)}function co(e){k(e)?e():e.unsubscribe()}var ke={onUnhandledError:null,onStoppedNotification:null,Promise:void 0,useDeprecatedSynchronousErrorHandling:!1,useDeprecatedNextContext:!1};var lt={setTimeout:function(e,t){for(var r=[],o=2;o<arguments.length;o++)r[o-2]=arguments[o];var n=lt.delegate;return n!=null&&n.setTimeout?n.setTimeout.apply(n,z([e,t],V(r))):setTimeout.apply(void 0,z([e,t],V(r)))},clearTimeout:function(e){var t=lt.delegate;return((t==null?void 0:t.clearTimeout)||clearTimeout)(e)},delegate:void 0};function Nt(e){lt.setTimeout(function(){var t=ke.onUnhandledError;if(t)t(e);else throw e})}function Se(){}var po=function(){return wr("C",void 0,void 0)}();function lo(e){return wr("E",void 0,e)}function mo(e){return wr("N",e,void 0)}function wr(e,t,r){return{kind:e,value:t,error:r}}var nt=null;function mt(e){if(ke.useDeprecatedSynchronousErrorHandling){var t=!nt;if(t&&(nt={errorThrown:!1,error:null}),e(),t){var r=nt,o=r.errorThrown,n=r.error;if(nt=null,o)throw n}}else e()}function fo(e){ke.useDeprecatedSynchronousErrorHandling&&nt&&(nt.errorThrown=!0,nt.error=e)}var Tt=function(e){se(t,e);function t(r){var o=e.call(this)||this;return o.isStopped=!1,r?(o.destination=r,Dt(r)&&r.add(o)):o.destination=Ii,o}return t.create=function(r,o,n){return new it(r,o,n)},t.prototype.next=function(r){this.isStopped?Sr(mo(r),this):this._next(r)},t.prototype.error=function(r){this.isStopped?Sr(lo(r),this):(this.isStopped=!0,this._error(r))},t.prototype.complete=function(){this.isStopped?Sr(po,this):(this.isStopped=!0,this._complete())},t.prototype.unsubscribe=function(){this.closed||(this.isStopped=!0,e.prototype.unsubscribe.call(this),this.destination=null)},t.prototype._next=function(r){this.destination.next(r)},t.prototype._error=function(r){try{this.destination.error(r)}finally{this.unsubscribe()}},t.prototype._complete=function(){try{this.destination.complete()}finally{this.unsubscribe()}},t}(Ie);var $i=Function.prototype.bind;function Tr(e,t){return $i.call(e,t)}var Ri=function(){function e(t){this.partialObserver=t}return e.prototype.next=function(t){var r=this.partialObserver;if(r.next)try{r.next(t)}catch(o){Vt(o)}},e.prototype.error=function(t){var r=this.partialObserver;if(r.error)try{r.error(t)}catch(o){Vt(o)}else Vt(t)},e.prototype.complete=function(){var t=this.partialObserver;if(t.complete)try{t.complete()}catch(r){Vt(r)}},e}(),it=function(e){se(t,e);function t(r,o,n){var i=e.call(this)||this,a;if(k(r)||!r)a={next:r!=null?r:void 0,error:o!=null?o:void 0,complete:n!=null?n:void 0};else{var s;i&&ke.useDeprecatedNextContext?(s=Object.create(r),s.unsubscribe=function(){return i.unsubscribe()},a={next:r.next&&Tr(r.next,s),error:r.error&&Tr(r.error,s),complete:r.complete&&Tr(r.complete,s)}):a=r}return i.destination=new Ri(a),i}return t}(Tt);function Vt(e){ke.useDeprecatedSynchronousErrorHandling?fo(e):Nt(e)}function Pi(e){throw e}function Sr(e,t){var r=ke.onStoppedNotification;r&&lt.setTimeout(function(){return r(e,t)})}var Ii={closed:!0,next:Se,error:Pi,complete:Se};var ft=function(){return typeof Symbol=="function"&&Symbol.observable||"@@observable"}();function ce(e){return e}function uo(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];return Or(e)}function Or(e){return e.length===0?ce:e.length===1?e[0]:function(r){return e.reduce(function(o,n){return n(o)},r)}}var j=function(){function e(t){t&&(this._subscribe=t)}return e.prototype.lift=function(t){var r=new e;return r.source=this,r.operator=t,r},e.prototype.subscribe=function(t,r,o){var n=this,i=ji(t)?t:new it(t,r,o);return mt(function(){var a=n,s=a.operator,c=a.source;i.add(s?s.call(i,c):c?n._subscribe(i):n._trySubscribe(i))}),i},e.prototype._trySubscribe=function(t){try{return this._subscribe(t)}catch(r){t.error(r)}},e.prototype.forEach=function(t,r){var o=this;return r=ho(r),new r(function(n,i){var a=new it({next:function(s){try{t(s)}catch(c){i(c),a.unsubscribe()}},error:i,complete:n});o.subscribe(a)})},e.prototype._subscribe=function(t){var r;return(r=this.source)===null||r===void 0?void 0:r.subscribe(t)},e.prototype[ft]=function(){return this},e.prototype.pipe=function(){for(var t=[],r=0;r<arguments.length;r++)t[r]=arguments[r];return Or(t)(this)},e.prototype.toPromise=function(t){var r=this;return t=ho(t),new t(function(o,n){var i;r.subscribe(function(a){return i=a},function(a){return n(a)},function(){return o(i)})})},e.create=function(t){return new e(t)},e}();function ho(e){var t;return(t=e!=null?e:ke.Promise)!==null&&t!==void 0?t:Promise}function Fi(e){return e&&k(e.next)&&k(e.error)&&k(e.complete)}function ji(e){return e&&e instanceof Tt||Fi(e)&&Dt(e)}function Ui(e){return k(e==null?void 0:e.lift)}function x(e){return function(t){if(Ui(t))return t.lift(function(r){try{return e(r,this)}catch(o){this.error(o)}});throw new TypeError("Unable to lift unknown Observable type")}}function S(e,t,r,o,n){return new Wi(e,t,r,o,n)}var Wi=function(e){se(t,e);function t(r,o,n,i,a,s){var c=e.call(this,r)||this;return c.onFinalize=a,c.shouldUnsubscribe=s,c._next=o?function(p){try{o(p)}catch(l){r.error(l)}}:e.prototype._next,c._error=i?function(p){try{i(p)}catch(l){r.error(l)}finally{this.unsubscribe()}}:e.prototype._error,c._complete=n?function(){try{n()}catch(p){r.error(p)}finally{this.unsubscribe()}}:e.prototype._complete,c}return t.prototype.unsubscribe=function(){var r;if(!this.shouldUnsubscribe||this.shouldUnsubscribe()){var o=this.closed;e.prototype.unsubscribe.call(this),!o&&((r=this.onFinalize)===null||r===void 0||r.call(this))}},t}(Tt);var ut={schedule:function(e){var t=requestAnimationFrame,r=cancelAnimationFrame,o=ut.delegate;o&&(t=o.requestAnimationFrame,r=o.cancelAnimationFrame);var n=t(function(i){r=void 0,e(i)});return new Ie(function(){return r==null?void 0:r(n)})},requestAnimationFrame:function(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=ut.delegate;return((r==null?void 0:r.requestAnimationFrame)||requestAnimationFrame).apply(void 0,z([],V(e)))},cancelAnimationFrame:function(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=ut.delegate;return((r==null?void 0:r.cancelAnimationFrame)||cancelAnimationFrame).apply(void 0,z([],V(e)))},delegate:void 0};var bo=pt(function(e){return function(){e(this),this.name="ObjectUnsubscribedError",this.message="object unsubscribed"}});var g=function(e){se(t,e);function t(){var r=e.call(this)||this;return r.closed=!1,r.currentObservers=null,r.observers=[],r.isStopped=!1,r.hasError=!1,r.thrownError=null,r}return t.prototype.lift=function(r){var o=new vo(this,this);return o.operator=r,o},t.prototype._throwIfClosed=function(){if(this.closed)throw new bo},t.prototype.next=function(r){var o=this;mt(function(){var n,i;if(o._throwIfClosed(),!o.isStopped){o.currentObservers||(o.currentObservers=Array.from(o.observers));try{for(var a=ue(o.currentObservers),s=a.next();!s.done;s=a.next()){var c=s.value;c.next(r)}}catch(p){n={error:p}}finally{try{s&&!s.done&&(i=a.return)&&i.call(a)}finally{if(n)throw n.error}}}})},t.prototype.error=function(r){var o=this;mt(function(){if(o._throwIfClosed(),!o.isStopped){o.hasError=o.isStopped=!0,o.thrownError=r;for(var n=o.observers;n.length;)n.shift().error(r)}})},t.prototype.complete=function(){var r=this;mt(function(){if(r._throwIfClosed(),!r.isStopped){r.isStopped=!0;for(var o=r.observers;o.length;)o.shift().complete()}})},t.prototype.unsubscribe=function(){this.isStopped=this.closed=!0,this.observers=this.currentObservers=null},Object.defineProperty(t.prototype,"observed",{get:function(){var r;return((r=this.observers)===null||r===void 0?void 0:r.length)>0},enumerable:!1,configurable:!0}),t.prototype._trySubscribe=function(r){return this._throwIfClosed(),e.prototype._trySubscribe.call(this,r)},t.prototype._subscribe=function(r){return this._throwIfClosed(),this._checkFinalizedStatuses(r),this._innerSubscribe(r)},t.prototype._innerSubscribe=function(r){var o=this,n=this,i=n.hasError,a=n.isStopped,s=n.observers;return i||a?Er:(this.currentObservers=null,s.push(r),new Ie(function(){o.currentObservers=null,Ve(s,r)}))},t.prototype._checkFinalizedStatuses=function(r){var o=this,n=o.hasError,i=o.thrownError,a=o.isStopped;n?r.error(i):a&&r.complete()},t.prototype.asObservable=function(){var r=new j;return r.source=this,r},t.create=function(r,o){return new vo(r,o)},t}(j);var vo=function(e){se(t,e);function t(r,o){var n=e.call(this)||this;return n.destination=r,n.source=o,n}return t.prototype.next=function(r){var o,n;(n=(o=this.destination)===null||o===void 0?void 0:o.next)===null||n===void 0||n.call(o,r)},t.prototype.error=function(r){var o,n;(n=(o=this.destination)===null||o===void 0?void 0:o.error)===null||n===void 0||n.call(o,r)},t.prototype.complete=function(){var r,o;(o=(r=this.destination)===null||r===void 0?void 0:r.complete)===null||o===void 0||o.call(r)},t.prototype._subscribe=function(r){var o,n;return(n=(o=this.source)===null||o===void 0?void 0:o.subscribe(r))!==null&&n!==void 0?n:Er},t}(g);var St={now:function(){return(St.delegate||Date).now()},delegate:void 0};var Ot=function(e){se(t,e);function t(r,o,n){r===void 0&&(r=1/0),o===void 0&&(o=1/0),n===void 0&&(n=St);var i=e.call(this)||this;return i._bufferSize=r,i._windowTime=o,i._timestampProvider=n,i._buffer=[],i._infiniteTimeWindow=!0,i._infiniteTimeWindow=o===1/0,i._bufferSize=Math.max(1,r),i._windowTime=Math.max(1,o),i}return t.prototype.next=function(r){var o=this,n=o.isStopped,i=o._buffer,a=o._infiniteTimeWindow,s=o._timestampProvider,c=o._windowTime;n||(i.push(r),!a&&i.push(s.now()+c)),this._trimBuffer(),e.prototype.next.call(this,r)},t.prototype._subscribe=function(r){this._throwIfClosed(),this._trimBuffer();for(var o=this._innerSubscribe(r),n=this,i=n._infiniteTimeWindow,a=n._buffer,s=a.slice(),c=0;c<s.length&&!r.closed;c+=i?1:2)r.next(s[c]);return this._checkFinalizedStatuses(r),o},t.prototype._trimBuffer=function(){var r=this,o=r._bufferSize,n=r._timestampProvider,i=r._buffer,a=r._infiniteTimeWindow,s=(a?1:2)*o;if(o<1/0&&s<i.length&&i.splice(0,i.length-s),!a){for(var c=n.now(),p=0,l=1;l<i.length&&i[l]<=c;l+=2)p=l;p&&i.splice(0,p+1)}},t}(g);var go=function(e){se(t,e);function t(r,o){return e.call(this)||this}return t.prototype.schedule=function(r,o){return o===void 0&&(o=0),this},t}(Ie);var Mt={setInterval:function(e,t){for(var r=[],o=2;o<arguments.length;o++)r[o-2]=arguments[o];var n=Mt.delegate;return n!=null&&n.setInterval?n.setInterval.apply(n,z([e,t],V(r))):setInterval.apply(void 0,z([e,t],V(r)))},clearInterval:function(e){var t=Mt.delegate;return((t==null?void 0:t.clearInterval)||clearInterval)(e)},delegate:void 0};var zt=function(e){se(t,e);function t(r,o){var n=e.call(this,r,o)||this;return n.scheduler=r,n.work=o,n.pending=!1,n}return t.prototype.schedule=function(r,o){var n;if(o===void 0&&(o=0),this.closed)return this;this.state=r;var i=this.id,a=this.scheduler;return i!=null&&(this.id=this.recycleAsyncId(a,i,o)),this.pending=!0,this.delay=o,this.id=(n=this.id)!==null&&n!==void 0?n:this.requestAsyncId(a,this.id,o),this},t.prototype.requestAsyncId=function(r,o,n){return n===void 0&&(n=0),Mt.setInterval(r.flush.bind(r,this),n)},t.prototype.recycleAsyncId=function(r,o,n){if(n===void 0&&(n=0),n!=null&&this.delay===n&&this.pending===!1)return o;o!=null&&Mt.clearInterval(o)},t.prototype.execute=function(r,o){if(this.closed)return new Error("executing a cancelled action");this.pending=!1;var n=this._execute(r,o);if(n)return n;this.pending===!1&&this.id!=null&&(this.id=this.recycleAsyncId(this.scheduler,this.id,null))},t.prototype._execute=function(r,o){var n=!1,i;try{this.work(r)}catch(a){n=!0,i=a||new Error("Scheduled action threw falsy error")}if(n)return this.unsubscribe(),i},t.prototype.unsubscribe=function(){if(!this.closed){var r=this,o=r.id,n=r.scheduler,i=n.actions;this.work=this.state=this.scheduler=null,this.pending=!1,Ve(i,this),o!=null&&(this.id=this.recycleAsyncId(n,o,null)),this.delay=null,e.prototype.unsubscribe.call(this)}},t}(go);var Mr=function(){function e(t,r){r===void 0&&(r=e.now),this.schedulerActionCtor=t,this.now=r}return e.prototype.schedule=function(t,r,o){return r===void 0&&(r=0),new this.schedulerActionCtor(this,t).schedule(o,r)},e.now=St.now,e}();var qt=function(e){se(t,e);function t(r,o){o===void 0&&(o=Mr.now);var n=e.call(this,r,o)||this;return n.actions=[],n._active=!1,n}return t.prototype.flush=function(r){var o=this.actions;if(this._active){o.push(r);return}var n;this._active=!0;do if(n=r.execute(r.state,r.delay))break;while(r=o.shift());if(this._active=!1,n){for(;r=o.shift();)r.unsubscribe();throw n}},t}(Mr);var ae=new qt(zt),Lr=ae;var xo=function(e){se(t,e);function t(r,o){var n=e.call(this,r,o)||this;return n.scheduler=r,n.work=o,n}return t.prototype.requestAsyncId=function(r,o,n){return n===void 0&&(n=0),n!==null&&n>0?e.prototype.requestAsyncId.call(this,r,o,n):(r.actions.push(this),r._scheduled||(r._scheduled=ut.requestAnimationFrame(function(){return r.flush(void 0)})))},t.prototype.recycleAsyncId=function(r,o,n){var i;if(n===void 0&&(n=0),n!=null?n>0:this.delay>0)return e.prototype.recycleAsyncId.call(this,r,o,n);var a=r.actions;o!=null&&((i=a[a.length-1])===null||i===void 0?void 0:i.id)!==o&&(ut.cancelAnimationFrame(o),r._scheduled=void 0)},t}(zt);var yo=function(e){se(t,e);function t(){return e!==null&&e.apply(this,arguments)||this}return t.prototype.flush=function(r){this._active=!0;var o=this._scheduled;this._scheduled=void 0;var n=this.actions,i;r=r||n.shift();do if(i=r.execute(r.state,r.delay))break;while((r=n[0])&&r.id===o&&n.shift());if(this._active=!1,i){for(;(r=n[0])&&r.id===o&&n.shift();)r.unsubscribe();throw i}},t}(qt);var de=new yo(xo);var L=new j(function(e){return e.complete()});function Kt(e){return e&&k(e.schedule)}function _r(e){return e[e.length-1]}function Je(e){return k(_r(e))?e.pop():void 0}function Ae(e){return Kt(_r(e))?e.pop():void 0}function Qt(e,t){return typeof _r(e)=="number"?e.pop():t}var dt=function(e){return e&&typeof e.length=="number"&&typeof e!="function"};function Yt(e){return k(e==null?void 0:e.then)}function Bt(e){return k(e[ft])}function Gt(e){return Symbol.asyncIterator&&k(e==null?void 0:e[Symbol.asyncIterator])}function Jt(e){return new TypeError("You provided "+(e!==null&&typeof e=="object"?"an invalid object":"'"+e+"'")+" where a stream was expected. You can provide an Observable, Promise, ReadableStream, Array, AsyncIterable, or Iterable.")}function Di(){return typeof Symbol!="function"||!Symbol.iterator?"@@iterator":Symbol.iterator}var Xt=Di();function Zt(e){return k(e==null?void 0:e[Xt])}function er(e){return ao(this,arguments,function(){var r,o,n,i;return Ut(this,function(a){switch(a.label){case 0:r=e.getReader(),a.label=1;case 1:a.trys.push([1,,9,10]),a.label=2;case 2:return[4,ot(r.read())];case 3:return o=a.sent(),n=o.value,i=o.done,i?[4,ot(void 0)]:[3,5];case 4:return[2,a.sent()];case 5:return[4,ot(n)];case 6:return[4,a.sent()];case 7:return a.sent(),[3,2];case 8:return[3,10];case 9:return r.releaseLock(),[7];case 10:return[2]}})})}function tr(e){return k(e==null?void 0:e.getReader)}function N(e){if(e instanceof j)return e;if(e!=null){if(Bt(e))return Ni(e);if(dt(e))return Vi(e);if(Yt(e))return zi(e);if(Gt(e))return Eo(e);if(Zt(e))return qi(e);if(tr(e))return Ki(e)}throw Jt(e)}function Ni(e){return new j(function(t){var r=e[ft]();if(k(r.subscribe))return r.subscribe(t);throw new TypeError("Provided object does not correctly implement Symbol.observable")})}function Vi(e){return new j(function(t){for(var r=0;r<e.length&&!t.closed;r++)t.next(e[r]);t.complete()})}function zi(e){return new j(function(t){e.then(function(r){t.closed||(t.next(r),t.complete())},function(r){return t.error(r)}).then(null,Nt)})}function qi(e){return new j(function(t){var r,o;try{for(var n=ue(e),i=n.next();!i.done;i=n.next()){var a=i.value;if(t.next(a),t.closed)return}}catch(s){r={error:s}}finally{try{i&&!i.done&&(o=n.return)&&o.call(n)}finally{if(r)throw r.error}}t.complete()})}function Eo(e){return new j(function(t){Qi(e,t).catch(function(r){return t.error(r)})})}function Ki(e){return Eo(er(e))}function Qi(e,t){var r,o,n,i;return io(this,void 0,void 0,function(){var a,s;return Ut(this,function(c){switch(c.label){case 0:c.trys.push([0,5,6,11]),r=so(e),c.label=1;case 1:return[4,r.next()];case 2:if(o=c.sent(),!!o.done)return[3,4];if(a=o.value,t.next(a),t.closed)return[2];c.label=3;case 3:return[3,1];case 4:return[3,11];case 5:return s=c.sent(),n={error:s},[3,11];case 6:return c.trys.push([6,,9,10]),o&&!o.done&&(i=r.return)?[4,i.call(r)]:[3,8];case 7:c.sent(),c.label=8;case 8:return[3,10];case 9:if(n)throw n.error;return[7];case 10:return[7];case 11:return t.complete(),[2]}})})}function xe(e,t,r,o,n){o===void 0&&(o=0),n===void 0&&(n=!1);var i=t.schedule(function(){r(),n?e.add(this.schedule(null,o)):this.unsubscribe()},o);if(e.add(i),!n)return i}function Oe(e,t){return t===void 0&&(t=0),x(function(r,o){r.subscribe(S(o,function(n){return xe(o,e,function(){return o.next(n)},t)},function(){return xe(o,e,function(){return o.complete()},t)},function(n){return xe(o,e,function(){return o.error(n)},t)}))})}function ze(e,t){return t===void 0&&(t=0),x(function(r,o){o.add(e.schedule(function(){return r.subscribe(o)},t))})}function wo(e,t){return N(e).pipe(ze(t),Oe(t))}function To(e,t){return N(e).pipe(ze(t),Oe(t))}function So(e,t){return new j(function(r){var o=0;return t.schedule(function(){o===e.length?r.complete():(r.next(e[o++]),r.closed||this.schedule())})})}function Oo(e,t){return new j(function(r){var o;return xe(r,t,function(){o=e[Xt](),xe(r,t,function(){var n,i,a;try{n=o.next(),i=n.value,a=n.done}catch(s){r.error(s);return}a?r.complete():r.next(i)},0,!0)}),function(){return k(o==null?void 0:o.return)&&o.return()}})}function rr(e,t){if(!e)throw new Error("Iterable cannot be null");return new j(function(r){xe(r,t,function(){var o=e[Symbol.asyncIterator]();xe(r,t,function(){o.next().then(function(n){n.done?r.complete():r.next(n.value)})},0,!0)})})}function Mo(e,t){return rr(er(e),t)}function Lo(e,t){if(e!=null){if(Bt(e))return wo(e,t);if(dt(e))return So(e,t);if(Yt(e))return To(e,t);if(Gt(e))return rr(e,t);if(Zt(e))return Oo(e,t);if(tr(e))return Mo(e,t)}throw Jt(e)}function fe(e,t){return t?Lo(e,t):N(e)}function $(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Ae(e);return fe(e,r)}function Ar(e,t){var r=k(e)?e:function(){return e},o=function(n){return n.error(r())};return new j(t?function(n){return t.schedule(o,0,n)}:o)}var or=pt(function(e){return function(){e(this),this.name="EmptyError",this.message="no elements in sequence"}});function _o(e){return e instanceof Date&&!isNaN(e)}function m(e,t){return x(function(r,o){var n=0;r.subscribe(S(o,function(i){o.next(e.call(t,i,n++))}))})}var Yi=Array.isArray;function Bi(e,t){return Yi(t)?e.apply(void 0,z([],V(t))):e(t)}function Xe(e){return m(function(t){return Bi(e,t)})}var Gi=Array.isArray,Ji=Object.getPrototypeOf,Xi=Object.prototype,Zi=Object.keys;function Ao(e){if(e.length===1){var t=e[0];if(Gi(t))return{args:t,keys:null};if(ea(t)){var r=Zi(t);return{args:r.map(function(o){return t[o]}),keys:r}}}return{args:e,keys:null}}function ea(e){return e&&typeof e=="object"&&Ji(e)===Xi}function Co(e,t){return e.reduce(function(r,o,n){return r[o]=t[n],r},{})}function Q(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Ae(e),o=Je(e),n=Ao(e),i=n.args,a=n.keys;if(i.length===0)return fe([],r);var s=new j(Cr(i,r,a?function(c){return Co(a,c)}:ce));return o?s.pipe(Xe(o)):s}function Cr(e,t,r){return r===void 0&&(r=ce),function(o){ko(t,function(){for(var n=e.length,i=new Array(n),a=n,s=n,c=function(l){ko(t,function(){var f=fe(e[l],t),u=!1;f.subscribe(S(o,function(h){i[l]=h,u||(u=!0,s--),s||o.next(r(i.slice()))},function(){--a||o.complete()}))},o)},p=0;p<n;p++)c(p)},o)}}function ko(e,t,r){e?xe(r,e,t):t()}function Ho(e,t,r,o,n,i,a,s){var c=[],p=0,l=0,f=!1,u=function(){f&&!c.length&&!p&&t.complete()},h=function(A){return p<o?w(A):c.push(A)},w=function(A){i&&t.next(A),p++;var Z=!1;N(r(A,l++)).subscribe(S(t,function(te){n==null||n(te),i?h(te):t.next(te)},function(){Z=!0},void 0,function(){if(Z)try{p--;for(var te=function(){var J=c.shift();a?xe(t,a,function(){return w(J)}):w(J)};c.length&&p<o;)te();u()}catch(J){t.error(J)}}))};return e.subscribe(S(t,h,function(){f=!0,u()})),function(){s==null||s()}}function re(e,t,r){return r===void 0&&(r=1/0),k(t)?re(function(o,n){return m(function(i,a){return t(o,i,n,a)})(N(e(o,n)))},r):(typeof t=="number"&&(r=t),x(function(o,n){return Ho(o,n,e,r)}))}function ht(e){return e===void 0&&(e=1/0),re(ce,e)}function $o(){return ht(1)}function Fe(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];return $o()(fe(e,Ae(e)))}function H(e){return new j(function(t){N(e()).subscribe(t)})}var ta=["addListener","removeListener"],ra=["addEventListener","removeEventListener"],oa=["on","off"];function d(e,t,r,o){if(k(r)&&(o=r,r=void 0),o)return d(e,t,r).pipe(Xe(o));var n=V(aa(e)?ra.map(function(s){return function(c){return e[s](t,c,r)}}):na(e)?ta.map(Ro(e,t)):ia(e)?oa.map(Ro(e,t)):[],2),i=n[0],a=n[1];if(!i&&dt(e))return re(function(s){return d(s,t,r)})(N(e));if(!i)throw new TypeError("Invalid event target");return new j(function(s){var c=function(){for(var p=[],l=0;l<arguments.length;l++)p[l]=arguments[l];return s.next(1<p.length?p:p[0])};return i(c),function(){return a(c)}})}function Ro(e,t){return function(r){return function(o){return e[r](t,o)}}}function na(e){return k(e.addListener)&&k(e.removeListener)}function ia(e){return k(e.on)&&k(e.off)}function aa(e){return k(e.addEventListener)&&k(e.removeEventListener)}function nr(e,t,r){return r?nr(e,t).pipe(Xe(r)):new j(function(o){var n=function(){for(var a=[],s=0;s<arguments.length;s++)a[s]=arguments[s];return o.next(a.length===1?a[0]:a)},i=e(n);return k(t)?function(){return t(n,i)}:void 0})}function Ze(e,t,r){e===void 0&&(e=0),r===void 0&&(r=Lr);var o=-1;return t!=null&&(Kt(t)?r=t:o=t),new j(function(n){var i=_o(e)?+e-r.now():e;i<0&&(i=0);var a=0;return r.schedule(function(){n.closed||(n.next(a++),0<=o?this.schedule(void 0,o):n.complete())},i)})}function T(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Ae(e),o=Qt(e,1/0),n=e;return n.length?n.length===1?N(n[0]):ht(o)(fe(n,r)):L}var qe=new j(Se);var sa=Array.isArray;function bt(e){return e.length===1&&sa(e[0])?e[0]:e}function b(e,t){return x(function(r,o){var n=0;r.subscribe(S(o,function(i){return e.call(t,i,n++)&&o.next(i)}))})}function Lt(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Je(e),o=bt(e);return o.length?new j(function(n){var i=o.map(function(){return[]}),a=o.map(function(){return!1});n.add(function(){i=a=null});for(var s=function(p){N(o[p]).subscribe(S(n,function(l){if(i[p].push(l),i.every(function(u){return u.length})){var f=i.map(function(u){return u.shift()});n.next(r?r.apply(void 0,z([],V(f))):f),i.some(function(u,h){return!u.length&&a[h]})&&n.complete()}},function(){a[p]=!0,!i[p].length&&n.complete()}))},c=0;!n.closed&&c<o.length;c++)s(c);return function(){i=a=null}}):L}function Po(e){return x(function(t,r){var o=!1,n=null,i=null,a=!1,s=function(){if(i==null||i.unsubscribe(),i=null,o){o=!1;var p=n;n=null,r.next(p)}a&&r.complete()},c=function(){i=null,a&&r.complete()};t.subscribe(S(r,function(p){o=!0,n=p,i||N(e(p)).subscribe(i=S(r,s,c))},function(){a=!0,(!o||!i||i.closed)&&r.complete()}))})}function Me(e,t){return t===void 0&&(t=ae),Po(function(){return Ze(e,t)})}function Ke(e,t){return t===void 0&&(t=null),t=t!=null?t:e,x(function(r,o){var n=[],i=0;r.subscribe(S(o,function(a){var s,c,p,l,f=null;i++%t===0&&n.push([]);try{for(var u=ue(n),h=u.next();!h.done;h=u.next()){var w=h.value;w.push(a),e<=w.length&&(f=f!=null?f:[],f.push(w))}}catch(te){s={error:te}}finally{try{h&&!h.done&&(c=u.return)&&c.call(u)}finally{if(s)throw s.error}}if(f)try{for(var A=ue(f),Z=A.next();!Z.done;Z=A.next()){var w=Z.value;Ve(n,w),o.next(w)}}catch(te){p={error:te}}finally{try{Z&&!Z.done&&(l=A.return)&&l.call(A)}finally{if(p)throw p.error}}},function(){var a,s;try{for(var c=ue(n),p=c.next();!p.done;p=c.next()){var l=p.value;o.next(l)}}catch(f){a={error:f}}finally{try{p&&!p.done&&(s=c.return)&&s.call(c)}finally{if(a)throw a.error}}o.complete()},void 0,function(){n=null}))})}function he(e){return x(function(t,r){var o=null,n=!1,i;o=t.subscribe(S(r,void 0,void 0,function(a){i=N(e(a,he(e)(t))),o?(o.unsubscribe(),o=null,i.subscribe(r)):n=!0})),n&&(o.unsubscribe(),o=null,i.subscribe(r))})}function Io(e,t,r,o,n){return function(i,a){var s=r,c=t,p=0;i.subscribe(S(a,function(l){var f=p++;c=s?e(c,l,f):(s=!0,l),o&&a.next(c)},n&&function(){s&&a.next(c),a.complete()}))}}function kr(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Je(e);return r?uo(kr.apply(void 0,z([],V(e))),Xe(r)):x(function(o,n){Cr(z([o],V(bt(e))))(n)})}function je(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];return kr.apply(void 0,z([],V(e)))}function be(e,t){return t===void 0&&(t=ae),x(function(r,o){var n=null,i=null,a=null,s=function(){if(n){n.unsubscribe(),n=null;var p=i;i=null,o.next(p)}};function c(){var p=a+e,l=t.now();if(l<p){n=this.schedule(void 0,p-l),o.add(n);return}s()}r.subscribe(S(o,function(p){i=p,a=t.now(),n||(n=t.schedule(c,e),o.add(n))},function(){s(),o.complete()},void 0,function(){i=n=null}))})}function Qe(e){return x(function(t,r){var o=!1;t.subscribe(S(r,function(n){o=!0,r.next(n)},function(){o||r.next(e),r.complete()}))})}function ye(e){return e<=0?function(){return L}:x(function(t,r){var o=0;t.subscribe(S(r,function(n){++o<=e&&(r.next(n),e<=o&&r.complete())}))})}function ee(){return x(function(e,t){e.subscribe(S(t,Se))})}function Fo(e){return m(function(){return e})}function Hr(e,t){return t?function(r){return Fe(t.pipe(ye(1),ee()),r.pipe(Hr(e)))}:re(function(r,o){return N(e(r,o)).pipe(ye(1),Fo(r))})}function Ye(e,t){t===void 0&&(t=ae);var r=Ze(e,t);return Hr(function(){return r})}function Y(e,t){return t===void 0&&(t=ce),e=e!=null?e:ca,x(function(r,o){var n,i=!0;r.subscribe(S(o,function(a){var s=t(a);(i||!e(n,s))&&(i=!1,n=s,o.next(a))}))})}function ca(e,t){return e===t}function X(e,t){return Y(function(r,o){return t?t(r[e],o[e]):r[e]===o[e]})}function jo(e){return e===void 0&&(e=pa),x(function(t,r){var o=!1;t.subscribe(S(r,function(n){o=!0,r.next(n)},function(){return o?r.complete():r.error(e())}))})}function pa(){return new or}function oe(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];return function(r){return Fe(r,$.apply(void 0,z([],V(e))))}}function _(e){return x(function(t,r){try{t.subscribe(r)}finally{r.add(e)}})}function He(e,t){var r=arguments.length>=2;return function(o){return o.pipe(e?b(function(n,i){return e(n,i,o)}):ce,ye(1),r?Qe(t):jo(function(){return new or}))}}function $r(e){return e<=0?function(){return L}:x(function(t,r){var o=[];t.subscribe(S(r,function(n){o.push(n),e<o.length&&o.shift()},function(){var n,i;try{for(var a=ue(o),s=a.next();!s.done;s=a.next()){var c=s.value;r.next(c)}}catch(p){n={error:p}}finally{try{s&&!s.done&&(i=a.return)&&i.call(a)}finally{if(n)throw n.error}}r.complete()},void 0,function(){o=null}))})}function Uo(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Ae(e),o=Qt(e,1/0);return e=bt(e),x(function(n,i){ht(o)(fe(z([n],V(e)),r)).subscribe(i)})}function $e(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];return Uo.apply(void 0,z([],V(e)))}function at(e){var t,r=1/0,o;return e!=null&&(typeof e=="object"?(t=e.count,r=t===void 0?1/0:t,o=e.delay):r=e),r<=0?function(){return L}:x(function(n,i){var a=0,s,c=function(){if(s==null||s.unsubscribe(),s=null,o!=null){var l=typeof o=="number"?Ze(o):N(o(a)),f=S(i,function(){f.unsubscribe(),p()});l.subscribe(f)}else p()},p=function(){var l=!1;s=n.subscribe(S(i,void 0,function(){++a<r?s?c():l=!0:i.complete()})),l&&c()};p()})}function Rr(e,t){return x(Io(e,t,arguments.length>=2,!0))}function le(e){e===void 0&&(e={});var t=e.connector,r=t===void 0?function(){return new g}:t,o=e.resetOnError,n=o===void 0?!0:o,i=e.resetOnComplete,a=i===void 0?!0:i,s=e.resetOnRefCountZero,c=s===void 0?!0:s;return function(p){var l,f,u,h=0,w=!1,A=!1,Z=function(){f==null||f.unsubscribe(),f=void 0},te=function(){Z(),l=u=void 0,w=A=!1},J=function(){var C=l;te(),C==null||C.unsubscribe()};return x(function(C,ct){h++,!A&&!w&&Z();var Ne=u=u!=null?u:r();ct.add(function(){h--,h===0&&!A&&!w&&(f=Pr(J,c))}),Ne.subscribe(ct),!l&&h>0&&(l=new it({next:function(Pe){return Ne.next(Pe)},error:function(Pe){A=!0,Z(),f=Pr(te,n,Pe),Ne.error(Pe)},complete:function(){w=!0,Z(),f=Pr(te,a),Ne.complete()}}),N(C).subscribe(l))})(p)}}function Pr(e,t){for(var r=[],o=2;o<arguments.length;o++)r[o-2]=arguments[o];if(t===!0){e();return}if(t!==!1){var n=new it({next:function(){n.unsubscribe(),e()}});return N(t.apply(void 0,z([],V(r)))).subscribe(n)}}function B(e,t,r){var o,n,i,a,s=!1;return e&&typeof e=="object"?(o=e.bufferSize,a=o===void 0?1/0:o,n=e.windowTime,t=n===void 0?1/0:n,i=e.refCount,s=i===void 0?!1:i,r=e.scheduler):a=e!=null?e:1/0,le({connector:function(){return new Ot(a,t,r)},resetOnError:!0,resetOnComplete:!1,resetOnRefCountZero:s})}function Le(e){return b(function(t,r){return e<=r})}function Ir(e){return x(function(t,r){var o=!1,n=S(r,function(){n==null||n.unsubscribe(),o=!0},Se);N(e).subscribe(n),t.subscribe(S(r,function(i){return o&&r.next(i)}))})}function q(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Ae(e);return x(function(o,n){(r?Fe(e,o,r):Fe(e,o)).subscribe(n)})}function v(e,t){return x(function(r,o){var n=null,i=0,a=!1,s=function(){return a&&!n&&o.complete()};r.subscribe(S(o,function(c){n==null||n.unsubscribe();var p=0,l=i++;N(e(c,l)).subscribe(n=S(o,function(f){return o.next(t?t(c,f,l,p++):f)},function(){n=null,s()}))},function(){a=!0,s()}))})}function U(e){return x(function(t,r){N(e).subscribe(S(r,function(){return r.complete()},Se)),!r.closed&&t.subscribe(r)})}function Fr(e,t){return t===void 0&&(t=!1),x(function(r,o){var n=0;r.subscribe(S(o,function(i){var a=e(i,n++);(a||t)&&o.next(i),!a&&o.complete()}))})}function y(e,t,r){var o=k(e)||t||r?{next:e,error:t,complete:r}:e;return o?x(function(n,i){var a;(a=o.subscribe)===null||a===void 0||a.call(o);var s=!0;n.subscribe(S(i,function(c){var p;(p=o.next)===null||p===void 0||p.call(o,c),i.next(c)},function(){var c;s=!1,(c=o.complete)===null||c===void 0||c.call(o),i.complete()},function(c){var p;s=!1,(p=o.error)===null||p===void 0||p.call(o,c),i.error(c)},function(){var c,p;s&&((c=o.unsubscribe)===null||c===void 0||c.call(o)),(p=o.finalize)===null||p===void 0||p.call(o)}))}):ce}function Wo(e,t){return x(function(r,o){var n=t!=null?t:{},i=n.leading,a=i===void 0?!0:i,s=n.trailing,c=s===void 0?!1:s,p=!1,l=null,f=null,u=!1,h=function(){f==null||f.unsubscribe(),f=null,c&&(Z(),u&&o.complete())},w=function(){f=null,u&&o.complete()},A=function(te){return f=N(e(te)).subscribe(S(o,h,w))},Z=function(){if(p){p=!1;var te=l;l=null,o.next(te),!u&&A(te)}};r.subscribe(S(o,function(te){p=!0,l=te,!(f&&!f.closed)&&(a?Z():A(te))},function(){u=!0,!(c&&p&&f&&!f.closed)&&o.complete()}))})}function _t(e,t,r){t===void 0&&(t=ae);var o=Ze(e,t);return Wo(function(){return o},r)}function ne(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Je(e);return x(function(o,n){for(var i=e.length,a=new Array(i),s=e.map(function(){return!1}),c=!1,p=function(f){N(e[f]).subscribe(S(n,function(u){a[f]=u,!c&&!s[f]&&(s[f]=!0,(c=s.every(ce))&&(s=null))},Se))},l=0;l<i;l++)p(l);o.subscribe(S(n,function(f){if(c){var u=z([f],V(a));n.next(r?r.apply(void 0,z([],V(u))):u)}}))})}function Do(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];return x(function(r,o){Lt.apply(void 0,z([r],V(e))).subscribe(o)})}function jr(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];return Do.apply(void 0,z([],V(e)))}function No(){let e=new Ot(1);return d(document,"DOMContentLoaded",{once:!0}).subscribe(()=>e.next(document)),e}function R(e,t=document){return Array.from(t.querySelectorAll(e))}function P(e,t=document){let r=me(e,t);if(typeof r=="undefined")throw new ReferenceError(`Missing element: expected "${e}" to be present`);return r}function me(e,t=document){return t.querySelector(e)||void 0}function Re(){var e,t,r,o;return(o=(r=(t=(e=document.activeElement)==null?void 0:e.shadowRoot)==null?void 0:t.activeElement)!=null?r:document.activeElement)!=null?o:void 0}var la=T(d(document.body,"focusin"),d(document.body,"focusout")).pipe(be(1),q(void 0),m(()=>Re()||document.body),B(1));function vt(e){return la.pipe(m(t=>e.contains(t)),Y())}function Vo(e,t){return T(d(e,"mouseenter").pipe(m(()=>!0)),d(e,"mouseleave").pipe(m(()=>!1))).pipe(t?be(t):ce,q(!1))}function Ue(e){return{x:e.offsetLeft,y:e.offsetTop}}function zo(e){return T(d(window,"load"),d(window,"resize")).pipe(Me(0,de),m(()=>Ue(e)),q(Ue(e)))}function ir(e){return{x:e.scrollLeft,y:e.scrollTop}}function et(e){return T(d(e,"scroll"),d(window,"resize")).pipe(Me(0,de),m(()=>ir(e)),q(ir(e)))}function qo(e,t){if(typeof t=="string"||typeof t=="number")e.innerHTML+=t.toString();else if(t instanceof Node)e.appendChild(t);else if(Array.isArray(t))for(let r of t)qo(e,r)}function E(e,t,...r){let o=document.createElement(e);if(t)for(let n of Object.keys(t))typeof t[n]!="undefined"&&(typeof t[n]!="boolean"?o.setAttribute(n,t[n]):o.setAttribute(n,""));for(let n of r)qo(o,n);return o}function ar(e){if(e>999){let t=+((e-950)%1e3>99);return`${((e+1e-6)/1e3).toFixed(t)}k`}else return e.toString()}function gt(e){let t=E("script",{src:e});return H(()=>(document.head.appendChild(t),T(d(t,"load"),d(t,"error").pipe(v(()=>Ar(()=>new ReferenceError(`Invalid script: ${e}`))))).pipe(m(()=>{}),_(()=>document.head.removeChild(t)),ye(1))))}var Ko=new g,ma=H(()=>typeof ResizeObserver=="undefined"?gt("https://unpkg.com/resize-observer-polyfill"):$(void 0)).pipe(m(()=>new ResizeObserver(e=>{for(let t of e)Ko.next(t)})),v(e=>T(qe,$(e)).pipe(_(()=>e.disconnect()))),B(1));function pe(e){return{width:e.offsetWidth,height:e.offsetHeight}}function Ee(e){return ma.pipe(y(t=>t.observe(e)),v(t=>Ko.pipe(b(({target:r})=>r===e),_(()=>t.unobserve(e)),m(()=>pe(e)))),q(pe(e)))}function xt(e){return{width:e.scrollWidth,height:e.scrollHeight}}function sr(e){let t=e.parentElement;for(;t&&(e.scrollWidth<=t.scrollWidth&&e.scrollHeight<=t.scrollHeight);)t=(e=t).parentElement;return t?e:void 0}var Qo=new g,fa=H(()=>$(new IntersectionObserver(e=>{for(let t of e)Qo.next(t)},{threshold:0}))).pipe(v(e=>T(qe,$(e)).pipe(_(()=>e.disconnect()))),B(1));function yt(e){return fa.pipe(y(t=>t.observe(e)),v(t=>Qo.pipe(b(({target:r})=>r===e),_(()=>t.unobserve(e)),m(({isIntersecting:r})=>r))))}function Yo(e,t=16){return et(e).pipe(m(({y:r})=>{let o=pe(e),n=xt(e);return r>=n.height-o.height-t}),Y())}var cr={drawer:P("[data-md-toggle=drawer]"),search:P("[data-md-toggle=search]")};function Bo(e){return cr[e].checked}function Be(e,t){cr[e].checked!==t&&cr[e].click()}function We(e){let t=cr[e];return d(t,"change").pipe(m(()=>t.checked),q(t.checked))}function ua(e,t){switch(e.constructor){case HTMLInputElement:return e.type==="radio"?/^Arrow/.test(t):!0;case HTMLSelectElement:case HTMLTextAreaElement:return!0;default:return e.isContentEditable}}function da(){return T(d(window,"compositionstart").pipe(m(()=>!0)),d(window,"compositionend").pipe(m(()=>!1))).pipe(q(!1))}function Go(){let e=d(window,"keydown").pipe(b(t=>!(t.metaKey||t.ctrlKey)),m(t=>({mode:Bo("search")?"search":"global",type:t.key,claim(){t.preventDefault(),t.stopPropagation()}})),b(({mode:t,type:r})=>{if(t==="global"){let o=Re();if(typeof o!="undefined")return!ua(o,r)}return!0}),le());return da().pipe(v(t=>t?L:e))}function ve(){return new URL(location.href)}function st(e,t=!1){if(G("navigation.instant")&&!t){let r=E("a",{href:e.href});document.body.appendChild(r),r.click(),r.remove()}else location.href=e.href}function Jo(){return new g}function Xo(){return location.hash.slice(1)}function Zo(e){let t=E("a",{href:e});t.addEventListener("click",r=>r.stopPropagation()),t.click()}function ha(e){return T(d(window,"hashchange"),e).pipe(m(Xo),q(Xo()),b(t=>t.length>0),B(1))}function en(e){return ha(e).pipe(m(t=>me(`[id="${t}"]`)),b(t=>typeof t!="undefined"))}function At(e){let t=matchMedia(e);return nr(r=>t.addListener(()=>r(t.matches))).pipe(q(t.matches))}function tn(){let e=matchMedia("print");return T(d(window,"beforeprint").pipe(m(()=>!0)),d(window,"afterprint").pipe(m(()=>!1))).pipe(q(e.matches))}function Ur(e,t){return e.pipe(v(r=>r?t():L))}function Wr(e,t){return new j(r=>{let o=new XMLHttpRequest;return o.open("GET",`${e}`),o.responseType="blob",o.addEventListener("load",()=>{o.status>=200&&o.status<300?(r.next(o.response),r.complete()):r.error(new Error(o.statusText))}),o.addEventListener("error",()=>{r.error(new Error("Network error"))}),o.addEventListener("abort",()=>{r.complete()}),typeof(t==null?void 0:t.progress$)!="undefined"&&(o.addEventListener("progress",n=>{var i;if(n.lengthComputable)t.progress$.next(n.loaded/n.total*100);else{let a=(i=o.getResponseHeader("Content-Length"))!=null?i:0;t.progress$.next(n.loaded/+a*100)}}),t.progress$.next(5)),o.send(),()=>o.abort()})}function De(e,t){return Wr(e,t).pipe(v(r=>r.text()),m(r=>JSON.parse(r)),B(1))}function rn(e,t){let r=new DOMParser;return Wr(e,t).pipe(v(o=>o.text()),m(o=>r.parseFromString(o,"text/html")),B(1))}function on(e,t){let r=new DOMParser;return Wr(e,t).pipe(v(o=>o.text()),m(o=>r.parseFromString(o,"text/xml")),B(1))}function nn(){return{x:Math.max(0,scrollX),y:Math.max(0,scrollY)}}function an(){return T(d(window,"scroll",{passive:!0}),d(window,"resize",{passive:!0})).pipe(m(nn),q(nn()))}function sn(){return{width:innerWidth,height:innerHeight}}function cn(){return d(window,"resize",{passive:!0}).pipe(m(sn),q(sn()))}function pn(){return Q([an(),cn()]).pipe(m(([e,t])=>({offset:e,size:t})),B(1))}function pr(e,{viewport$:t,header$:r}){let o=t.pipe(X("size")),n=Q([o,r]).pipe(m(()=>Ue(e)));return Q([r,t,n]).pipe(m(([{height:i},{offset:a,size:s},{x:c,y:p}])=>({offset:{x:a.x-c,y:a.y-p+i},size:s})))}function ba(e){return d(e,"message",t=>t.data)}function va(e){let t=new g;return t.subscribe(r=>e.postMessage(r)),t}function ln(e,t=new Worker(e)){let r=ba(t),o=va(t),n=new g;n.subscribe(o);let i=o.pipe(ee(),oe(!0));return n.pipe(ee(),$e(r.pipe(U(i))),le())}var ga=P("#__config"),Et=JSON.parse(ga.textContent);Et.base=`${new URL(Et.base,ve())}`;function we(){return Et}function G(e){return Et.features.includes(e)}function ge(e,t){return typeof t!="undefined"?Et.translations[e].replace("#",t.toString()):Et.translations[e]}function Te(e,t=document){return P(`[data-md-component=${e}]`,t)}function ie(e,t=document){return R(`[data-md-component=${e}]`,t)}function xa(e){let t=P(".md-typeset > :first-child",e);return d(t,"click",{once:!0}).pipe(m(()=>P(".md-typeset",e)),m(r=>({hash:__md_hash(r.innerHTML)})))}function mn(e){if(!G("announce.dismiss")||!e.childElementCount)return L;if(!e.hidden){let t=P(".md-typeset",e);__md_hash(t.innerHTML)===__md_get("__announce")&&(e.hidden=!0)}return H(()=>{let t=new g;return t.subscribe(({hash:r})=>{e.hidden=!0,__md_set("__announce",r)}),xa(e).pipe(y(r=>t.next(r)),_(()=>t.complete()),m(r=>F({ref:e},r)))})}function ya(e,{target$:t}){return t.pipe(m(r=>({hidden:r!==e})))}function fn(e,t){let r=new g;return r.subscribe(({hidden:o})=>{e.hidden=o}),ya(e,t).pipe(y(o=>r.next(o)),_(()=>r.complete()),m(o=>F({ref:e},o)))}function Ct(e,t){return t==="inline"?E("div",{class:"md-tooltip md-tooltip--inline",id:e,role:"tooltip"},E("div",{class:"md-tooltip__inner md-typeset"})):E("div",{class:"md-tooltip",id:e,role:"tooltip"},E("div",{class:"md-tooltip__inner md-typeset"}))}function un(e,t){if(t=t?`${t}_annotation_${e}`:void 0,t){let r=t?`#${t}`:void 0;return E("aside",{class:"md-annotation",tabIndex:0},Ct(t),E("a",{href:r,class:"md-annotation__index",tabIndex:-1},E("span",{"data-md-annotation-id":e})))}else return E("aside",{class:"md-annotation",tabIndex:0},Ct(t),E("span",{class:"md-annotation__index",tabIndex:-1},E("span",{"data-md-annotation-id":e})))}function dn(e){return E("button",{class:"md-clipboard md-icon",title:ge("clipboard.copy"),"data-clipboard-target":`#${e} > code`})}function Dr(e,t){let r=t&2,o=t&1,n=Object.keys(e.terms).filter(c=>!e.terms[c]).reduce((c,p)=>[...c,E("del",null,p)," "],[]).slice(0,-1),i=we(),a=new URL(e.location,i.base);G("search.highlight")&&a.searchParams.set("h",Object.entries(e.terms).filter(([,c])=>c).reduce((c,[p])=>`${c} ${p}`.trim(),""));let{tags:s}=we();return E("a",{href:`${a}`,class:"md-search-result__link",tabIndex:-1},E("article",{class:"md-search-result__article md-typeset","data-md-score":e.score.toFixed(2)},r>0&&E("div",{class:"md-search-result__icon md-icon"}),r>0&&E("h1",null,e.title),r<=0&&E("h2",null,e.title),o>0&&e.text.length>0&&e.text,e.tags&&e.tags.map(c=>{let p=s?c in s?`md-tag-icon md-tag--${s[c]}`:"md-tag-icon":"";return E("span",{class:`md-tag ${p}`},c)}),o>0&&n.length>0&&E("p",{class:"md-search-result__terms"},ge("search.result.term.missing"),": ",...n)))}function hn(e){let t=e[0].score,r=[...e],o=we(),n=r.findIndex(l=>!`${new URL(l.location,o.base)}`.includes("#")),[i]=r.splice(n,1),a=r.findIndex(l=>l.score<t);a===-1&&(a=r.length);let s=r.slice(0,a),c=r.slice(a),p=[Dr(i,2|+(!n&&a===0)),...s.map(l=>Dr(l,1)),...c.length?[E("details",{class:"md-search-result__more"},E("summary",{tabIndex:-1},E("div",null,c.length>0&&c.length===1?ge("search.result.more.one"):ge("search.result.more.other",c.length))),...c.map(l=>Dr(l,1)))]:[]];return E("li",{class:"md-search-result__item"},p)}function bn(e){return E("ul",{class:"md-source__facts"},Object.entries(e).map(([t,r])=>E("li",{class:`md-source__fact md-source__fact--${t}`},typeof r=="number"?ar(r):r)))}function Nr(e){let t=`tabbed-control tabbed-control--${e}`;return E("div",{class:t,hidden:!0},E("button",{class:"tabbed-button",tabIndex:-1,"aria-hidden":"true"}))}function vn(e){return E("div",{class:"md-typeset__scrollwrap"},E("div",{class:"md-typeset__table"},e))}function Ea(e){let t=we(),r=new URL(`../${e.version}/`,t.base);return E("li",{class:"md-version__item"},E("a",{href:`${r}`,class:"md-version__link"},e.title))}function gn(e,t){return e=e.filter(r=>{var o;return!((o=r.properties)!=null&&o.hidden)}),E("div",{class:"md-version"},E("button",{class:"md-version__current","aria-label":ge("select.version")},t.title),E("ul",{class:"md-version__list"},e.map(Ea)))}var wa=0;function Ta(e,t){document.body.append(e);let{width:r}=pe(e);e.style.setProperty("--md-tooltip-width",`${r}px`),e.remove();let o=sr(t),n=typeof o!="undefined"?et(o):$({x:0,y:0}),i=T(vt(t),Vo(t)).pipe(Y());return Q([i,n]).pipe(m(([a,s])=>{let{x:c,y:p}=Ue(t),l=pe(t),f=t.closest("table");return f&&t.parentElement&&(c+=f.offsetLeft+t.parentElement.offsetLeft,p+=f.offsetTop+t.parentElement.offsetTop),{active:a,offset:{x:c-s.x+l.width/2-r/2,y:p-s.y+l.height+8}}}))}function Ge(e){let t=e.title;if(!t.length)return L;let r=`__tooltip_${wa++}`,o=Ct(r,"inline"),n=P(".md-typeset",o);return n.innerHTML=t,H(()=>{let i=new g;return i.subscribe({next({offset:a}){o.style.setProperty("--md-tooltip-x",`${a.x}px`),o.style.setProperty("--md-tooltip-y",`${a.y}px`)},complete(){o.style.removeProperty("--md-tooltip-x"),o.style.removeProperty("--md-tooltip-y")}}),T(i.pipe(b(({active:a})=>a)),i.pipe(be(250),b(({active:a})=>!a))).subscribe({next({active:a}){a?(e.insertAdjacentElement("afterend",o),e.setAttribute("aria-describedby",r),e.removeAttribute("title")):(o.remove(),e.removeAttribute("aria-describedby"),e.setAttribute("title",t))},complete(){o.remove(),e.removeAttribute("aria-describedby"),e.setAttribute("title",t)}}),i.pipe(Me(16,de)).subscribe(({active:a})=>{o.classList.toggle("md-tooltip--active",a)}),i.pipe(_t(125,de),b(()=>!!e.offsetParent),m(()=>e.offsetParent.getBoundingClientRect()),m(({x:a})=>a)).subscribe({next(a){a?o.style.setProperty("--md-tooltip-0",`${-a}px`):o.style.removeProperty("--md-tooltip-0")},complete(){o.style.removeProperty("--md-tooltip-0")}}),Ta(o,e).pipe(y(a=>i.next(a)),_(()=>i.complete()),m(a=>F({ref:e},a)))}).pipe(ze(ae))}function Sa(e,t){let r=H(()=>Q([zo(e),et(t)])).pipe(m(([{x:o,y:n},i])=>{let{width:a,height:s}=pe(e);return{x:o-i.x+a/2,y:n-i.y+s/2}}));return vt(e).pipe(v(o=>r.pipe(m(n=>({active:o,offset:n})),ye(+!o||1/0))))}function xn(e,t,{target$:r}){let[o,n]=Array.from(e.children);return H(()=>{let i=new g,a=i.pipe(ee(),oe(!0));return i.subscribe({next({offset:s}){e.style.setProperty("--md-tooltip-x",`${s.x}px`),e.style.setProperty("--md-tooltip-y",`${s.y}px`)},complete(){e.style.removeProperty("--md-tooltip-x"),e.style.removeProperty("--md-tooltip-y")}}),yt(e).pipe(U(a)).subscribe(s=>{e.toggleAttribute("data-md-visible",s)}),T(i.pipe(b(({active:s})=>s)),i.pipe(be(250),b(({active:s})=>!s))).subscribe({next({active:s}){s?e.prepend(o):o.remove()},complete(){e.prepend(o)}}),i.pipe(Me(16,de)).subscribe(({active:s})=>{o.classList.toggle("md-tooltip--active",s)}),i.pipe(_t(125,de),b(()=>!!e.offsetParent),m(()=>e.offsetParent.getBoundingClientRect()),m(({x:s})=>s)).subscribe({next(s){s?e.style.setProperty("--md-tooltip-0",`${-s}px`):e.style.removeProperty("--md-tooltip-0")},complete(){e.style.removeProperty("--md-tooltip-0")}}),d(n,"click").pipe(U(a),b(s=>!(s.metaKey||s.ctrlKey))).subscribe(s=>{s.stopPropagation(),s.preventDefault()}),d(n,"mousedown").pipe(U(a),ne(i)).subscribe(([s,{active:c}])=>{var p;if(s.button!==0||s.metaKey||s.ctrlKey)s.preventDefault();else if(c){s.preventDefault();let l=e.parentElement.closest(".md-annotation");l instanceof HTMLElement?l.focus():(p=Re())==null||p.blur()}}),r.pipe(U(a),b(s=>s===o),Ye(125)).subscribe(()=>e.focus()),Sa(e,t).pipe(y(s=>i.next(s)),_(()=>i.complete()),m(s=>F({ref:e},s)))})}function Oa(e){return e.tagName==="CODE"?R(".c, .c1, .cm",e):[e]}function Ma(e){let t=[];for(let r of Oa(e)){let o=[],n=document.createNodeIterator(r,NodeFilter.SHOW_TEXT);for(let i=n.nextNode();i;i=n.nextNode())o.push(i);for(let i of o){let a;for(;a=/(\(\d+\))(!)?/.exec(i.textContent);){let[,s,c]=a;if(typeof c=="undefined"){let p=i.splitText(a.index);i=p.splitText(s.length),t.push(p)}else{i.textContent=s,t.push(i);break}}}}return t}function yn(e,t){t.append(...Array.from(e.childNodes))}function lr(e,t,{target$:r,print$:o}){let n=t.closest("[id]"),i=n==null?void 0:n.id,a=new Map;for(let s of Ma(t)){let[,c]=s.textContent.match(/\((\d+)\)/);me(`:scope > li:nth-child(${c})`,e)&&(a.set(c,un(c,i)),s.replaceWith(a.get(c)))}return a.size===0?L:H(()=>{let s=new g,c=s.pipe(ee(),oe(!0)),p=[];for(let[l,f]of a)p.push([P(".md-typeset",f),P(`:scope > li:nth-child(${l})`,e)]);return o.pipe(U(c)).subscribe(l=>{e.hidden=!l,e.classList.toggle("md-annotation-list",l);for(let[f,u]of p)l?yn(f,u):yn(u,f)}),T(...[...a].map(([,l])=>xn(l,t,{target$:r}))).pipe(_(()=>s.complete()),le())})}function En(e){if(e.nextElementSibling){let t=e.nextElementSibling;if(t.tagName==="OL")return t;if(t.tagName==="P"&&!t.children.length)return En(t)}}function wn(e,t){return H(()=>{let r=En(e);return typeof r!="undefined"?lr(r,e,t):L})}var Tn=jt(zr());var La=0;function Sn(e){if(e.nextElementSibling){let t=e.nextElementSibling;if(t.tagName==="OL")return t;if(t.tagName==="P"&&!t.children.length)return Sn(t)}}function _a(e){return Ee(e).pipe(m(({width:t})=>({scrollable:xt(e).width>t})),X("scrollable"))}function On(e,t){let{matches:r}=matchMedia("(hover)"),o=H(()=>{let n=new g,i=n.pipe($r(1));n.subscribe(({scrollable:c})=>{c&&r?e.setAttribute("tabindex","0"):e.removeAttribute("tabindex")});let a=[];if(Tn.default.isSupported()&&(e.closest(".copy")||G("content.code.copy")&&!e.closest(".no-copy"))){let c=e.closest("pre");c.id=`__code_${La++}`;let p=dn(c.id);c.insertBefore(p,e),G("content.tooltips")&&a.push(Ge(p))}let s=e.closest(".highlight");if(s instanceof HTMLElement){let c=Sn(s);if(typeof c!="undefined"&&(s.classList.contains("annotate")||G("content.code.annotate"))){let p=lr(c,e,t);a.push(Ee(s).pipe(U(i),m(({width:l,height:f})=>l&&f),Y(),v(l=>l?p:L)))}}return _a(e).pipe(y(c=>n.next(c)),_(()=>n.complete()),m(c=>F({ref:e},c)),$e(...a))});return G("content.lazy")?yt(e).pipe(b(n=>n),ye(1),v(()=>o)):o}function Aa(e,{target$:t,print$:r}){let o=!0;return T(t.pipe(m(n=>n.closest("details:not([open])")),b(n=>e===n),m(()=>({action:"open",reveal:!0}))),r.pipe(b(n=>n||!o),y(()=>o=e.open),m(n=>({action:n?"open":"close"}))))}function Mn(e,t){return H(()=>{let r=new g;return r.subscribe(({action:o,reveal:n})=>{e.toggleAttribute("open",o==="open"),n&&e.scrollIntoView()}),Aa(e,t).pipe(y(o=>r.next(o)),_(()=>r.complete()),m(o=>F({ref:e},o)))})}var Ln=".node circle,.node ellipse,.node path,.node polygon,.node rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}marker{fill:var(--md-mermaid-edge-color)!important}.edgeLabel .label rect{fill:#0000}.label{color:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.label foreignObject{line-height:normal;overflow:visible}.label div .edgeLabel{color:var(--md-mermaid-label-fg-color)}.edgeLabel,.edgeLabel rect,.label div .edgeLabel{background-color:var(--md-mermaid-label-bg-color)}.edgeLabel,.edgeLabel rect{fill:var(--md-mermaid-label-bg-color);color:var(--md-mermaid-edge-color)}.edgePath .path,.flowchart-link{stroke:var(--md-mermaid-edge-color);stroke-width:.05rem}.edgePath .arrowheadPath{fill:var(--md-mermaid-edge-color);stroke:none}.cluster rect{fill:var(--md-default-fg-color--lightest);stroke:var(--md-default-fg-color--lighter)}.cluster span{color:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}g #flowchart-circleEnd,g #flowchart-circleStart,g #flowchart-crossEnd,g #flowchart-crossStart,g #flowchart-pointEnd,g #flowchart-pointStart{stroke:none}g.classGroup line,g.classGroup rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}g.classGroup text{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.classLabel .box{fill:var(--md-mermaid-label-bg-color);background-color:var(--md-mermaid-label-bg-color);opacity:1}.classLabel .label{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.node .divider{stroke:var(--md-mermaid-node-fg-color)}.relation{stroke:var(--md-mermaid-edge-color)}.cardinality{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.cardinality text{fill:inherit!important}defs #classDiagram-compositionEnd,defs #classDiagram-compositionStart,defs #classDiagram-dependencyEnd,defs #classDiagram-dependencyStart,defs #classDiagram-extensionEnd,defs #classDiagram-extensionStart{fill:var(--md-mermaid-edge-color)!important;stroke:var(--md-mermaid-edge-color)!important}defs #classDiagram-aggregationEnd,defs #classDiagram-aggregationStart{fill:var(--md-mermaid-label-bg-color)!important;stroke:var(--md-mermaid-edge-color)!important}g.stateGroup rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}g.stateGroup .state-title{fill:var(--md-mermaid-label-fg-color)!important;font-family:var(--md-mermaid-font-family)}g.stateGroup .composit{fill:var(--md-mermaid-label-bg-color)}.nodeLabel,.nodeLabel p{color:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.node circle.state-end,.node circle.state-start,.start-state{fill:var(--md-mermaid-edge-color);stroke:none}.end-state-inner,.end-state-outer{fill:var(--md-mermaid-edge-color)}.end-state-inner,.node circle.state-end{stroke:var(--md-mermaid-label-bg-color)}.transition{stroke:var(--md-mermaid-edge-color)}[id^=state-fork] rect,[id^=state-join] rect{fill:var(--md-mermaid-edge-color)!important;stroke:none!important}.statediagram-cluster.statediagram-cluster .inner{fill:var(--md-default-bg-color)}.statediagram-cluster rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}.statediagram-state rect.divider{fill:var(--md-default-fg-color--lightest);stroke:var(--md-default-fg-color--lighter)}defs #statediagram-barbEnd{stroke:var(--md-mermaid-edge-color)}.attributeBoxEven,.attributeBoxOdd{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}.entityBox{fill:var(--md-mermaid-label-bg-color);stroke:var(--md-mermaid-node-fg-color)}.entityLabel{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.relationshipLabelBox{fill:var(--md-mermaid-label-bg-color);fill-opacity:1;background-color:var(--md-mermaid-label-bg-color);opacity:1}.relationshipLabel{fill:var(--md-mermaid-label-fg-color)}.relationshipLine{stroke:var(--md-mermaid-edge-color)}defs #ONE_OR_MORE_END *,defs #ONE_OR_MORE_START *,defs #ONLY_ONE_END *,defs #ONLY_ONE_START *,defs #ZERO_OR_MORE_END *,defs #ZERO_OR_MORE_START *,defs #ZERO_OR_ONE_END *,defs #ZERO_OR_ONE_START *{stroke:var(--md-mermaid-edge-color)!important}defs #ZERO_OR_MORE_END circle,defs #ZERO_OR_MORE_START circle{fill:var(--md-mermaid-label-bg-color)}.actor{fill:var(--md-mermaid-sequence-actor-bg-color);stroke:var(--md-mermaid-sequence-actor-border-color)}text.actor>tspan{fill:var(--md-mermaid-sequence-actor-fg-color);font-family:var(--md-mermaid-font-family)}line{stroke:var(--md-mermaid-sequence-actor-line-color)}.actor-man circle,.actor-man line{fill:var(--md-mermaid-sequence-actorman-bg-color);stroke:var(--md-mermaid-sequence-actorman-line-color)}.messageLine0,.messageLine1{stroke:var(--md-mermaid-sequence-message-line-color)}.note{fill:var(--md-mermaid-sequence-note-bg-color);stroke:var(--md-mermaid-sequence-note-border-color)}.loopText,.loopText>tspan,.messageText,.noteText>tspan{stroke:none;font-family:var(--md-mermaid-font-family)!important}.messageText{fill:var(--md-mermaid-sequence-message-fg-color)}.loopText,.loopText>tspan{fill:var(--md-mermaid-sequence-loop-fg-color)}.noteText>tspan{fill:var(--md-mermaid-sequence-note-fg-color)}#arrowhead path{fill:var(--md-mermaid-sequence-message-line-color);stroke:none}.loopLine{fill:var(--md-mermaid-sequence-loop-bg-color);stroke:var(--md-mermaid-sequence-loop-border-color)}.labelBox{fill:var(--md-mermaid-sequence-label-bg-color);stroke:none}.labelText,.labelText>span{fill:var(--md-mermaid-sequence-label-fg-color);font-family:var(--md-mermaid-font-family)}.sequenceNumber{fill:var(--md-mermaid-sequence-number-fg-color)}rect.rect{fill:var(--md-mermaid-sequence-box-bg-color);stroke:none}rect.rect+text.text{fill:var(--md-mermaid-sequence-box-fg-color)}defs #sequencenumber{fill:var(--md-mermaid-sequence-number-bg-color)!important}";var qr,ka=0;function Ha(){return typeof mermaid=="undefined"||mermaid instanceof Element?gt("https://unpkg.com/mermaid@10.7.0/dist/mermaid.min.js"):$(void 0)}function _n(e){return e.classList.remove("mermaid"),qr||(qr=Ha().pipe(y(()=>mermaid.initialize({startOnLoad:!1,themeCSS:Ln,sequence:{actorFontSize:"16px",messageFontSize:"16px",noteFontSize:"16px"}})),m(()=>{}),B(1))),qr.subscribe(()=>ro(this,null,function*(){e.classList.add("mermaid");let t=`__mermaid_${ka++}`,r=E("div",{class:"mermaid"}),o=e.textContent,{svg:n,fn:i}=yield mermaid.render(t,o),a=r.attachShadow({mode:"closed"});a.innerHTML=n,e.replaceWith(r),i==null||i(a)})),qr.pipe(m(()=>({ref:e})))}var An=E("table");function Cn(e){return e.replaceWith(An),An.replaceWith(vn(e)),$({ref:e})}function $a(e){let t=e.find(r=>r.checked)||e[0];return T(...e.map(r=>d(r,"change").pipe(m(()=>P(`label[for="${r.id}"]`))))).pipe(q(P(`label[for="${t.id}"]`)),m(r=>({active:r})))}function kn(e,{viewport$:t,target$:r}){let o=P(".tabbed-labels",e),n=R(":scope > input",e),i=Nr("prev");e.append(i);let a=Nr("next");return e.append(a),H(()=>{let s=new g,c=s.pipe(ee(),oe(!0));Q([s,Ee(e)]).pipe(U(c),Me(1,de)).subscribe({next([{active:p},l]){let f=Ue(p),{width:u}=pe(p);e.style.setProperty("--md-indicator-x",`${f.x}px`),e.style.setProperty("--md-indicator-width",`${u}px`);let h=ir(o);(f.x<h.x||f.x+u>h.x+l.width)&&o.scrollTo({left:Math.max(0,f.x-16),behavior:"smooth"})},complete(){e.style.removeProperty("--md-indicator-x"),e.style.removeProperty("--md-indicator-width")}}),Q([et(o),Ee(o)]).pipe(U(c)).subscribe(([p,l])=>{let f=xt(o);i.hidden=p.x<16,a.hidden=p.x>f.width-l.width-16}),T(d(i,"click").pipe(m(()=>-1)),d(a,"click").pipe(m(()=>1))).pipe(U(c)).subscribe(p=>{let{width:l}=pe(o);o.scrollBy({left:l*p,behavior:"smooth"})}),r.pipe(U(c),b(p=>n.includes(p))).subscribe(p=>p.click()),o.classList.add("tabbed-labels--linked");for(let p of n){let l=P(`label[for="${p.id}"]`);l.replaceChildren(E("a",{href:`#${l.htmlFor}`,tabIndex:-1},...Array.from(l.childNodes))),d(l.firstElementChild,"click").pipe(U(c),b(f=>!(f.metaKey||f.ctrlKey)),y(f=>{f.preventDefault(),f.stopPropagation()})).subscribe(()=>{history.replaceState({},"",`#${l.htmlFor}`),l.click()})}return G("content.tabs.link")&&s.pipe(Le(1),ne(t)).subscribe(([{active:p},{offset:l}])=>{let f=p.innerText.trim();if(p.hasAttribute("data-md-switching"))p.removeAttribute("data-md-switching");else{let u=e.offsetTop-l.y;for(let w of R("[data-tabs]"))for(let A of R(":scope > input",w)){let Z=P(`label[for="${A.id}"]`);if(Z!==p&&Z.innerText.trim()===f){Z.setAttribute("data-md-switching",""),A.click();break}}window.scrollTo({top:e.offsetTop-u});let h=__md_get("__tabs")||[];__md_set("__tabs",[...new Set([f,...h])])}}),s.pipe(U(c)).subscribe(()=>{for(let p of R("audio, video",e))p.pause()}),$a(n).pipe(y(p=>s.next(p)),_(()=>s.complete()),m(p=>F({ref:e},p)))}).pipe(ze(ae))}function Hn(e,{viewport$:t,target$:r,print$:o}){return T(...R(".annotate:not(.highlight)",e).map(n=>wn(n,{target$:r,print$:o})),...R("pre:not(.mermaid) > code",e).map(n=>On(n,{target$:r,print$:o})),...R("pre.mermaid",e).map(n=>_n(n)),...R("table:not([class])",e).map(n=>Cn(n)),...R("details",e).map(n=>Mn(n,{target$:r,print$:o})),...R("[data-tabs]",e).map(n=>kn(n,{viewport$:t,target$:r})),...R("[title]",e).filter(()=>G("content.tooltips")).map(n=>Ge(n)))}function Ra(e,{alert$:t}){return t.pipe(v(r=>T($(!0),$(!1).pipe(Ye(2e3))).pipe(m(o=>({message:r,active:o})))))}function $n(e,t){let r=P(".md-typeset",e);return H(()=>{let o=new g;return o.subscribe(({message:n,active:i})=>{e.classList.toggle("md-dialog--active",i),r.textContent=n}),Ra(e,t).pipe(y(n=>o.next(n)),_(()=>o.complete()),m(n=>F({ref:e},n)))})}function Pa({viewport$:e}){if(!G("header.autohide"))return $(!1);let t=e.pipe(m(({offset:{y:n}})=>n),Ke(2,1),m(([n,i])=>[n<i,i]),X(0)),r=Q([e,t]).pipe(b(([{offset:n},[,i]])=>Math.abs(i-n.y)>100),m(([,[n]])=>n),Y()),o=We("search");return Q([e,o]).pipe(m(([{offset:n},i])=>n.y>400&&!i),Y(),v(n=>n?r:$(!1)),q(!1))}function Rn(e,t){return H(()=>Q([Ee(e),Pa(t)])).pipe(m(([{height:r},o])=>({height:r,hidden:o})),Y((r,o)=>r.height===o.height&&r.hidden===o.hidden),B(1))}function Pn(e,{header$:t,main$:r}){return H(()=>{let o=new g,n=o.pipe(ee(),oe(!0));o.pipe(X("active"),je(t)).subscribe(([{active:a},{hidden:s}])=>{e.classList.toggle("md-header--shadow",a&&!s),e.hidden=s});let i=fe(R("[title]",e)).pipe(b(()=>G("content.tooltips")),re(a=>Ge(a)));return r.subscribe(o),t.pipe(U(n),m(a=>F({ref:e},a)),$e(i.pipe(U(n))))})}function Ia(e,{viewport$:t,header$:r}){return pr(e,{viewport$:t,header$:r}).pipe(m(({offset:{y:o}})=>{let{height:n}=pe(e);return{active:o>=n}}),X("active"))}function In(e,t){return H(()=>{let r=new g;r.subscribe({next({active:n}){e.classList.toggle("md-header__title--active",n)},complete(){e.classList.remove("md-header__title--active")}});let o=me(".md-content h1");return typeof o=="undefined"?L:Ia(o,t).pipe(y(n=>r.next(n)),_(()=>r.complete()),m(n=>F({ref:e},n)))})}function Fn(e,{viewport$:t,header$:r}){let o=r.pipe(m(({height:i})=>i),Y()),n=o.pipe(v(()=>Ee(e).pipe(m(({height:i})=>({top:e.offsetTop,bottom:e.offsetTop+i})),X("bottom"))));return Q([o,n,t]).pipe(m(([i,{top:a,bottom:s},{offset:{y:c},size:{height:p}}])=>(p=Math.max(0,p-Math.max(0,a-c,i)-Math.max(0,p+c-s)),{offset:a-i,height:p,active:a-i<=c})),Y((i,a)=>i.offset===a.offset&&i.height===a.height&&i.active===a.active))}function Fa(e){let t=__md_get("__palette")||{index:e.findIndex(o=>matchMedia(o.getAttribute("data-md-color-media")).matches)},r=Math.max(0,Math.min(t.index,e.length-1));return $(...e).pipe(re(o=>d(o,"change").pipe(m(()=>o))),q(e[r]),m(o=>({index:e.indexOf(o),color:{media:o.getAttribute("data-md-color-media"),scheme:o.getAttribute("data-md-color-scheme"),primary:o.getAttribute("data-md-color-primary"),accent:o.getAttribute("data-md-color-accent")}})),B(1))}function jn(e){let t=R("input",e),r=E("meta",{name:"theme-color"});document.head.appendChild(r);let o=E("meta",{name:"color-scheme"});document.head.appendChild(o);let n=At("(prefers-color-scheme: light)");return H(()=>{let i=new g;return i.subscribe(a=>{if(document.body.setAttribute("data-md-color-switching",""),a.color.media==="(prefers-color-scheme)"){let s=matchMedia("(prefers-color-scheme: light)"),c=document.querySelector(s.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']");a.color.scheme=c.getAttribute("data-md-color-scheme"),a.color.primary=c.getAttribute("data-md-color-primary"),a.color.accent=c.getAttribute("data-md-color-accent")}for(let[s,c]of Object.entries(a.color))document.body.setAttribute(`data-md-color-${s}`,c);for(let s=0;s<t.length;s++){let c=t[s].nextElementSibling;c instanceof HTMLElement&&(c.hidden=a.index!==s)}__md_set("__palette",a)}),d(e,"keydown").pipe(b(a=>a.key==="Enter"),ne(i,(a,s)=>s)).subscribe(({index:a})=>{a=(a+1)%t.length,t[a].click(),t[a].focus()}),i.pipe(m(()=>{let a=Te("header"),s=window.getComputedStyle(a);return o.content=s.colorScheme,s.backgroundColor.match(/\d+/g).map(c=>(+c).toString(16).padStart(2,"0")).join("")})).subscribe(a=>r.content=`#${a}`),i.pipe(Oe(ae)).subscribe(()=>{document.body.removeAttribute("data-md-color-switching")}),Fa(t).pipe(U(n.pipe(Le(1))),at(),y(a=>i.next(a)),_(()=>i.complete()),m(a=>F({ref:e},a)))})}function Un(e,{progress$:t}){return H(()=>{let r=new g;return r.subscribe(({value:o})=>{e.style.setProperty("--md-progress-value",`${o}`)}),t.pipe(y(o=>r.next({value:o})),_(()=>r.complete()),m(o=>({ref:e,value:o})))})}var Kr=jt(zr());function ja(e){e.setAttribute("data-md-copying","");let t=e.closest("[data-copy]"),r=t?t.getAttribute("data-copy"):e.innerText;return e.removeAttribute("data-md-copying"),r.trimEnd()}function Wn({alert$:e}){Kr.default.isSupported()&&new j(t=>{new Kr.default("[data-clipboard-target], [data-clipboard-text]",{text:r=>r.getAttribute("data-clipboard-text")||ja(P(r.getAttribute("data-clipboard-target")))}).on("success",r=>t.next(r))}).pipe(y(t=>{t.trigger.focus()}),m(()=>ge("clipboard.copied"))).subscribe(e)}function Dn(e,t){return e.protocol=t.protocol,e.hostname=t.hostname,e}function Ua(e,t){let r=new Map;for(let o of R("url",e)){let n=P("loc",o),i=[Dn(new URL(n.textContent),t)];r.set(`${i[0]}`,i);for(let a of R("[rel=alternate]",o)){let s=a.getAttribute("href");s!=null&&i.push(Dn(new URL(s),t))}}return r}function mr(e){return on(new URL("sitemap.xml",e)).pipe(m(t=>Ua(t,new URL(e))),he(()=>$(new Map)))}function Wa(e,t){if(!(e.target instanceof Element))return L;let r=e.target.closest("a");if(r===null)return L;if(r.target||e.metaKey||e.ctrlKey)return L;let o=new URL(r.href);return o.search=o.hash="",t.has(`${o}`)?(e.preventDefault(),$(new URL(r.href))):L}function Nn(e){let t=new Map;for(let r of R(":scope > *",e.head))t.set(r.outerHTML,r);return t}function Vn(e){for(let t of R("[href], [src]",e))for(let r of["href","src"]){let o=t.getAttribute(r);if(o&&!/^(?:[a-z]+:)?\/\//i.test(o)){t[r]=t[r];break}}return $(e)}function Da(e){for(let o of["[data-md-component=announce]","[data-md-component=container]","[data-md-component=header-topic]","[data-md-component=outdated]","[data-md-component=logo]","[data-md-component=skip]",...G("navigation.tabs.sticky")?["[data-md-component=tabs]"]:[]]){let n=me(o),i=me(o,e);typeof n!="undefined"&&typeof i!="undefined"&&n.replaceWith(i)}let t=Nn(document);for(let[o,n]of Nn(e))t.has(o)?t.delete(o):document.head.appendChild(n);for(let o of t.values()){let n=o.getAttribute("name");n!=="theme-color"&&n!=="color-scheme"&&o.remove()}let r=Te("container");return Fe(R("script",r)).pipe(v(o=>{let n=e.createElement("script");if(o.src){for(let i of o.getAttributeNames())n.setAttribute(i,o.getAttribute(i));return o.replaceWith(n),new j(i=>{n.onload=()=>i.complete()})}else return n.textContent=o.textContent,o.replaceWith(n),L}),ee(),oe(document))}function zn({location$:e,viewport$:t,progress$:r}){let o=we();if(location.protocol==="file:")return L;let n=mr(o.base);$(document).subscribe(Vn);let i=d(document.body,"click").pipe(je(n),v(([c,p])=>Wa(c,p)),le()),a=d(window,"popstate").pipe(m(ve),le());i.pipe(ne(t)).subscribe(([c,{offset:p}])=>{history.replaceState(p,""),history.pushState(null,"",c)}),T(i,a).subscribe(e);let s=e.pipe(X("pathname"),v(c=>rn(c,{progress$:r}).pipe(he(()=>(st(c,!0),L)))),v(Vn),v(Da),le());return T(s.pipe(ne(e,(c,p)=>p)),e.pipe(X("pathname"),v(()=>e),X("hash")),e.pipe(Y((c,p)=>c.pathname===p.pathname&&c.hash===p.hash),v(()=>i),y(()=>history.back()))).subscribe(c=>{var p,l;history.state!==null||!c.hash?window.scrollTo(0,(l=(p=history.state)==null?void 0:p.y)!=null?l:0):(history.scrollRestoration="auto",Zo(c.hash),history.scrollRestoration="manual")}),e.subscribe(()=>{history.scrollRestoration="manual"}),d(window,"beforeunload").subscribe(()=>{history.scrollRestoration="auto"}),t.pipe(X("offset"),be(100)).subscribe(({offset:c})=>{history.replaceState(c,"")}),s}var Qn=jt(Kn());function Yn(e){let t=e.separator.split("|").map(n=>n.replace(/(\(\?[!=<][^)]+\))/g,"").length===0?"\uFFFD":n).join("|"),r=new RegExp(t,"img"),o=(n,i,a)=>`${i}<mark data-md-highlight>${a}</mark>`;return n=>{n=n.replace(/[\s*+\-:~^]+/g," ").trim();let i=new RegExp(`(^|${e.separator}|)(${n.replace(/[|\\{}()[\]^$+*?.-]/g,"\\$&").replace(r,"|")})`,"img");return a=>(0,Qn.default)(a).replace(i,o).replace(/<\/mark>(\s+)<mark[^>]*>/img,"$1")}}function Ht(e){return e.type===1}function fr(e){return e.type===3}function Bn(e,t){let r=ln(e);return T($(location.protocol!=="file:"),We("search")).pipe(He(o=>o),v(()=>t)).subscribe(({config:o,docs:n})=>r.next({type:0,data:{config:o,docs:n,options:{suggest:G("search.suggest")}}})),r}function Gn({document$:e}){let t=we(),r=De(new URL("../versions.json",t.base)).pipe(he(()=>L)),o=r.pipe(m(n=>{let[,i]=t.base.match(/([^/]+)\/?$/);return n.find(({version:a,aliases:s})=>a===i||s.includes(i))||n[0]}));r.pipe(m(n=>new Map(n.map(i=>[`${new URL(`../${i.version}/`,t.base)}`,i]))),v(n=>d(document.body,"click").pipe(b(i=>!i.metaKey&&!i.ctrlKey),ne(o),v(([i,a])=>{if(i.target instanceof Element){let s=i.target.closest("a");if(s&&!s.target&&n.has(s.href)){let c=s.href;return!i.target.closest(".md-version")&&n.get(c)===a?L:(i.preventDefault(),$(c))}}return L}),v(i=>{let{version:a}=n.get(i);return mr(new URL(i)).pipe(m(s=>{let p=ve().href.replace(t.base,"");return s.has(p.split("#")[0])?new URL(`../${a}/${p}`,t.base):new URL(i)}))})))).subscribe(n=>st(n,!0)),Q([r,o]).subscribe(([n,i])=>{P(".md-header__topic").appendChild(gn(n,i))}),e.pipe(v(()=>o)).subscribe(n=>{var a;let i=__md_get("__outdated",sessionStorage);if(i===null){i=!0;let s=((a=t.version)==null?void 0:a.default)||"latest";Array.isArray(s)||(s=[s]);e:for(let c of s)for(let p of n.aliases.concat(n.version))if(new RegExp(c,"i").test(p)){i=!1;break e}__md_set("__outdated",i,sessionStorage)}if(i)for(let s of ie("outdated"))s.hidden=!1})}function Ka(e,{worker$:t}){let{searchParams:r}=ve();r.has("q")&&(Be("search",!0),e.value=r.get("q"),e.focus(),We("search").pipe(He(i=>!i)).subscribe(()=>{let i=ve();i.searchParams.delete("q"),history.replaceState({},"",`${i}`)}));let o=vt(e),n=T(t.pipe(He(Ht)),d(e,"keyup"),o).pipe(m(()=>e.value),Y());return Q([n,o]).pipe(m(([i,a])=>({value:i,focus:a})),B(1))}function Jn(e,{worker$:t}){let r=new g,o=r.pipe(ee(),oe(!0));Q([t.pipe(He(Ht)),r],(i,a)=>a).pipe(X("value")).subscribe(({value:i})=>t.next({type:2,data:i})),r.pipe(X("focus")).subscribe(({focus:i})=>{i&&Be("search",i)}),d(e.form,"reset").pipe(U(o)).subscribe(()=>e.focus());let n=P("header [for=__search]");return d(n,"click").subscribe(()=>e.focus()),Ka(e,{worker$:t}).pipe(y(i=>r.next(i)),_(()=>r.complete()),m(i=>F({ref:e},i)),B(1))}function Xn(e,{worker$:t,query$:r}){let o=new g,n=Yo(e.parentElement).pipe(b(Boolean)),i=e.parentElement,a=P(":scope > :first-child",e),s=P(":scope > :last-child",e);We("search").subscribe(l=>s.setAttribute("role",l?"list":"presentation")),o.pipe(ne(r),Ir(t.pipe(He(Ht)))).subscribe(([{items:l},{value:f}])=>{switch(l.length){case 0:a.textContent=f.length?ge("search.result.none"):ge("search.result.placeholder");break;case 1:a.textContent=ge("search.result.one");break;default:let u=ar(l.length);a.textContent=ge("search.result.other",u)}});let c=o.pipe(y(()=>s.innerHTML=""),v(({items:l})=>T($(...l.slice(0,10)),$(...l.slice(10)).pipe(Ke(4),jr(n),v(([f])=>f)))),m(hn),le());return c.subscribe(l=>s.appendChild(l)),c.pipe(re(l=>{let f=me("details",l);return typeof f=="undefined"?L:d(f,"toggle").pipe(U(o),m(()=>f))})).subscribe(l=>{l.open===!1&&l.offsetTop<=i.scrollTop&&i.scrollTo({top:l.offsetTop})}),t.pipe(b(fr),m(({data:l})=>l)).pipe(y(l=>o.next(l)),_(()=>o.complete()),m(l=>F({ref:e},l)))}function Qa(e,{query$:t}){return t.pipe(m(({value:r})=>{let o=ve();return o.hash="",r=r.replace(/\s+/g,"+").replace(/&/g,"%26").replace(/=/g,"%3D"),o.search=`q=${r}`,{url:o}}))}function Zn(e,t){let r=new g,o=r.pipe(ee(),oe(!0));return r.subscribe(({url:n})=>{e.setAttribute("data-clipboard-text",e.href),e.href=`${n}`}),d(e,"click").pipe(U(o)).subscribe(n=>n.preventDefault()),Qa(e,t).pipe(y(n=>r.next(n)),_(()=>r.complete()),m(n=>F({ref:e},n)))}function ei(e,{worker$:t,keyboard$:r}){let o=new g,n=Te("search-query"),i=T(d(n,"keydown"),d(n,"focus")).pipe(Oe(ae),m(()=>n.value),Y());return o.pipe(je(i),m(([{suggest:s},c])=>{let p=c.split(/([\s-]+)/);if(s!=null&&s.length&&p[p.length-1]){let l=s[s.length-1];l.startsWith(p[p.length-1])&&(p[p.length-1]=l)}else p.length=0;return p})).subscribe(s=>e.innerHTML=s.join("").replace(/\s/g,"&nbsp;")),r.pipe(b(({mode:s})=>s==="search")).subscribe(s=>{switch(s.type){case"ArrowRight":e.innerText.length&&n.selectionStart===n.value.length&&(n.value=e.innerText);break}}),t.pipe(b(fr),m(({data:s})=>s)).pipe(y(s=>o.next(s)),_(()=>o.complete()),m(()=>({ref:e})))}function ti(e,{index$:t,keyboard$:r}){let o=we();try{let n=Bn(o.search,t),i=Te("search-query",e),a=Te("search-result",e);d(e,"click").pipe(b(({target:c})=>c instanceof Element&&!!c.closest("a"))).subscribe(()=>Be("search",!1)),r.pipe(b(({mode:c})=>c==="search")).subscribe(c=>{let p=Re();switch(c.type){case"Enter":if(p===i){let l=new Map;for(let f of R(":first-child [href]",a)){let u=f.firstElementChild;l.set(f,parseFloat(u.getAttribute("data-md-score")))}if(l.size){let[[f]]=[...l].sort(([,u],[,h])=>h-u);f.click()}c.claim()}break;case"Escape":case"Tab":Be("search",!1),i.blur();break;case"ArrowUp":case"ArrowDown":if(typeof p=="undefined")i.focus();else{let l=[i,...R(":not(details) > [href], summary, details[open] [href]",a)],f=Math.max(0,(Math.max(0,l.indexOf(p))+l.length+(c.type==="ArrowUp"?-1:1))%l.length);l[f].focus()}c.claim();break;default:i!==Re()&&i.focus()}}),r.pipe(b(({mode:c})=>c==="global")).subscribe(c=>{switch(c.type){case"f":case"s":case"/":i.focus(),i.select(),c.claim();break}});let s=Jn(i,{worker$:n});return T(s,Xn(a,{worker$:n,query$:s})).pipe($e(...ie("search-share",e).map(c=>Zn(c,{query$:s})),...ie("search-suggest",e).map(c=>ei(c,{worker$:n,keyboard$:r}))))}catch(n){return e.hidden=!0,qe}}function ri(e,{index$:t,location$:r}){return Q([t,r.pipe(q(ve()),b(o=>!!o.searchParams.get("h")))]).pipe(m(([o,n])=>Yn(o.config)(n.searchParams.get("h"))),m(o=>{var a;let n=new Map,i=document.createNodeIterator(e,NodeFilter.SHOW_TEXT);for(let s=i.nextNode();s;s=i.nextNode())if((a=s.parentElement)!=null&&a.offsetHeight){let c=s.textContent,p=o(c);p.length>c.length&&n.set(s,p)}for(let[s,c]of n){let{childNodes:p}=E("span",null,c);s.replaceWith(...Array.from(p))}return{ref:e,nodes:n}}))}function Ya(e,{viewport$:t,main$:r}){let o=e.closest(".md-grid"),n=o.offsetTop-o.parentElement.offsetTop;return Q([r,t]).pipe(m(([{offset:i,height:a},{offset:{y:s}}])=>(a=a+Math.min(n,Math.max(0,s-i))-n,{height:a,locked:s>=i+n})),Y((i,a)=>i.height===a.height&&i.locked===a.locked))}function Qr(e,o){var n=o,{header$:t}=n,r=to(n,["header$"]);let i=P(".md-sidebar__scrollwrap",e),{y:a}=Ue(i);return H(()=>{let s=new g,c=s.pipe(ee(),oe(!0)),p=s.pipe(Me(0,de));return p.pipe(ne(t)).subscribe({next([{height:l},{height:f}]){i.style.height=`${l-2*a}px`,e.style.top=`${f}px`},complete(){i.style.height="",e.style.top=""}}),p.pipe(He()).subscribe(()=>{for(let l of R(".md-nav__link--active[href]",e)){if(!l.clientHeight)continue;let f=l.closest(".md-sidebar__scrollwrap");if(typeof f!="undefined"){let u=l.offsetTop-f.offsetTop,{height:h}=pe(f);f.scrollTo({top:u-h/2})}}}),fe(R("label[tabindex]",e)).pipe(re(l=>d(l,"click").pipe(Oe(ae),m(()=>l),U(c)))).subscribe(l=>{let f=P(`[id="${l.htmlFor}"]`);P(`[aria-labelledby="${l.id}"]`).setAttribute("aria-expanded",`${f.checked}`)}),Ya(e,r).pipe(y(l=>s.next(l)),_(()=>s.complete()),m(l=>F({ref:e},l)))})}function oi(e,t){if(typeof t!="undefined"){let r=`https://api.github.com/repos/${e}/${t}`;return Lt(De(`${r}/releases/latest`).pipe(he(()=>L),m(o=>({version:o.tag_name})),Qe({})),De(r).pipe(he(()=>L),m(o=>({stars:o.stargazers_count,forks:o.forks_count})),Qe({}))).pipe(m(([o,n])=>F(F({},o),n)))}else{let r=`https://api.github.com/users/${e}`;return De(r).pipe(m(o=>({repositories:o.public_repos})),Qe({}))}}function ni(e,t){let r=`https://${e}/api/v4/projects/${encodeURIComponent(t)}`;return De(r).pipe(he(()=>L),m(({star_count:o,forks_count:n})=>({stars:o,forks:n})),Qe({}))}function ii(e){let t=e.match(/^.+github\.com\/([^/]+)\/?([^/]+)?/i);if(t){let[,r,o]=t;return oi(r,o)}if(t=e.match(/^.+?([^/]*gitlab[^/]+)\/(.+?)\/?$/i),t){let[,r,o]=t;return ni(r,o)}return L}var Ba;function Ga(e){return Ba||(Ba=H(()=>{let t=__md_get("__source",sessionStorage);if(t)return $(t);if(ie("consent").length){let o=__md_get("__consent");if(!(o&&o.github))return L}return ii(e.href).pipe(y(o=>__md_set("__source",o,sessionStorage)))}).pipe(he(()=>L),b(t=>Object.keys(t).length>0),m(t=>({facts:t})),B(1)))}function ai(e){let t=P(":scope > :last-child",e);return H(()=>{let r=new g;return r.subscribe(({facts:o})=>{t.appendChild(bn(o)),t.classList.add("md-source__repository--active")}),Ga(e).pipe(y(o=>r.next(o)),_(()=>r.complete()),m(o=>F({ref:e},o)))})}function Ja(e,{viewport$:t,header$:r}){return Ee(document.body).pipe(v(()=>pr(e,{header$:r,viewport$:t})),m(({offset:{y:o}})=>({hidden:o>=10})),X("hidden"))}function si(e,t){return H(()=>{let r=new g;return r.subscribe({next({hidden:o}){e.hidden=o},complete(){e.hidden=!1}}),(G("navigation.tabs.sticky")?$({hidden:!1}):Ja(e,t)).pipe(y(o=>r.next(o)),_(()=>r.complete()),m(o=>F({ref:e},o)))})}function Xa(e,{viewport$:t,header$:r}){let o=new Map,n=R(".md-nav__link",e);for(let s of n){let c=decodeURIComponent(s.hash.substring(1)),p=me(`[id="${c}"]`);typeof p!="undefined"&&o.set(s,p)}let i=r.pipe(X("height"),m(({height:s})=>{let c=Te("main"),p=P(":scope > :first-child",c);return s+.8*(p.offsetTop-c.offsetTop)}),le());return Ee(document.body).pipe(X("height"),v(s=>H(()=>{let c=[];return $([...o].reduce((p,[l,f])=>{for(;c.length&&o.get(c[c.length-1]).tagName>=f.tagName;)c.pop();let u=f.offsetTop;for(;!u&&f.parentElement;)f=f.parentElement,u=f.offsetTop;let h=f.offsetParent;for(;h;h=h.offsetParent)u+=h.offsetTop;return p.set([...c=[...c,l]].reverse(),u)},new Map))}).pipe(m(c=>new Map([...c].sort(([,p],[,l])=>p-l))),je(i),v(([c,p])=>t.pipe(Rr(([l,f],{offset:{y:u},size:h})=>{let w=u+h.height>=Math.floor(s.height);for(;f.length;){let[,A]=f[0];if(A-p<u||w)l=[...l,f.shift()];else break}for(;l.length;){let[,A]=l[l.length-1];if(A-p>=u&&!w)f=[l.pop(),...f];else break}return[l,f]},[[],[...c]]),Y((l,f)=>l[0]===f[0]&&l[1]===f[1])))))).pipe(m(([s,c])=>({prev:s.map(([p])=>p),next:c.map(([p])=>p)})),q({prev:[],next:[]}),Ke(2,1),m(([s,c])=>s.prev.length<c.prev.length?{prev:c.prev.slice(Math.max(0,s.prev.length-1),c.prev.length),next:[]}:{prev:c.prev.slice(-1),next:c.next.slice(0,c.next.length-s.next.length)}))}function ci(e,{viewport$:t,header$:r,main$:o,target$:n}){return H(()=>{let i=new g,a=i.pipe(ee(),oe(!0));if(i.subscribe(({prev:s,next:c})=>{for(let[p]of c)p.classList.remove("md-nav__link--passed"),p.classList.remove("md-nav__link--active");for(let[p,[l]]of s.entries())l.classList.add("md-nav__link--passed"),l.classList.toggle("md-nav__link--active",p===s.length-1)}),G("toc.follow")){let s=T(t.pipe(be(1),m(()=>{})),t.pipe(be(250),m(()=>"smooth")));i.pipe(b(({prev:c})=>c.length>0),je(o.pipe(Oe(ae))),ne(s)).subscribe(([[{prev:c}],p])=>{let[l]=c[c.length-1];if(l.offsetHeight){let f=sr(l);if(typeof f!="undefined"){let u=l.offsetTop-f.offsetTop,{height:h}=pe(f);f.scrollTo({top:u-h/2,behavior:p})}}})}return G("navigation.tracking")&&t.pipe(U(a),X("offset"),be(250),Le(1),U(n.pipe(Le(1))),at({delay:250}),ne(i)).subscribe(([,{prev:s}])=>{let c=ve(),p=s[s.length-1];if(p&&p.length){let[l]=p,{hash:f}=new URL(l.href);c.hash!==f&&(c.hash=f,history.replaceState({},"",`${c}`))}else c.hash="",history.replaceState({},"",`${c}`)}),Xa(e,{viewport$:t,header$:r}).pipe(y(s=>i.next(s)),_(()=>i.complete()),m(s=>F({ref:e},s)))})}function Za(e,{viewport$:t,main$:r,target$:o}){let n=t.pipe(m(({offset:{y:a}})=>a),Ke(2,1),m(([a,s])=>a>s&&s>0),Y()),i=r.pipe(m(({active:a})=>a));return Q([i,n]).pipe(m(([a,s])=>!(a&&s)),Y(),U(o.pipe(Le(1))),oe(!0),at({delay:250}),m(a=>({hidden:a})))}function pi(e,{viewport$:t,header$:r,main$:o,target$:n}){let i=new g,a=i.pipe(ee(),oe(!0));return i.subscribe({next({hidden:s}){e.hidden=s,s?(e.setAttribute("tabindex","-1"),e.blur()):e.removeAttribute("tabindex")},complete(){e.style.top="",e.hidden=!0,e.removeAttribute("tabindex")}}),r.pipe(U(a),X("height")).subscribe(({height:s})=>{e.style.top=`${s+16}px`}),d(e,"click").subscribe(s=>{s.preventDefault(),window.scrollTo({top:0})}),Za(e,{viewport$:t,main$:o,target$:n}).pipe(y(s=>i.next(s)),_(()=>i.complete()),m(s=>F({ref:e},s)))}function li({document$:e}){e.pipe(v(()=>R(".md-ellipsis")),re(t=>yt(t).pipe(U(e.pipe(Le(1))),b(r=>r),m(()=>t),ye(1))),b(t=>t.offsetWidth<t.scrollWidth),re(t=>{let r=t.innerText,o=t.closest("a")||t;return o.title=r,Ge(o).pipe(U(e.pipe(Le(1))),_(()=>o.removeAttribute("title")))})).subscribe(),e.pipe(v(()=>R(".md-status")),re(t=>Ge(t))).subscribe()}function mi({document$:e,tablet$:t}){e.pipe(v(()=>R(".md-toggle--indeterminate")),y(r=>{r.indeterminate=!0,r.checked=!1}),re(r=>d(r,"change").pipe(Fr(()=>r.classList.contains("md-toggle--indeterminate")),m(()=>r))),ne(t)).subscribe(([r,o])=>{r.classList.remove("md-toggle--indeterminate"),o&&(r.checked=!1)})}function es(){return/(iPad|iPhone|iPod)/.test(navigator.userAgent)}function fi({document$:e}){e.pipe(v(()=>R("[data-md-scrollfix]")),y(t=>t.removeAttribute("data-md-scrollfix")),b(es),re(t=>d(t,"touchstart").pipe(m(()=>t)))).subscribe(t=>{let r=t.scrollTop;r===0?t.scrollTop=1:r+t.offsetHeight===t.scrollHeight&&(t.scrollTop=r-1)})}function ui({viewport$:e,tablet$:t}){Q([We("search"),t]).pipe(m(([r,o])=>r&&!o),v(r=>$(r).pipe(Ye(r?400:100))),ne(e)).subscribe(([r,{offset:{y:o}}])=>{if(r)document.body.setAttribute("data-md-scrolllock",""),document.body.style.top=`-${o}px`;else{let n=-1*parseInt(document.body.style.top,10);document.body.removeAttribute("data-md-scrolllock"),document.body.style.top="",n&&window.scrollTo(0,n)}})}Object.entries||(Object.entries=function(e){let t=[];for(let r of Object.keys(e))t.push([r,e[r]]);return t});Object.values||(Object.values=function(e){let t=[];for(let r of Object.keys(e))t.push(e[r]);return t});typeof Element!="undefined"&&(Element.prototype.scrollTo||(Element.prototype.scrollTo=function(e,t){typeof e=="object"?(this.scrollLeft=e.left,this.scrollTop=e.top):(this.scrollLeft=e,this.scrollTop=t)}),Element.prototype.replaceWith||(Element.prototype.replaceWith=function(...e){let t=this.parentNode;if(t){e.length===0&&t.removeChild(this);for(let r=e.length-1;r>=0;r--){let o=e[r];typeof o=="string"?o=document.createTextNode(o):o.parentNode&&o.parentNode.removeChild(o),r?t.insertBefore(this.previousSibling,o):t.replaceChild(o,this)}}}));function ts(){return location.protocol==="file:"?gt(`${new URL("search/search_index.js",Yr.base)}`).pipe(m(()=>__index),B(1)):De(new URL("search/search_index.json",Yr.base))}document.documentElement.classList.remove("no-js");document.documentElement.classList.add("js");var rt=No(),Rt=Jo(),wt=en(Rt),Br=Go(),_e=pn(),ur=At("(min-width: 960px)"),hi=At("(min-width: 1220px)"),bi=tn(),Yr=we(),vi=document.forms.namedItem("search")?ts():qe,Gr=new g;Wn({alert$:Gr});var Jr=new g;G("navigation.instant")&&zn({location$:Rt,viewport$:_e,progress$:Jr}).subscribe(rt);var di;((di=Yr.version)==null?void 0:di.provider)==="mike"&&Gn({document$:rt});T(Rt,wt).pipe(Ye(125)).subscribe(()=>{Be("drawer",!1),Be("search",!1)});Br.pipe(b(({mode:e})=>e==="global")).subscribe(e=>{switch(e.type){case"p":case",":let t=me("link[rel=prev]");typeof t!="undefined"&&st(t);break;case"n":case".":let r=me("link[rel=next]");typeof r!="undefined"&&st(r);break;case"Enter":let o=Re();o instanceof HTMLLabelElement&&o.click()}});li({document$:rt});mi({document$:rt,tablet$:ur});fi({document$:rt});ui({viewport$:_e,tablet$:ur});var tt=Rn(Te("header"),{viewport$:_e}),$t=rt.pipe(m(()=>Te("main")),v(e=>Fn(e,{viewport$:_e,header$:tt})),B(1)),rs=T(...ie("consent").map(e=>fn(e,{target$:wt})),...ie("dialog").map(e=>$n(e,{alert$:Gr})),...ie("header").map(e=>Pn(e,{viewport$:_e,header$:tt,main$:$t})),...ie("palette").map(e=>jn(e)),...ie("progress").map(e=>Un(e,{progress$:Jr})),...ie("search").map(e=>ti(e,{index$:vi,keyboard$:Br})),...ie("source").map(e=>ai(e))),os=H(()=>T(...ie("announce").map(e=>mn(e)),...ie("content").map(e=>Hn(e,{viewport$:_e,target$:wt,print$:bi})),...ie("content").map(e=>G("search.highlight")?ri(e,{index$:vi,location$:Rt}):L),...ie("header-title").map(e=>In(e,{viewport$:_e,header$:tt})),...ie("sidebar").map(e=>e.getAttribute("data-md-type")==="navigation"?Ur(hi,()=>Qr(e,{viewport$:_e,header$:tt,main$:$t})):Ur(ur,()=>Qr(e,{viewport$:_e,header$:tt,main$:$t}))),...ie("tabs").map(e=>si(e,{viewport$:_e,header$:tt})),...ie("toc").map(e=>ci(e,{viewport$:_e,header$:tt,main$:$t,target$:wt})),...ie("top").map(e=>pi(e,{viewport$:_e,header$:tt,main$:$t,target$:wt})))),gi=rt.pipe(v(()=>os),$e(rs),B(1));gi.subscribe();window.document$=rt;window.location$=Rt;window.target$=wt;window.keyboard$=Br;window.viewport$=_e;window.tablet$=ur;window.screen$=hi;window.print$=bi;window.alert$=Gr;window.progress$=Jr;window.component$=gi;})();
+//# sourceMappingURL=bundle.1e8ae164.min.js.map
+
diff --git a/assets/javascripts/bundle.1e8ae164.min.js.map b/assets/javascripts/bundle.1e8ae164.min.js.map
new file mode 100644
index 0000000000..6c33b8e8e6
--- /dev/null
+++ b/assets/javascripts/bundle.1e8ae164.min.js.map
@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["node_modules/focus-visible/dist/focus-visible.js", "node_modules/clipboard/dist/clipboard.js", "node_modules/escape-html/index.js", "src/templates/assets/javascripts/bundle.ts", "node_modules/rxjs/node_modules/tslib/tslib.es6.js", "node_modules/rxjs/src/internal/util/isFunction.ts", "node_modules/rxjs/src/internal/util/createErrorClass.ts", "node_modules/rxjs/src/internal/util/UnsubscriptionError.ts", "node_modules/rxjs/src/internal/util/arrRemove.ts", "node_modules/rxjs/src/internal/Subscription.ts", "node_modules/rxjs/src/internal/config.ts", "node_modules/rxjs/src/internal/scheduler/timeoutProvider.ts", "node_modules/rxjs/src/internal/util/reportUnhandledError.ts", "node_modules/rxjs/src/internal/util/noop.ts", "node_modules/rxjs/src/internal/NotificationFactories.ts", "node_modules/rxjs/src/internal/util/errorContext.ts", "node_modules/rxjs/src/internal/Subscriber.ts", "node_modules/rxjs/src/internal/symbol/observable.ts", "node_modules/rxjs/src/internal/util/identity.ts", "node_modules/rxjs/src/internal/util/pipe.ts", "node_modules/rxjs/src/internal/Observable.ts", "node_modules/rxjs/src/internal/util/lift.ts", "node_modules/rxjs/src/internal/operators/OperatorSubscriber.ts", "node_modules/rxjs/src/internal/scheduler/animationFrameProvider.ts", "node_modules/rxjs/src/internal/util/ObjectUnsubscribedError.ts", "node_modules/rxjs/src/internal/Subject.ts", "node_modules/rxjs/src/internal/scheduler/dateTimestampProvider.ts", "node_modules/rxjs/src/internal/ReplaySubject.ts", "node_modules/rxjs/src/internal/scheduler/Action.ts", "node_modules/rxjs/src/internal/scheduler/intervalProvider.ts", "node_modules/rxjs/src/internal/scheduler/AsyncAction.ts", "node_modules/rxjs/src/internal/Scheduler.ts", "node_modules/rxjs/src/internal/scheduler/AsyncScheduler.ts", "node_modules/rxjs/src/internal/scheduler/async.ts", "node_modules/rxjs/src/internal/scheduler/AnimationFrameAction.ts", "node_modules/rxjs/src/internal/scheduler/AnimationFrameScheduler.ts", "node_modules/rxjs/src/internal/scheduler/animationFrame.ts", "node_modules/rxjs/src/internal/observable/empty.ts", "node_modules/rxjs/src/internal/util/isScheduler.ts", "node_modules/rxjs/src/internal/util/args.ts", "node_modules/rxjs/src/internal/util/isArrayLike.ts", "node_modules/rxjs/src/internal/util/isPromise.ts", "node_modules/rxjs/src/internal/util/isInteropObservable.ts", "node_modules/rxjs/src/internal/util/isAsyncIterable.ts", "node_modules/rxjs/src/internal/util/throwUnobservableError.ts", "node_modules/rxjs/src/internal/symbol/iterator.ts", "node_modules/rxjs/src/internal/util/isIterable.ts", "node_modules/rxjs/src/internal/util/isReadableStreamLike.ts", "node_modules/rxjs/src/internal/observable/innerFrom.ts", "node_modules/rxjs/src/internal/util/executeSchedule.ts", "node_modules/rxjs/src/internal/operators/observeOn.ts", "node_modules/rxjs/src/internal/operators/subscribeOn.ts", "node_modules/rxjs/src/internal/scheduled/scheduleObservable.ts", "node_modules/rxjs/src/internal/scheduled/schedulePromise.ts", "node_modules/rxjs/src/internal/scheduled/scheduleArray.ts", "node_modules/rxjs/src/internal/scheduled/scheduleIterable.ts", "node_modules/rxjs/src/internal/scheduled/scheduleAsyncIterable.ts", "node_modules/rxjs/src/internal/scheduled/scheduleReadableStreamLike.ts", "node_modules/rxjs/src/internal/scheduled/scheduled.ts", "node_modules/rxjs/src/internal/observable/from.ts", "node_modules/rxjs/src/internal/observable/of.ts", "node_modules/rxjs/src/internal/observable/throwError.ts", "node_modules/rxjs/src/internal/util/EmptyError.ts", "node_modules/rxjs/src/internal/util/isDate.ts", "node_modules/rxjs/src/internal/operators/map.ts", "node_modules/rxjs/src/internal/util/mapOneOrManyArgs.ts", "node_modules/rxjs/src/internal/util/argsArgArrayOrObject.ts", "node_modules/rxjs/src/internal/util/createObject.ts", "node_modules/rxjs/src/internal/observable/combineLatest.ts", "node_modules/rxjs/src/internal/operators/mergeInternals.ts", "node_modules/rxjs/src/internal/operators/mergeMap.ts", "node_modules/rxjs/src/internal/operators/mergeAll.ts", "node_modules/rxjs/src/internal/operators/concatAll.ts", "node_modules/rxjs/src/internal/observable/concat.ts", "node_modules/rxjs/src/internal/observable/defer.ts", "node_modules/rxjs/src/internal/observable/fromEvent.ts", "node_modules/rxjs/src/internal/observable/fromEventPattern.ts", "node_modules/rxjs/src/internal/observable/timer.ts", "node_modules/rxjs/src/internal/observable/merge.ts", "node_modules/rxjs/src/internal/observable/never.ts", "node_modules/rxjs/src/internal/util/argsOrArgArray.ts", "node_modules/rxjs/src/internal/operators/filter.ts", "node_modules/rxjs/src/internal/observable/zip.ts", "node_modules/rxjs/src/internal/operators/audit.ts", "node_modules/rxjs/src/internal/operators/auditTime.ts", "node_modules/rxjs/src/internal/operators/bufferCount.ts", "node_modules/rxjs/src/internal/operators/catchError.ts", "node_modules/rxjs/src/internal/operators/scanInternals.ts", "node_modules/rxjs/src/internal/operators/combineLatest.ts", "node_modules/rxjs/src/internal/operators/combineLatestWith.ts", "node_modules/rxjs/src/internal/operators/debounceTime.ts", "node_modules/rxjs/src/internal/operators/defaultIfEmpty.ts", "node_modules/rxjs/src/internal/operators/take.ts", "node_modules/rxjs/src/internal/operators/ignoreElements.ts", "node_modules/rxjs/src/internal/operators/mapTo.ts", "node_modules/rxjs/src/internal/operators/delayWhen.ts", "node_modules/rxjs/src/internal/operators/delay.ts", "node_modules/rxjs/src/internal/operators/distinctUntilChanged.ts", "node_modules/rxjs/src/internal/operators/distinctUntilKeyChanged.ts", "node_modules/rxjs/src/internal/operators/throwIfEmpty.ts", "node_modules/rxjs/src/internal/operators/endWith.ts", "node_modules/rxjs/src/internal/operators/finalize.ts", "node_modules/rxjs/src/internal/operators/first.ts", "node_modules/rxjs/src/internal/operators/takeLast.ts", "node_modules/rxjs/src/internal/operators/merge.ts", "node_modules/rxjs/src/internal/operators/mergeWith.ts", "node_modules/rxjs/src/internal/operators/repeat.ts", "node_modules/rxjs/src/internal/operators/scan.ts", "node_modules/rxjs/src/internal/operators/share.ts", "node_modules/rxjs/src/internal/operators/shareReplay.ts", "node_modules/rxjs/src/internal/operators/skip.ts", "node_modules/rxjs/src/internal/operators/skipUntil.ts", "node_modules/rxjs/src/internal/operators/startWith.ts", "node_modules/rxjs/src/internal/operators/switchMap.ts", "node_modules/rxjs/src/internal/operators/takeUntil.ts", "node_modules/rxjs/src/internal/operators/takeWhile.ts", "node_modules/rxjs/src/internal/operators/tap.ts", "node_modules/rxjs/src/internal/operators/throttle.ts", "node_modules/rxjs/src/internal/operators/throttleTime.ts", "node_modules/rxjs/src/internal/operators/withLatestFrom.ts", "node_modules/rxjs/src/internal/operators/zip.ts", "node_modules/rxjs/src/internal/operators/zipWith.ts", "src/templates/assets/javascripts/browser/document/index.ts", "src/templates/assets/javascripts/browser/element/_/index.ts", "src/templates/assets/javascripts/browser/element/focus/index.ts", "src/templates/assets/javascripts/browser/element/hover/index.ts", "src/templates/assets/javascripts/browser/element/offset/_/index.ts", "src/templates/assets/javascripts/browser/element/offset/content/index.ts", "src/templates/assets/javascripts/utilities/h/index.ts", "src/templates/assets/javascripts/utilities/round/index.ts", "src/templates/assets/javascripts/browser/script/index.ts", "src/templates/assets/javascripts/browser/element/size/_/index.ts", "src/templates/assets/javascripts/browser/element/size/content/index.ts", "src/templates/assets/javascripts/browser/element/visibility/index.ts", "src/templates/assets/javascripts/browser/toggle/index.ts", "src/templates/assets/javascripts/browser/keyboard/index.ts", "src/templates/assets/javascripts/browser/location/_/index.ts", "src/templates/assets/javascripts/browser/location/hash/index.ts", "src/templates/assets/javascripts/browser/media/index.ts", "src/templates/assets/javascripts/browser/request/index.ts", "src/templates/assets/javascripts/browser/viewport/offset/index.ts", "src/templates/assets/javascripts/browser/viewport/size/index.ts", "src/templates/assets/javascripts/browser/viewport/_/index.ts", "src/templates/assets/javascripts/browser/viewport/at/index.ts", "src/templates/assets/javascripts/browser/worker/index.ts", "src/templates/assets/javascripts/_/index.ts", "src/templates/assets/javascripts/components/_/index.ts", "src/templates/assets/javascripts/components/announce/index.ts", "src/templates/assets/javascripts/components/consent/index.ts", "src/templates/assets/javascripts/templates/tooltip/index.tsx", "src/templates/assets/javascripts/templates/annotation/index.tsx", "src/templates/assets/javascripts/templates/clipboard/index.tsx", "src/templates/assets/javascripts/templates/search/index.tsx", "src/templates/assets/javascripts/templates/source/index.tsx", "src/templates/assets/javascripts/templates/tabbed/index.tsx", "src/templates/assets/javascripts/templates/table/index.tsx", "src/templates/assets/javascripts/templates/version/index.tsx", "src/templates/assets/javascripts/components/tooltip/index.ts", "src/templates/assets/javascripts/components/content/annotation/_/index.ts", "src/templates/assets/javascripts/components/content/annotation/list/index.ts", "src/templates/assets/javascripts/components/content/annotation/block/index.ts", "src/templates/assets/javascripts/components/content/code/_/index.ts", "src/templates/assets/javascripts/components/content/details/index.ts", "src/templates/assets/javascripts/components/content/mermaid/index.css", "src/templates/assets/javascripts/components/content/mermaid/index.ts", "src/templates/assets/javascripts/components/content/table/index.ts", "src/templates/assets/javascripts/components/content/tabs/index.ts", "src/templates/assets/javascripts/components/content/_/index.ts", "src/templates/assets/javascripts/components/dialog/index.ts", "src/templates/assets/javascripts/components/header/_/index.ts", "src/templates/assets/javascripts/components/header/title/index.ts", "src/templates/assets/javascripts/components/main/index.ts", "src/templates/assets/javascripts/components/palette/index.ts", "src/templates/assets/javascripts/components/progress/index.ts", "src/templates/assets/javascripts/integrations/clipboard/index.ts", "src/templates/assets/javascripts/integrations/sitemap/index.ts", "src/templates/assets/javascripts/integrations/instant/index.ts", "src/templates/assets/javascripts/integrations/search/highlighter/index.ts", "src/templates/assets/javascripts/integrations/search/worker/message/index.ts", "src/templates/assets/javascripts/integrations/search/worker/_/index.ts", "src/templates/assets/javascripts/integrations/version/index.ts", "src/templates/assets/javascripts/components/search/query/index.ts", "src/templates/assets/javascripts/components/search/result/index.ts", "src/templates/assets/javascripts/components/search/share/index.ts", "src/templates/assets/javascripts/components/search/suggest/index.ts", "src/templates/assets/javascripts/components/search/_/index.ts", "src/templates/assets/javascripts/components/search/highlight/index.ts", "src/templates/assets/javascripts/components/sidebar/index.ts", "src/templates/assets/javascripts/components/source/facts/github/index.ts", "src/templates/assets/javascripts/components/source/facts/gitlab/index.ts", "src/templates/assets/javascripts/components/source/facts/_/index.ts", "src/templates/assets/javascripts/components/source/_/index.ts", "src/templates/assets/javascripts/components/tabs/index.ts", "src/templates/assets/javascripts/components/toc/index.ts", "src/templates/assets/javascripts/components/top/index.ts", "src/templates/assets/javascripts/patches/ellipsis/index.ts", "src/templates/assets/javascripts/patches/indeterminate/index.ts", "src/templates/assets/javascripts/patches/scrollfix/index.ts", "src/templates/assets/javascripts/patches/scrolllock/index.ts", "src/templates/assets/javascripts/polyfills/index.ts"],
+  "sourcesContent": ["(function (global, factory) {\n  typeof exports === 'object' && typeof module !== 'undefined' ? factory() :\n  typeof define === 'function' && define.amd ? define(factory) :\n  (factory());\n}(this, (function () { 'use strict';\n\n  /**\n   * Applies the :focus-visible polyfill at the given scope.\n   * A scope in this case is either the top-level Document or a Shadow Root.\n   *\n   * @param {(Document|ShadowRoot)} scope\n   * @see https://github.com/WICG/focus-visible\n   */\n  function applyFocusVisiblePolyfill(scope) {\n    var hadKeyboardEvent = true;\n    var hadFocusVisibleRecently = false;\n    var hadFocusVisibleRecentlyTimeout = null;\n\n    var inputTypesAllowlist = {\n      text: true,\n      search: true,\n      url: true,\n      tel: true,\n      email: true,\n      password: true,\n      number: true,\n      date: true,\n      month: true,\n      week: true,\n      time: true,\n      datetime: true,\n      'datetime-local': true\n    };\n\n    /**\n     * Helper function for legacy browsers and iframes which sometimes focus\n     * elements like document, body, and non-interactive SVG.\n     * @param {Element} el\n     */\n    function isValidFocusTarget(el) {\n      if (\n        el &&\n        el !== document &&\n        el.nodeName !== 'HTML' &&\n        el.nodeName !== 'BODY' &&\n        'classList' in el &&\n        'contains' in el.classList\n      ) {\n        return true;\n      }\n      return false;\n    }\n\n    /**\n     * Computes whether the given element should automatically trigger the\n     * `focus-visible` class being added, i.e. whether it should always match\n     * `:focus-visible` when focused.\n     * @param {Element} el\n     * @return {boolean}\n     */\n    function focusTriggersKeyboardModality(el) {\n      var type = el.type;\n      var tagName = el.tagName;\n\n      if (tagName === 'INPUT' && inputTypesAllowlist[type] && !el.readOnly) {\n        return true;\n      }\n\n      if (tagName === 'TEXTAREA' && !el.readOnly) {\n        return true;\n      }\n\n      if (el.isContentEditable) {\n        return true;\n      }\n\n      return false;\n    }\n\n    /**\n     * Add the `focus-visible` class to the given element if it was not added by\n     * the author.\n     * @param {Element} el\n     */\n    function addFocusVisibleClass(el) {\n      if (el.classList.contains('focus-visible')) {\n        return;\n      }\n      el.classList.add('focus-visible');\n      el.setAttribute('data-focus-visible-added', '');\n    }\n\n    /**\n     * Remove the `focus-visible` class from the given element if it was not\n     * originally added by the author.\n     * @param {Element} el\n     */\n    function removeFocusVisibleClass(el) {\n      if (!el.hasAttribute('data-focus-visible-added')) {\n        return;\n      }\n      el.classList.remove('focus-visible');\n      el.removeAttribute('data-focus-visible-added');\n    }\n\n    /**\n     * If the most recent user interaction was via the keyboard;\n     * and the key press did not include a meta, alt/option, or control key;\n     * then the modality is keyboard. Otherwise, the modality is not keyboard.\n     * Apply `focus-visible` to any current active element and keep track\n     * of our keyboard modality state with `hadKeyboardEvent`.\n     * @param {KeyboardEvent} e\n     */\n    function onKeyDown(e) {\n      if (e.metaKey || e.altKey || e.ctrlKey) {\n        return;\n      }\n\n      if (isValidFocusTarget(scope.activeElement)) {\n        addFocusVisibleClass(scope.activeElement);\n      }\n\n      hadKeyboardEvent = true;\n    }\n\n    /**\n     * If at any point a user clicks with a pointing device, ensure that we change\n     * the modality away from keyboard.\n     * This avoids the situation where a user presses a key on an already focused\n     * element, and then clicks on a different element, focusing it with a\n     * pointing device, while we still think we're in keyboard modality.\n     * @param {Event} e\n     */\n    function onPointerDown(e) {\n      hadKeyboardEvent = false;\n    }\n\n    /**\n     * On `focus`, add the `focus-visible` class to the target if:\n     * - the target received focus as a result of keyboard navigation, or\n     * - the event target is an element that will likely require interaction\n     *   via the keyboard (e.g. a text box)\n     * @param {Event} e\n     */\n    function onFocus(e) {\n      // Prevent IE from focusing the document or HTML element.\n      if (!isValidFocusTarget(e.target)) {\n        return;\n      }\n\n      if (hadKeyboardEvent || focusTriggersKeyboardModality(e.target)) {\n        addFocusVisibleClass(e.target);\n      }\n    }\n\n    /**\n     * On `blur`, remove the `focus-visible` class from the target.\n     * @param {Event} e\n     */\n    function onBlur(e) {\n      if (!isValidFocusTarget(e.target)) {\n        return;\n      }\n\n      if (\n        e.target.classList.contains('focus-visible') ||\n        e.target.hasAttribute('data-focus-visible-added')\n      ) {\n        // To detect a tab/window switch, we look for a blur event followed\n        // rapidly by a visibility change.\n        // If we don't see a visibility change within 100ms, it's probably a\n        // regular focus change.\n        hadFocusVisibleRecently = true;\n        window.clearTimeout(hadFocusVisibleRecentlyTimeout);\n        hadFocusVisibleRecentlyTimeout = window.setTimeout(function() {\n          hadFocusVisibleRecently = false;\n        }, 100);\n        removeFocusVisibleClass(e.target);\n      }\n    }\n\n    /**\n     * If the user changes tabs, keep track of whether or not the previously\n     * focused element had .focus-visible.\n     * @param {Event} e\n     */\n    function onVisibilityChange(e) {\n      if (document.visibilityState === 'hidden') {\n        // If the tab becomes active again, the browser will handle calling focus\n        // on the element (Safari actually calls it twice).\n        // If this tab change caused a blur on an element with focus-visible,\n        // re-apply the class when the user switches back to the tab.\n        if (hadFocusVisibleRecently) {\n          hadKeyboardEvent = true;\n        }\n        addInitialPointerMoveListeners();\n      }\n    }\n\n    /**\n     * Add a group of listeners to detect usage of any pointing devices.\n     * These listeners will be added when the polyfill first loads, and anytime\n     * the window is blurred, so that they are active when the window regains\n     * focus.\n     */\n    function addInitialPointerMoveListeners() {\n      document.addEventListener('mousemove', onInitialPointerMove);\n      document.addEventListener('mousedown', onInitialPointerMove);\n      document.addEventListener('mouseup', onInitialPointerMove);\n      document.addEventListener('pointermove', onInitialPointerMove);\n      document.addEventListener('pointerdown', onInitialPointerMove);\n      document.addEventListener('pointerup', onInitialPointerMove);\n      document.addEventListener('touchmove', onInitialPointerMove);\n      document.addEventListener('touchstart', onInitialPointerMove);\n      document.addEventListener('touchend', onInitialPointerMove);\n    }\n\n    function removeInitialPointerMoveListeners() {\n      document.removeEventListener('mousemove', onInitialPointerMove);\n      document.removeEventListener('mousedown', onInitialPointerMove);\n      document.removeEventListener('mouseup', onInitialPointerMove);\n      document.removeEventListener('pointermove', onInitialPointerMove);\n      document.removeEventListener('pointerdown', onInitialPointerMove);\n      document.removeEventListener('pointerup', onInitialPointerMove);\n      document.removeEventListener('touchmove', onInitialPointerMove);\n      document.removeEventListener('touchstart', onInitialPointerMove);\n      document.removeEventListener('touchend', onInitialPointerMove);\n    }\n\n    /**\n     * When the polfyill first loads, assume the user is in keyboard modality.\n     * If any event is received from a pointing device (e.g. mouse, pointer,\n     * touch), turn off keyboard modality.\n     * This accounts for situations where focus enters the page from the URL bar.\n     * @param {Event} e\n     */\n    function onInitialPointerMove(e) {\n      // Work around a Safari quirk that fires a mousemove on <html> whenever the\n      // window blurs, even if you're tabbing out of the page. \u00AF\\_(\u30C4)_/\u00AF\n      if (e.target.nodeName && e.target.nodeName.toLowerCase() === 'html') {\n        return;\n      }\n\n      hadKeyboardEvent = false;\n      removeInitialPointerMoveListeners();\n    }\n\n    // For some kinds of state, we are interested in changes at the global scope\n    // only. For example, global pointer input, global key presses and global\n    // visibility change should affect the state at every scope:\n    document.addEventListener('keydown', onKeyDown, true);\n    document.addEventListener('mousedown', onPointerDown, true);\n    document.addEventListener('pointerdown', onPointerDown, true);\n    document.addEventListener('touchstart', onPointerDown, true);\n    document.addEventListener('visibilitychange', onVisibilityChange, true);\n\n    addInitialPointerMoveListeners();\n\n    // For focus and blur, we specifically care about state changes in the local\n    // scope. This is because focus / blur events that originate from within a\n    // shadow root are not re-dispatched from the host element if it was already\n    // the active element in its own scope:\n    scope.addEventListener('focus', onFocus, true);\n    scope.addEventListener('blur', onBlur, true);\n\n    // We detect that a node is a ShadowRoot by ensuring that it is a\n    // DocumentFragment and also has a host property. This check covers native\n    // implementation and polyfill implementation transparently. If we only cared\n    // about the native implementation, we could just check if the scope was\n    // an instance of a ShadowRoot.\n    if (scope.nodeType === Node.DOCUMENT_FRAGMENT_NODE && scope.host) {\n      // Since a ShadowRoot is a special kind of DocumentFragment, it does not\n      // have a root element to add a class to. So, we add this attribute to the\n      // host element instead:\n      scope.host.setAttribute('data-js-focus-visible', '');\n    } else if (scope.nodeType === Node.DOCUMENT_NODE) {\n      document.documentElement.classList.add('js-focus-visible');\n      document.documentElement.setAttribute('data-js-focus-visible', '');\n    }\n  }\n\n  // It is important to wrap all references to global window and document in\n  // these checks to support server-side rendering use cases\n  // @see https://github.com/WICG/focus-visible/issues/199\n  if (typeof window !== 'undefined' && typeof document !== 'undefined') {\n    // Make the polyfill helper globally available. This can be used as a signal\n    // to interested libraries that wish to coordinate with the polyfill for e.g.,\n    // applying the polyfill to a shadow root:\n    window.applyFocusVisiblePolyfill = applyFocusVisiblePolyfill;\n\n    // Notify interested libraries of the polyfill's presence, in case the\n    // polyfill was loaded lazily:\n    var event;\n\n    try {\n      event = new CustomEvent('focus-visible-polyfill-ready');\n    } catch (error) {\n      // IE11 does not support using CustomEvent as a constructor directly:\n      event = document.createEvent('CustomEvent');\n      event.initCustomEvent('focus-visible-polyfill-ready', false, false, {});\n    }\n\n    window.dispatchEvent(event);\n  }\n\n  if (typeof document !== 'undefined') {\n    // Apply the polyfill to the global document, so that no JavaScript\n    // coordination is required to use the polyfill in the top-level document:\n    applyFocusVisiblePolyfill(document);\n  }\n\n})));\n", "/*!\n * clipboard.js v2.0.11\n * https://clipboardjs.com/\n *\n * Licensed MIT \u00A9 Zeno Rocha\n */\n(function webpackUniversalModuleDefinition(root, factory) {\n\tif(typeof exports === 'object' && typeof module === 'object')\n\t\tmodule.exports = factory();\n\telse if(typeof define === 'function' && define.amd)\n\t\tdefine([], factory);\n\telse if(typeof exports === 'object')\n\t\texports[\"ClipboardJS\"] = factory();\n\telse\n\t\troot[\"ClipboardJS\"] = factory();\n})(this, function() {\nreturn /******/ (function() { // webpackBootstrap\n/******/ \tvar __webpack_modules__ = ({\n\n/***/ 686:\n/***/ (function(__unused_webpack_module, __webpack_exports__, __webpack_require__) {\n\n\"use strict\";\n\n// EXPORTS\n__webpack_require__.d(__webpack_exports__, {\n  \"default\": function() { return /* binding */ clipboard; }\n});\n\n// EXTERNAL MODULE: ./node_modules/tiny-emitter/index.js\nvar tiny_emitter = __webpack_require__(279);\nvar tiny_emitter_default = /*#__PURE__*/__webpack_require__.n(tiny_emitter);\n// EXTERNAL MODULE: ./node_modules/good-listener/src/listen.js\nvar listen = __webpack_require__(370);\nvar listen_default = /*#__PURE__*/__webpack_require__.n(listen);\n// EXTERNAL MODULE: ./node_modules/select/src/select.js\nvar src_select = __webpack_require__(817);\nvar select_default = /*#__PURE__*/__webpack_require__.n(src_select);\n;// CONCATENATED MODULE: ./src/common/command.js\n/**\n * Executes a given operation type.\n * @param {String} type\n * @return {Boolean}\n */\nfunction command(type) {\n  try {\n    return document.execCommand(type);\n  } catch (err) {\n    return false;\n  }\n}\n;// CONCATENATED MODULE: ./src/actions/cut.js\n\n\n/**\n * Cut action wrapper.\n * @param {String|HTMLElement} target\n * @return {String}\n */\n\nvar ClipboardActionCut = function ClipboardActionCut(target) {\n  var selectedText = select_default()(target);\n  command('cut');\n  return selectedText;\n};\n\n/* harmony default export */ var actions_cut = (ClipboardActionCut);\n;// CONCATENATED MODULE: ./src/common/create-fake-element.js\n/**\n * Creates a fake textarea element with a value.\n * @param {String} value\n * @return {HTMLElement}\n */\nfunction createFakeElement(value) {\n  var isRTL = document.documentElement.getAttribute('dir') === 'rtl';\n  var fakeElement = document.createElement('textarea'); // Prevent zooming on iOS\n\n  fakeElement.style.fontSize = '12pt'; // Reset box model\n\n  fakeElement.style.border = '0';\n  fakeElement.style.padding = '0';\n  fakeElement.style.margin = '0'; // Move element out of screen horizontally\n\n  fakeElement.style.position = 'absolute';\n  fakeElement.style[isRTL ? 'right' : 'left'] = '-9999px'; // Move element to the same position vertically\n\n  var yPosition = window.pageYOffset || document.documentElement.scrollTop;\n  fakeElement.style.top = \"\".concat(yPosition, \"px\");\n  fakeElement.setAttribute('readonly', '');\n  fakeElement.value = value;\n  return fakeElement;\n}\n;// CONCATENATED MODULE: ./src/actions/copy.js\n\n\n\n/**\n * Create fake copy action wrapper using a fake element.\n * @param {String} target\n * @param {Object} options\n * @return {String}\n */\n\nvar fakeCopyAction = function fakeCopyAction(value, options) {\n  var fakeElement = createFakeElement(value);\n  options.container.appendChild(fakeElement);\n  var selectedText = select_default()(fakeElement);\n  command('copy');\n  fakeElement.remove();\n  return selectedText;\n};\n/**\n * Copy action wrapper.\n * @param {String|HTMLElement} target\n * @param {Object} options\n * @return {String}\n */\n\n\nvar ClipboardActionCopy = function ClipboardActionCopy(target) {\n  var options = arguments.length > 1 && arguments[1] !== undefined ? arguments[1] : {\n    container: document.body\n  };\n  var selectedText = '';\n\n  if (typeof target === 'string') {\n    selectedText = fakeCopyAction(target, options);\n  } else if (target instanceof HTMLInputElement && !['text', 'search', 'url', 'tel', 'password'].includes(target === null || target === void 0 ? void 0 : target.type)) {\n    // If input type doesn't support `setSelectionRange`. Simulate it. https://developer.mozilla.org/en-US/docs/Web/API/HTMLInputElement/setSelectionRange\n    selectedText = fakeCopyAction(target.value, options);\n  } else {\n    selectedText = select_default()(target);\n    command('copy');\n  }\n\n  return selectedText;\n};\n\n/* harmony default export */ var actions_copy = (ClipboardActionCopy);\n;// CONCATENATED MODULE: ./src/actions/default.js\nfunction _typeof(obj) { \"@babel/helpers - typeof\"; if (typeof Symbol === \"function\" && typeof Symbol.iterator === \"symbol\") { _typeof = function _typeof(obj) { return typeof obj; }; } else { _typeof = function _typeof(obj) { return obj && typeof Symbol === \"function\" && obj.constructor === Symbol && obj !== Symbol.prototype ? \"symbol\" : typeof obj; }; } return _typeof(obj); }\n\n\n\n/**\n * Inner function which performs selection from either `text` or `target`\n * properties and then executes copy or cut operations.\n * @param {Object} options\n */\n\nvar ClipboardActionDefault = function ClipboardActionDefault() {\n  var options = arguments.length > 0 && arguments[0] !== undefined ? arguments[0] : {};\n  // Defines base properties passed from constructor.\n  var _options$action = options.action,\n      action = _options$action === void 0 ? 'copy' : _options$action,\n      container = options.container,\n      target = options.target,\n      text = options.text; // Sets the `action` to be performed which can be either 'copy' or 'cut'.\n\n  if (action !== 'copy' && action !== 'cut') {\n    throw new Error('Invalid \"action\" value, use either \"copy\" or \"cut\"');\n  } // Sets the `target` property using an element that will be have its content copied.\n\n\n  if (target !== undefined) {\n    if (target && _typeof(target) === 'object' && target.nodeType === 1) {\n      if (action === 'copy' && target.hasAttribute('disabled')) {\n        throw new Error('Invalid \"target\" attribute. Please use \"readonly\" instead of \"disabled\" attribute');\n      }\n\n      if (action === 'cut' && (target.hasAttribute('readonly') || target.hasAttribute('disabled'))) {\n        throw new Error('Invalid \"target\" attribute. You can\\'t cut text from elements with \"readonly\" or \"disabled\" attributes');\n      }\n    } else {\n      throw new Error('Invalid \"target\" value, use a valid Element');\n    }\n  } // Define selection strategy based on `text` property.\n\n\n  if (text) {\n    return actions_copy(text, {\n      container: container\n    });\n  } // Defines which selection strategy based on `target` property.\n\n\n  if (target) {\n    return action === 'cut' ? actions_cut(target) : actions_copy(target, {\n      container: container\n    });\n  }\n};\n\n/* harmony default export */ var actions_default = (ClipboardActionDefault);\n;// CONCATENATED MODULE: ./src/clipboard.js\nfunction clipboard_typeof(obj) { \"@babel/helpers - typeof\"; if (typeof Symbol === \"function\" && typeof Symbol.iterator === \"symbol\") { clipboard_typeof = function _typeof(obj) { return typeof obj; }; } else { clipboard_typeof = function _typeof(obj) { return obj && typeof Symbol === \"function\" && obj.constructor === Symbol && obj !== Symbol.prototype ? \"symbol\" : typeof obj; }; } return clipboard_typeof(obj); }\n\nfunction _classCallCheck(instance, Constructor) { if (!(instance instanceof Constructor)) { throw new TypeError(\"Cannot call a class as a function\"); } }\n\nfunction _defineProperties(target, props) { for (var i = 0; i < props.length; i++) { var descriptor = props[i]; descriptor.enumerable = descriptor.enumerable || false; descriptor.configurable = true; if (\"value\" in descriptor) descriptor.writable = true; Object.defineProperty(target, descriptor.key, descriptor); } }\n\nfunction _createClass(Constructor, protoProps, staticProps) { if (protoProps) _defineProperties(Constructor.prototype, protoProps); if (staticProps) _defineProperties(Constructor, staticProps); return Constructor; }\n\nfunction _inherits(subClass, superClass) { if (typeof superClass !== \"function\" && superClass !== null) { throw new TypeError(\"Super expression must either be null or a function\"); } subClass.prototype = Object.create(superClass && superClass.prototype, { constructor: { value: subClass, writable: true, configurable: true } }); if (superClass) _setPrototypeOf(subClass, superClass); }\n\nfunction _setPrototypeOf(o, p) { _setPrototypeOf = Object.setPrototypeOf || function _setPrototypeOf(o, p) { o.__proto__ = p; return o; }; return _setPrototypeOf(o, p); }\n\nfunction _createSuper(Derived) { var hasNativeReflectConstruct = _isNativeReflectConstruct(); return function _createSuperInternal() { var Super = _getPrototypeOf(Derived), result; if (hasNativeReflectConstruct) { var NewTarget = _getPrototypeOf(this).constructor; result = Reflect.construct(Super, arguments, NewTarget); } else { result = Super.apply(this, arguments); } return _possibleConstructorReturn(this, result); }; }\n\nfunction _possibleConstructorReturn(self, call) { if (call && (clipboard_typeof(call) === \"object\" || typeof call === \"function\")) { return call; } return _assertThisInitialized(self); }\n\nfunction _assertThisInitialized(self) { if (self === void 0) { throw new ReferenceError(\"this hasn't been initialised - super() hasn't been called\"); } return self; }\n\nfunction _isNativeReflectConstruct() { if (typeof Reflect === \"undefined\" || !Reflect.construct) return false; if (Reflect.construct.sham) return false; if (typeof Proxy === \"function\") return true; try { Date.prototype.toString.call(Reflect.construct(Date, [], function () {})); return true; } catch (e) { return false; } }\n\nfunction _getPrototypeOf(o) { _getPrototypeOf = Object.setPrototypeOf ? Object.getPrototypeOf : function _getPrototypeOf(o) { return o.__proto__ || Object.getPrototypeOf(o); }; return _getPrototypeOf(o); }\n\n\n\n\n\n\n/**\n * Helper function to retrieve attribute value.\n * @param {String} suffix\n * @param {Element} element\n */\n\nfunction getAttributeValue(suffix, element) {\n  var attribute = \"data-clipboard-\".concat(suffix);\n\n  if (!element.hasAttribute(attribute)) {\n    return;\n  }\n\n  return element.getAttribute(attribute);\n}\n/**\n * Base class which takes one or more elements, adds event listeners to them,\n * and instantiates a new `ClipboardAction` on each click.\n */\n\n\nvar Clipboard = /*#__PURE__*/function (_Emitter) {\n  _inherits(Clipboard, _Emitter);\n\n  var _super = _createSuper(Clipboard);\n\n  /**\n   * @param {String|HTMLElement|HTMLCollection|NodeList} trigger\n   * @param {Object} options\n   */\n  function Clipboard(trigger, options) {\n    var _this;\n\n    _classCallCheck(this, Clipboard);\n\n    _this = _super.call(this);\n\n    _this.resolveOptions(options);\n\n    _this.listenClick(trigger);\n\n    return _this;\n  }\n  /**\n   * Defines if attributes would be resolved using internal setter functions\n   * or custom functions that were passed in the constructor.\n   * @param {Object} options\n   */\n\n\n  _createClass(Clipboard, [{\n    key: \"resolveOptions\",\n    value: function resolveOptions() {\n      var options = arguments.length > 0 && arguments[0] !== undefined ? arguments[0] : {};\n      this.action = typeof options.action === 'function' ? options.action : this.defaultAction;\n      this.target = typeof options.target === 'function' ? options.target : this.defaultTarget;\n      this.text = typeof options.text === 'function' ? options.text : this.defaultText;\n      this.container = clipboard_typeof(options.container) === 'object' ? options.container : document.body;\n    }\n    /**\n     * Adds a click event listener to the passed trigger.\n     * @param {String|HTMLElement|HTMLCollection|NodeList} trigger\n     */\n\n  }, {\n    key: \"listenClick\",\n    value: function listenClick(trigger) {\n      var _this2 = this;\n\n      this.listener = listen_default()(trigger, 'click', function (e) {\n        return _this2.onClick(e);\n      });\n    }\n    /**\n     * Defines a new `ClipboardAction` on each click event.\n     * @param {Event} e\n     */\n\n  }, {\n    key: \"onClick\",\n    value: function onClick(e) {\n      var trigger = e.delegateTarget || e.currentTarget;\n      var action = this.action(trigger) || 'copy';\n      var text = actions_default({\n        action: action,\n        container: this.container,\n        target: this.target(trigger),\n        text: this.text(trigger)\n      }); // Fires an event based on the copy operation result.\n\n      this.emit(text ? 'success' : 'error', {\n        action: action,\n        text: text,\n        trigger: trigger,\n        clearSelection: function clearSelection() {\n          if (trigger) {\n            trigger.focus();\n          }\n\n          window.getSelection().removeAllRanges();\n        }\n      });\n    }\n    /**\n     * Default `action` lookup function.\n     * @param {Element} trigger\n     */\n\n  }, {\n    key: \"defaultAction\",\n    value: function defaultAction(trigger) {\n      return getAttributeValue('action', trigger);\n    }\n    /**\n     * Default `target` lookup function.\n     * @param {Element} trigger\n     */\n\n  }, {\n    key: \"defaultTarget\",\n    value: function defaultTarget(trigger) {\n      var selector = getAttributeValue('target', trigger);\n\n      if (selector) {\n        return document.querySelector(selector);\n      }\n    }\n    /**\n     * Allow fire programmatically a copy action\n     * @param {String|HTMLElement} target\n     * @param {Object} options\n     * @returns Text copied.\n     */\n\n  }, {\n    key: \"defaultText\",\n\n    /**\n     * Default `text` lookup function.\n     * @param {Element} trigger\n     */\n    value: function defaultText(trigger) {\n      return getAttributeValue('text', trigger);\n    }\n    /**\n     * Destroy lifecycle.\n     */\n\n  }, {\n    key: \"destroy\",\n    value: function destroy() {\n      this.listener.destroy();\n    }\n  }], [{\n    key: \"copy\",\n    value: function copy(target) {\n      var options = arguments.length > 1 && arguments[1] !== undefined ? arguments[1] : {\n        container: document.body\n      };\n      return actions_copy(target, options);\n    }\n    /**\n     * Allow fire programmatically a cut action\n     * @param {String|HTMLElement} target\n     * @returns Text cutted.\n     */\n\n  }, {\n    key: \"cut\",\n    value: function cut(target) {\n      return actions_cut(target);\n    }\n    /**\n     * Returns the support of the given action, or all actions if no action is\n     * given.\n     * @param {String} [action]\n     */\n\n  }, {\n    key: \"isSupported\",\n    value: function isSupported() {\n      var action = arguments.length > 0 && arguments[0] !== undefined ? arguments[0] : ['copy', 'cut'];\n      var actions = typeof action === 'string' ? [action] : action;\n      var support = !!document.queryCommandSupported;\n      actions.forEach(function (action) {\n        support = support && !!document.queryCommandSupported(action);\n      });\n      return support;\n    }\n  }]);\n\n  return Clipboard;\n}((tiny_emitter_default()));\n\n/* harmony default export */ var clipboard = (Clipboard);\n\n/***/ }),\n\n/***/ 828:\n/***/ (function(module) {\n\nvar DOCUMENT_NODE_TYPE = 9;\n\n/**\n * A polyfill for Element.matches()\n */\nif (typeof Element !== 'undefined' && !Element.prototype.matches) {\n    var proto = Element.prototype;\n\n    proto.matches = proto.matchesSelector ||\n                    proto.mozMatchesSelector ||\n                    proto.msMatchesSelector ||\n                    proto.oMatchesSelector ||\n                    proto.webkitMatchesSelector;\n}\n\n/**\n * Finds the closest parent that matches a selector.\n *\n * @param {Element} element\n * @param {String} selector\n * @return {Function}\n */\nfunction closest (element, selector) {\n    while (element && element.nodeType !== DOCUMENT_NODE_TYPE) {\n        if (typeof element.matches === 'function' &&\n            element.matches(selector)) {\n          return element;\n        }\n        element = element.parentNode;\n    }\n}\n\nmodule.exports = closest;\n\n\n/***/ }),\n\n/***/ 438:\n/***/ (function(module, __unused_webpack_exports, __webpack_require__) {\n\nvar closest = __webpack_require__(828);\n\n/**\n * Delegates event to a selector.\n *\n * @param {Element} element\n * @param {String} selector\n * @param {String} type\n * @param {Function} callback\n * @param {Boolean} useCapture\n * @return {Object}\n */\nfunction _delegate(element, selector, type, callback, useCapture) {\n    var listenerFn = listener.apply(this, arguments);\n\n    element.addEventListener(type, listenerFn, useCapture);\n\n    return {\n        destroy: function() {\n            element.removeEventListener(type, listenerFn, useCapture);\n        }\n    }\n}\n\n/**\n * Delegates event to a selector.\n *\n * @param {Element|String|Array} [elements]\n * @param {String} selector\n * @param {String} type\n * @param {Function} callback\n * @param {Boolean} useCapture\n * @return {Object}\n */\nfunction delegate(elements, selector, type, callback, useCapture) {\n    // Handle the regular Element usage\n    if (typeof elements.addEventListener === 'function') {\n        return _delegate.apply(null, arguments);\n    }\n\n    // Handle Element-less usage, it defaults to global delegation\n    if (typeof type === 'function') {\n        // Use `document` as the first parameter, then apply arguments\n        // This is a short way to .unshift `arguments` without running into deoptimizations\n        return _delegate.bind(null, document).apply(null, arguments);\n    }\n\n    // Handle Selector-based usage\n    if (typeof elements === 'string') {\n        elements = document.querySelectorAll(elements);\n    }\n\n    // Handle Array-like based usage\n    return Array.prototype.map.call(elements, function (element) {\n        return _delegate(element, selector, type, callback, useCapture);\n    });\n}\n\n/**\n * Finds closest match and invokes callback.\n *\n * @param {Element} element\n * @param {String} selector\n * @param {String} type\n * @param {Function} callback\n * @return {Function}\n */\nfunction listener(element, selector, type, callback) {\n    return function(e) {\n        e.delegateTarget = closest(e.target, selector);\n\n        if (e.delegateTarget) {\n            callback.call(element, e);\n        }\n    }\n}\n\nmodule.exports = delegate;\n\n\n/***/ }),\n\n/***/ 879:\n/***/ (function(__unused_webpack_module, exports) {\n\n/**\n * Check if argument is a HTML element.\n *\n * @param {Object} value\n * @return {Boolean}\n */\nexports.node = function(value) {\n    return value !== undefined\n        && value instanceof HTMLElement\n        && value.nodeType === 1;\n};\n\n/**\n * Check if argument is a list of HTML elements.\n *\n * @param {Object} value\n * @return {Boolean}\n */\nexports.nodeList = function(value) {\n    var type = Object.prototype.toString.call(value);\n\n    return value !== undefined\n        && (type === '[object NodeList]' || type === '[object HTMLCollection]')\n        && ('length' in value)\n        && (value.length === 0 || exports.node(value[0]));\n};\n\n/**\n * Check if argument is a string.\n *\n * @param {Object} value\n * @return {Boolean}\n */\nexports.string = function(value) {\n    return typeof value === 'string'\n        || value instanceof String;\n};\n\n/**\n * Check if argument is a function.\n *\n * @param {Object} value\n * @return {Boolean}\n */\nexports.fn = function(value) {\n    var type = Object.prototype.toString.call(value);\n\n    return type === '[object Function]';\n};\n\n\n/***/ }),\n\n/***/ 370:\n/***/ (function(module, __unused_webpack_exports, __webpack_require__) {\n\nvar is = __webpack_require__(879);\nvar delegate = __webpack_require__(438);\n\n/**\n * Validates all params and calls the right\n * listener function based on its target type.\n *\n * @param {String|HTMLElement|HTMLCollection|NodeList} target\n * @param {String} type\n * @param {Function} callback\n * @return {Object}\n */\nfunction listen(target, type, callback) {\n    if (!target && !type && !callback) {\n        throw new Error('Missing required arguments');\n    }\n\n    if (!is.string(type)) {\n        throw new TypeError('Second argument must be a String');\n    }\n\n    if (!is.fn(callback)) {\n        throw new TypeError('Third argument must be a Function');\n    }\n\n    if (is.node(target)) {\n        return listenNode(target, type, callback);\n    }\n    else if (is.nodeList(target)) {\n        return listenNodeList(target, type, callback);\n    }\n    else if (is.string(target)) {\n        return listenSelector(target, type, callback);\n    }\n    else {\n        throw new TypeError('First argument must be a String, HTMLElement, HTMLCollection, or NodeList');\n    }\n}\n\n/**\n * Adds an event listener to a HTML element\n * and returns a remove listener function.\n *\n * @param {HTMLElement} node\n * @param {String} type\n * @param {Function} callback\n * @return {Object}\n */\nfunction listenNode(node, type, callback) {\n    node.addEventListener(type, callback);\n\n    return {\n        destroy: function() {\n            node.removeEventListener(type, callback);\n        }\n    }\n}\n\n/**\n * Add an event listener to a list of HTML elements\n * and returns a remove listener function.\n *\n * @param {NodeList|HTMLCollection} nodeList\n * @param {String} type\n * @param {Function} callback\n * @return {Object}\n */\nfunction listenNodeList(nodeList, type, callback) {\n    Array.prototype.forEach.call(nodeList, function(node) {\n        node.addEventListener(type, callback);\n    });\n\n    return {\n        destroy: function() {\n            Array.prototype.forEach.call(nodeList, function(node) {\n                node.removeEventListener(type, callback);\n            });\n        }\n    }\n}\n\n/**\n * Add an event listener to a selector\n * and returns a remove listener function.\n *\n * @param {String} selector\n * @param {String} type\n * @param {Function} callback\n * @return {Object}\n */\nfunction listenSelector(selector, type, callback) {\n    return delegate(document.body, selector, type, callback);\n}\n\nmodule.exports = listen;\n\n\n/***/ }),\n\n/***/ 817:\n/***/ (function(module) {\n\nfunction select(element) {\n    var selectedText;\n\n    if (element.nodeName === 'SELECT') {\n        element.focus();\n\n        selectedText = element.value;\n    }\n    else if (element.nodeName === 'INPUT' || element.nodeName === 'TEXTAREA') {\n        var isReadOnly = element.hasAttribute('readonly');\n\n        if (!isReadOnly) {\n            element.setAttribute('readonly', '');\n        }\n\n        element.select();\n        element.setSelectionRange(0, element.value.length);\n\n        if (!isReadOnly) {\n            element.removeAttribute('readonly');\n        }\n\n        selectedText = element.value;\n    }\n    else {\n        if (element.hasAttribute('contenteditable')) {\n            element.focus();\n        }\n\n        var selection = window.getSelection();\n        var range = document.createRange();\n\n        range.selectNodeContents(element);\n        selection.removeAllRanges();\n        selection.addRange(range);\n\n        selectedText = selection.toString();\n    }\n\n    return selectedText;\n}\n\nmodule.exports = select;\n\n\n/***/ }),\n\n/***/ 279:\n/***/ (function(module) {\n\nfunction E () {\n  // Keep this empty so it's easier to inherit from\n  // (via https://github.com/lipsmack from https://github.com/scottcorgan/tiny-emitter/issues/3)\n}\n\nE.prototype = {\n  on: function (name, callback, ctx) {\n    var e = this.e || (this.e = {});\n\n    (e[name] || (e[name] = [])).push({\n      fn: callback,\n      ctx: ctx\n    });\n\n    return this;\n  },\n\n  once: function (name, callback, ctx) {\n    var self = this;\n    function listener () {\n      self.off(name, listener);\n      callback.apply(ctx, arguments);\n    };\n\n    listener._ = callback\n    return this.on(name, listener, ctx);\n  },\n\n  emit: function (name) {\n    var data = [].slice.call(arguments, 1);\n    var evtArr = ((this.e || (this.e = {}))[name] || []).slice();\n    var i = 0;\n    var len = evtArr.length;\n\n    for (i; i < len; i++) {\n      evtArr[i].fn.apply(evtArr[i].ctx, data);\n    }\n\n    return this;\n  },\n\n  off: function (name, callback) {\n    var e = this.e || (this.e = {});\n    var evts = e[name];\n    var liveEvents = [];\n\n    if (evts && callback) {\n      for (var i = 0, len = evts.length; i < len; i++) {\n        if (evts[i].fn !== callback && evts[i].fn._ !== callback)\n          liveEvents.push(evts[i]);\n      }\n    }\n\n    // Remove event from queue to prevent memory leak\n    // Suggested by https://github.com/lazd\n    // Ref: https://github.com/scottcorgan/tiny-emitter/commit/c6ebfaa9bc973b33d110a84a307742b7cf94c953#commitcomment-5024910\n\n    (liveEvents.length)\n      ? e[name] = liveEvents\n      : delete e[name];\n\n    return this;\n  }\n};\n\nmodule.exports = E;\nmodule.exports.TinyEmitter = E;\n\n\n/***/ })\n\n/******/ \t});\n/************************************************************************/\n/******/ \t// The module cache\n/******/ \tvar __webpack_module_cache__ = {};\n/******/ \t\n/******/ \t// The require function\n/******/ \tfunction __webpack_require__(moduleId) {\n/******/ \t\t// Check if module is in cache\n/******/ \t\tif(__webpack_module_cache__[moduleId]) {\n/******/ \t\t\treturn __webpack_module_cache__[moduleId].exports;\n/******/ \t\t}\n/******/ \t\t// Create a new module (and put it into the cache)\n/******/ \t\tvar module = __webpack_module_cache__[moduleId] = {\n/******/ \t\t\t// no module.id needed\n/******/ \t\t\t// no module.loaded needed\n/******/ \t\t\texports: {}\n/******/ \t\t};\n/******/ \t\n/******/ \t\t// Execute the module function\n/******/ \t\t__webpack_modules__[moduleId](module, module.exports, __webpack_require__);\n/******/ \t\n/******/ \t\t// Return the exports of the module\n/******/ \t\treturn module.exports;\n/******/ \t}\n/******/ \t\n/************************************************************************/\n/******/ \t/* webpack/runtime/compat get default export */\n/******/ \t!function() {\n/******/ \t\t// getDefaultExport function for compatibility with non-harmony modules\n/******/ \t\t__webpack_require__.n = function(module) {\n/******/ \t\t\tvar getter = module && module.__esModule ?\n/******/ \t\t\t\tfunction() { return module['default']; } :\n/******/ \t\t\t\tfunction() { return module; };\n/******/ \t\t\t__webpack_require__.d(getter, { a: getter });\n/******/ \t\t\treturn getter;\n/******/ \t\t};\n/******/ \t}();\n/******/ \t\n/******/ \t/* webpack/runtime/define property getters */\n/******/ \t!function() {\n/******/ \t\t// define getter functions for harmony exports\n/******/ \t\t__webpack_require__.d = function(exports, definition) {\n/******/ \t\t\tfor(var key in definition) {\n/******/ \t\t\t\tif(__webpack_require__.o(definition, key) && !__webpack_require__.o(exports, key)) {\n/******/ \t\t\t\t\tObject.defineProperty(exports, key, { enumerable: true, get: definition[key] });\n/******/ \t\t\t\t}\n/******/ \t\t\t}\n/******/ \t\t};\n/******/ \t}();\n/******/ \t\n/******/ \t/* webpack/runtime/hasOwnProperty shorthand */\n/******/ \t!function() {\n/******/ \t\t__webpack_require__.o = function(obj, prop) { return Object.prototype.hasOwnProperty.call(obj, prop); }\n/******/ \t}();\n/******/ \t\n/************************************************************************/\n/******/ \t// module exports must be returned from runtime so entry inlining is disabled\n/******/ \t// startup\n/******/ \t// Load entry module and return exports\n/******/ \treturn __webpack_require__(686);\n/******/ })()\n.default;\n});", "/*!\n * escape-html\n * Copyright(c) 2012-2013 TJ Holowaychuk\n * Copyright(c) 2015 Andreas Lubbe\n * Copyright(c) 2015 Tiancheng \"Timothy\" Gu\n * MIT Licensed\n */\n\n'use strict';\n\n/**\n * Module variables.\n * @private\n */\n\nvar matchHtmlRegExp = /[\"'&<>]/;\n\n/**\n * Module exports.\n * @public\n */\n\nmodule.exports = escapeHtml;\n\n/**\n * Escape special characters in the given string of html.\n *\n * @param  {string} string The string to escape for inserting into HTML\n * @return {string}\n * @public\n */\n\nfunction escapeHtml(string) {\n  var str = '' + string;\n  var match = matchHtmlRegExp.exec(str);\n\n  if (!match) {\n    return str;\n  }\n\n  var escape;\n  var html = '';\n  var index = 0;\n  var lastIndex = 0;\n\n  for (index = match.index; index < str.length; index++) {\n    switch (str.charCodeAt(index)) {\n      case 34: // \"\n        escape = '&quot;';\n        break;\n      case 38: // &\n        escape = '&amp;';\n        break;\n      case 39: // '\n        escape = '&#39;';\n        break;\n      case 60: // <\n        escape = '&lt;';\n        break;\n      case 62: // >\n        escape = '&gt;';\n        break;\n      default:\n        continue;\n    }\n\n    if (lastIndex !== index) {\n      html += str.substring(lastIndex, index);\n    }\n\n    lastIndex = index + 1;\n    html += escape;\n  }\n\n  return lastIndex !== index\n    ? html + str.substring(lastIndex, index)\n    : html;\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport \"focus-visible\"\n\nimport {\n  EMPTY,\n  NEVER,\n  Observable,\n  Subject,\n  defer,\n  delay,\n  filter,\n  map,\n  merge,\n  mergeWith,\n  shareReplay,\n  switchMap\n} from \"rxjs\"\n\nimport { configuration, feature } from \"./_\"\nimport {\n  at,\n  getActiveElement,\n  getOptionalElement,\n  requestJSON,\n  setLocation,\n  setToggle,\n  watchDocument,\n  watchKeyboard,\n  watchLocation,\n  watchLocationTarget,\n  watchMedia,\n  watchPrint,\n  watchScript,\n  watchViewport\n} from \"./browser\"\nimport {\n  getComponentElement,\n  getComponentElements,\n  mountAnnounce,\n  mountBackToTop,\n  mountConsent,\n  mountContent,\n  mountDialog,\n  mountHeader,\n  mountHeaderTitle,\n  mountPalette,\n  mountProgress,\n  mountSearch,\n  mountSearchHiglight,\n  mountSidebar,\n  mountSource,\n  mountTableOfContents,\n  mountTabs,\n  watchHeader,\n  watchMain\n} from \"./components\"\nimport {\n  SearchIndex,\n  setupClipboardJS,\n  setupInstantNavigation,\n  setupVersionSelector\n} from \"./integrations\"\nimport {\n  patchEllipsis,\n  patchIndeterminate,\n  patchScrollfix,\n  patchScrolllock\n} from \"./patches\"\nimport \"./polyfills\"\n\n/* ----------------------------------------------------------------------------\n * Functions - @todo refactor\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch search index\n *\n * @returns Search index observable\n */\nfunction fetchSearchIndex(): Observable<SearchIndex> {\n  if (location.protocol === \"file:\") {\n    return watchScript(\n      `${new URL(\"search/search_index.js\", config.base)}`\n    )\n      .pipe(\n        // @ts-ignore - @todo fix typings\n        map(() => __index),\n        shareReplay(1)\n      )\n  } else {\n    return requestJSON<SearchIndex>(\n      new URL(\"search/search_index.json\", config.base)\n    )\n  }\n}\n\n/* ----------------------------------------------------------------------------\n * Application\n * ------------------------------------------------------------------------- */\n\n/* Yay, JavaScript is available */\ndocument.documentElement.classList.remove(\"no-js\")\ndocument.documentElement.classList.add(\"js\")\n\n/* Set up navigation observables and subjects */\nconst document$ = watchDocument()\nconst location$ = watchLocation()\nconst target$   = watchLocationTarget(location$)\nconst keyboard$ = watchKeyboard()\n\n/* Set up media observables */\nconst viewport$ = watchViewport()\nconst tablet$   = watchMedia(\"(min-width: 960px)\")\nconst screen$   = watchMedia(\"(min-width: 1220px)\")\nconst print$    = watchPrint()\n\n/* Retrieve search index, if search is enabled */\nconst config = configuration()\nconst index$ = document.forms.namedItem(\"search\")\n  ? fetchSearchIndex()\n  : NEVER\n\n/* Set up Clipboard.js integration */\nconst alert$ = new Subject<string>()\nsetupClipboardJS({ alert$ })\n\n/* Set up progress indicator */\nconst progress$ = new Subject<number>()\n\n/* Set up instant navigation, if enabled */\nif (feature(\"navigation.instant\"))\n  setupInstantNavigation({ location$, viewport$, progress$ })\n    .subscribe(document$)\n\n/* Set up version selector */\nif (config.version?.provider === \"mike\")\n  setupVersionSelector({ document$ })\n\n/* Always close drawer and search on navigation */\nmerge(location$, target$)\n  .pipe(\n    delay(125)\n  )\n    .subscribe(() => {\n      setToggle(\"drawer\", false)\n      setToggle(\"search\", false)\n    })\n\n/* Set up global keyboard handlers */\nkeyboard$\n  .pipe(\n    filter(({ mode }) => mode === \"global\")\n  )\n    .subscribe(key => {\n      switch (key.type) {\n\n        /* Go to previous page */\n        case \"p\":\n        case \",\":\n          const prev = getOptionalElement<HTMLLinkElement>(\"link[rel=prev]\")\n          if (typeof prev !== \"undefined\")\n            setLocation(prev)\n          break\n\n        /* Go to next page */\n        case \"n\":\n        case \".\":\n          const next = getOptionalElement<HTMLLinkElement>(\"link[rel=next]\")\n          if (typeof next !== \"undefined\")\n            setLocation(next)\n          break\n\n        /* Expand navigation, see https://bit.ly/3ZjG5io */\n        case \"Enter\":\n          const active = getActiveElement()\n          if (active instanceof HTMLLabelElement)\n            active.click()\n      }\n    })\n\n/* Set up patches */\npatchEllipsis({ document$ })\npatchIndeterminate({ document$, tablet$ })\npatchScrollfix({ document$ })\npatchScrolllock({ viewport$, tablet$ })\n\n/* Set up header and main area observable */\nconst header$ = watchHeader(getComponentElement(\"header\"), { viewport$ })\nconst main$ = document$\n  .pipe(\n    map(() => getComponentElement(\"main\")),\n    switchMap(el => watchMain(el, { viewport$, header$ })),\n    shareReplay(1)\n  )\n\n/* Set up control component observables */\nconst control$ = merge(\n\n  /* Consent */\n  ...getComponentElements(\"consent\")\n    .map(el => mountConsent(el, { target$ })),\n\n  /* Dialog */\n  ...getComponentElements(\"dialog\")\n    .map(el => mountDialog(el, { alert$ })),\n\n  /* Header */\n  ...getComponentElements(\"header\")\n    .map(el => mountHeader(el, { viewport$, header$, main$ })),\n\n  /* Color palette */\n  ...getComponentElements(\"palette\")\n    .map(el => mountPalette(el)),\n\n  /* Progress bar */\n  ...getComponentElements(\"progress\")\n    .map(el => mountProgress(el, { progress$ })),\n\n  /* Search */\n  ...getComponentElements(\"search\")\n    .map(el => mountSearch(el, { index$, keyboard$ })),\n\n  /* Repository information */\n  ...getComponentElements(\"source\")\n    .map(el => mountSource(el))\n)\n\n/* Set up content component observables */\nconst content$ = defer(() => merge(\n\n  /* Announcement bar */\n  ...getComponentElements(\"announce\")\n    .map(el => mountAnnounce(el)),\n\n  /* Content */\n  ...getComponentElements(\"content\")\n    .map(el => mountContent(el, { viewport$, target$, print$ })),\n\n  /* Search highlighting */\n  ...getComponentElements(\"content\")\n    .map(el => feature(\"search.highlight\")\n      ? mountSearchHiglight(el, { index$, location$ })\n      : EMPTY\n    ),\n\n  /* Header title */\n  ...getComponentElements(\"header-title\")\n    .map(el => mountHeaderTitle(el, { viewport$, header$ })),\n\n  /* Sidebar */\n  ...getComponentElements(\"sidebar\")\n    .map(el => el.getAttribute(\"data-md-type\") === \"navigation\"\n      ? at(screen$, () => mountSidebar(el, { viewport$, header$, main$ }))\n      : at(tablet$, () => mountSidebar(el, { viewport$, header$, main$ }))\n    ),\n\n  /* Navigation tabs */\n  ...getComponentElements(\"tabs\")\n    .map(el => mountTabs(el, { viewport$, header$ })),\n\n  /* Table of contents */\n  ...getComponentElements(\"toc\")\n    .map(el => mountTableOfContents(el, {\n      viewport$, header$, main$, target$\n    })),\n\n  /* Back-to-top button */\n  ...getComponentElements(\"top\")\n    .map(el => mountBackToTop(el, { viewport$, header$, main$, target$ }))\n))\n\n/* Set up component observables */\nconst component$ = document$\n  .pipe(\n    switchMap(() => content$),\n    mergeWith(control$),\n    shareReplay(1)\n  )\n\n/* Subscribe to all components */\ncomponent$.subscribe()\n\n/* ----------------------------------------------------------------------------\n * Exports\n * ------------------------------------------------------------------------- */\n\nwindow.document$  = document$          /* Document observable */\nwindow.location$  = location$          /* Location subject */\nwindow.target$    = target$            /* Location target observable */\nwindow.keyboard$  = keyboard$          /* Keyboard observable */\nwindow.viewport$  = viewport$          /* Viewport observable */\nwindow.tablet$    = tablet$            /* Media tablet observable */\nwindow.screen$    = screen$            /* Media screen observable */\nwindow.print$     = print$             /* Media print observable */\nwindow.alert$     = alert$             /* Alert subject */\nwindow.progress$  = progress$          /* Progress indicator subject */\nwindow.component$ = component$         /* Component observable */\n", "/*! *****************************************************************************\r\nCopyright (c) Microsoft Corporation.\r\n\r\nPermission to use, copy, modify, and/or distribute this software for any\r\npurpose with or without fee is hereby granted.\r\n\r\nTHE SOFTWARE IS PROVIDED \"AS IS\" AND THE AUTHOR DISCLAIMS ALL WARRANTIES WITH\r\nREGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY\r\nAND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT,\r\nINDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM\r\nLOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR\r\nOTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR\r\nPERFORMANCE OF THIS SOFTWARE.\r\n***************************************************************************** */\r\n/* global Reflect, Promise */\r\n\r\nvar extendStatics = function(d, b) {\r\n    extendStatics = Object.setPrototypeOf ||\r\n        ({ __proto__: [] } instanceof Array && function (d, b) { d.__proto__ = b; }) ||\r\n        function (d, b) { for (var p in b) if (Object.prototype.hasOwnProperty.call(b, p)) d[p] = b[p]; };\r\n    return extendStatics(d, b);\r\n};\r\n\r\nexport function __extends(d, b) {\r\n    if (typeof b !== \"function\" && b !== null)\r\n        throw new TypeError(\"Class extends value \" + String(b) + \" is not a constructor or null\");\r\n    extendStatics(d, b);\r\n    function __() { this.constructor = d; }\r\n    d.prototype = b === null ? Object.create(b) : (__.prototype = b.prototype, new __());\r\n}\r\n\r\nexport var __assign = function() {\r\n    __assign = Object.assign || function __assign(t) {\r\n        for (var s, i = 1, n = arguments.length; i < n; i++) {\r\n            s = arguments[i];\r\n            for (var p in s) if (Object.prototype.hasOwnProperty.call(s, p)) t[p] = s[p];\r\n        }\r\n        return t;\r\n    }\r\n    return __assign.apply(this, arguments);\r\n}\r\n\r\nexport function __rest(s, e) {\r\n    var t = {};\r\n    for (var p in s) if (Object.prototype.hasOwnProperty.call(s, p) && e.indexOf(p) < 0)\r\n        t[p] = s[p];\r\n    if (s != null && typeof Object.getOwnPropertySymbols === \"function\")\r\n        for (var i = 0, p = Object.getOwnPropertySymbols(s); i < p.length; i++) {\r\n            if (e.indexOf(p[i]) < 0 && Object.prototype.propertyIsEnumerable.call(s, p[i]))\r\n                t[p[i]] = s[p[i]];\r\n        }\r\n    return t;\r\n}\r\n\r\nexport function __decorate(decorators, target, key, desc) {\r\n    var c = arguments.length, r = c < 3 ? target : desc === null ? desc = Object.getOwnPropertyDescriptor(target, key) : desc, d;\r\n    if (typeof Reflect === \"object\" && typeof Reflect.decorate === \"function\") r = Reflect.decorate(decorators, target, key, desc);\r\n    else for (var i = decorators.length - 1; i >= 0; i--) if (d = decorators[i]) r = (c < 3 ? d(r) : c > 3 ? d(target, key, r) : d(target, key)) || r;\r\n    return c > 3 && r && Object.defineProperty(target, key, r), r;\r\n}\r\n\r\nexport function __param(paramIndex, decorator) {\r\n    return function (target, key) { decorator(target, key, paramIndex); }\r\n}\r\n\r\nexport function __metadata(metadataKey, metadataValue) {\r\n    if (typeof Reflect === \"object\" && typeof Reflect.metadata === \"function\") return Reflect.metadata(metadataKey, metadataValue);\r\n}\r\n\r\nexport function __awaiter(thisArg, _arguments, P, generator) {\r\n    function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }\r\n    return new (P || (P = Promise))(function (resolve, reject) {\r\n        function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }\r\n        function rejected(value) { try { step(generator[\"throw\"](value)); } catch (e) { reject(e); } }\r\n        function step(result) { result.done ? resolve(result.value) : adopt(result.value).then(fulfilled, rejected); }\r\n        step((generator = generator.apply(thisArg, _arguments || [])).next());\r\n    });\r\n}\r\n\r\nexport function __generator(thisArg, body) {\r\n    var _ = { label: 0, sent: function() { if (t[0] & 1) throw t[1]; return t[1]; }, trys: [], ops: [] }, f, y, t, g;\r\n    return g = { next: verb(0), \"throw\": verb(1), \"return\": verb(2) }, typeof Symbol === \"function\" && (g[Symbol.iterator] = function() { return this; }), g;\r\n    function verb(n) { return function (v) { return step([n, v]); }; }\r\n    function step(op) {\r\n        if (f) throw new TypeError(\"Generator is already executing.\");\r\n        while (_) try {\r\n            if (f = 1, y && (t = op[0] & 2 ? y[\"return\"] : op[0] ? y[\"throw\"] || ((t = y[\"return\"]) && t.call(y), 0) : y.next) && !(t = t.call(y, op[1])).done) return t;\r\n            if (y = 0, t) op = [op[0] & 2, t.value];\r\n            switch (op[0]) {\r\n                case 0: case 1: t = op; break;\r\n                case 4: _.label++; return { value: op[1], done: false };\r\n                case 5: _.label++; y = op[1]; op = [0]; continue;\r\n                case 7: op = _.ops.pop(); _.trys.pop(); continue;\r\n                default:\r\n                    if (!(t = _.trys, t = t.length > 0 && t[t.length - 1]) && (op[0] === 6 || op[0] === 2)) { _ = 0; continue; }\r\n                    if (op[0] === 3 && (!t || (op[1] > t[0] && op[1] < t[3]))) { _.label = op[1]; break; }\r\n                    if (op[0] === 6 && _.label < t[1]) { _.label = t[1]; t = op; break; }\r\n                    if (t && _.label < t[2]) { _.label = t[2]; _.ops.push(op); break; }\r\n                    if (t[2]) _.ops.pop();\r\n                    _.trys.pop(); continue;\r\n            }\r\n            op = body.call(thisArg, _);\r\n        } catch (e) { op = [6, e]; y = 0; } finally { f = t = 0; }\r\n        if (op[0] & 5) throw op[1]; return { value: op[0] ? op[1] : void 0, done: true };\r\n    }\r\n}\r\n\r\nexport var __createBinding = Object.create ? (function(o, m, k, k2) {\r\n    if (k2 === undefined) k2 = k;\r\n    Object.defineProperty(o, k2, { enumerable: true, get: function() { return m[k]; } });\r\n}) : (function(o, m, k, k2) {\r\n    if (k2 === undefined) k2 = k;\r\n    o[k2] = m[k];\r\n});\r\n\r\nexport function __exportStar(m, o) {\r\n    for (var p in m) if (p !== \"default\" && !Object.prototype.hasOwnProperty.call(o, p)) __createBinding(o, m, p);\r\n}\r\n\r\nexport function __values(o) {\r\n    var s = typeof Symbol === \"function\" && Symbol.iterator, m = s && o[s], i = 0;\r\n    if (m) return m.call(o);\r\n    if (o && typeof o.length === \"number\") return {\r\n        next: function () {\r\n            if (o && i >= o.length) o = void 0;\r\n            return { value: o && o[i++], done: !o };\r\n        }\r\n    };\r\n    throw new TypeError(s ? \"Object is not iterable.\" : \"Symbol.iterator is not defined.\");\r\n}\r\n\r\nexport function __read(o, n) {\r\n    var m = typeof Symbol === \"function\" && o[Symbol.iterator];\r\n    if (!m) return o;\r\n    var i = m.call(o), r, ar = [], e;\r\n    try {\r\n        while ((n === void 0 || n-- > 0) && !(r = i.next()).done) ar.push(r.value);\r\n    }\r\n    catch (error) { e = { error: error }; }\r\n    finally {\r\n        try {\r\n            if (r && !r.done && (m = i[\"return\"])) m.call(i);\r\n        }\r\n        finally { if (e) throw e.error; }\r\n    }\r\n    return ar;\r\n}\r\n\r\n/** @deprecated */\r\nexport function __spread() {\r\n    for (var ar = [], i = 0; i < arguments.length; i++)\r\n        ar = ar.concat(__read(arguments[i]));\r\n    return ar;\r\n}\r\n\r\n/** @deprecated */\r\nexport function __spreadArrays() {\r\n    for (var s = 0, i = 0, il = arguments.length; i < il; i++) s += arguments[i].length;\r\n    for (var r = Array(s), k = 0, i = 0; i < il; i++)\r\n        for (var a = arguments[i], j = 0, jl = a.length; j < jl; j++, k++)\r\n            r[k] = a[j];\r\n    return r;\r\n}\r\n\r\nexport function __spreadArray(to, from, pack) {\r\n    if (pack || arguments.length === 2) for (var i = 0, l = from.length, ar; i < l; i++) {\r\n        if (ar || !(i in from)) {\r\n            if (!ar) ar = Array.prototype.slice.call(from, 0, i);\r\n            ar[i] = from[i];\r\n        }\r\n    }\r\n    return to.concat(ar || Array.prototype.slice.call(from));\r\n}\r\n\r\nexport function __await(v) {\r\n    return this instanceof __await ? (this.v = v, this) : new __await(v);\r\n}\r\n\r\nexport function __asyncGenerator(thisArg, _arguments, generator) {\r\n    if (!Symbol.asyncIterator) throw new TypeError(\"Symbol.asyncIterator is not defined.\");\r\n    var g = generator.apply(thisArg, _arguments || []), i, q = [];\r\n    return i = {}, verb(\"next\"), verb(\"throw\"), verb(\"return\"), i[Symbol.asyncIterator] = function () { return this; }, i;\r\n    function verb(n) { if (g[n]) i[n] = function (v) { return new Promise(function (a, b) { q.push([n, v, a, b]) > 1 || resume(n, v); }); }; }\r\n    function resume(n, v) { try { step(g[n](v)); } catch (e) { settle(q[0][3], e); } }\r\n    function step(r) { r.value instanceof __await ? Promise.resolve(r.value.v).then(fulfill, reject) : settle(q[0][2], r); }\r\n    function fulfill(value) { resume(\"next\", value); }\r\n    function reject(value) { resume(\"throw\", value); }\r\n    function settle(f, v) { if (f(v), q.shift(), q.length) resume(q[0][0], q[0][1]); }\r\n}\r\n\r\nexport function __asyncDelegator(o) {\r\n    var i, p;\r\n    return i = {}, verb(\"next\"), verb(\"throw\", function (e) { throw e; }), verb(\"return\"), i[Symbol.iterator] = function () { return this; }, i;\r\n    function verb(n, f) { i[n] = o[n] ? function (v) { return (p = !p) ? { value: __await(o[n](v)), done: n === \"return\" } : f ? f(v) : v; } : f; }\r\n}\r\n\r\nexport function __asyncValues(o) {\r\n    if (!Symbol.asyncIterator) throw new TypeError(\"Symbol.asyncIterator is not defined.\");\r\n    var m = o[Symbol.asyncIterator], i;\r\n    return m ? m.call(o) : (o = typeof __values === \"function\" ? __values(o) : o[Symbol.iterator](), i = {}, verb(\"next\"), verb(\"throw\"), verb(\"return\"), i[Symbol.asyncIterator] = function () { return this; }, i);\r\n    function verb(n) { i[n] = o[n] && function (v) { return new Promise(function (resolve, reject) { v = o[n](v), settle(resolve, reject, v.done, v.value); }); }; }\r\n    function settle(resolve, reject, d, v) { Promise.resolve(v).then(function(v) { resolve({ value: v, done: d }); }, reject); }\r\n}\r\n\r\nexport function __makeTemplateObject(cooked, raw) {\r\n    if (Object.defineProperty) { Object.defineProperty(cooked, \"raw\", { value: raw }); } else { cooked.raw = raw; }\r\n    return cooked;\r\n};\r\n\r\nvar __setModuleDefault = Object.create ? (function(o, v) {\r\n    Object.defineProperty(o, \"default\", { enumerable: true, value: v });\r\n}) : function(o, v) {\r\n    o[\"default\"] = v;\r\n};\r\n\r\nexport function __importStar(mod) {\r\n    if (mod && mod.__esModule) return mod;\r\n    var result = {};\r\n    if (mod != null) for (var k in mod) if (k !== \"default\" && Object.prototype.hasOwnProperty.call(mod, k)) __createBinding(result, mod, k);\r\n    __setModuleDefault(result, mod);\r\n    return result;\r\n}\r\n\r\nexport function __importDefault(mod) {\r\n    return (mod && mod.__esModule) ? mod : { default: mod };\r\n}\r\n\r\nexport function __classPrivateFieldGet(receiver, state, kind, f) {\r\n    if (kind === \"a\" && !f) throw new TypeError(\"Private accessor was defined without a getter\");\r\n    if (typeof state === \"function\" ? receiver !== state || !f : !state.has(receiver)) throw new TypeError(\"Cannot read private member from an object whose class did not declare it\");\r\n    return kind === \"m\" ? f : kind === \"a\" ? f.call(receiver) : f ? f.value : state.get(receiver);\r\n}\r\n\r\nexport function __classPrivateFieldSet(receiver, state, value, kind, f) {\r\n    if (kind === \"m\") throw new TypeError(\"Private method is not writable\");\r\n    if (kind === \"a\" && !f) throw new TypeError(\"Private accessor was defined without a setter\");\r\n    if (typeof state === \"function\" ? receiver !== state || !f : !state.has(receiver)) throw new TypeError(\"Cannot write private member to an object whose class did not declare it\");\r\n    return (kind === \"a\" ? f.call(receiver, value) : f ? f.value = value : state.set(receiver, value)), value;\r\n}\r\n", "/**\n * Returns true if the object is a function.\n * @param value The value to check\n */\nexport function isFunction(value: any): value is (...args: any[]) => any {\n  return typeof value === 'function';\n}\n", "/**\n * Used to create Error subclasses until the community moves away from ES5.\n *\n * This is because compiling from TypeScript down to ES5 has issues with subclassing Errors\n * as well as other built-in types: https://github.com/Microsoft/TypeScript/issues/12123\n *\n * @param createImpl A factory function to create the actual constructor implementation. The returned\n * function should be a named function that calls `_super` internally.\n */\nexport function createErrorClass<T>(createImpl: (_super: any) => any): T {\n  const _super = (instance: any) => {\n    Error.call(instance);\n    instance.stack = new Error().stack;\n  };\n\n  const ctorFunc = createImpl(_super);\n  ctorFunc.prototype = Object.create(Error.prototype);\n  ctorFunc.prototype.constructor = ctorFunc;\n  return ctorFunc;\n}\n", "import { createErrorClass } from './createErrorClass';\n\nexport interface UnsubscriptionError extends Error {\n  readonly errors: any[];\n}\n\nexport interface UnsubscriptionErrorCtor {\n  /**\n   * @deprecated Internal implementation detail. Do not construct error instances.\n   * Cannot be tagged as internal: https://github.com/ReactiveX/rxjs/issues/6269\n   */\n  new (errors: any[]): UnsubscriptionError;\n}\n\n/**\n * An error thrown when one or more errors have occurred during the\n * `unsubscribe` of a {@link Subscription}.\n */\nexport const UnsubscriptionError: UnsubscriptionErrorCtor = createErrorClass(\n  (_super) =>\n    function UnsubscriptionErrorImpl(this: any, errors: (Error | string)[]) {\n      _super(this);\n      this.message = errors\n        ? `${errors.length} errors occurred during unsubscription:\n${errors.map((err, i) => `${i + 1}) ${err.toString()}`).join('\\n  ')}`\n        : '';\n      this.name = 'UnsubscriptionError';\n      this.errors = errors;\n    }\n);\n", "/**\n * Removes an item from an array, mutating it.\n * @param arr The array to remove the item from\n * @param item The item to remove\n */\nexport function arrRemove<T>(arr: T[] | undefined | null, item: T) {\n  if (arr) {\n    const index = arr.indexOf(item);\n    0 <= index && arr.splice(index, 1);\n  }\n}\n", "import { isFunction } from './util/isFunction';\nimport { UnsubscriptionError } from './util/UnsubscriptionError';\nimport { SubscriptionLike, TeardownLogic, Unsubscribable } from './types';\nimport { arrRemove } from './util/arrRemove';\n\n/**\n * Represents a disposable resource, such as the execution of an Observable. A\n * Subscription has one important method, `unsubscribe`, that takes no argument\n * and just disposes the resource held by the subscription.\n *\n * Additionally, subscriptions may be grouped together through the `add()`\n * method, which will attach a child Subscription to the current Subscription.\n * When a Subscription is unsubscribed, all its children (and its grandchildren)\n * will be unsubscribed as well.\n *\n * @class Subscription\n */\nexport class Subscription implements SubscriptionLike {\n  /** @nocollapse */\n  public static EMPTY = (() => {\n    const empty = new Subscription();\n    empty.closed = true;\n    return empty;\n  })();\n\n  /**\n   * A flag to indicate whether this Subscription has already been unsubscribed.\n   */\n  public closed = false;\n\n  private _parentage: Subscription[] | Subscription | null = null;\n\n  /**\n   * The list of registered finalizers to execute upon unsubscription. Adding and removing from this\n   * list occurs in the {@link #add} and {@link #remove} methods.\n   */\n  private _finalizers: Exclude<TeardownLogic, void>[] | null = null;\n\n  /**\n   * @param initialTeardown A function executed first as part of the finalization\n   * process that is kicked off when {@link #unsubscribe} is called.\n   */\n  constructor(private initialTeardown?: () => void) {}\n\n  /**\n   * Disposes the resources held by the subscription. May, for instance, cancel\n   * an ongoing Observable execution or cancel any other type of work that\n   * started when the Subscription was created.\n   * @return {void}\n   */\n  unsubscribe(): void {\n    let errors: any[] | undefined;\n\n    if (!this.closed) {\n      this.closed = true;\n\n      // Remove this from it's parents.\n      const { _parentage } = this;\n      if (_parentage) {\n        this._parentage = null;\n        if (Array.isArray(_parentage)) {\n          for (const parent of _parentage) {\n            parent.remove(this);\n          }\n        } else {\n          _parentage.remove(this);\n        }\n      }\n\n      const { initialTeardown: initialFinalizer } = this;\n      if (isFunction(initialFinalizer)) {\n        try {\n          initialFinalizer();\n        } catch (e) {\n          errors = e instanceof UnsubscriptionError ? e.errors : [e];\n        }\n      }\n\n      const { _finalizers } = this;\n      if (_finalizers) {\n        this._finalizers = null;\n        for (const finalizer of _finalizers) {\n          try {\n            execFinalizer(finalizer);\n          } catch (err) {\n            errors = errors ?? [];\n            if (err instanceof UnsubscriptionError) {\n              errors = [...errors, ...err.errors];\n            } else {\n              errors.push(err);\n            }\n          }\n        }\n      }\n\n      if (errors) {\n        throw new UnsubscriptionError(errors);\n      }\n    }\n  }\n\n  /**\n   * Adds a finalizer to this subscription, so that finalization will be unsubscribed/called\n   * when this subscription is unsubscribed. If this subscription is already {@link #closed},\n   * because it has already been unsubscribed, then whatever finalizer is passed to it\n   * will automatically be executed (unless the finalizer itself is also a closed subscription).\n   *\n   * Closed Subscriptions cannot be added as finalizers to any subscription. Adding a closed\n   * subscription to a any subscription will result in no operation. (A noop).\n   *\n   * Adding a subscription to itself, or adding `null` or `undefined` will not perform any\n   * operation at all. (A noop).\n   *\n   * `Subscription` instances that are added to this instance will automatically remove themselves\n   * if they are unsubscribed. Functions and {@link Unsubscribable} objects that you wish to remove\n   * will need to be removed manually with {@link #remove}\n   *\n   * @param teardown The finalization logic to add to this subscription.\n   */\n  add(teardown: TeardownLogic): void {\n    // Only add the finalizer if it's not undefined\n    // and don't add a subscription to itself.\n    if (teardown && teardown !== this) {\n      if (this.closed) {\n        // If this subscription is already closed,\n        // execute whatever finalizer is handed to it automatically.\n        execFinalizer(teardown);\n      } else {\n        if (teardown instanceof Subscription) {\n          // We don't add closed subscriptions, and we don't add the same subscription\n          // twice. Subscription unsubscribe is idempotent.\n          if (teardown.closed || teardown._hasParent(this)) {\n            return;\n          }\n          teardown._addParent(this);\n        }\n        (this._finalizers = this._finalizers ?? []).push(teardown);\n      }\n    }\n  }\n\n  /**\n   * Checks to see if a this subscription already has a particular parent.\n   * This will signal that this subscription has already been added to the parent in question.\n   * @param parent the parent to check for\n   */\n  private _hasParent(parent: Subscription) {\n    const { _parentage } = this;\n    return _parentage === parent || (Array.isArray(_parentage) && _parentage.includes(parent));\n  }\n\n  /**\n   * Adds a parent to this subscription so it can be removed from the parent if it\n   * unsubscribes on it's own.\n   *\n   * NOTE: THIS ASSUMES THAT {@link _hasParent} HAS ALREADY BEEN CHECKED.\n   * @param parent The parent subscription to add\n   */\n  private _addParent(parent: Subscription) {\n    const { _parentage } = this;\n    this._parentage = Array.isArray(_parentage) ? (_parentage.push(parent), _parentage) : _parentage ? [_parentage, parent] : parent;\n  }\n\n  /**\n   * Called on a child when it is removed via {@link #remove}.\n   * @param parent The parent to remove\n   */\n  private _removeParent(parent: Subscription) {\n    const { _parentage } = this;\n    if (_parentage === parent) {\n      this._parentage = null;\n    } else if (Array.isArray(_parentage)) {\n      arrRemove(_parentage, parent);\n    }\n  }\n\n  /**\n   * Removes a finalizer from this subscription that was previously added with the {@link #add} method.\n   *\n   * Note that `Subscription` instances, when unsubscribed, will automatically remove themselves\n   * from every other `Subscription` they have been added to. This means that using the `remove` method\n   * is not a common thing and should be used thoughtfully.\n   *\n   * If you add the same finalizer instance of a function or an unsubscribable object to a `Subscription` instance\n   * more than once, you will need to call `remove` the same number of times to remove all instances.\n   *\n   * All finalizer instances are removed to free up memory upon unsubscription.\n   *\n   * @param teardown The finalizer to remove from this subscription\n   */\n  remove(teardown: Exclude<TeardownLogic, void>): void {\n    const { _finalizers } = this;\n    _finalizers && arrRemove(_finalizers, teardown);\n\n    if (teardown instanceof Subscription) {\n      teardown._removeParent(this);\n    }\n  }\n}\n\nexport const EMPTY_SUBSCRIPTION = Subscription.EMPTY;\n\nexport function isSubscription(value: any): value is Subscription {\n  return (\n    value instanceof Subscription ||\n    (value && 'closed' in value && isFunction(value.remove) && isFunction(value.add) && isFunction(value.unsubscribe))\n  );\n}\n\nfunction execFinalizer(finalizer: Unsubscribable | (() => void)) {\n  if (isFunction(finalizer)) {\n    finalizer();\n  } else {\n    finalizer.unsubscribe();\n  }\n}\n", "import { Subscriber } from './Subscriber';\nimport { ObservableNotification } from './types';\n\n/**\n * The {@link GlobalConfig} object for RxJS. It is used to configure things\n * like how to react on unhandled errors.\n */\nexport const config: GlobalConfig = {\n  onUnhandledError: null,\n  onStoppedNotification: null,\n  Promise: undefined,\n  useDeprecatedSynchronousErrorHandling: false,\n  useDeprecatedNextContext: false,\n};\n\n/**\n * The global configuration object for RxJS, used to configure things\n * like how to react on unhandled errors. Accessible via {@link config}\n * object.\n */\nexport interface GlobalConfig {\n  /**\n   * A registration point for unhandled errors from RxJS. These are errors that\n   * cannot were not handled by consuming code in the usual subscription path. For\n   * example, if you have this configured, and you subscribe to an observable without\n   * providing an error handler, errors from that subscription will end up here. This\n   * will _always_ be called asynchronously on another job in the runtime. This is because\n   * we do not want errors thrown in this user-configured handler to interfere with the\n   * behavior of the library.\n   */\n  onUnhandledError: ((err: any) => void) | null;\n\n  /**\n   * A registration point for notifications that cannot be sent to subscribers because they\n   * have completed, errored or have been explicitly unsubscribed. By default, next, complete\n   * and error notifications sent to stopped subscribers are noops. However, sometimes callers\n   * might want a different behavior. For example, with sources that attempt to report errors\n   * to stopped subscribers, a caller can configure RxJS to throw an unhandled error instead.\n   * This will _always_ be called asynchronously on another job in the runtime. This is because\n   * we do not want errors thrown in this user-configured handler to interfere with the\n   * behavior of the library.\n   */\n  onStoppedNotification: ((notification: ObservableNotification<any>, subscriber: Subscriber<any>) => void) | null;\n\n  /**\n   * The promise constructor used by default for {@link Observable#toPromise toPromise} and {@link Observable#forEach forEach}\n   * methods.\n   *\n   * @deprecated As of version 8, RxJS will no longer support this sort of injection of a\n   * Promise constructor. If you need a Promise implementation other than native promises,\n   * please polyfill/patch Promise as you see appropriate. Will be removed in v8.\n   */\n  Promise?: PromiseConstructorLike;\n\n  /**\n   * If true, turns on synchronous error rethrowing, which is a deprecated behavior\n   * in v6 and higher. This behavior enables bad patterns like wrapping a subscribe\n   * call in a try/catch block. It also enables producer interference, a nasty bug\n   * where a multicast can be broken for all observers by a downstream consumer with\n   * an unhandled error. DO NOT USE THIS FLAG UNLESS IT'S NEEDED TO BUY TIME\n   * FOR MIGRATION REASONS.\n   *\n   * @deprecated As of version 8, RxJS will no longer support synchronous throwing\n   * of unhandled errors. All errors will be thrown on a separate call stack to prevent bad\n   * behaviors described above. Will be removed in v8.\n   */\n  useDeprecatedSynchronousErrorHandling: boolean;\n\n  /**\n   * If true, enables an as-of-yet undocumented feature from v5: The ability to access\n   * `unsubscribe()` via `this` context in `next` functions created in observers passed\n   * to `subscribe`.\n   *\n   * This is being removed because the performance was severely problematic, and it could also cause\n   * issues when types other than POJOs are passed to subscribe as subscribers, as they will likely have\n   * their `this` context overwritten.\n   *\n   * @deprecated As of version 8, RxJS will no longer support altering the\n   * context of next functions provided as part of an observer to Subscribe. Instead,\n   * you will have access to a subscription or a signal or token that will allow you to do things like\n   * unsubscribe and test closed status. Will be removed in v8.\n   */\n  useDeprecatedNextContext: boolean;\n}\n", "import type { TimerHandle } from './timerHandle';\ntype SetTimeoutFunction = (handler: () => void, timeout?: number, ...args: any[]) => TimerHandle;\ntype ClearTimeoutFunction = (handle: TimerHandle) => void;\n\ninterface TimeoutProvider {\n  setTimeout: SetTimeoutFunction;\n  clearTimeout: ClearTimeoutFunction;\n  delegate:\n    | {\n        setTimeout: SetTimeoutFunction;\n        clearTimeout: ClearTimeoutFunction;\n      }\n    | undefined;\n}\n\nexport const timeoutProvider: TimeoutProvider = {\n  // When accessing the delegate, use the variable rather than `this` so that\n  // the functions can be called without being bound to the provider.\n  setTimeout(handler: () => void, timeout?: number, ...args) {\n    const { delegate } = timeoutProvider;\n    if (delegate?.setTimeout) {\n      return delegate.setTimeout(handler, timeout, ...args);\n    }\n    return setTimeout(handler, timeout, ...args);\n  },\n  clearTimeout(handle) {\n    const { delegate } = timeoutProvider;\n    return (delegate?.clearTimeout || clearTimeout)(handle as any);\n  },\n  delegate: undefined,\n};\n", "import { config } from '../config';\nimport { timeoutProvider } from '../scheduler/timeoutProvider';\n\n/**\n * Handles an error on another job either with the user-configured {@link onUnhandledError},\n * or by throwing it on that new job so it can be picked up by `window.onerror`, `process.on('error')`, etc.\n *\n * This should be called whenever there is an error that is out-of-band with the subscription\n * or when an error hits a terminal boundary of the subscription and no error handler was provided.\n *\n * @param err the error to report\n */\nexport function reportUnhandledError(err: any) {\n  timeoutProvider.setTimeout(() => {\n    const { onUnhandledError } = config;\n    if (onUnhandledError) {\n      // Execute the user-configured error handler.\n      onUnhandledError(err);\n    } else {\n      // Throw so it is picked up by the runtime's uncaught error mechanism.\n      throw err;\n    }\n  });\n}\n", "/* tslint:disable:no-empty */\nexport function noop() { }\n", "import { CompleteNotification, NextNotification, ErrorNotification } from './types';\n\n/**\n * A completion object optimized for memory use and created to be the\n * same \"shape\" as other notifications in v8.\n * @internal\n */\nexport const COMPLETE_NOTIFICATION = (() => createNotification('C', undefined, undefined) as CompleteNotification)();\n\n/**\n * Internal use only. Creates an optimized error notification that is the same \"shape\"\n * as other notifications.\n * @internal\n */\nexport function errorNotification(error: any): ErrorNotification {\n  return createNotification('E', undefined, error) as any;\n}\n\n/**\n * Internal use only. Creates an optimized next notification that is the same \"shape\"\n * as other notifications.\n * @internal\n */\nexport function nextNotification<T>(value: T) {\n  return createNotification('N', value, undefined) as NextNotification<T>;\n}\n\n/**\n * Ensures that all notifications created internally have the same \"shape\" in v8.\n *\n * TODO: This is only exported to support a crazy legacy test in `groupBy`.\n * @internal\n */\nexport function createNotification(kind: 'N' | 'E' | 'C', value: any, error: any) {\n  return {\n    kind,\n    value,\n    error,\n  };\n}\n", "import { config } from '../config';\n\nlet context: { errorThrown: boolean; error: any } | null = null;\n\n/**\n * Handles dealing with errors for super-gross mode. Creates a context, in which\n * any synchronously thrown errors will be passed to {@link captureError}. Which\n * will record the error such that it will be rethrown after the call back is complete.\n * TODO: Remove in v8\n * @param cb An immediately executed function.\n */\nexport function errorContext(cb: () => void) {\n  if (config.useDeprecatedSynchronousErrorHandling) {\n    const isRoot = !context;\n    if (isRoot) {\n      context = { errorThrown: false, error: null };\n    }\n    cb();\n    if (isRoot) {\n      const { errorThrown, error } = context!;\n      context = null;\n      if (errorThrown) {\n        throw error;\n      }\n    }\n  } else {\n    // This is the general non-deprecated path for everyone that\n    // isn't crazy enough to use super-gross mode (useDeprecatedSynchronousErrorHandling)\n    cb();\n  }\n}\n\n/**\n * Captures errors only in super-gross mode.\n * @param err the error to capture\n */\nexport function captureError(err: any) {\n  if (config.useDeprecatedSynchronousErrorHandling && context) {\n    context.errorThrown = true;\n    context.error = err;\n  }\n}\n", "import { isFunction } from './util/isFunction';\nimport { Observer, ObservableNotification } from './types';\nimport { isSubscription, Subscription } from './Subscription';\nimport { config } from './config';\nimport { reportUnhandledError } from './util/reportUnhandledError';\nimport { noop } from './util/noop';\nimport { nextNotification, errorNotification, COMPLETE_NOTIFICATION } from './NotificationFactories';\nimport { timeoutProvider } from './scheduler/timeoutProvider';\nimport { captureError } from './util/errorContext';\n\n/**\n * Implements the {@link Observer} interface and extends the\n * {@link Subscription} class. While the {@link Observer} is the public API for\n * consuming the values of an {@link Observable}, all Observers get converted to\n * a Subscriber, in order to provide Subscription-like capabilities such as\n * `unsubscribe`. Subscriber is a common type in RxJS, and crucial for\n * implementing operators, but it is rarely used as a public API.\n *\n * @class Subscriber<T>\n */\nexport class Subscriber<T> extends Subscription implements Observer<T> {\n  /**\n   * A static factory for a Subscriber, given a (potentially partial) definition\n   * of an Observer.\n   * @param next The `next` callback of an Observer.\n   * @param error The `error` callback of an\n   * Observer.\n   * @param complete The `complete` callback of an\n   * Observer.\n   * @return A Subscriber wrapping the (partially defined)\n   * Observer represented by the given arguments.\n   * @nocollapse\n   * @deprecated Do not use. Will be removed in v8. There is no replacement for this\n   * method, and there is no reason to be creating instances of `Subscriber` directly.\n   * If you have a specific use case, please file an issue.\n   */\n  static create<T>(next?: (x?: T) => void, error?: (e?: any) => void, complete?: () => void): Subscriber<T> {\n    return new SafeSubscriber(next, error, complete);\n  }\n\n  /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n  protected isStopped: boolean = false;\n  /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n  protected destination: Subscriber<any> | Observer<any>; // this `any` is the escape hatch to erase extra type param (e.g. R)\n\n  /**\n   * @deprecated Internal implementation detail, do not use directly. Will be made internal in v8.\n   * There is no reason to directly create an instance of Subscriber. This type is exported for typings reasons.\n   */\n  constructor(destination?: Subscriber<any> | Observer<any>) {\n    super();\n    if (destination) {\n      this.destination = destination;\n      // Automatically chain subscriptions together here.\n      // if destination is a Subscription, then it is a Subscriber.\n      if (isSubscription(destination)) {\n        destination.add(this);\n      }\n    } else {\n      this.destination = EMPTY_OBSERVER;\n    }\n  }\n\n  /**\n   * The {@link Observer} callback to receive notifications of type `next` from\n   * the Observable, with a value. The Observable may call this method 0 or more\n   * times.\n   * @param {T} [value] The `next` value.\n   * @return {void}\n   */\n  next(value?: T): void {\n    if (this.isStopped) {\n      handleStoppedNotification(nextNotification(value), this);\n    } else {\n      this._next(value!);\n    }\n  }\n\n  /**\n   * The {@link Observer} callback to receive notifications of type `error` from\n   * the Observable, with an attached `Error`. Notifies the Observer that\n   * the Observable has experienced an error condition.\n   * @param {any} [err] The `error` exception.\n   * @return {void}\n   */\n  error(err?: any): void {\n    if (this.isStopped) {\n      handleStoppedNotification(errorNotification(err), this);\n    } else {\n      this.isStopped = true;\n      this._error(err);\n    }\n  }\n\n  /**\n   * The {@link Observer} callback to receive a valueless notification of type\n   * `complete` from the Observable. Notifies the Observer that the Observable\n   * has finished sending push-based notifications.\n   * @return {void}\n   */\n  complete(): void {\n    if (this.isStopped) {\n      handleStoppedNotification(COMPLETE_NOTIFICATION, this);\n    } else {\n      this.isStopped = true;\n      this._complete();\n    }\n  }\n\n  unsubscribe(): void {\n    if (!this.closed) {\n      this.isStopped = true;\n      super.unsubscribe();\n      this.destination = null!;\n    }\n  }\n\n  protected _next(value: T): void {\n    this.destination.next(value);\n  }\n\n  protected _error(err: any): void {\n    try {\n      this.destination.error(err);\n    } finally {\n      this.unsubscribe();\n    }\n  }\n\n  protected _complete(): void {\n    try {\n      this.destination.complete();\n    } finally {\n      this.unsubscribe();\n    }\n  }\n}\n\n/**\n * This bind is captured here because we want to be able to have\n * compatibility with monoid libraries that tend to use a method named\n * `bind`. In particular, a library called Monio requires this.\n */\nconst _bind = Function.prototype.bind;\n\nfunction bind<Fn extends (...args: any[]) => any>(fn: Fn, thisArg: any): Fn {\n  return _bind.call(fn, thisArg);\n}\n\n/**\n * Internal optimization only, DO NOT EXPOSE.\n * @internal\n */\nclass ConsumerObserver<T> implements Observer<T> {\n  constructor(private partialObserver: Partial<Observer<T>>) {}\n\n  next(value: T): void {\n    const { partialObserver } = this;\n    if (partialObserver.next) {\n      try {\n        partialObserver.next(value);\n      } catch (error) {\n        handleUnhandledError(error);\n      }\n    }\n  }\n\n  error(err: any): void {\n    const { partialObserver } = this;\n    if (partialObserver.error) {\n      try {\n        partialObserver.error(err);\n      } catch (error) {\n        handleUnhandledError(error);\n      }\n    } else {\n      handleUnhandledError(err);\n    }\n  }\n\n  complete(): void {\n    const { partialObserver } = this;\n    if (partialObserver.complete) {\n      try {\n        partialObserver.complete();\n      } catch (error) {\n        handleUnhandledError(error);\n      }\n    }\n  }\n}\n\nexport class SafeSubscriber<T> extends Subscriber<T> {\n  constructor(\n    observerOrNext?: Partial<Observer<T>> | ((value: T) => void) | null,\n    error?: ((e?: any) => void) | null,\n    complete?: (() => void) | null\n  ) {\n    super();\n\n    let partialObserver: Partial<Observer<T>>;\n    if (isFunction(observerOrNext) || !observerOrNext) {\n      // The first argument is a function, not an observer. The next\n      // two arguments *could* be observers, or they could be empty.\n      partialObserver = {\n        next: (observerOrNext ?? undefined) as (((value: T) => void) | undefined),\n        error: error ?? undefined,\n        complete: complete ?? undefined,\n      };\n    } else {\n      // The first argument is a partial observer.\n      let context: any;\n      if (this && config.useDeprecatedNextContext) {\n        // This is a deprecated path that made `this.unsubscribe()` available in\n        // next handler functions passed to subscribe. This only exists behind a flag\n        // now, as it is *very* slow.\n        context = Object.create(observerOrNext);\n        context.unsubscribe = () => this.unsubscribe();\n        partialObserver = {\n          next: observerOrNext.next && bind(observerOrNext.next, context),\n          error: observerOrNext.error && bind(observerOrNext.error, context),\n          complete: observerOrNext.complete && bind(observerOrNext.complete, context),\n        };\n      } else {\n        // The \"normal\" path. Just use the partial observer directly.\n        partialObserver = observerOrNext;\n      }\n    }\n\n    // Wrap the partial observer to ensure it's a full observer, and\n    // make sure proper error handling is accounted for.\n    this.destination = new ConsumerObserver(partialObserver);\n  }\n}\n\nfunction handleUnhandledError(error: any) {\n  if (config.useDeprecatedSynchronousErrorHandling) {\n    captureError(error);\n  } else {\n    // Ideal path, we report this as an unhandled error,\n    // which is thrown on a new call stack.\n    reportUnhandledError(error);\n  }\n}\n\n/**\n * An error handler used when no error handler was supplied\n * to the SafeSubscriber -- meaning no error handler was supplied\n * do the `subscribe` call on our observable.\n * @param err The error to handle\n */\nfunction defaultErrorHandler(err: any) {\n  throw err;\n}\n\n/**\n * A handler for notifications that cannot be sent to a stopped subscriber.\n * @param notification The notification being sent\n * @param subscriber The stopped subscriber\n */\nfunction handleStoppedNotification(notification: ObservableNotification<any>, subscriber: Subscriber<any>) {\n  const { onStoppedNotification } = config;\n  onStoppedNotification && timeoutProvider.setTimeout(() => onStoppedNotification(notification, subscriber));\n}\n\n/**\n * The observer used as a stub for subscriptions where the user did not\n * pass any arguments to `subscribe`. Comes with the default error handling\n * behavior.\n */\nexport const EMPTY_OBSERVER: Readonly<Observer<any>> & { closed: true } = {\n  closed: true,\n  next: noop,\n  error: defaultErrorHandler,\n  complete: noop,\n};\n", "/**\n * Symbol.observable or a string \"@@observable\". Used for interop\n *\n * @deprecated We will no longer be exporting this symbol in upcoming versions of RxJS.\n * Instead polyfill and use Symbol.observable directly *or* use https://www.npmjs.com/package/symbol-observable\n */\nexport const observable: string | symbol = (() => (typeof Symbol === 'function' && Symbol.observable) || '@@observable')();\n", "/**\n * This function takes one parameter and just returns it. Simply put,\n * this is like `<T>(x: T): T => x`.\n *\n * ## Examples\n *\n * This is useful in some cases when using things like `mergeMap`\n *\n * ```ts\n * import { interval, take, map, range, mergeMap, identity } from 'rxjs';\n *\n * const source$ = interval(1000).pipe(take(5));\n *\n * const result$ = source$.pipe(\n *   map(i => range(i)),\n *   mergeMap(identity) // same as mergeMap(x => x)\n * );\n *\n * result$.subscribe({\n *   next: console.log\n * });\n * ```\n *\n * Or when you want to selectively apply an operator\n *\n * ```ts\n * import { interval, take, identity } from 'rxjs';\n *\n * const shouldLimit = () => Math.random() < 0.5;\n *\n * const source$ = interval(1000);\n *\n * const result$ = source$.pipe(shouldLimit() ? take(5) : identity);\n *\n * result$.subscribe({\n *   next: console.log\n * });\n * ```\n *\n * @param x Any value that is returned by this function\n * @returns The value passed as the first parameter to this function\n */\nexport function identity<T>(x: T): T {\n  return x;\n}\n", "import { identity } from './identity';\nimport { UnaryFunction } from '../types';\n\nexport function pipe(): typeof identity;\nexport function pipe<T, A>(fn1: UnaryFunction<T, A>): UnaryFunction<T, A>;\nexport function pipe<T, A, B>(fn1: UnaryFunction<T, A>, fn2: UnaryFunction<A, B>): UnaryFunction<T, B>;\nexport function pipe<T, A, B, C>(fn1: UnaryFunction<T, A>, fn2: UnaryFunction<A, B>, fn3: UnaryFunction<B, C>): UnaryFunction<T, C>;\nexport function pipe<T, A, B, C, D>(\n  fn1: UnaryFunction<T, A>,\n  fn2: UnaryFunction<A, B>,\n  fn3: UnaryFunction<B, C>,\n  fn4: UnaryFunction<C, D>\n): UnaryFunction<T, D>;\nexport function pipe<T, A, B, C, D, E>(\n  fn1: UnaryFunction<T, A>,\n  fn2: UnaryFunction<A, B>,\n  fn3: UnaryFunction<B, C>,\n  fn4: UnaryFunction<C, D>,\n  fn5: UnaryFunction<D, E>\n): UnaryFunction<T, E>;\nexport function pipe<T, A, B, C, D, E, F>(\n  fn1: UnaryFunction<T, A>,\n  fn2: UnaryFunction<A, B>,\n  fn3: UnaryFunction<B, C>,\n  fn4: UnaryFunction<C, D>,\n  fn5: UnaryFunction<D, E>,\n  fn6: UnaryFunction<E, F>\n): UnaryFunction<T, F>;\nexport function pipe<T, A, B, C, D, E, F, G>(\n  fn1: UnaryFunction<T, A>,\n  fn2: UnaryFunction<A, B>,\n  fn3: UnaryFunction<B, C>,\n  fn4: UnaryFunction<C, D>,\n  fn5: UnaryFunction<D, E>,\n  fn6: UnaryFunction<E, F>,\n  fn7: UnaryFunction<F, G>\n): UnaryFunction<T, G>;\nexport function pipe<T, A, B, C, D, E, F, G, H>(\n  fn1: UnaryFunction<T, A>,\n  fn2: UnaryFunction<A, B>,\n  fn3: UnaryFunction<B, C>,\n  fn4: UnaryFunction<C, D>,\n  fn5: UnaryFunction<D, E>,\n  fn6: UnaryFunction<E, F>,\n  fn7: UnaryFunction<F, G>,\n  fn8: UnaryFunction<G, H>\n): UnaryFunction<T, H>;\nexport function pipe<T, A, B, C, D, E, F, G, H, I>(\n  fn1: UnaryFunction<T, A>,\n  fn2: UnaryFunction<A, B>,\n  fn3: UnaryFunction<B, C>,\n  fn4: UnaryFunction<C, D>,\n  fn5: UnaryFunction<D, E>,\n  fn6: UnaryFunction<E, F>,\n  fn7: UnaryFunction<F, G>,\n  fn8: UnaryFunction<G, H>,\n  fn9: UnaryFunction<H, I>\n): UnaryFunction<T, I>;\nexport function pipe<T, A, B, C, D, E, F, G, H, I>(\n  fn1: UnaryFunction<T, A>,\n  fn2: UnaryFunction<A, B>,\n  fn3: UnaryFunction<B, C>,\n  fn4: UnaryFunction<C, D>,\n  fn5: UnaryFunction<D, E>,\n  fn6: UnaryFunction<E, F>,\n  fn7: UnaryFunction<F, G>,\n  fn8: UnaryFunction<G, H>,\n  fn9: UnaryFunction<H, I>,\n  ...fns: UnaryFunction<any, any>[]\n): UnaryFunction<T, unknown>;\n\n/**\n * pipe() can be called on one or more functions, each of which can take one argument (\"UnaryFunction\")\n * and uses it to return a value.\n * It returns a function that takes one argument, passes it to the first UnaryFunction, and then\n * passes the result to the next one, passes that result to the next one, and so on.  \n */\nexport function pipe(...fns: Array<UnaryFunction<any, any>>): UnaryFunction<any, any> {\n  return pipeFromArray(fns);\n}\n\n/** @internal */\nexport function pipeFromArray<T, R>(fns: Array<UnaryFunction<T, R>>): UnaryFunction<T, R> {\n  if (fns.length === 0) {\n    return identity as UnaryFunction<any, any>;\n  }\n\n  if (fns.length === 1) {\n    return fns[0];\n  }\n\n  return function piped(input: T): R {\n    return fns.reduce((prev: any, fn: UnaryFunction<T, R>) => fn(prev), input as any);\n  };\n}\n", "import { Operator } from './Operator';\nimport { SafeSubscriber, Subscriber } from './Subscriber';\nimport { isSubscription, Subscription } from './Subscription';\nimport { TeardownLogic, OperatorFunction, Subscribable, Observer } from './types';\nimport { observable as Symbol_observable } from './symbol/observable';\nimport { pipeFromArray } from './util/pipe';\nimport { config } from './config';\nimport { isFunction } from './util/isFunction';\nimport { errorContext } from './util/errorContext';\n\n/**\n * A representation of any set of values over any amount of time. This is the most basic building block\n * of RxJS.\n *\n * @class Observable<T>\n */\nexport class Observable<T> implements Subscribable<T> {\n  /**\n   * @deprecated Internal implementation detail, do not use directly. Will be made internal in v8.\n   */\n  source: Observable<any> | undefined;\n\n  /**\n   * @deprecated Internal implementation detail, do not use directly. Will be made internal in v8.\n   */\n  operator: Operator<any, T> | undefined;\n\n  /**\n   * @constructor\n   * @param {Function} subscribe the function that is called when the Observable is\n   * initially subscribed to. This function is given a Subscriber, to which new values\n   * can be `next`ed, or an `error` method can be called to raise an error, or\n   * `complete` can be called to notify of a successful completion.\n   */\n  constructor(subscribe?: (this: Observable<T>, subscriber: Subscriber<T>) => TeardownLogic) {\n    if (subscribe) {\n      this._subscribe = subscribe;\n    }\n  }\n\n  // HACK: Since TypeScript inherits static properties too, we have to\n  // fight against TypeScript here so Subject can have a different static create signature\n  /**\n   * Creates a new Observable by calling the Observable constructor\n   * @owner Observable\n   * @method create\n   * @param {Function} subscribe? the subscriber function to be passed to the Observable constructor\n   * @return {Observable} a new observable\n   * @nocollapse\n   * @deprecated Use `new Observable()` instead. Will be removed in v8.\n   */\n  static create: (...args: any[]) => any = <T>(subscribe?: (subscriber: Subscriber<T>) => TeardownLogic) => {\n    return new Observable<T>(subscribe);\n  };\n\n  /**\n   * Creates a new Observable, with this Observable instance as the source, and the passed\n   * operator defined as the new observable's operator.\n   * @method lift\n   * @param operator the operator defining the operation to take on the observable\n   * @return a new observable with the Operator applied\n   * @deprecated Internal implementation detail, do not use directly. Will be made internal in v8.\n   * If you have implemented an operator using `lift`, it is recommended that you create an\n   * operator by simply returning `new Observable()` directly. See \"Creating new operators from\n   * scratch\" section here: https://rxjs.dev/guide/operators\n   */\n  lift<R>(operator?: Operator<T, R>): Observable<R> {\n    const observable = new Observable<R>();\n    observable.source = this;\n    observable.operator = operator;\n    return observable;\n  }\n\n  subscribe(observerOrNext?: Partial<Observer<T>> | ((value: T) => void)): Subscription;\n  /** @deprecated Instead of passing separate callback arguments, use an observer argument. Signatures taking separate callback arguments will be removed in v8. Details: https://rxjs.dev/deprecations/subscribe-arguments */\n  subscribe(next?: ((value: T) => void) | null, error?: ((error: any) => void) | null, complete?: (() => void) | null): Subscription;\n  /**\n   * Invokes an execution of an Observable and registers Observer handlers for notifications it will emit.\n   *\n   * <span class=\"informal\">Use it when you have all these Observables, but still nothing is happening.</span>\n   *\n   * `subscribe` is not a regular operator, but a method that calls Observable's internal `subscribe` function. It\n   * might be for example a function that you passed to Observable's constructor, but most of the time it is\n   * a library implementation, which defines what will be emitted by an Observable, and when it be will emitted. This means\n   * that calling `subscribe` is actually the moment when Observable starts its work, not when it is created, as it is often\n   * the thought.\n   *\n   * Apart from starting the execution of an Observable, this method allows you to listen for values\n   * that an Observable emits, as well as for when it completes or errors. You can achieve this in two\n   * of the following ways.\n   *\n   * The first way is creating an object that implements {@link Observer} interface. It should have methods\n   * defined by that interface, but note that it should be just a regular JavaScript object, which you can create\n   * yourself in any way you want (ES6 class, classic function constructor, object literal etc.). In particular, do\n   * not attempt to use any RxJS implementation details to create Observers - you don't need them. Remember also\n   * that your object does not have to implement all methods. If you find yourself creating a method that doesn't\n   * do anything, you can simply omit it. Note however, if the `error` method is not provided and an error happens,\n   * it will be thrown asynchronously. Errors thrown asynchronously cannot be caught using `try`/`catch`. Instead,\n   * use the {@link onUnhandledError} configuration option or use a runtime handler (like `window.onerror` or\n   * `process.on('error)`) to be notified of unhandled errors. Because of this, it's recommended that you provide\n   * an `error` method to avoid missing thrown errors.\n   *\n   * The second way is to give up on Observer object altogether and simply provide callback functions in place of its methods.\n   * This means you can provide three functions as arguments to `subscribe`, where the first function is equivalent\n   * of a `next` method, the second of an `error` method and the third of a `complete` method. Just as in case of an Observer,\n   * if you do not need to listen for something, you can omit a function by passing `undefined` or `null`,\n   * since `subscribe` recognizes these functions by where they were placed in function call. When it comes\n   * to the `error` function, as with an Observer, if not provided, errors emitted by an Observable will be thrown asynchronously.\n   *\n   * You can, however, subscribe with no parameters at all. This may be the case where you're not interested in terminal events\n   * and you also handled emissions internally by using operators (e.g. using `tap`).\n   *\n   * Whichever style of calling `subscribe` you use, in both cases it returns a Subscription object.\n   * This object allows you to call `unsubscribe` on it, which in turn will stop the work that an Observable does and will clean\n   * up all resources that an Observable used. Note that cancelling a subscription will not call `complete` callback\n   * provided to `subscribe` function, which is reserved for a regular completion signal that comes from an Observable.\n   *\n   * Remember that callbacks provided to `subscribe` are not guaranteed to be called asynchronously.\n   * It is an Observable itself that decides when these functions will be called. For example {@link of}\n   * by default emits all its values synchronously. Always check documentation for how given Observable\n   * will behave when subscribed and if its default behavior can be modified with a `scheduler`.\n   *\n   * #### Examples\n   *\n   * Subscribe with an {@link guide/observer Observer}\n   *\n   * ```ts\n   * import { of } from 'rxjs';\n   *\n   * const sumObserver = {\n   *   sum: 0,\n   *   next(value) {\n   *     console.log('Adding: ' + value);\n   *     this.sum = this.sum + value;\n   *   },\n   *   error() {\n   *     // We actually could just remove this method,\n   *     // since we do not really care about errors right now.\n   *   },\n   *   complete() {\n   *     console.log('Sum equals: ' + this.sum);\n   *   }\n   * };\n   *\n   * of(1, 2, 3) // Synchronously emits 1, 2, 3 and then completes.\n   *   .subscribe(sumObserver);\n   *\n   * // Logs:\n   * // 'Adding: 1'\n   * // 'Adding: 2'\n   * // 'Adding: 3'\n   * // 'Sum equals: 6'\n   * ```\n   *\n   * Subscribe with functions ({@link deprecations/subscribe-arguments deprecated})\n   *\n   * ```ts\n   * import { of } from 'rxjs'\n   *\n   * let sum = 0;\n   *\n   * of(1, 2, 3).subscribe(\n   *   value => {\n   *     console.log('Adding: ' + value);\n   *     sum = sum + value;\n   *   },\n   *   undefined,\n   *   () => console.log('Sum equals: ' + sum)\n   * );\n   *\n   * // Logs:\n   * // 'Adding: 1'\n   * // 'Adding: 2'\n   * // 'Adding: 3'\n   * // 'Sum equals: 6'\n   * ```\n   *\n   * Cancel a subscription\n   *\n   * ```ts\n   * import { interval } from 'rxjs';\n   *\n   * const subscription = interval(1000).subscribe({\n   *   next(num) {\n   *     console.log(num)\n   *   },\n   *   complete() {\n   *     // Will not be called, even when cancelling subscription.\n   *     console.log('completed!');\n   *   }\n   * });\n   *\n   * setTimeout(() => {\n   *   subscription.unsubscribe();\n   *   console.log('unsubscribed!');\n   * }, 2500);\n   *\n   * // Logs:\n   * // 0 after 1s\n   * // 1 after 2s\n   * // 'unsubscribed!' after 2.5s\n   * ```\n   *\n   * @param {Observer|Function} observerOrNext (optional) Either an observer with methods to be called,\n   * or the first of three possible handlers, which is the handler for each value emitted from the subscribed\n   * Observable.\n   * @param {Function} error (optional) A handler for a terminal event resulting from an error. If no error handler is provided,\n   * the error will be thrown asynchronously as unhandled.\n   * @param {Function} complete (optional) A handler for a terminal event resulting from successful completion.\n   * @return {Subscription} a subscription reference to the registered handlers\n   * @method subscribe\n   */\n  subscribe(\n    observerOrNext?: Partial<Observer<T>> | ((value: T) => void) | null,\n    error?: ((error: any) => void) | null,\n    complete?: (() => void) | null\n  ): Subscription {\n    const subscriber = isSubscriber(observerOrNext) ? observerOrNext : new SafeSubscriber(observerOrNext, error, complete);\n\n    errorContext(() => {\n      const { operator, source } = this;\n      subscriber.add(\n        operator\n          ? // We're dealing with a subscription in the\n            // operator chain to one of our lifted operators.\n            operator.call(subscriber, source)\n          : source\n          ? // If `source` has a value, but `operator` does not, something that\n            // had intimate knowledge of our API, like our `Subject`, must have\n            // set it. We're going to just call `_subscribe` directly.\n            this._subscribe(subscriber)\n          : // In all other cases, we're likely wrapping a user-provided initializer\n            // function, so we need to catch errors and handle them appropriately.\n            this._trySubscribe(subscriber)\n      );\n    });\n\n    return subscriber;\n  }\n\n  /** @internal */\n  protected _trySubscribe(sink: Subscriber<T>): TeardownLogic {\n    try {\n      return this._subscribe(sink);\n    } catch (err) {\n      // We don't need to return anything in this case,\n      // because it's just going to try to `add()` to a subscription\n      // above.\n      sink.error(err);\n    }\n  }\n\n  /**\n   * Used as a NON-CANCELLABLE means of subscribing to an observable, for use with\n   * APIs that expect promises, like `async/await`. You cannot unsubscribe from this.\n   *\n   * **WARNING**: Only use this with observables you *know* will complete. If the source\n   * observable does not complete, you will end up with a promise that is hung up, and\n   * potentially all of the state of an async function hanging out in memory. To avoid\n   * this situation, look into adding something like {@link timeout}, {@link take},\n   * {@link takeWhile}, or {@link takeUntil} amongst others.\n   *\n   * #### Example\n   *\n   * ```ts\n   * import { interval, take } from 'rxjs';\n   *\n   * const source$ = interval(1000).pipe(take(4));\n   *\n   * async function getTotal() {\n   *   let total = 0;\n   *\n   *   await source$.forEach(value => {\n   *     total += value;\n   *     console.log('observable -> ' + value);\n   *   });\n   *\n   *   return total;\n   * }\n   *\n   * getTotal().then(\n   *   total => console.log('Total: ' + total)\n   * );\n   *\n   * // Expected:\n   * // 'observable -> 0'\n   * // 'observable -> 1'\n   * // 'observable -> 2'\n   * // 'observable -> 3'\n   * // 'Total: 6'\n   * ```\n   *\n   * @param next a handler for each value emitted by the observable\n   * @return a promise that either resolves on observable completion or\n   *  rejects with the handled error\n   */\n  forEach(next: (value: T) => void): Promise<void>;\n\n  /**\n   * @param next a handler for each value emitted by the observable\n   * @param promiseCtor a constructor function used to instantiate the Promise\n   * @return a promise that either resolves on observable completion or\n   *  rejects with the handled error\n   * @deprecated Passing a Promise constructor will no longer be available\n   * in upcoming versions of RxJS. This is because it adds weight to the library, for very\n   * little benefit. If you need this functionality, it is recommended that you either\n   * polyfill Promise, or you create an adapter to convert the returned native promise\n   * to whatever promise implementation you wanted. Will be removed in v8.\n   */\n  forEach(next: (value: T) => void, promiseCtor: PromiseConstructorLike): Promise<void>;\n\n  forEach(next: (value: T) => void, promiseCtor?: PromiseConstructorLike): Promise<void> {\n    promiseCtor = getPromiseCtor(promiseCtor);\n\n    return new promiseCtor<void>((resolve, reject) => {\n      const subscriber = new SafeSubscriber<T>({\n        next: (value) => {\n          try {\n            next(value);\n          } catch (err) {\n            reject(err);\n            subscriber.unsubscribe();\n          }\n        },\n        error: reject,\n        complete: resolve,\n      });\n      this.subscribe(subscriber);\n    }) as Promise<void>;\n  }\n\n  /** @internal */\n  protected _subscribe(subscriber: Subscriber<any>): TeardownLogic {\n    return this.source?.subscribe(subscriber);\n  }\n\n  /**\n   * An interop point defined by the es7-observable spec https://github.com/zenparsing/es-observable\n   * @method Symbol.observable\n   * @return {Observable} this instance of the observable\n   */\n  [Symbol_observable]() {\n    return this;\n  }\n\n  /* tslint:disable:max-line-length */\n  pipe(): Observable<T>;\n  pipe<A>(op1: OperatorFunction<T, A>): Observable<A>;\n  pipe<A, B>(op1: OperatorFunction<T, A>, op2: OperatorFunction<A, B>): Observable<B>;\n  pipe<A, B, C>(op1: OperatorFunction<T, A>, op2: OperatorFunction<A, B>, op3: OperatorFunction<B, C>): Observable<C>;\n  pipe<A, B, C, D>(\n    op1: OperatorFunction<T, A>,\n    op2: OperatorFunction<A, B>,\n    op3: OperatorFunction<B, C>,\n    op4: OperatorFunction<C, D>\n  ): Observable<D>;\n  pipe<A, B, C, D, E>(\n    op1: OperatorFunction<T, A>,\n    op2: OperatorFunction<A, B>,\n    op3: OperatorFunction<B, C>,\n    op4: OperatorFunction<C, D>,\n    op5: OperatorFunction<D, E>\n  ): Observable<E>;\n  pipe<A, B, C, D, E, F>(\n    op1: OperatorFunction<T, A>,\n    op2: OperatorFunction<A, B>,\n    op3: OperatorFunction<B, C>,\n    op4: OperatorFunction<C, D>,\n    op5: OperatorFunction<D, E>,\n    op6: OperatorFunction<E, F>\n  ): Observable<F>;\n  pipe<A, B, C, D, E, F, G>(\n    op1: OperatorFunction<T, A>,\n    op2: OperatorFunction<A, B>,\n    op3: OperatorFunction<B, C>,\n    op4: OperatorFunction<C, D>,\n    op5: OperatorFunction<D, E>,\n    op6: OperatorFunction<E, F>,\n    op7: OperatorFunction<F, G>\n  ): Observable<G>;\n  pipe<A, B, C, D, E, F, G, H>(\n    op1: OperatorFunction<T, A>,\n    op2: OperatorFunction<A, B>,\n    op3: OperatorFunction<B, C>,\n    op4: OperatorFunction<C, D>,\n    op5: OperatorFunction<D, E>,\n    op6: OperatorFunction<E, F>,\n    op7: OperatorFunction<F, G>,\n    op8: OperatorFunction<G, H>\n  ): Observable<H>;\n  pipe<A, B, C, D, E, F, G, H, I>(\n    op1: OperatorFunction<T, A>,\n    op2: OperatorFunction<A, B>,\n    op3: OperatorFunction<B, C>,\n    op4: OperatorFunction<C, D>,\n    op5: OperatorFunction<D, E>,\n    op6: OperatorFunction<E, F>,\n    op7: OperatorFunction<F, G>,\n    op8: OperatorFunction<G, H>,\n    op9: OperatorFunction<H, I>\n  ): Observable<I>;\n  pipe<A, B, C, D, E, F, G, H, I>(\n    op1: OperatorFunction<T, A>,\n    op2: OperatorFunction<A, B>,\n    op3: OperatorFunction<B, C>,\n    op4: OperatorFunction<C, D>,\n    op5: OperatorFunction<D, E>,\n    op6: OperatorFunction<E, F>,\n    op7: OperatorFunction<F, G>,\n    op8: OperatorFunction<G, H>,\n    op9: OperatorFunction<H, I>,\n    ...operations: OperatorFunction<any, any>[]\n  ): Observable<unknown>;\n  /* tslint:enable:max-line-length */\n\n  /**\n   * Used to stitch together functional operators into a chain.\n   * @method pipe\n   * @return {Observable} the Observable result of all of the operators having\n   * been called in the order they were passed in.\n   *\n   * ## Example\n   *\n   * ```ts\n   * import { interval, filter, map, scan } from 'rxjs';\n   *\n   * interval(1000)\n   *   .pipe(\n   *     filter(x => x % 2 === 0),\n   *     map(x => x + x),\n   *     scan((acc, x) => acc + x)\n   *   )\n   *   .subscribe(x => console.log(x));\n   * ```\n   */\n  pipe(...operations: OperatorFunction<any, any>[]): Observable<any> {\n    return pipeFromArray(operations)(this);\n  }\n\n  /* tslint:disable:max-line-length */\n  /** @deprecated Replaced with {@link firstValueFrom} and {@link lastValueFrom}. Will be removed in v8. Details: https://rxjs.dev/deprecations/to-promise */\n  toPromise(): Promise<T | undefined>;\n  /** @deprecated Replaced with {@link firstValueFrom} and {@link lastValueFrom}. Will be removed in v8. Details: https://rxjs.dev/deprecations/to-promise */\n  toPromise(PromiseCtor: typeof Promise): Promise<T | undefined>;\n  /** @deprecated Replaced with {@link firstValueFrom} and {@link lastValueFrom}. Will be removed in v8. Details: https://rxjs.dev/deprecations/to-promise */\n  toPromise(PromiseCtor: PromiseConstructorLike): Promise<T | undefined>;\n  /* tslint:enable:max-line-length */\n\n  /**\n   * Subscribe to this Observable and get a Promise resolving on\n   * `complete` with the last emission (if any).\n   *\n   * **WARNING**: Only use this with observables you *know* will complete. If the source\n   * observable does not complete, you will end up with a promise that is hung up, and\n   * potentially all of the state of an async function hanging out in memory. To avoid\n   * this situation, look into adding something like {@link timeout}, {@link take},\n   * {@link takeWhile}, or {@link takeUntil} amongst others.\n   *\n   * @method toPromise\n   * @param [promiseCtor] a constructor function used to instantiate\n   * the Promise\n   * @return A Promise that resolves with the last value emit, or\n   * rejects on an error. If there were no emissions, Promise\n   * resolves with undefined.\n   * @deprecated Replaced with {@link firstValueFrom} and {@link lastValueFrom}. Will be removed in v8. Details: https://rxjs.dev/deprecations/to-promise\n   */\n  toPromise(promiseCtor?: PromiseConstructorLike): Promise<T | undefined> {\n    promiseCtor = getPromiseCtor(promiseCtor);\n\n    return new promiseCtor((resolve, reject) => {\n      let value: T | undefined;\n      this.subscribe(\n        (x: T) => (value = x),\n        (err: any) => reject(err),\n        () => resolve(value)\n      );\n    }) as Promise<T | undefined>;\n  }\n}\n\n/**\n * Decides between a passed promise constructor from consuming code,\n * A default configured promise constructor, and the native promise\n * constructor and returns it. If nothing can be found, it will throw\n * an error.\n * @param promiseCtor The optional promise constructor to passed by consuming code\n */\nfunction getPromiseCtor(promiseCtor: PromiseConstructorLike | undefined) {\n  return promiseCtor ?? config.Promise ?? Promise;\n}\n\nfunction isObserver<T>(value: any): value is Observer<T> {\n  return value && isFunction(value.next) && isFunction(value.error) && isFunction(value.complete);\n}\n\nfunction isSubscriber<T>(value: any): value is Subscriber<T> {\n  return (value && value instanceof Subscriber) || (isObserver(value) && isSubscription(value));\n}\n", "import { Observable } from '../Observable';\nimport { Subscriber } from '../Subscriber';\nimport { OperatorFunction } from '../types';\nimport { isFunction } from './isFunction';\n\n/**\n * Used to determine if an object is an Observable with a lift function.\n */\nexport function hasLift(source: any): source is { lift: InstanceType<typeof Observable>['lift'] } {\n  return isFunction(source?.lift);\n}\n\n/**\n * Creates an `OperatorFunction`. Used to define operators throughout the library in a concise way.\n * @param init The logic to connect the liftedSource to the subscriber at the moment of subscription.\n */\nexport function operate<T, R>(\n  init: (liftedSource: Observable<T>, subscriber: Subscriber<R>) => (() => void) | void\n): OperatorFunction<T, R> {\n  return (source: Observable<T>) => {\n    if (hasLift(source)) {\n      return source.lift(function (this: Subscriber<R>, liftedSource: Observable<T>) {\n        try {\n          return init(liftedSource, this);\n        } catch (err) {\n          this.error(err);\n        }\n      });\n    }\n    throw new TypeError('Unable to lift unknown Observable type');\n  };\n}\n", "import { Subscriber } from '../Subscriber';\n\n/**\n * Creates an instance of an `OperatorSubscriber`.\n * @param destination The downstream subscriber.\n * @param onNext Handles next values, only called if this subscriber is not stopped or closed. Any\n * error that occurs in this function is caught and sent to the `error` method of this subscriber.\n * @param onError Handles errors from the subscription, any errors that occur in this handler are caught\n * and send to the `destination` error handler.\n * @param onComplete Handles completion notification from the subscription. Any errors that occur in\n * this handler are sent to the `destination` error handler.\n * @param onFinalize Additional teardown logic here. This will only be called on teardown if the\n * subscriber itself is not already closed. This is called after all other teardown logic is executed.\n */\nexport function createOperatorSubscriber<T>(\n  destination: Subscriber<any>,\n  onNext?: (value: T) => void,\n  onComplete?: () => void,\n  onError?: (err: any) => void,\n  onFinalize?: () => void\n): Subscriber<T> {\n  return new OperatorSubscriber(destination, onNext, onComplete, onError, onFinalize);\n}\n\n/**\n * A generic helper for allowing operators to be created with a Subscriber and\n * use closures to capture necessary state from the operator function itself.\n */\nexport class OperatorSubscriber<T> extends Subscriber<T> {\n  /**\n   * Creates an instance of an `OperatorSubscriber`.\n   * @param destination The downstream subscriber.\n   * @param onNext Handles next values, only called if this subscriber is not stopped or closed. Any\n   * error that occurs in this function is caught and sent to the `error` method of this subscriber.\n   * @param onError Handles errors from the subscription, any errors that occur in this handler are caught\n   * and send to the `destination` error handler.\n   * @param onComplete Handles completion notification from the subscription. Any errors that occur in\n   * this handler are sent to the `destination` error handler.\n   * @param onFinalize Additional finalization logic here. This will only be called on finalization if the\n   * subscriber itself is not already closed. This is called after all other finalization logic is executed.\n   * @param shouldUnsubscribe An optional check to see if an unsubscribe call should truly unsubscribe.\n   * NOTE: This currently **ONLY** exists to support the strange behavior of {@link groupBy}, where unsubscription\n   * to the resulting observable does not actually disconnect from the source if there are active subscriptions\n   * to any grouped observable. (DO NOT EXPOSE OR USE EXTERNALLY!!!)\n   */\n  constructor(\n    destination: Subscriber<any>,\n    onNext?: (value: T) => void,\n    onComplete?: () => void,\n    onError?: (err: any) => void,\n    private onFinalize?: () => void,\n    private shouldUnsubscribe?: () => boolean\n  ) {\n    // It's important - for performance reasons - that all of this class's\n    // members are initialized and that they are always initialized in the same\n    // order. This will ensure that all OperatorSubscriber instances have the\n    // same hidden class in V8. This, in turn, will help keep the number of\n    // hidden classes involved in property accesses within the base class as\n    // low as possible. If the number of hidden classes involved exceeds four,\n    // the property accesses will become megamorphic and performance penalties\n    // will be incurred - i.e. inline caches won't be used.\n    //\n    // The reasons for ensuring all instances have the same hidden class are\n    // further discussed in this blog post from Benedikt Meurer:\n    // https://benediktmeurer.de/2018/03/23/impact-of-polymorphism-on-component-based-frameworks-like-react/\n    super(destination);\n    this._next = onNext\n      ? function (this: OperatorSubscriber<T>, value: T) {\n          try {\n            onNext(value);\n          } catch (err) {\n            destination.error(err);\n          }\n        }\n      : super._next;\n    this._error = onError\n      ? function (this: OperatorSubscriber<T>, err: any) {\n          try {\n            onError(err);\n          } catch (err) {\n            // Send any errors that occur down stream.\n            destination.error(err);\n          } finally {\n            // Ensure finalization.\n            this.unsubscribe();\n          }\n        }\n      : super._error;\n    this._complete = onComplete\n      ? function (this: OperatorSubscriber<T>) {\n          try {\n            onComplete();\n          } catch (err) {\n            // Send any errors that occur down stream.\n            destination.error(err);\n          } finally {\n            // Ensure finalization.\n            this.unsubscribe();\n          }\n        }\n      : super._complete;\n  }\n\n  unsubscribe() {\n    if (!this.shouldUnsubscribe || this.shouldUnsubscribe()) {\n      const { closed } = this;\n      super.unsubscribe();\n      // Execute additional teardown if we have any and we didn't already do so.\n      !closed && this.onFinalize?.();\n    }\n  }\n}\n", "import { Subscription } from '../Subscription';\n\ninterface AnimationFrameProvider {\n  schedule(callback: FrameRequestCallback): Subscription;\n  requestAnimationFrame: typeof requestAnimationFrame;\n  cancelAnimationFrame: typeof cancelAnimationFrame;\n  delegate:\n    | {\n        requestAnimationFrame: typeof requestAnimationFrame;\n        cancelAnimationFrame: typeof cancelAnimationFrame;\n      }\n    | undefined;\n}\n\nexport const animationFrameProvider: AnimationFrameProvider = {\n  // When accessing the delegate, use the variable rather than `this` so that\n  // the functions can be called without being bound to the provider.\n  schedule(callback) {\n    let request = requestAnimationFrame;\n    let cancel: typeof cancelAnimationFrame | undefined = cancelAnimationFrame;\n    const { delegate } = animationFrameProvider;\n    if (delegate) {\n      request = delegate.requestAnimationFrame;\n      cancel = delegate.cancelAnimationFrame;\n    }\n    const handle = request((timestamp) => {\n      // Clear the cancel function. The request has been fulfilled, so\n      // attempting to cancel the request upon unsubscription would be\n      // pointless.\n      cancel = undefined;\n      callback(timestamp);\n    });\n    return new Subscription(() => cancel?.(handle));\n  },\n  requestAnimationFrame(...args) {\n    const { delegate } = animationFrameProvider;\n    return (delegate?.requestAnimationFrame || requestAnimationFrame)(...args);\n  },\n  cancelAnimationFrame(...args) {\n    const { delegate } = animationFrameProvider;\n    return (delegate?.cancelAnimationFrame || cancelAnimationFrame)(...args);\n  },\n  delegate: undefined,\n};\n", "import { createErrorClass } from './createErrorClass';\n\nexport interface ObjectUnsubscribedError extends Error {}\n\nexport interface ObjectUnsubscribedErrorCtor {\n  /**\n   * @deprecated Internal implementation detail. Do not construct error instances.\n   * Cannot be tagged as internal: https://github.com/ReactiveX/rxjs/issues/6269\n   */\n  new (): ObjectUnsubscribedError;\n}\n\n/**\n * An error thrown when an action is invalid because the object has been\n * unsubscribed.\n *\n * @see {@link Subject}\n * @see {@link BehaviorSubject}\n *\n * @class ObjectUnsubscribedError\n */\nexport const ObjectUnsubscribedError: ObjectUnsubscribedErrorCtor = createErrorClass(\n  (_super) =>\n    function ObjectUnsubscribedErrorImpl(this: any) {\n      _super(this);\n      this.name = 'ObjectUnsubscribedError';\n      this.message = 'object unsubscribed';\n    }\n);\n", "import { Operator } from './Operator';\nimport { Observable } from './Observable';\nimport { Subscriber } from './Subscriber';\nimport { Subscription, EMPTY_SUBSCRIPTION } from './Subscription';\nimport { Observer, SubscriptionLike, TeardownLogic } from './types';\nimport { ObjectUnsubscribedError } from './util/ObjectUnsubscribedError';\nimport { arrRemove } from './util/arrRemove';\nimport { errorContext } from './util/errorContext';\n\n/**\n * A Subject is a special type of Observable that allows values to be\n * multicasted to many Observers. Subjects are like EventEmitters.\n *\n * Every Subject is an Observable and an Observer. You can subscribe to a\n * Subject, and you can call next to feed values as well as error and complete.\n */\nexport class Subject<T> extends Observable<T> implements SubscriptionLike {\n  closed = false;\n\n  private currentObservers: Observer<T>[] | null = null;\n\n  /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n  observers: Observer<T>[] = [];\n  /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n  isStopped = false;\n  /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n  hasError = false;\n  /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n  thrownError: any = null;\n\n  /**\n   * Creates a \"subject\" by basically gluing an observer to an observable.\n   *\n   * @nocollapse\n   * @deprecated Recommended you do not use. Will be removed at some point in the future. Plans for replacement still under discussion.\n   */\n  static create: (...args: any[]) => any = <T>(destination: Observer<T>, source: Observable<T>): AnonymousSubject<T> => {\n    return new AnonymousSubject<T>(destination, source);\n  };\n\n  constructor() {\n    // NOTE: This must be here to obscure Observable's constructor.\n    super();\n  }\n\n  /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n  lift<R>(operator: Operator<T, R>): Observable<R> {\n    const subject = new AnonymousSubject(this, this);\n    subject.operator = operator as any;\n    return subject as any;\n  }\n\n  /** @internal */\n  protected _throwIfClosed() {\n    if (this.closed) {\n      throw new ObjectUnsubscribedError();\n    }\n  }\n\n  next(value: T) {\n    errorContext(() => {\n      this._throwIfClosed();\n      if (!this.isStopped) {\n        if (!this.currentObservers) {\n          this.currentObservers = Array.from(this.observers);\n        }\n        for (const observer of this.currentObservers) {\n          observer.next(value);\n        }\n      }\n    });\n  }\n\n  error(err: any) {\n    errorContext(() => {\n      this._throwIfClosed();\n      if (!this.isStopped) {\n        this.hasError = this.isStopped = true;\n        this.thrownError = err;\n        const { observers } = this;\n        while (observers.length) {\n          observers.shift()!.error(err);\n        }\n      }\n    });\n  }\n\n  complete() {\n    errorContext(() => {\n      this._throwIfClosed();\n      if (!this.isStopped) {\n        this.isStopped = true;\n        const { observers } = this;\n        while (observers.length) {\n          observers.shift()!.complete();\n        }\n      }\n    });\n  }\n\n  unsubscribe() {\n    this.isStopped = this.closed = true;\n    this.observers = this.currentObservers = null!;\n  }\n\n  get observed() {\n    return this.observers?.length > 0;\n  }\n\n  /** @internal */\n  protected _trySubscribe(subscriber: Subscriber<T>): TeardownLogic {\n    this._throwIfClosed();\n    return super._trySubscribe(subscriber);\n  }\n\n  /** @internal */\n  protected _subscribe(subscriber: Subscriber<T>): Subscription {\n    this._throwIfClosed();\n    this._checkFinalizedStatuses(subscriber);\n    return this._innerSubscribe(subscriber);\n  }\n\n  /** @internal */\n  protected _innerSubscribe(subscriber: Subscriber<any>) {\n    const { hasError, isStopped, observers } = this;\n    if (hasError || isStopped) {\n      return EMPTY_SUBSCRIPTION;\n    }\n    this.currentObservers = null;\n    observers.push(subscriber);\n    return new Subscription(() => {\n      this.currentObservers = null;\n      arrRemove(observers, subscriber);\n    });\n  }\n\n  /** @internal */\n  protected _checkFinalizedStatuses(subscriber: Subscriber<any>) {\n    const { hasError, thrownError, isStopped } = this;\n    if (hasError) {\n      subscriber.error(thrownError);\n    } else if (isStopped) {\n      subscriber.complete();\n    }\n  }\n\n  /**\n   * Creates a new Observable with this Subject as the source. You can do this\n   * to create custom Observer-side logic of the Subject and conceal it from\n   * code that uses the Observable.\n   * @return {Observable} Observable that the Subject casts to\n   */\n  asObservable(): Observable<T> {\n    const observable: any = new Observable<T>();\n    observable.source = this;\n    return observable;\n  }\n}\n\n/**\n * @class AnonymousSubject<T>\n */\nexport class AnonymousSubject<T> extends Subject<T> {\n  constructor(\n    /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n    public destination?: Observer<T>,\n    source?: Observable<T>\n  ) {\n    super();\n    this.source = source;\n  }\n\n  next(value: T) {\n    this.destination?.next?.(value);\n  }\n\n  error(err: any) {\n    this.destination?.error?.(err);\n  }\n\n  complete() {\n    this.destination?.complete?.();\n  }\n\n  /** @internal */\n  protected _subscribe(subscriber: Subscriber<T>): Subscription {\n    return this.source?.subscribe(subscriber) ?? EMPTY_SUBSCRIPTION;\n  }\n}\n", "import { TimestampProvider } from '../types';\n\ninterface DateTimestampProvider extends TimestampProvider {\n  delegate: TimestampProvider | undefined;\n}\n\nexport const dateTimestampProvider: DateTimestampProvider = {\n  now() {\n    // Use the variable rather than `this` so that the function can be called\n    // without being bound to the provider.\n    return (dateTimestampProvider.delegate || Date).now();\n  },\n  delegate: undefined,\n};\n", "import { Subject } from './Subject';\nimport { TimestampProvider } from './types';\nimport { Subscriber } from './Subscriber';\nimport { Subscription } from './Subscription';\nimport { dateTimestampProvider } from './scheduler/dateTimestampProvider';\n\n/**\n * A variant of {@link Subject} that \"replays\" old values to new subscribers by emitting them when they first subscribe.\n *\n * `ReplaySubject` has an internal buffer that will store a specified number of values that it has observed. Like `Subject`,\n * `ReplaySubject` \"observes\" values by having them passed to its `next` method. When it observes a value, it will store that\n * value for a time determined by the configuration of the `ReplaySubject`, as passed to its constructor.\n *\n * When a new subscriber subscribes to the `ReplaySubject` instance, it will synchronously emit all values in its buffer in\n * a First-In-First-Out (FIFO) manner. The `ReplaySubject` will also complete, if it has observed completion; and it will\n * error if it has observed an error.\n *\n * There are two main configuration items to be concerned with:\n *\n * 1. `bufferSize` - This will determine how many items are stored in the buffer, defaults to infinite.\n * 2. `windowTime` - The amount of time to hold a value in the buffer before removing it from the buffer.\n *\n * Both configurations may exist simultaneously. So if you would like to buffer a maximum of 3 values, as long as the values\n * are less than 2 seconds old, you could do so with a `new ReplaySubject(3, 2000)`.\n *\n * ### Differences with BehaviorSubject\n *\n * `BehaviorSubject` is similar to `new ReplaySubject(1)`, with a couple of exceptions:\n *\n * 1. `BehaviorSubject` comes \"primed\" with a single value upon construction.\n * 2. `ReplaySubject` will replay values, even after observing an error, where `BehaviorSubject` will not.\n *\n * @see {@link Subject}\n * @see {@link BehaviorSubject}\n * @see {@link shareReplay}\n */\nexport class ReplaySubject<T> extends Subject<T> {\n  private _buffer: (T | number)[] = [];\n  private _infiniteTimeWindow = true;\n\n  /**\n   * @param bufferSize The size of the buffer to replay on subscription\n   * @param windowTime The amount of time the buffered items will stay buffered\n   * @param timestampProvider An object with a `now()` method that provides the current timestamp. This is used to\n   * calculate the amount of time something has been buffered.\n   */\n  constructor(\n    private _bufferSize = Infinity,\n    private _windowTime = Infinity,\n    private _timestampProvider: TimestampProvider = dateTimestampProvider\n  ) {\n    super();\n    this._infiniteTimeWindow = _windowTime === Infinity;\n    this._bufferSize = Math.max(1, _bufferSize);\n    this._windowTime = Math.max(1, _windowTime);\n  }\n\n  next(value: T): void {\n    const { isStopped, _buffer, _infiniteTimeWindow, _timestampProvider, _windowTime } = this;\n    if (!isStopped) {\n      _buffer.push(value);\n      !_infiniteTimeWindow && _buffer.push(_timestampProvider.now() + _windowTime);\n    }\n    this._trimBuffer();\n    super.next(value);\n  }\n\n  /** @internal */\n  protected _subscribe(subscriber: Subscriber<T>): Subscription {\n    this._throwIfClosed();\n    this._trimBuffer();\n\n    const subscription = this._innerSubscribe(subscriber);\n\n    const { _infiniteTimeWindow, _buffer } = this;\n    // We use a copy here, so reentrant code does not mutate our array while we're\n    // emitting it to a new subscriber.\n    const copy = _buffer.slice();\n    for (let i = 0; i < copy.length && !subscriber.closed; i += _infiniteTimeWindow ? 1 : 2) {\n      subscriber.next(copy[i] as T);\n    }\n\n    this._checkFinalizedStatuses(subscriber);\n\n    return subscription;\n  }\n\n  private _trimBuffer() {\n    const { _bufferSize, _timestampProvider, _buffer, _infiniteTimeWindow } = this;\n    // If we don't have an infinite buffer size, and we're over the length,\n    // use splice to truncate the old buffer values off. Note that we have to\n    // double the size for instances where we're not using an infinite time window\n    // because we're storing the values and the timestamps in the same array.\n    const adjustedBufferSize = (_infiniteTimeWindow ? 1 : 2) * _bufferSize;\n    _bufferSize < Infinity && adjustedBufferSize < _buffer.length && _buffer.splice(0, _buffer.length - adjustedBufferSize);\n\n    // Now, if we're not in an infinite time window, remove all values where the time is\n    // older than what is allowed.\n    if (!_infiniteTimeWindow) {\n      const now = _timestampProvider.now();\n      let last = 0;\n      // Search the array for the first timestamp that isn't expired and\n      // truncate the buffer up to that point.\n      for (let i = 1; i < _buffer.length && (_buffer[i] as number) <= now; i += 2) {\n        last = i;\n      }\n      last && _buffer.splice(0, last + 1);\n    }\n  }\n}\n", "import { Scheduler } from '../Scheduler';\nimport { Subscription } from '../Subscription';\nimport { SchedulerAction } from '../types';\n\n/**\n * A unit of work to be executed in a `scheduler`. An action is typically\n * created from within a {@link SchedulerLike} and an RxJS user does not need to concern\n * themselves about creating and manipulating an Action.\n *\n * ```ts\n * class Action<T> extends Subscription {\n *   new (scheduler: Scheduler, work: (state?: T) => void);\n *   schedule(state?: T, delay: number = 0): Subscription;\n * }\n * ```\n *\n * @class Action<T>\n */\nexport class Action<T> extends Subscription {\n  constructor(scheduler: Scheduler, work: (this: SchedulerAction<T>, state?: T) => void) {\n    super();\n  }\n  /**\n   * Schedules this action on its parent {@link SchedulerLike} for execution. May be passed\n   * some context object, `state`. May happen at some point in the future,\n   * according to the `delay` parameter, if specified.\n   * @param {T} [state] Some contextual data that the `work` function uses when\n   * called by the Scheduler.\n   * @param {number} [delay] Time to wait before executing the work, where the\n   * time unit is implicit and defined by the Scheduler.\n   * @return {void}\n   */\n  public schedule(state?: T, delay: number = 0): Subscription {\n    return this;\n  }\n}\n", "import type { TimerHandle } from './timerHandle';\ntype SetIntervalFunction = (handler: () => void, timeout?: number, ...args: any[]) => TimerHandle;\ntype ClearIntervalFunction = (handle: TimerHandle) => void;\n\ninterface IntervalProvider {\n  setInterval: SetIntervalFunction;\n  clearInterval: ClearIntervalFunction;\n  delegate:\n    | {\n        setInterval: SetIntervalFunction;\n        clearInterval: ClearIntervalFunction;\n      }\n    | undefined;\n}\n\nexport const intervalProvider: IntervalProvider = {\n  // When accessing the delegate, use the variable rather than `this` so that\n  // the functions can be called without being bound to the provider.\n  setInterval(handler: () => void, timeout?: number, ...args) {\n    const { delegate } = intervalProvider;\n    if (delegate?.setInterval) {\n      return delegate.setInterval(handler, timeout, ...args);\n    }\n    return setInterval(handler, timeout, ...args);\n  },\n  clearInterval(handle) {\n    const { delegate } = intervalProvider;\n    return (delegate?.clearInterval || clearInterval)(handle as any);\n  },\n  delegate: undefined,\n};\n", "import { Action } from './Action';\nimport { SchedulerAction } from '../types';\nimport { Subscription } from '../Subscription';\nimport { AsyncScheduler } from './AsyncScheduler';\nimport { intervalProvider } from './intervalProvider';\nimport { arrRemove } from '../util/arrRemove';\nimport { TimerHandle } from './timerHandle';\n\nexport class AsyncAction<T> extends Action<T> {\n  public id: TimerHandle | undefined;\n  public state?: T;\n  // @ts-ignore: Property has no initializer and is not definitely assigned\n  public delay: number;\n  protected pending: boolean = false;\n\n  constructor(protected scheduler: AsyncScheduler, protected work: (this: SchedulerAction<T>, state?: T) => void) {\n    super(scheduler, work);\n  }\n\n  public schedule(state?: T, delay: number = 0): Subscription {\n    if (this.closed) {\n      return this;\n    }\n\n    // Always replace the current state with the new state.\n    this.state = state;\n\n    const id = this.id;\n    const scheduler = this.scheduler;\n\n    //\n    // Important implementation note:\n    //\n    // Actions only execute once by default, unless rescheduled from within the\n    // scheduled callback. This allows us to implement single and repeat\n    // actions via the same code path, without adding API surface area, as well\n    // as mimic traditional recursion but across asynchronous boundaries.\n    //\n    // However, JS runtimes and timers distinguish between intervals achieved by\n    // serial `setTimeout` calls vs. a single `setInterval` call. An interval of\n    // serial `setTimeout` calls can be individually delayed, which delays\n    // scheduling the next `setTimeout`, and so on. `setInterval` attempts to\n    // guarantee the interval callback will be invoked more precisely to the\n    // interval period, regardless of load.\n    //\n    // Therefore, we use `setInterval` to schedule single and repeat actions.\n    // If the action reschedules itself with the same delay, the interval is not\n    // canceled. If the action doesn't reschedule, or reschedules with a\n    // different delay, the interval will be canceled after scheduled callback\n    // execution.\n    //\n    if (id != null) {\n      this.id = this.recycleAsyncId(scheduler, id, delay);\n    }\n\n    // Set the pending flag indicating that this action has been scheduled, or\n    // has recursively rescheduled itself.\n    this.pending = true;\n\n    this.delay = delay;\n    // If this action has already an async Id, don't request a new one.\n    this.id = this.id ?? this.requestAsyncId(scheduler, this.id, delay);\n\n    return this;\n  }\n\n  protected requestAsyncId(scheduler: AsyncScheduler, _id?: TimerHandle, delay: number = 0): TimerHandle {\n    return intervalProvider.setInterval(scheduler.flush.bind(scheduler, this), delay);\n  }\n\n  protected recycleAsyncId(_scheduler: AsyncScheduler, id?: TimerHandle, delay: number | null = 0): TimerHandle | undefined {\n    // If this action is rescheduled with the same delay time, don't clear the interval id.\n    if (delay != null && this.delay === delay && this.pending === false) {\n      return id;\n    }\n    // Otherwise, if the action's delay time is different from the current delay,\n    // or the action has been rescheduled before it's executed, clear the interval id\n    if (id != null) {\n      intervalProvider.clearInterval(id);\n    }\n\n    return undefined;\n  }\n\n  /**\n   * Immediately executes this action and the `work` it contains.\n   * @return {any}\n   */\n  public execute(state: T, delay: number): any {\n    if (this.closed) {\n      return new Error('executing a cancelled action');\n    }\n\n    this.pending = false;\n    const error = this._execute(state, delay);\n    if (error) {\n      return error;\n    } else if (this.pending === false && this.id != null) {\n      // Dequeue if the action didn't reschedule itself. Don't call\n      // unsubscribe(), because the action could reschedule later.\n      // For example:\n      // ```\n      // scheduler.schedule(function doWork(counter) {\n      //   /* ... I'm a busy worker bee ... */\n      //   var originalAction = this;\n      //   /* wait 100ms before rescheduling the action */\n      //   setTimeout(function () {\n      //     originalAction.schedule(counter + 1);\n      //   }, 100);\n      // }, 1000);\n      // ```\n      this.id = this.recycleAsyncId(this.scheduler, this.id, null);\n    }\n  }\n\n  protected _execute(state: T, _delay: number): any {\n    let errored: boolean = false;\n    let errorValue: any;\n    try {\n      this.work(state);\n    } catch (e) {\n      errored = true;\n      // HACK: Since code elsewhere is relying on the \"truthiness\" of the\n      // return here, we can't have it return \"\" or 0 or false.\n      // TODO: Clean this up when we refactor schedulers mid-version-8 or so.\n      errorValue = e ? e : new Error('Scheduled action threw falsy error');\n    }\n    if (errored) {\n      this.unsubscribe();\n      return errorValue;\n    }\n  }\n\n  unsubscribe() {\n    if (!this.closed) {\n      const { id, scheduler } = this;\n      const { actions } = scheduler;\n\n      this.work = this.state = this.scheduler = null!;\n      this.pending = false;\n\n      arrRemove(actions, this);\n      if (id != null) {\n        this.id = this.recycleAsyncId(scheduler, id, null);\n      }\n\n      this.delay = null!;\n      super.unsubscribe();\n    }\n  }\n}\n", "import { Action } from './scheduler/Action';\nimport { Subscription } from './Subscription';\nimport { SchedulerLike, SchedulerAction } from './types';\nimport { dateTimestampProvider } from './scheduler/dateTimestampProvider';\n\n/**\n * An execution context and a data structure to order tasks and schedule their\n * execution. Provides a notion of (potentially virtual) time, through the\n * `now()` getter method.\n *\n * Each unit of work in a Scheduler is called an `Action`.\n *\n * ```ts\n * class Scheduler {\n *   now(): number;\n *   schedule(work, delay?, state?): Subscription;\n * }\n * ```\n *\n * @class Scheduler\n * @deprecated Scheduler is an internal implementation detail of RxJS, and\n * should not be used directly. Rather, create your own class and implement\n * {@link SchedulerLike}. Will be made internal in v8.\n */\nexport class Scheduler implements SchedulerLike {\n  public static now: () => number = dateTimestampProvider.now;\n\n  constructor(private schedulerActionCtor: typeof Action, now: () => number = Scheduler.now) {\n    this.now = now;\n  }\n\n  /**\n   * A getter method that returns a number representing the current time\n   * (at the time this function was called) according to the scheduler's own\n   * internal clock.\n   * @return {number} A number that represents the current time. May or may not\n   * have a relation to wall-clock time. May or may not refer to a time unit\n   * (e.g. milliseconds).\n   */\n  public now: () => number;\n\n  /**\n   * Schedules a function, `work`, for execution. May happen at some point in\n   * the future, according to the `delay` parameter, if specified. May be passed\n   * some context object, `state`, which will be passed to the `work` function.\n   *\n   * The given arguments will be processed an stored as an Action object in a\n   * queue of actions.\n   *\n   * @param {function(state: ?T): ?Subscription} work A function representing a\n   * task, or some unit of work to be executed by the Scheduler.\n   * @param {number} [delay] Time to wait before executing the work, where the\n   * time unit is implicit and defined by the Scheduler itself.\n   * @param {T} [state] Some contextual data that the `work` function uses when\n   * called by the Scheduler.\n   * @return {Subscription} A subscription in order to be able to unsubscribe\n   * the scheduled work.\n   */\n  public schedule<T>(work: (this: SchedulerAction<T>, state?: T) => void, delay: number = 0, state?: T): Subscription {\n    return new this.schedulerActionCtor<T>(this, work).schedule(state, delay);\n  }\n}\n", "import { Scheduler } from '../Scheduler';\nimport { Action } from './Action';\nimport { AsyncAction } from './AsyncAction';\nimport { TimerHandle } from './timerHandle';\n\nexport class AsyncScheduler extends Scheduler {\n  public actions: Array<AsyncAction<any>> = [];\n  /**\n   * A flag to indicate whether the Scheduler is currently executing a batch of\n   * queued actions.\n   * @type {boolean}\n   * @internal\n   */\n  public _active: boolean = false;\n  /**\n   * An internal ID used to track the latest asynchronous task such as those\n   * coming from `setTimeout`, `setInterval`, `requestAnimationFrame`, and\n   * others.\n   * @type {any}\n   * @internal\n   */\n  public _scheduled: TimerHandle | undefined;\n\n  constructor(SchedulerAction: typeof Action, now: () => number = Scheduler.now) {\n    super(SchedulerAction, now);\n  }\n\n  public flush(action: AsyncAction<any>): void {\n    const { actions } = this;\n\n    if (this._active) {\n      actions.push(action);\n      return;\n    }\n\n    let error: any;\n    this._active = true;\n\n    do {\n      if ((error = action.execute(action.state, action.delay))) {\n        break;\n      }\n    } while ((action = actions.shift()!)); // exhaust the scheduler queue\n\n    this._active = false;\n\n    if (error) {\n      while ((action = actions.shift()!)) {\n        action.unsubscribe();\n      }\n      throw error;\n    }\n  }\n}\n", "import { AsyncAction } from './AsyncAction';\nimport { AsyncScheduler } from './AsyncScheduler';\n\n/**\n *\n * Async Scheduler\n *\n * <span class=\"informal\">Schedule task as if you used setTimeout(task, duration)</span>\n *\n * `async` scheduler schedules tasks asynchronously, by putting them on the JavaScript\n * event loop queue. It is best used to delay tasks in time or to schedule tasks repeating\n * in intervals.\n *\n * If you just want to \"defer\" task, that is to perform it right after currently\n * executing synchronous code ends (commonly achieved by `setTimeout(deferredTask, 0)`),\n * better choice will be the {@link asapScheduler} scheduler.\n *\n * ## Examples\n * Use async scheduler to delay task\n * ```ts\n * import { asyncScheduler } from 'rxjs';\n *\n * const task = () => console.log('it works!');\n *\n * asyncScheduler.schedule(task, 2000);\n *\n * // After 2 seconds logs:\n * // \"it works!\"\n * ```\n *\n * Use async scheduler to repeat task in intervals\n * ```ts\n * import { asyncScheduler } from 'rxjs';\n *\n * function task(state) {\n *   console.log(state);\n *   this.schedule(state + 1, 1000); // `this` references currently executing Action,\n *                                   // which we reschedule with new state and delay\n * }\n *\n * asyncScheduler.schedule(task, 3000, 0);\n *\n * // Logs:\n * // 0 after 3s\n * // 1 after 4s\n * // 2 after 5s\n * // 3 after 6s\n * ```\n */\n\nexport const asyncScheduler = new AsyncScheduler(AsyncAction);\n\n/**\n * @deprecated Renamed to {@link asyncScheduler}. Will be removed in v8.\n */\nexport const async = asyncScheduler;\n", "import { AsyncAction } from './AsyncAction';\nimport { AnimationFrameScheduler } from './AnimationFrameScheduler';\nimport { SchedulerAction } from '../types';\nimport { animationFrameProvider } from './animationFrameProvider';\nimport { TimerHandle } from './timerHandle';\n\nexport class AnimationFrameAction<T> extends AsyncAction<T> {\n  constructor(protected scheduler: AnimationFrameScheduler, protected work: (this: SchedulerAction<T>, state?: T) => void) {\n    super(scheduler, work);\n  }\n\n  protected requestAsyncId(scheduler: AnimationFrameScheduler, id?: TimerHandle, delay: number = 0): TimerHandle {\n    // If delay is greater than 0, request as an async action.\n    if (delay !== null && delay > 0) {\n      return super.requestAsyncId(scheduler, id, delay);\n    }\n    // Push the action to the end of the scheduler queue.\n    scheduler.actions.push(this);\n    // If an animation frame has already been requested, don't request another\n    // one. If an animation frame hasn't been requested yet, request one. Return\n    // the current animation frame request id.\n    return scheduler._scheduled || (scheduler._scheduled = animationFrameProvider.requestAnimationFrame(() => scheduler.flush(undefined)));\n  }\n\n  protected recycleAsyncId(scheduler: AnimationFrameScheduler, id?: TimerHandle, delay: number = 0): TimerHandle | undefined {\n    // If delay exists and is greater than 0, or if the delay is null (the\n    // action wasn't rescheduled) but was originally scheduled as an async\n    // action, then recycle as an async action.\n    if (delay != null ? delay > 0 : this.delay > 0) {\n      return super.recycleAsyncId(scheduler, id, delay);\n    }\n    // If the scheduler queue has no remaining actions with the same async id,\n    // cancel the requested animation frame and set the scheduled flag to\n    // undefined so the next AnimationFrameAction will request its own.\n    const { actions } = scheduler;\n    if (id != null && actions[actions.length - 1]?.id !== id) {\n      animationFrameProvider.cancelAnimationFrame(id as number);\n      scheduler._scheduled = undefined;\n    }\n    // Return undefined so the action knows to request a new async id if it's rescheduled.\n    return undefined;\n  }\n}\n", "import { AsyncAction } from './AsyncAction';\nimport { AsyncScheduler } from './AsyncScheduler';\n\nexport class AnimationFrameScheduler extends AsyncScheduler {\n  public flush(action?: AsyncAction<any>): void {\n    this._active = true;\n    // The async id that effects a call to flush is stored in _scheduled.\n    // Before executing an action, it's necessary to check the action's async\n    // id to determine whether it's supposed to be executed in the current\n    // flush.\n    // Previous implementations of this method used a count to determine this,\n    // but that was unsound, as actions that are unsubscribed - i.e. cancelled -\n    // are removed from the actions array and that can shift actions that are\n    // scheduled to be executed in a subsequent flush into positions at which\n    // they are executed within the current flush.\n    const flushId = this._scheduled;\n    this._scheduled = undefined;\n\n    const { actions } = this;\n    let error: any;\n    action = action || actions.shift()!;\n\n    do {\n      if ((error = action.execute(action.state, action.delay))) {\n        break;\n      }\n    } while ((action = actions[0]) && action.id === flushId && actions.shift());\n\n    this._active = false;\n\n    if (error) {\n      while ((action = actions[0]) && action.id === flushId && actions.shift()) {\n        action.unsubscribe();\n      }\n      throw error;\n    }\n  }\n}\n", "import { AnimationFrameAction } from './AnimationFrameAction';\nimport { AnimationFrameScheduler } from './AnimationFrameScheduler';\n\n/**\n *\n * Animation Frame Scheduler\n *\n * <span class=\"informal\">Perform task when `window.requestAnimationFrame` would fire</span>\n *\n * When `animationFrame` scheduler is used with delay, it will fall back to {@link asyncScheduler} scheduler\n * behaviour.\n *\n * Without delay, `animationFrame` scheduler can be used to create smooth browser animations.\n * It makes sure scheduled task will happen just before next browser content repaint,\n * thus performing animations as efficiently as possible.\n *\n * ## Example\n * Schedule div height animation\n * ```ts\n * // html: <div style=\"background: #0ff;\"></div>\n * import { animationFrameScheduler } from 'rxjs';\n *\n * const div = document.querySelector('div');\n *\n * animationFrameScheduler.schedule(function(height) {\n *   div.style.height = height + \"px\";\n *\n *   this.schedule(height + 1);  // `this` references currently executing Action,\n *                               // which we reschedule with new state\n * }, 0, 0);\n *\n * // You will see a div element growing in height\n * ```\n */\n\nexport const animationFrameScheduler = new AnimationFrameScheduler(AnimationFrameAction);\n\n/**\n * @deprecated Renamed to {@link animationFrameScheduler}. Will be removed in v8.\n */\nexport const animationFrame = animationFrameScheduler;\n", "import { Observable } from '../Observable';\nimport { SchedulerLike } from '../types';\n\n/**\n * A simple Observable that emits no items to the Observer and immediately\n * emits a complete notification.\n *\n * <span class=\"informal\">Just emits 'complete', and nothing else.</span>\n *\n * ![](empty.png)\n *\n * A simple Observable that only emits the complete notification. It can be used\n * for composing with other Observables, such as in a {@link mergeMap}.\n *\n * ## Examples\n *\n * Log complete notification\n *\n * ```ts\n * import { EMPTY } from 'rxjs';\n *\n * EMPTY.subscribe({\n *   next: () => console.log('Next'),\n *   complete: () => console.log('Complete!')\n * });\n *\n * // Outputs\n * // Complete!\n * ```\n *\n * Emit the number 7, then complete\n *\n * ```ts\n * import { EMPTY, startWith } from 'rxjs';\n *\n * const result = EMPTY.pipe(startWith(7));\n * result.subscribe(x => console.log(x));\n *\n * // Outputs\n * // 7\n * ```\n *\n * Map and flatten only odd numbers to the sequence `'a'`, `'b'`, `'c'`\n *\n * ```ts\n * import { interval, mergeMap, of, EMPTY } from 'rxjs';\n *\n * const interval$ = interval(1000);\n * const result = interval$.pipe(\n *   mergeMap(x => x % 2 === 1 ? of('a', 'b', 'c') : EMPTY),\n * );\n * result.subscribe(x => console.log(x));\n *\n * // Results in the following to the console:\n * // x is equal to the count on the interval, e.g. (0, 1, 2, 3, ...)\n * // x will occur every 1000ms\n * // if x % 2 is equal to 1, print a, b, c (each on its own)\n * // if x % 2 is not equal to 1, nothing will be output\n * ```\n *\n * @see {@link Observable}\n * @see {@link NEVER}\n * @see {@link of}\n * @see {@link throwError}\n */\nexport const EMPTY = new Observable<never>((subscriber) => subscriber.complete());\n\n/**\n * @param scheduler A {@link SchedulerLike} to use for scheduling\n * the emission of the complete notification.\n * @deprecated Replaced with the {@link EMPTY} constant or {@link scheduled} (e.g. `scheduled([], scheduler)`). Will be removed in v8.\n */\nexport function empty(scheduler?: SchedulerLike) {\n  return scheduler ? emptyScheduled(scheduler) : EMPTY;\n}\n\nfunction emptyScheduled(scheduler: SchedulerLike) {\n  return new Observable<never>((subscriber) => scheduler.schedule(() => subscriber.complete()));\n}\n", "import { SchedulerLike } from '../types';\nimport { isFunction } from './isFunction';\n\nexport function isScheduler(value: any): value is SchedulerLike {\n  return value && isFunction(value.schedule);\n}\n", "import { SchedulerLike } from '../types';\nimport { isFunction } from './isFunction';\nimport { isScheduler } from './isScheduler';\n\nfunction last<T>(arr: T[]): T | undefined {\n  return arr[arr.length - 1];\n}\n\nexport function popResultSelector(args: any[]): ((...args: unknown[]) => unknown) | undefined {\n  return isFunction(last(args)) ? args.pop() : undefined;\n}\n\nexport function popScheduler(args: any[]): SchedulerLike | undefined {\n  return isScheduler(last(args)) ? args.pop() : undefined;\n}\n\nexport function popNumber(args: any[], defaultValue: number): number {\n  return typeof last(args) === 'number' ? args.pop()! : defaultValue;\n}\n", "export const isArrayLike = (<T>(x: any): x is ArrayLike<T> => x && typeof x.length === 'number' && typeof x !== 'function');", "import { isFunction } from \"./isFunction\";\n\n/**\n * Tests to see if the object is \"thennable\".\n * @param value the object to test\n */\nexport function isPromise(value: any): value is PromiseLike<any> {\n  return isFunction(value?.then);\n}\n", "import { InteropObservable } from '../types';\nimport { observable as Symbol_observable } from '../symbol/observable';\nimport { isFunction } from './isFunction';\n\n/** Identifies an input as being Observable (but not necessary an Rx Observable) */\nexport function isInteropObservable(input: any): input is InteropObservable<any> {\n  return isFunction(input[Symbol_observable]);\n}\n", "import { isFunction } from './isFunction';\n\nexport function isAsyncIterable<T>(obj: any): obj is AsyncIterable<T> {\n  return Symbol.asyncIterator && isFunction(obj?.[Symbol.asyncIterator]);\n}\n", "/**\n * Creates the TypeError to throw if an invalid object is passed to `from` or `scheduled`.\n * @param input The object that was passed.\n */\nexport function createInvalidObservableTypeError(input: any) {\n  // TODO: We should create error codes that can be looked up, so this can be less verbose.\n  return new TypeError(\n    `You provided ${\n      input !== null && typeof input === 'object' ? 'an invalid object' : `'${input}'`\n    } where a stream was expected. You can provide an Observable, Promise, ReadableStream, Array, AsyncIterable, or Iterable.`\n  );\n}\n", "export function getSymbolIterator(): symbol {\n  if (typeof Symbol !== 'function' || !Symbol.iterator) {\n    return '@@iterator' as any;\n  }\n\n  return Symbol.iterator;\n}\n\nexport const iterator = getSymbolIterator();\n", "import { iterator as Symbol_iterator } from '../symbol/iterator';\nimport { isFunction } from './isFunction';\n\n/** Identifies an input as being an Iterable */\nexport function isIterable(input: any): input is Iterable<any> {\n  return isFunction(input?.[Symbol_iterator]);\n}\n", "import { ReadableStreamLike } from '../types';\nimport { isFunction } from './isFunction';\n\nexport async function* readableStreamLikeToAsyncGenerator<T>(readableStream: ReadableStreamLike<T>): AsyncGenerator<T> {\n  const reader = readableStream.getReader();\n  try {\n    while (true) {\n      const { value, done } = await reader.read();\n      if (done) {\n        return;\n      }\n      yield value!;\n    }\n  } finally {\n    reader.releaseLock();\n  }\n}\n\nexport function isReadableStreamLike<T>(obj: any): obj is ReadableStreamLike<T> {\n  // We don't want to use instanceof checks because they would return\n  // false for instances from another Realm, like an <iframe>.\n  return isFunction(obj?.getReader);\n}\n", "import { isArrayLike } from '../util/isArrayLike';\nimport { isPromise } from '../util/isPromise';\nimport { Observable } from '../Observable';\nimport { ObservableInput, ObservedValueOf, ReadableStreamLike } from '../types';\nimport { isInteropObservable } from '../util/isInteropObservable';\nimport { isAsyncIterable } from '../util/isAsyncIterable';\nimport { createInvalidObservableTypeError } from '../util/throwUnobservableError';\nimport { isIterable } from '../util/isIterable';\nimport { isReadableStreamLike, readableStreamLikeToAsyncGenerator } from '../util/isReadableStreamLike';\nimport { Subscriber } from '../Subscriber';\nimport { isFunction } from '../util/isFunction';\nimport { reportUnhandledError } from '../util/reportUnhandledError';\nimport { observable as Symbol_observable } from '../symbol/observable';\n\nexport function innerFrom<O extends ObservableInput<any>>(input: O): Observable<ObservedValueOf<O>>;\nexport function innerFrom<T>(input: ObservableInput<T>): Observable<T> {\n  if (input instanceof Observable) {\n    return input;\n  }\n  if (input != null) {\n    if (isInteropObservable(input)) {\n      return fromInteropObservable(input);\n    }\n    if (isArrayLike(input)) {\n      return fromArrayLike(input);\n    }\n    if (isPromise(input)) {\n      return fromPromise(input);\n    }\n    if (isAsyncIterable(input)) {\n      return fromAsyncIterable(input);\n    }\n    if (isIterable(input)) {\n      return fromIterable(input);\n    }\n    if (isReadableStreamLike(input)) {\n      return fromReadableStreamLike(input);\n    }\n  }\n\n  throw createInvalidObservableTypeError(input);\n}\n\n/**\n * Creates an RxJS Observable from an object that implements `Symbol.observable`.\n * @param obj An object that properly implements `Symbol.observable`.\n */\nexport function fromInteropObservable<T>(obj: any) {\n  return new Observable((subscriber: Subscriber<T>) => {\n    const obs = obj[Symbol_observable]();\n    if (isFunction(obs.subscribe)) {\n      return obs.subscribe(subscriber);\n    }\n    // Should be caught by observable subscribe function error handling.\n    throw new TypeError('Provided object does not correctly implement Symbol.observable');\n  });\n}\n\n/**\n * Synchronously emits the values of an array like and completes.\n * This is exported because there are creation functions and operators that need to\n * make direct use of the same logic, and there's no reason to make them run through\n * `from` conditionals because we *know* they're dealing with an array.\n * @param array The array to emit values from\n */\nexport function fromArrayLike<T>(array: ArrayLike<T>) {\n  return new Observable((subscriber: Subscriber<T>) => {\n    // Loop over the array and emit each value. Note two things here:\n    // 1. We're making sure that the subscriber is not closed on each loop.\n    //    This is so we don't continue looping over a very large array after\n    //    something like a `take`, `takeWhile`, or other synchronous unsubscription\n    //    has already unsubscribed.\n    // 2. In this form, reentrant code can alter that array we're looping over.\n    //    This is a known issue, but considered an edge case. The alternative would\n    //    be to copy the array before executing the loop, but this has\n    //    performance implications.\n    for (let i = 0; i < array.length && !subscriber.closed; i++) {\n      subscriber.next(array[i]);\n    }\n    subscriber.complete();\n  });\n}\n\nexport function fromPromise<T>(promise: PromiseLike<T>) {\n  return new Observable((subscriber: Subscriber<T>) => {\n    promise\n      .then(\n        (value) => {\n          if (!subscriber.closed) {\n            subscriber.next(value);\n            subscriber.complete();\n          }\n        },\n        (err: any) => subscriber.error(err)\n      )\n      .then(null, reportUnhandledError);\n  });\n}\n\nexport function fromIterable<T>(iterable: Iterable<T>) {\n  return new Observable((subscriber: Subscriber<T>) => {\n    for (const value of iterable) {\n      subscriber.next(value);\n      if (subscriber.closed) {\n        return;\n      }\n    }\n    subscriber.complete();\n  });\n}\n\nexport function fromAsyncIterable<T>(asyncIterable: AsyncIterable<T>) {\n  return new Observable((subscriber: Subscriber<T>) => {\n    process(asyncIterable, subscriber).catch((err) => subscriber.error(err));\n  });\n}\n\nexport function fromReadableStreamLike<T>(readableStream: ReadableStreamLike<T>) {\n  return fromAsyncIterable(readableStreamLikeToAsyncGenerator(readableStream));\n}\n\nasync function process<T>(asyncIterable: AsyncIterable<T>, subscriber: Subscriber<T>) {\n  for await (const value of asyncIterable) {\n    subscriber.next(value);\n    // A side-effect may have closed our subscriber,\n    // check before the next iteration.\n    if (subscriber.closed) {\n      return;\n    }\n  }\n  subscriber.complete();\n}\n", "import { Subscription } from '../Subscription';\nimport { SchedulerAction, SchedulerLike } from '../types';\n\nexport function executeSchedule(\n  parentSubscription: Subscription,\n  scheduler: SchedulerLike,\n  work: () => void,\n  delay: number,\n  repeat: true\n): void;\nexport function executeSchedule(\n  parentSubscription: Subscription,\n  scheduler: SchedulerLike,\n  work: () => void,\n  delay?: number,\n  repeat?: false\n): Subscription;\n\nexport function executeSchedule(\n  parentSubscription: Subscription,\n  scheduler: SchedulerLike,\n  work: () => void,\n  delay = 0,\n  repeat = false\n): Subscription | void {\n  const scheduleSubscription = scheduler.schedule(function (this: SchedulerAction<any>) {\n    work();\n    if (repeat) {\n      parentSubscription.add(this.schedule(null, delay));\n    } else {\n      this.unsubscribe();\n    }\n  }, delay);\n\n  parentSubscription.add(scheduleSubscription);\n\n  if (!repeat) {\n    // Because user-land scheduler implementations are unlikely to properly reuse\n    // Actions for repeat scheduling, we can't trust that the returned subscription\n    // will control repeat subscription scenarios. So we're trying to avoid using them\n    // incorrectly within this library.\n    return scheduleSubscription;\n  }\n}\n", "/** @prettier */\nimport { MonoTypeOperatorFunction, SchedulerLike } from '../types';\nimport { executeSchedule } from '../util/executeSchedule';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * Re-emits all notifications from source Observable with specified scheduler.\n *\n * <span class=\"informal\">Ensure a specific scheduler is used, from outside of an Observable.</span>\n *\n * `observeOn` is an operator that accepts a scheduler as a first parameter, which will be used to reschedule\n * notifications emitted by the source Observable. It might be useful, if you do not have control over\n * internal scheduler of a given Observable, but want to control when its values are emitted nevertheless.\n *\n * Returned Observable emits the same notifications (nexted values, complete and error events) as the source Observable,\n * but rescheduled with provided scheduler. Note that this doesn't mean that source Observables internal\n * scheduler will be replaced in any way. Original scheduler still will be used, but when the source Observable emits\n * notification, it will be immediately scheduled again - this time with scheduler passed to `observeOn`.\n * An anti-pattern would be calling `observeOn` on Observable that emits lots of values synchronously, to split\n * that emissions into asynchronous chunks. For this to happen, scheduler would have to be passed into the source\n * Observable directly (usually into the operator that creates it). `observeOn` simply delays notifications a\n * little bit more, to ensure that they are emitted at expected moments.\n *\n * As a matter of fact, `observeOn` accepts second parameter, which specifies in milliseconds with what delay notifications\n * will be emitted. The main difference between {@link delay} operator and `observeOn` is that `observeOn`\n * will delay all notifications - including error notifications - while `delay` will pass through error\n * from source Observable immediately when it is emitted. In general it is highly recommended to use `delay` operator\n * for any kind of delaying of values in the stream, while using `observeOn` to specify which scheduler should be used\n * for notification emissions in general.\n *\n * ## Example\n *\n * Ensure values in subscribe are called just before browser repaint\n *\n * ```ts\n * import { interval, observeOn, animationFrameScheduler } from 'rxjs';\n *\n * const someDiv = document.createElement('div');\n * someDiv.style.cssText = 'width: 200px;background: #09c';\n * document.body.appendChild(someDiv);\n * const intervals = interval(10);      // Intervals are scheduled\n *                                      // with async scheduler by default...\n * intervals.pipe(\n *   observeOn(animationFrameScheduler) // ...but we will observe on animationFrame\n * )                                    // scheduler to ensure smooth animation.\n * .subscribe(val => {\n *   someDiv.style.height = val + 'px';\n * });\n * ```\n *\n * @see {@link delay}\n *\n * @param scheduler Scheduler that will be used to reschedule notifications from source Observable.\n * @param delay Number of milliseconds that states with what delay every notification should be rescheduled.\n * @return A function that returns an Observable that emits the same\n * notifications as the source Observable, but with provided scheduler.\n */\nexport function observeOn<T>(scheduler: SchedulerLike, delay = 0): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value) => executeSchedule(subscriber, scheduler, () => subscriber.next(value), delay),\n        () => executeSchedule(subscriber, scheduler, () => subscriber.complete(), delay),\n        (err) => executeSchedule(subscriber, scheduler, () => subscriber.error(err), delay)\n      )\n    );\n  });\n}\n", "import { MonoTypeOperatorFunction, SchedulerLike } from '../types';\nimport { operate } from '../util/lift';\n\n/**\n * Asynchronously subscribes Observers to this Observable on the specified {@link SchedulerLike}.\n *\n * With `subscribeOn` you can decide what type of scheduler a specific Observable will be using when it is subscribed to.\n *\n * Schedulers control the speed and order of emissions to observers from an Observable stream.\n *\n * ![](subscribeOn.png)\n *\n * ## Example\n *\n * Given the following code:\n *\n * ```ts\n * import { of, merge } from 'rxjs';\n *\n * const a = of(1, 2, 3);\n * const b = of(4, 5, 6);\n *\n * merge(a, b).subscribe(console.log);\n *\n * // Outputs\n * // 1\n * // 2\n * // 3\n * // 4\n * // 5\n * // 6\n * ```\n *\n * Both Observable `a` and `b` will emit their values directly and synchronously once they are subscribed to.\n *\n * If we instead use the `subscribeOn` operator declaring that we want to use the {@link asyncScheduler} for values emitted by Observable `a`:\n *\n * ```ts\n * import { of, subscribeOn, asyncScheduler, merge } from 'rxjs';\n *\n * const a = of(1, 2, 3).pipe(subscribeOn(asyncScheduler));\n * const b = of(4, 5, 6);\n *\n * merge(a, b).subscribe(console.log);\n *\n * // Outputs\n * // 4\n * // 5\n * // 6\n * // 1\n * // 2\n * // 3\n * ```\n *\n * The reason for this is that Observable `b` emits its values directly and synchronously like before\n * but the emissions from `a` are scheduled on the event loop because we are now using the {@link asyncScheduler} for that specific Observable.\n *\n * @param scheduler The {@link SchedulerLike} to perform subscription actions on.\n * @param delay A delay to pass to the scheduler to delay subscriptions\n * @return A function that returns an Observable modified so that its\n * subscriptions happen on the specified {@link SchedulerLike}.\n */\nexport function subscribeOn<T>(scheduler: SchedulerLike, delay: number = 0): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    subscriber.add(scheduler.schedule(() => source.subscribe(subscriber), delay));\n  });\n}\n", "import { innerFrom } from '../observable/innerFrom';\nimport { observeOn } from '../operators/observeOn';\nimport { subscribeOn } from '../operators/subscribeOn';\nimport { InteropObservable, SchedulerLike } from '../types';\n\nexport function scheduleObservable<T>(input: InteropObservable<T>, scheduler: SchedulerLike) {\n  return innerFrom(input).pipe(subscribeOn(scheduler), observeOn(scheduler));\n}\n", "import { innerFrom } from '../observable/innerFrom';\nimport { observeOn } from '../operators/observeOn';\nimport { subscribeOn } from '../operators/subscribeOn';\nimport { SchedulerLike } from '../types';\n\nexport function schedulePromise<T>(input: PromiseLike<T>, scheduler: SchedulerLike) {\n  return innerFrom(input).pipe(subscribeOn(scheduler), observeOn(scheduler));\n}\n", "import { Observable } from '../Observable';\nimport { SchedulerLike } from '../types';\n\nexport function scheduleArray<T>(input: ArrayLike<T>, scheduler: SchedulerLike) {\n  return new Observable<T>((subscriber) => {\n    // The current array index.\n    let i = 0;\n    // Start iterating over the array like on a schedule.\n    return scheduler.schedule(function () {\n      if (i === input.length) {\n        // If we have hit the end of the array like in the\n        // previous job, we can complete.\n        subscriber.complete();\n      } else {\n        // Otherwise let's next the value at the current index,\n        // then increment our index.\n        subscriber.next(input[i++]);\n        // If the last emission didn't cause us to close the subscriber\n        // (via take or some side effect), reschedule the job and we'll\n        // make another pass.\n        if (!subscriber.closed) {\n          this.schedule();\n        }\n      }\n    });\n  });\n}\n", "import { Observable } from '../Observable';\nimport { SchedulerLike } from '../types';\nimport { iterator as Symbol_iterator } from '../symbol/iterator';\nimport { isFunction } from '../util/isFunction';\nimport { executeSchedule } from '../util/executeSchedule';\n\n/**\n * Used in {@link scheduled} to create an observable from an Iterable.\n * @param input The iterable to create an observable from\n * @param scheduler The scheduler to use\n */\nexport function scheduleIterable<T>(input: Iterable<T>, scheduler: SchedulerLike) {\n  return new Observable<T>((subscriber) => {\n    let iterator: Iterator<T, T>;\n\n    // Schedule the initial creation of the iterator from\n    // the iterable. This is so the code in the iterable is\n    // not called until the scheduled job fires.\n    executeSchedule(subscriber, scheduler, () => {\n      // Create the iterator.\n      iterator = (input as any)[Symbol_iterator]();\n\n      executeSchedule(\n        subscriber,\n        scheduler,\n        () => {\n          let value: T;\n          let done: boolean | undefined;\n          try {\n            // Pull the value out of the iterator\n            ({ value, done } = iterator.next());\n          } catch (err) {\n            // We got an error while pulling from the iterator\n            subscriber.error(err);\n            return;\n          }\n\n          if (done) {\n            // If it is \"done\" we just complete. This mimics the\n            // behavior of JavaScript's `for..of` consumption of\n            // iterables, which will not emit the value from an iterator\n            // result of `{ done: true: value: 'here' }`.\n            subscriber.complete();\n          } else {\n            // The iterable is not done, emit the value.\n            subscriber.next(value);\n          }\n        },\n        0,\n        true\n      );\n    });\n\n    // During finalization, if we see this iterator has a `return` method,\n    // then we know it is a Generator, and not just an Iterator. So we call\n    // the `return()` function. This will ensure that any `finally { }` blocks\n    // inside of the generator we can hit will be hit properly.\n    return () => isFunction(iterator?.return) && iterator.return();\n  });\n}\n", "import { SchedulerLike } from '../types';\nimport { Observable } from '../Observable';\nimport { executeSchedule } from '../util/executeSchedule';\n\nexport function scheduleAsyncIterable<T>(input: AsyncIterable<T>, scheduler: SchedulerLike) {\n  if (!input) {\n    throw new Error('Iterable cannot be null');\n  }\n  return new Observable<T>((subscriber) => {\n    executeSchedule(subscriber, scheduler, () => {\n      const iterator = input[Symbol.asyncIterator]();\n      executeSchedule(\n        subscriber,\n        scheduler,\n        () => {\n          iterator.next().then((result) => {\n            if (result.done) {\n              // This will remove the subscriptions from\n              // the parent subscription.\n              subscriber.complete();\n            } else {\n              subscriber.next(result.value);\n            }\n          });\n        },\n        0,\n        true\n      );\n    });\n  });\n}\n", "import { SchedulerLike, ReadableStreamLike } from '../types';\nimport { Observable } from '../Observable';\nimport { scheduleAsyncIterable } from './scheduleAsyncIterable';\nimport { readableStreamLikeToAsyncGenerator } from '../util/isReadableStreamLike';\n\nexport function scheduleReadableStreamLike<T>(input: ReadableStreamLike<T>, scheduler: SchedulerLike): Observable<T> {\n  return scheduleAsyncIterable(readableStreamLikeToAsyncGenerator(input), scheduler);\n}\n", "import { scheduleObservable } from './scheduleObservable';\nimport { schedulePromise } from './schedulePromise';\nimport { scheduleArray } from './scheduleArray';\nimport { scheduleIterable } from './scheduleIterable';\nimport { scheduleAsyncIterable } from './scheduleAsyncIterable';\nimport { isInteropObservable } from '../util/isInteropObservable';\nimport { isPromise } from '../util/isPromise';\nimport { isArrayLike } from '../util/isArrayLike';\nimport { isIterable } from '../util/isIterable';\nimport { ObservableInput, SchedulerLike } from '../types';\nimport { Observable } from '../Observable';\nimport { isAsyncIterable } from '../util/isAsyncIterable';\nimport { createInvalidObservableTypeError } from '../util/throwUnobservableError';\nimport { isReadableStreamLike } from '../util/isReadableStreamLike';\nimport { scheduleReadableStreamLike } from './scheduleReadableStreamLike';\n\n/**\n * Converts from a common {@link ObservableInput} type to an observable where subscription and emissions\n * are scheduled on the provided scheduler.\n *\n * @see {@link from}\n * @see {@link of}\n *\n * @param input The observable, array, promise, iterable, etc you would like to schedule\n * @param scheduler The scheduler to use to schedule the subscription and emissions from\n * the returned observable.\n */\nexport function scheduled<T>(input: ObservableInput<T>, scheduler: SchedulerLike): Observable<T> {\n  if (input != null) {\n    if (isInteropObservable(input)) {\n      return scheduleObservable(input, scheduler);\n    }\n    if (isArrayLike(input)) {\n      return scheduleArray(input, scheduler);\n    }\n    if (isPromise(input)) {\n      return schedulePromise(input, scheduler);\n    }\n    if (isAsyncIterable(input)) {\n      return scheduleAsyncIterable(input, scheduler);\n    }\n    if (isIterable(input)) {\n      return scheduleIterable(input, scheduler);\n    }\n    if (isReadableStreamLike(input)) {\n      return scheduleReadableStreamLike(input, scheduler);\n    }\n  }\n  throw createInvalidObservableTypeError(input);\n}\n", "import { Observable } from '../Observable';\nimport { ObservableInput, SchedulerLike, ObservedValueOf } from '../types';\nimport { scheduled } from '../scheduled/scheduled';\nimport { innerFrom } from './innerFrom';\n\nexport function from<O extends ObservableInput<any>>(input: O): Observable<ObservedValueOf<O>>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function from<O extends ObservableInput<any>>(input: O, scheduler: SchedulerLike | undefined): Observable<ObservedValueOf<O>>;\n\n/**\n * Creates an Observable from an Array, an array-like object, a Promise, an iterable object, or an Observable-like object.\n *\n * <span class=\"informal\">Converts almost anything to an Observable.</span>\n *\n * ![](from.png)\n *\n * `from` converts various other objects and data types into Observables. It also converts a Promise, an array-like, or an\n * <a href=\"https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Iteration_protocols#iterable\" target=\"_blank\">iterable</a>\n * object into an Observable that emits the items in that promise, array, or iterable. A String, in this context, is treated\n * as an array of characters. Observable-like objects (contains a function named with the ES2015 Symbol for Observable) can also be\n * converted through this operator.\n *\n * ## Examples\n *\n * Converts an array to an Observable\n *\n * ```ts\n * import { from } from 'rxjs';\n *\n * const array = [10, 20, 30];\n * const result = from(array);\n *\n * result.subscribe(x => console.log(x));\n *\n * // Logs:\n * // 10\n * // 20\n * // 30\n * ```\n *\n * Convert an infinite iterable (from a generator) to an Observable\n *\n * ```ts\n * import { from, take } from 'rxjs';\n *\n * function* generateDoubles(seed) {\n *    let i = seed;\n *    while (true) {\n *      yield i;\n *      i = 2 * i; // double it\n *    }\n * }\n *\n * const iterator = generateDoubles(3);\n * const result = from(iterator).pipe(take(10));\n *\n * result.subscribe(x => console.log(x));\n *\n * // Logs:\n * // 3\n * // 6\n * // 12\n * // 24\n * // 48\n * // 96\n * // 192\n * // 384\n * // 768\n * // 1536\n * ```\n *\n * With `asyncScheduler`\n *\n * ```ts\n * import { from, asyncScheduler } from 'rxjs';\n *\n * console.log('start');\n *\n * const array = [10, 20, 30];\n * const result = from(array, asyncScheduler);\n *\n * result.subscribe(x => console.log(x));\n *\n * console.log('end');\n *\n * // Logs:\n * // 'start'\n * // 'end'\n * // 10\n * // 20\n * // 30\n * ```\n *\n * @see {@link fromEvent}\n * @see {@link fromEventPattern}\n *\n * @param {ObservableInput<T>} A subscription object, a Promise, an Observable-like,\n * an Array, an iterable, or an array-like object to be converted.\n * @param {SchedulerLike} An optional {@link SchedulerLike} on which to schedule the emission of values.\n * @return {Observable<T>}\n */\nexport function from<T>(input: ObservableInput<T>, scheduler?: SchedulerLike): Observable<T> {\n  return scheduler ? scheduled(input, scheduler) : innerFrom(input);\n}\n", "import { SchedulerLike, ValueFromArray } from '../types';\nimport { Observable } from '../Observable';\nimport { popScheduler } from '../util/args';\nimport { from } from './from';\n\n// Devs are more likely to pass null or undefined than they are a scheduler\n// without accompanying values. To make things easier for (naughty) devs who\n// use the `strictNullChecks: false` TypeScript compiler option, these\n// overloads with explicit null and undefined values are included.\n\nexport function of(value: null): Observable<null>;\nexport function of(value: undefined): Observable<undefined>;\n\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function of(scheduler: SchedulerLike): Observable<never>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function of<A extends readonly unknown[]>(...valuesAndScheduler: [...A, SchedulerLike]): Observable<ValueFromArray<A>>;\n\nexport function of(): Observable<never>;\n/** @deprecated Do not specify explicit type parameters. Signatures with type parameters that cannot be inferred will be removed in v8. */\nexport function of<T>(): Observable<T>;\nexport function of<T>(value: T): Observable<T>;\nexport function of<A extends readonly unknown[]>(...values: A): Observable<ValueFromArray<A>>;\n\n/**\n * Converts the arguments to an observable sequence.\n *\n * <span class=\"informal\">Each argument becomes a `next` notification.</span>\n *\n * ![](of.png)\n *\n * Unlike {@link from}, it does not do any flattening and emits each argument in whole\n * as a separate `next` notification.\n *\n * ## Examples\n *\n * Emit the values `10, 20, 30`\n *\n * ```ts\n * import { of } from 'rxjs';\n *\n * of(10, 20, 30)\n *   .subscribe({\n *     next: value => console.log('next:', value),\n *     error: err => console.log('error:', err),\n *     complete: () => console.log('the end'),\n *   });\n *\n * // Outputs\n * // next: 10\n * // next: 20\n * // next: 30\n * // the end\n * ```\n *\n * Emit the array `[1, 2, 3]`\n *\n * ```ts\n * import { of } from 'rxjs';\n *\n * of([1, 2, 3])\n *   .subscribe({\n *     next: value => console.log('next:', value),\n *     error: err => console.log('error:', err),\n *     complete: () => console.log('the end'),\n *   });\n *\n * // Outputs\n * // next: [1, 2, 3]\n * // the end\n * ```\n *\n * @see {@link from}\n * @see {@link range}\n *\n * @param {...T} values A comma separated list of arguments you want to be emitted\n * @return {Observable} An Observable that emits the arguments\n * described above and then completes.\n */\nexport function of<T>(...args: Array<T | SchedulerLike>): Observable<T> {\n  const scheduler = popScheduler(args);\n  return from(args as T[], scheduler);\n}\n", "import { Observable } from '../Observable';\nimport { Subscriber } from '../Subscriber';\nimport { SchedulerLike } from '../types';\nimport { isFunction } from '../util/isFunction';\n\n/**\n * Creates an observable that will create an error instance and push it to the consumer as an error\n * immediately upon subscription.\n *\n * <span class=\"informal\">Just errors and does nothing else</span>\n *\n * ![](throw.png)\n *\n * This creation function is useful for creating an observable that will create an error and error every\n * time it is subscribed to. Generally, inside of most operators when you might want to return an errored\n * observable, this is unnecessary. In most cases, such as in the inner return of {@link concatMap},\n * {@link mergeMap}, {@link defer}, and many others, you can simply throw the error, and RxJS will pick\n * that up and notify the consumer of the error.\n *\n * ## Example\n *\n * Create a simple observable that will create a new error with a timestamp and log it\n * and the message every time you subscribe to it\n *\n * ```ts\n * import { throwError } from 'rxjs';\n *\n * let errorCount = 0;\n *\n * const errorWithTimestamp$ = throwError(() => {\n *   const error: any = new Error(`This is error number ${ ++errorCount }`);\n *   error.timestamp = Date.now();\n *   return error;\n * });\n *\n * errorWithTimestamp$.subscribe({\n *   error: err => console.log(err.timestamp, err.message)\n * });\n *\n * errorWithTimestamp$.subscribe({\n *   error: err => console.log(err.timestamp, err.message)\n * });\n *\n * // Logs the timestamp and a new error message for each subscription\n * ```\n *\n * ### Unnecessary usage\n *\n * Using `throwError` inside of an operator or creation function\n * with a callback, is usually not necessary\n *\n * ```ts\n * import { of, concatMap, timer, throwError } from 'rxjs';\n *\n * const delays$ = of(1000, 2000, Infinity, 3000);\n *\n * delays$.pipe(\n *   concatMap(ms => {\n *     if (ms < 10000) {\n *       return timer(ms);\n *     } else {\n *       // This is probably overkill.\n *       return throwError(() => new Error(`Invalid time ${ ms }`));\n *     }\n *   })\n * )\n * .subscribe({\n *   next: console.log,\n *   error: console.error\n * });\n * ```\n *\n * You can just throw the error instead\n *\n * ```ts\n * import { of, concatMap, timer } from 'rxjs';\n *\n * const delays$ = of(1000, 2000, Infinity, 3000);\n *\n * delays$.pipe(\n *   concatMap(ms => {\n *     if (ms < 10000) {\n *       return timer(ms);\n *     } else {\n *       // Cleaner and easier to read for most folks.\n *       throw new Error(`Invalid time ${ ms }`);\n *     }\n *   })\n * )\n * .subscribe({\n *   next: console.log,\n *   error: console.error\n * });\n * ```\n *\n * @param errorFactory A factory function that will create the error instance that is pushed.\n */\nexport function throwError(errorFactory: () => any): Observable<never>;\n\n/**\n * Returns an observable that will error with the specified error immediately upon subscription.\n *\n * @param error The error instance to emit\n * @deprecated Support for passing an error value will be removed in v8. Instead, pass a factory function to `throwError(() => new Error('test'))`. This is\n * because it will create the error at the moment it should be created and capture a more appropriate stack trace. If\n * for some reason you need to create the error ahead of time, you can still do that: `const err = new Error('test'); throwError(() => err);`.\n */\nexport function throwError(error: any): Observable<never>;\n\n/**\n * Notifies the consumer of an error using a given scheduler by scheduling it at delay `0` upon subscription.\n *\n * @param errorOrErrorFactory An error instance or error factory\n * @param scheduler A scheduler to use to schedule the error notification\n * @deprecated The `scheduler` parameter will be removed in v8.\n * Use `throwError` in combination with {@link observeOn}: `throwError(() => new Error('test')).pipe(observeOn(scheduler));`.\n * Details: https://rxjs.dev/deprecations/scheduler-argument\n */\nexport function throwError(errorOrErrorFactory: any, scheduler: SchedulerLike): Observable<never>;\n\nexport function throwError(errorOrErrorFactory: any, scheduler?: SchedulerLike): Observable<never> {\n  const errorFactory = isFunction(errorOrErrorFactory) ? errorOrErrorFactory : () => errorOrErrorFactory;\n  const init = (subscriber: Subscriber<never>) => subscriber.error(errorFactory());\n  return new Observable(scheduler ? (subscriber) => scheduler.schedule(init as any, 0, subscriber) : init);\n}\n", "import { createErrorClass } from './createErrorClass';\n\nexport interface EmptyError extends Error {}\n\nexport interface EmptyErrorCtor {\n  /**\n   * @deprecated Internal implementation detail. Do not construct error instances.\n   * Cannot be tagged as internal: https://github.com/ReactiveX/rxjs/issues/6269\n   */\n  new (): EmptyError;\n}\n\n/**\n * An error thrown when an Observable or a sequence was queried but has no\n * elements.\n *\n * @see {@link first}\n * @see {@link last}\n * @see {@link single}\n * @see {@link firstValueFrom}\n * @see {@link lastValueFrom}\n *\n * @class EmptyError\n */\nexport const EmptyError: EmptyErrorCtor = createErrorClass((_super) => function EmptyErrorImpl(this: any) {\n  _super(this);\n  this.name = 'EmptyError';\n  this.message = 'no elements in sequence';\n});\n", "/**\n * Checks to see if a value is not only a `Date` object,\n * but a *valid* `Date` object that can be converted to a\n * number. For example, `new Date('blah')` is indeed an\n * `instanceof Date`, however it cannot be converted to a\n * number.\n */\nexport function isValidDate(value: any): value is Date {\n  return value instanceof Date && !isNaN(value as any);\n}\n", "import { OperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\nexport function map<T, R>(project: (value: T, index: number) => R): OperatorFunction<T, R>;\n/** @deprecated Use a closure instead of a `thisArg`. Signatures accepting a `thisArg` will be removed in v8. */\nexport function map<T, R, A>(project: (this: A, value: T, index: number) => R, thisArg: A): OperatorFunction<T, R>;\n\n/**\n * Applies a given `project` function to each value emitted by the source\n * Observable, and emits the resulting values as an Observable.\n *\n * <span class=\"informal\">Like [Array.prototype.map()](https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array/map),\n * it passes each source value through a transformation function to get\n * corresponding output values.</span>\n *\n * ![](map.png)\n *\n * Similar to the well known `Array.prototype.map` function, this operator\n * applies a projection to each value and emits that projection in the output\n * Observable.\n *\n * ## Example\n *\n * Map every click to the `clientX` position of that click\n *\n * ```ts\n * import { fromEvent, map } from 'rxjs';\n *\n * const clicks = fromEvent<PointerEvent>(document, 'click');\n * const positions = clicks.pipe(map(ev => ev.clientX));\n *\n * positions.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link mapTo}\n * @see {@link pluck}\n *\n * @param {function(value: T, index: number): R} project The function to apply\n * to each `value` emitted by the source Observable. The `index` parameter is\n * the number `i` for the i-th emission that has happened since the\n * subscription, starting from the number `0`.\n * @param {any} [thisArg] An optional argument to define what `this` is in the\n * `project` function.\n * @return A function that returns an Observable that emits the values from the\n * source Observable transformed by the given `project` function.\n */\nexport function map<T, R>(project: (value: T, index: number) => R, thisArg?: any): OperatorFunction<T, R> {\n  return operate((source, subscriber) => {\n    // The index of the value from the source. Used with projection.\n    let index = 0;\n    // Subscribe to the source, all errors and completions are sent along\n    // to the consumer.\n    source.subscribe(\n      createOperatorSubscriber(subscriber, (value: T) => {\n        // Call the projection function with the appropriate this context,\n        // and send the resulting value to the consumer.\n        subscriber.next(project.call(thisArg, value, index++));\n      })\n    );\n  });\n}\n", "import { OperatorFunction } from \"../types\";\nimport { map } from \"../operators/map\";\n\nconst { isArray } = Array;\n\nfunction callOrApply<T, R>(fn: ((...values: T[]) => R), args: T|T[]): R {\n    return isArray(args) ? fn(...args) : fn(args);\n}\n\n/**\n * Used in several -- mostly deprecated -- situations where we need to \n * apply a list of arguments or a single argument to a result selector.\n */\nexport function mapOneOrManyArgs<T, R>(fn: ((...values: T[]) => R)): OperatorFunction<T|T[], R> {\n    return map(args => callOrApply(fn, args))\n}", "const { isArray } = Array;\nconst { getPrototypeOf, prototype: objectProto, keys: getKeys } = Object;\n\n/**\n * Used in functions where either a list of arguments, a single array of arguments, or a\n * dictionary of arguments can be returned. Returns an object with an `args` property with\n * the arguments in an array, if it is a dictionary, it will also return the `keys` in another\n * property.\n */\nexport function argsArgArrayOrObject<T, O extends Record<string, T>>(args: T[] | [O] | [T[]]): { args: T[]; keys: string[] | null } {\n  if (args.length === 1) {\n    const first = args[0];\n    if (isArray(first)) {\n      return { args: first, keys: null };\n    }\n    if (isPOJO(first)) {\n      const keys = getKeys(first);\n      return {\n        args: keys.map((key) => first[key]),\n        keys,\n      };\n    }\n  }\n\n  return { args: args as T[], keys: null };\n}\n\nfunction isPOJO(obj: any): obj is object {\n  return obj && typeof obj === 'object' && getPrototypeOf(obj) === objectProto;\n}\n", "export function createObject(keys: string[], values: any[]) {\n  return keys.reduce((result, key, i) => ((result[key] = values[i]), result), {} as any);\n}\n", "import { Observable } from '../Observable';\nimport { ObservableInput, SchedulerLike, ObservedValueOf, ObservableInputTuple } from '../types';\nimport { argsArgArrayOrObject } from '../util/argsArgArrayOrObject';\nimport { Subscriber } from '../Subscriber';\nimport { from } from './from';\nimport { identity } from '../util/identity';\nimport { Subscription } from '../Subscription';\nimport { mapOneOrManyArgs } from '../util/mapOneOrManyArgs';\nimport { popResultSelector, popScheduler } from '../util/args';\nimport { createObject } from '../util/createObject';\nimport { createOperatorSubscriber } from '../operators/OperatorSubscriber';\nimport { AnyCatcher } from '../AnyCatcher';\nimport { executeSchedule } from '../util/executeSchedule';\n\n// combineLatest(any)\n// We put this first because we need to catch cases where the user has supplied\n// _exactly `any`_ as the argument. Since `any` literally matches _anything_,\n// we don't want it to randomly hit one of the other type signatures below,\n// as we have no idea at build-time what type we should be returning when given an any.\n\n/**\n * You have passed `any` here, we can't figure out if it is\n * an array or an object, so you're getting `unknown`. Use better types.\n * @param arg Something typed as `any`\n */\nexport function combineLatest<T extends AnyCatcher>(arg: T): Observable<unknown>;\n\n// combineLatest([a, b, c])\nexport function combineLatest(sources: []): Observable<never>;\nexport function combineLatest<A extends readonly unknown[]>(sources: readonly [...ObservableInputTuple<A>]): Observable<A>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `combineLatestAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function combineLatest<A extends readonly unknown[], R>(\n  sources: readonly [...ObservableInputTuple<A>],\n  resultSelector: (...values: A) => R,\n  scheduler: SchedulerLike\n): Observable<R>;\nexport function combineLatest<A extends readonly unknown[], R>(\n  sources: readonly [...ObservableInputTuple<A>],\n  resultSelector: (...values: A) => R\n): Observable<R>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `combineLatestAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function combineLatest<A extends readonly unknown[]>(\n  sources: readonly [...ObservableInputTuple<A>],\n  scheduler: SchedulerLike\n): Observable<A>;\n\n// combineLatest(a, b, c)\n/** @deprecated Pass an array of sources instead. The rest-parameters signature will be removed in v8. Details: https://rxjs.dev/deprecations/array-argument */\nexport function combineLatest<A extends readonly unknown[]>(...sources: [...ObservableInputTuple<A>]): Observable<A>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `combineLatestAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function combineLatest<A extends readonly unknown[], R>(\n  ...sourcesAndResultSelectorAndScheduler: [...ObservableInputTuple<A>, (...values: A) => R, SchedulerLike]\n): Observable<R>;\n/** @deprecated Pass an array of sources instead. The rest-parameters signature will be removed in v8. Details: https://rxjs.dev/deprecations/array-argument */\nexport function combineLatest<A extends readonly unknown[], R>(\n  ...sourcesAndResultSelector: [...ObservableInputTuple<A>, (...values: A) => R]\n): Observable<R>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `combineLatestAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function combineLatest<A extends readonly unknown[]>(\n  ...sourcesAndScheduler: [...ObservableInputTuple<A>, SchedulerLike]\n): Observable<A>;\n\n// combineLatest({a, b, c})\nexport function combineLatest(sourcesObject: { [K in any]: never }): Observable<never>;\nexport function combineLatest<T extends Record<string, ObservableInput<any>>>(\n  sourcesObject: T\n): Observable<{ [K in keyof T]: ObservedValueOf<T[K]> }>;\n\n/**\n * Combines multiple Observables to create an Observable whose values are\n * calculated from the latest values of each of its input Observables.\n *\n * <span class=\"informal\">Whenever any input Observable emits a value, it\n * computes a formula using the latest values from all the inputs, then emits\n * the output of that formula.</span>\n *\n * ![](combineLatest.png)\n *\n * `combineLatest` combines the values from all the Observables passed in the\n * observables array. This is done by subscribing to each Observable in order and,\n * whenever any Observable emits, collecting an array of the most recent\n * values from each Observable. So if you pass `n` Observables to this operator,\n * the returned Observable will always emit an array of `n` values, in an order\n * corresponding to the order of the passed Observables (the value from the first Observable\n * will be at index 0 of the array and so on).\n *\n * Static version of `combineLatest` accepts an array of Observables. Note that an array of\n * Observables is a good choice, if you don't know beforehand how many Observables\n * you will combine. Passing an empty array will result in an Observable that\n * completes immediately.\n *\n * To ensure the output array always has the same length, `combineLatest` will\n * actually wait for all input Observables to emit at least once,\n * before it starts emitting results. This means if some Observable emits\n * values before other Observables started emitting, all these values but the last\n * will be lost. On the other hand, if some Observable does not emit a value but\n * completes, resulting Observable will complete at the same moment without\n * emitting anything, since it will now be impossible to include a value from the\n * completed Observable in the resulting array. Also, if some input Observable does\n * not emit any value and never completes, `combineLatest` will also never emit\n * and never complete, since, again, it will wait for all streams to emit some\n * value.\n *\n * If at least one Observable was passed to `combineLatest` and all passed Observables\n * emitted something, the resulting Observable will complete when all combined\n * streams complete. So even if some Observable completes, the result of\n * `combineLatest` will still emit values when other Observables do. In case\n * of a completed Observable, its value from now on will always be the last\n * emitted value. On the other hand, if any Observable errors, `combineLatest`\n * will error immediately as well, and all other Observables will be unsubscribed.\n *\n * ## Examples\n *\n * Combine two timer Observables\n *\n * ```ts\n * import { timer, combineLatest } from 'rxjs';\n *\n * const firstTimer = timer(0, 1000); // emit 0, 1, 2... after every second, starting from now\n * const secondTimer = timer(500, 1000); // emit 0, 1, 2... after every second, starting 0,5s from now\n * const combinedTimers = combineLatest([firstTimer, secondTimer]);\n * combinedTimers.subscribe(value => console.log(value));\n * // Logs\n * // [0, 0] after 0.5s\n * // [1, 0] after 1s\n * // [1, 1] after 1.5s\n * // [2, 1] after 2s\n * ```\n *\n * Combine a dictionary of Observables\n *\n * ```ts\n * import { of, delay, startWith, combineLatest } from 'rxjs';\n *\n * const observables = {\n *   a: of(1).pipe(delay(1000), startWith(0)),\n *   b: of(5).pipe(delay(5000), startWith(0)),\n *   c: of(10).pipe(delay(10000), startWith(0))\n * };\n * const combined = combineLatest(observables);\n * combined.subscribe(value => console.log(value));\n * // Logs\n * // { a: 0, b: 0, c: 0 } immediately\n * // { a: 1, b: 0, c: 0 } after 1s\n * // { a: 1, b: 5, c: 0 } after 5s\n * // { a: 1, b: 5, c: 10 } after 10s\n * ```\n *\n * Combine an array of Observables\n *\n * ```ts\n * import { of, delay, startWith, combineLatest } from 'rxjs';\n *\n * const observables = [1, 5, 10].map(\n *   n => of(n).pipe(\n *     delay(n * 1000), // emit 0 and then emit n after n seconds\n *     startWith(0)\n *   )\n * );\n * const combined = combineLatest(observables);\n * combined.subscribe(value => console.log(value));\n * // Logs\n * // [0, 0, 0] immediately\n * // [1, 0, 0] after 1s\n * // [1, 5, 0] after 5s\n * // [1, 5, 10] after 10s\n * ```\n *\n * Use map operator to dynamically calculate the Body-Mass Index\n *\n * ```ts\n * import { of, combineLatest, map } from 'rxjs';\n *\n * const weight = of(70, 72, 76, 79, 75);\n * const height = of(1.76, 1.77, 1.78);\n * const bmi = combineLatest([weight, height]).pipe(\n *   map(([w, h]) => w / (h * h)),\n * );\n * bmi.subscribe(x => console.log('BMI is ' + x));\n *\n * // With output to console:\n * // BMI is 24.212293388429753\n * // BMI is 23.93948099205209\n * // BMI is 23.671253629592222\n * ```\n *\n * @see {@link combineLatestAll}\n * @see {@link merge}\n * @see {@link withLatestFrom}\n *\n * @param {ObservableInput} [observables] An array of input Observables to combine with each other.\n * An array of Observables must be given as the first argument.\n * @param {function} [project] An optional function to project the values from\n * the combined latest values into a new value on the output Observable.\n * @param {SchedulerLike} [scheduler=null] The {@link SchedulerLike} to use for subscribing to\n * each input Observable.\n * @return {Observable} An Observable of projected values from the most recent\n * values from each input Observable, or an array of the most recent values from\n * each input Observable.\n */\nexport function combineLatest<O extends ObservableInput<any>, R>(...args: any[]): Observable<R> | Observable<ObservedValueOf<O>[]> {\n  const scheduler = popScheduler(args);\n  const resultSelector = popResultSelector(args);\n\n  const { args: observables, keys } = argsArgArrayOrObject(args);\n\n  if (observables.length === 0) {\n    // If no observables are passed, or someone has passed an empty array\n    // of observables, or even an empty object POJO, we need to just\n    // complete (EMPTY), but we have to honor the scheduler provided if any.\n    return from([], scheduler as any);\n  }\n\n  const result = new Observable<ObservedValueOf<O>[]>(\n    combineLatestInit(\n      observables as ObservableInput<ObservedValueOf<O>>[],\n      scheduler,\n      keys\n        ? // A handler for scrubbing the array of args into a dictionary.\n          (values) => createObject(keys, values)\n        : // A passthrough to just return the array\n          identity\n    )\n  );\n\n  return resultSelector ? (result.pipe(mapOneOrManyArgs(resultSelector)) as Observable<R>) : result;\n}\n\nexport function combineLatestInit(\n  observables: ObservableInput<any>[],\n  scheduler?: SchedulerLike,\n  valueTransform: (values: any[]) => any = identity\n) {\n  return (subscriber: Subscriber<any>) => {\n    // The outer subscription. We're capturing this in a function\n    // because we may have to schedule it.\n    maybeSchedule(\n      scheduler,\n      () => {\n        const { length } = observables;\n        // A store for the values each observable has emitted so far. We match observable to value on index.\n        const values = new Array(length);\n        // The number of currently active subscriptions, as they complete, we decrement this number to see if\n        // we are all done combining values, so we can complete the result.\n        let active = length;\n        // The number of inner sources that still haven't emitted the first value\n        // We need to track this because all sources need to emit one value in order\n        // to start emitting values.\n        let remainingFirstValues = length;\n        // The loop to kick off subscription. We're keying everything on index `i` to relate the observables passed\n        // in to the slot in the output array or the key in the array of keys in the output dictionary.\n        for (let i = 0; i < length; i++) {\n          maybeSchedule(\n            scheduler,\n            () => {\n              const source = from(observables[i], scheduler as any);\n              let hasFirstValue = false;\n              source.subscribe(\n                createOperatorSubscriber(\n                  subscriber,\n                  (value) => {\n                    // When we get a value, record it in our set of values.\n                    values[i] = value;\n                    if (!hasFirstValue) {\n                      // If this is our first value, record that.\n                      hasFirstValue = true;\n                      remainingFirstValues--;\n                    }\n                    if (!remainingFirstValues) {\n                      // We're not waiting for any more\n                      // first values, so we can emit!\n                      subscriber.next(valueTransform(values.slice()));\n                    }\n                  },\n                  () => {\n                    if (!--active) {\n                      // We only complete the result if we have no more active\n                      // inner observables.\n                      subscriber.complete();\n                    }\n                  }\n                )\n              );\n            },\n            subscriber\n          );\n        }\n      },\n      subscriber\n    );\n  };\n}\n\n/**\n * A small utility to handle the couple of locations where we want to schedule if a scheduler was provided,\n * but we don't if there was no scheduler.\n */\nfunction maybeSchedule(scheduler: SchedulerLike | undefined, execute: () => void, subscription: Subscription) {\n  if (scheduler) {\n    executeSchedule(subscription, scheduler, execute);\n  } else {\n    execute();\n  }\n}\n", "import { Observable } from '../Observable';\nimport { innerFrom } from '../observable/innerFrom';\nimport { Subscriber } from '../Subscriber';\nimport { ObservableInput, SchedulerLike } from '../types';\nimport { executeSchedule } from '../util/executeSchedule';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * A process embodying the general \"merge\" strategy. This is used in\n * `mergeMap` and `mergeScan` because the logic is otherwise nearly identical.\n * @param source The original source observable\n * @param subscriber The consumer subscriber\n * @param project The projection function to get our inner sources\n * @param concurrent The number of concurrent inner subscriptions\n * @param onBeforeNext Additional logic to apply before nexting to our consumer\n * @param expand If `true` this will perform an \"expand\" strategy, which differs only\n * in that it recurses, and the inner subscription must be schedule-able.\n * @param innerSubScheduler A scheduler to use to schedule inner subscriptions,\n * this is to support the expand strategy, mostly, and should be deprecated\n */\nexport function mergeInternals<T, R>(\n  source: Observable<T>,\n  subscriber: Subscriber<R>,\n  project: (value: T, index: number) => ObservableInput<R>,\n  concurrent: number,\n  onBeforeNext?: (innerValue: R) => void,\n  expand?: boolean,\n  innerSubScheduler?: SchedulerLike,\n  additionalFinalizer?: () => void\n) {\n  // Buffered values, in the event of going over our concurrency limit\n  const buffer: T[] = [];\n  // The number of active inner subscriptions.\n  let active = 0;\n  // An index to pass to our accumulator function\n  let index = 0;\n  // Whether or not the outer source has completed.\n  let isComplete = false;\n\n  /**\n   * Checks to see if we can complete our result or not.\n   */\n  const checkComplete = () => {\n    // If the outer has completed, and nothing is left in the buffer,\n    // and we don't have any active inner subscriptions, then we can\n    // Emit the state and complete.\n    if (isComplete && !buffer.length && !active) {\n      subscriber.complete();\n    }\n  };\n\n  // If we're under our concurrency limit, just start the inner subscription, otherwise buffer and wait.\n  const outerNext = (value: T) => (active < concurrent ? doInnerSub(value) : buffer.push(value));\n\n  const doInnerSub = (value: T) => {\n    // If we're expanding, we need to emit the outer values and the inner values\n    // as the inners will \"become outers\" in a way as they are recursively fed\n    // back to the projection mechanism.\n    expand && subscriber.next(value as any);\n\n    // Increment the number of active subscriptions so we can track it\n    // against our concurrency limit later.\n    active++;\n\n    // A flag used to show that the inner observable completed.\n    // This is checked during finalization to see if we should\n    // move to the next item in the buffer, if there is on.\n    let innerComplete = false;\n\n    // Start our inner subscription.\n    innerFrom(project(value, index++)).subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (innerValue) => {\n          // `mergeScan` has additional handling here. For example\n          // taking the inner value and updating state.\n          onBeforeNext?.(innerValue);\n\n          if (expand) {\n            // If we're expanding, then just recurse back to our outer\n            // handler. It will emit the value first thing.\n            outerNext(innerValue as any);\n          } else {\n            // Otherwise, emit the inner value.\n            subscriber.next(innerValue);\n          }\n        },\n        () => {\n          // Flag that we have completed, so we know to check the buffer\n          // during finalization.\n          innerComplete = true;\n        },\n        // Errors are passed to the destination.\n        undefined,\n        () => {\n          // During finalization, if the inner completed (it wasn't errored or\n          // cancelled), then we want to try the next item in the buffer if\n          // there is one.\n          if (innerComplete) {\n            // We have to wrap this in a try/catch because it happens during\n            // finalization, possibly asynchronously, and we want to pass\n            // any errors that happen (like in a projection function) to\n            // the outer Subscriber.\n            try {\n              // INNER SOURCE COMPLETE\n              // Decrement the active count to ensure that the next time\n              // we try to call `doInnerSub`, the number is accurate.\n              active--;\n              // If we have more values in the buffer, try to process those\n              // Note that this call will increment `active` ahead of the\n              // next conditional, if there were any more inner subscriptions\n              // to start.\n              while (buffer.length && active < concurrent) {\n                const bufferedValue = buffer.shift()!;\n                // Particularly for `expand`, we need to check to see if a scheduler was provided\n                // for when we want to start our inner subscription. Otherwise, we just start\n                // are next inner subscription.\n                if (innerSubScheduler) {\n                  executeSchedule(subscriber, innerSubScheduler, () => doInnerSub(bufferedValue));\n                } else {\n                  doInnerSub(bufferedValue);\n                }\n              }\n              // Check to see if we can complete, and complete if so.\n              checkComplete();\n            } catch (err) {\n              subscriber.error(err);\n            }\n          }\n        }\n      )\n    );\n  };\n\n  // Subscribe to our source observable.\n  source.subscribe(\n    createOperatorSubscriber(subscriber, outerNext, () => {\n      // Outer completed, make a note of it, and check to see if we can complete everything.\n      isComplete = true;\n      checkComplete();\n    })\n  );\n\n  // Additional finalization (for when the destination is torn down).\n  // Other finalization is added implicitly via subscription above.\n  return () => {\n    additionalFinalizer?.();\n  };\n}\n", "import { ObservableInput, OperatorFunction, ObservedValueOf } from '../types';\nimport { map } from './map';\nimport { innerFrom } from '../observable/innerFrom';\nimport { operate } from '../util/lift';\nimport { mergeInternals } from './mergeInternals';\nimport { isFunction } from '../util/isFunction';\n\n/* tslint:disable:max-line-length */\nexport function mergeMap<T, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O,\n  concurrent?: number\n): OperatorFunction<T, ObservedValueOf<O>>;\n/** @deprecated The `resultSelector` parameter will be removed in v8. Use an inner `map` instead. Details: https://rxjs.dev/deprecations/resultSelector */\nexport function mergeMap<T, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O,\n  resultSelector: undefined,\n  concurrent?: number\n): OperatorFunction<T, ObservedValueOf<O>>;\n/** @deprecated The `resultSelector` parameter will be removed in v8. Use an inner `map` instead. Details: https://rxjs.dev/deprecations/resultSelector */\nexport function mergeMap<T, R, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O,\n  resultSelector: (outerValue: T, innerValue: ObservedValueOf<O>, outerIndex: number, innerIndex: number) => R,\n  concurrent?: number\n): OperatorFunction<T, R>;\n/* tslint:enable:max-line-length */\n\n/**\n * Projects each source value to an Observable which is merged in the output\n * Observable.\n *\n * <span class=\"informal\">Maps each value to an Observable, then flattens all of\n * these inner Observables using {@link mergeAll}.</span>\n *\n * ![](mergeMap.png)\n *\n * Returns an Observable that emits items based on applying a function that you\n * supply to each item emitted by the source Observable, where that function\n * returns an Observable, and then merging those resulting Observables and\n * emitting the results of this merger.\n *\n * ## Example\n *\n * Map and flatten each letter to an Observable ticking every 1 second\n *\n * ```ts\n * import { of, mergeMap, interval, map } from 'rxjs';\n *\n * const letters = of('a', 'b', 'c');\n * const result = letters.pipe(\n *   mergeMap(x => interval(1000).pipe(map(i => x + i)))\n * );\n *\n * result.subscribe(x => console.log(x));\n *\n * // Results in the following:\n * // a0\n * // b0\n * // c0\n * // a1\n * // b1\n * // c1\n * // continues to list a, b, c every second with respective ascending integers\n * ```\n *\n * @see {@link concatMap}\n * @see {@link exhaustMap}\n * @see {@link merge}\n * @see {@link mergeAll}\n * @see {@link mergeMapTo}\n * @see {@link mergeScan}\n * @see {@link switchMap}\n *\n * @param {function(value: T, ?index: number): ObservableInput} project A function\n * that, when applied to an item emitted by the source Observable, returns an\n * Observable.\n * @param {number} [concurrent=Infinity] Maximum number of input\n * Observables being subscribed to concurrently.\n * @return A function that returns an Observable that emits the result of\n * applying the projection function (and the optional deprecated\n * `resultSelector`) to each item emitted by the source Observable and merging\n * the results of the Observables obtained from this transformation.\n */\nexport function mergeMap<T, R, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O,\n  resultSelector?: ((outerValue: T, innerValue: ObservedValueOf<O>, outerIndex: number, innerIndex: number) => R) | number,\n  concurrent: number = Infinity\n): OperatorFunction<T, ObservedValueOf<O> | R> {\n  if (isFunction(resultSelector)) {\n    // DEPRECATED PATH\n    return mergeMap((a, i) => map((b: any, ii: number) => resultSelector(a, b, i, ii))(innerFrom(project(a, i))), concurrent);\n  } else if (typeof resultSelector === 'number') {\n    concurrent = resultSelector;\n  }\n\n  return operate((source, subscriber) => mergeInternals(source, subscriber, project, concurrent));\n}\n", "import { mergeMap } from './mergeMap';\nimport { identity } from '../util/identity';\nimport { OperatorFunction, ObservableInput, ObservedValueOf } from '../types';\n\n/**\n * Converts a higher-order Observable into a first-order Observable which\n * concurrently delivers all values that are emitted on the inner Observables.\n *\n * <span class=\"informal\">Flattens an Observable-of-Observables.</span>\n *\n * ![](mergeAll.png)\n *\n * `mergeAll` subscribes to an Observable that emits Observables, also known as\n * a higher-order Observable. Each time it observes one of these emitted inner\n * Observables, it subscribes to that and delivers all the values from the\n * inner Observable on the output Observable. The output Observable only\n * completes once all inner Observables have completed. Any error delivered by\n * a inner Observable will be immediately emitted on the output Observable.\n *\n * ## Examples\n *\n * Spawn a new interval Observable for each click event, and blend their outputs as one Observable\n *\n * ```ts\n * import { fromEvent, map, interval, mergeAll } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const higherOrder = clicks.pipe(map(() => interval(1000)));\n * const firstOrder = higherOrder.pipe(mergeAll());\n *\n * firstOrder.subscribe(x => console.log(x));\n * ```\n *\n * Count from 0 to 9 every second for each click, but only allow 2 concurrent timers\n *\n * ```ts\n * import { fromEvent, map, interval, take, mergeAll } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const higherOrder = clicks.pipe(\n *   map(() => interval(1000).pipe(take(10)))\n * );\n * const firstOrder = higherOrder.pipe(mergeAll(2));\n *\n * firstOrder.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link combineLatestAll}\n * @see {@link concatAll}\n * @see {@link exhaustAll}\n * @see {@link merge}\n * @see {@link mergeMap}\n * @see {@link mergeMapTo}\n * @see {@link mergeScan}\n * @see {@link switchAll}\n * @see {@link switchMap}\n * @see {@link zipAll}\n *\n * @param {number} [concurrent=Infinity] Maximum number of inner\n * Observables being subscribed to concurrently.\n * @return A function that returns an Observable that emits values coming from\n * all the inner Observables emitted by the source Observable.\n */\nexport function mergeAll<O extends ObservableInput<any>>(concurrent: number = Infinity): OperatorFunction<O, ObservedValueOf<O>> {\n  return mergeMap(identity, concurrent);\n}\n", "import { mergeAll } from './mergeAll';\nimport { OperatorFunction, ObservableInput, ObservedValueOf } from '../types';\n\n/**\n * Converts a higher-order Observable into a first-order Observable by\n * concatenating the inner Observables in order.\n *\n * <span class=\"informal\">Flattens an Observable-of-Observables by putting one\n * inner Observable after the other.</span>\n *\n * ![](concatAll.svg)\n *\n * Joins every Observable emitted by the source (a higher-order Observable), in\n * a serial fashion. It subscribes to each inner Observable only after the\n * previous inner Observable has completed, and merges all of their values into\n * the returned observable.\n *\n * __Warning:__ If the source Observable emits Observables quickly and\n * endlessly, and the inner Observables it emits generally complete slower than\n * the source emits, you can run into memory issues as the incoming Observables\n * collect in an unbounded buffer.\n *\n * Note: `concatAll` is equivalent to `mergeAll` with concurrency parameter set\n * to `1`.\n *\n * ## Example\n *\n * For each click event, tick every second from 0 to 3, with no concurrency\n *\n * ```ts\n * import { fromEvent, map, interval, take, concatAll } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const higherOrder = clicks.pipe(\n *   map(() => interval(1000).pipe(take(4)))\n * );\n * const firstOrder = higherOrder.pipe(concatAll());\n * firstOrder.subscribe(x => console.log(x));\n *\n * // Results in the following:\n * // (results are not concurrent)\n * // For every click on the \"document\" it will emit values 0 to 3 spaced\n * // on a 1000ms interval\n * // one click = 1000ms-> 0 -1000ms-> 1 -1000ms-> 2 -1000ms-> 3\n * ```\n *\n * @see {@link combineLatestAll}\n * @see {@link concat}\n * @see {@link concatMap}\n * @see {@link concatMapTo}\n * @see {@link exhaustAll}\n * @see {@link mergeAll}\n * @see {@link switchAll}\n * @see {@link switchMap}\n * @see {@link zipAll}\n *\n * @return A function that returns an Observable emitting values from all the\n * inner Observables concatenated.\n */\nexport function concatAll<O extends ObservableInput<any>>(): OperatorFunction<O, ObservedValueOf<O>> {\n  return mergeAll(1);\n}\n", "import { Observable } from '../Observable';\nimport { ObservableInputTuple, SchedulerLike } from '../types';\nimport { concatAll } from '../operators/concatAll';\nimport { popScheduler } from '../util/args';\nimport { from } from './from';\n\nexport function concat<T extends readonly unknown[]>(...inputs: [...ObservableInputTuple<T>]): Observable<T[number]>;\nexport function concat<T extends readonly unknown[]>(\n  ...inputsAndScheduler: [...ObservableInputTuple<T>, SchedulerLike]\n): Observable<T[number]>;\n\n/**\n * Creates an output Observable which sequentially emits all values from the first given\n * Observable and then moves on to the next.\n *\n * <span class=\"informal\">Concatenates multiple Observables together by\n * sequentially emitting their values, one Observable after the other.</span>\n *\n * ![](concat.png)\n *\n * `concat` joins multiple Observables together, by subscribing to them one at a time and\n * merging their results into the output Observable. You can pass either an array of\n * Observables, or put them directly as arguments. Passing an empty array will result\n * in Observable that completes immediately.\n *\n * `concat` will subscribe to first input Observable and emit all its values, without\n * changing or affecting them in any way. When that Observable completes, it will\n * subscribe to then next Observable passed and, again, emit its values. This will be\n * repeated, until the operator runs out of Observables. When last input Observable completes,\n * `concat` will complete as well. At any given moment only one Observable passed to operator\n * emits values. If you would like to emit values from passed Observables concurrently, check out\n * {@link merge} instead, especially with optional `concurrent` parameter. As a matter of fact,\n * `concat` is an equivalent of `merge` operator with `concurrent` parameter set to `1`.\n *\n * Note that if some input Observable never completes, `concat` will also never complete\n * and Observables following the one that did not complete will never be subscribed. On the other\n * hand, if some Observable simply completes immediately after it is subscribed, it will be\n * invisible for `concat`, which will just move on to the next Observable.\n *\n * If any Observable in chain errors, instead of passing control to the next Observable,\n * `concat` will error immediately as well. Observables that would be subscribed after\n * the one that emitted error, never will.\n *\n * If you pass to `concat` the same Observable many times, its stream of values\n * will be \"replayed\" on every subscription, which means you can repeat given Observable\n * as many times as you like. If passing the same Observable to `concat` 1000 times becomes tedious,\n * you can always use {@link repeat}.\n *\n * ## Examples\n *\n * Concatenate a timer counting from 0 to 3 with a synchronous sequence from 1 to 10\n *\n * ```ts\n * import { interval, take, range, concat } from 'rxjs';\n *\n * const timer = interval(1000).pipe(take(4));\n * const sequence = range(1, 10);\n * const result = concat(timer, sequence);\n * result.subscribe(x => console.log(x));\n *\n * // results in:\n * // 0 -1000ms-> 1 -1000ms-> 2 -1000ms-> 3 -immediate-> 1 ... 10\n * ```\n *\n * Concatenate 3 Observables\n *\n * ```ts\n * import { interval, take, concat } from 'rxjs';\n *\n * const timer1 = interval(1000).pipe(take(10));\n * const timer2 = interval(2000).pipe(take(6));\n * const timer3 = interval(500).pipe(take(10));\n *\n * const result = concat(timer1, timer2, timer3);\n * result.subscribe(x => console.log(x));\n *\n * // results in the following:\n * // (Prints to console sequentially)\n * // -1000ms-> 0 -1000ms-> 1 -1000ms-> ... 9\n * // -2000ms-> 0 -2000ms-> 1 -2000ms-> ... 5\n * // -500ms-> 0 -500ms-> 1 -500ms-> ... 9\n * ```\n *\n * Concatenate the same Observable to repeat it\n *\n * ```ts\n * import { interval, take, concat } from 'rxjs';\n *\n * const timer = interval(1000).pipe(take(2));\n *\n * concat(timer, timer) // concatenating the same Observable!\n *   .subscribe({\n *     next: value => console.log(value),\n *     complete: () => console.log('...and it is done!')\n *   });\n *\n * // Logs:\n * // 0 after 1s\n * // 1 after 2s\n * // 0 after 3s\n * // 1 after 4s\n * // '...and it is done!' also after 4s\n * ```\n *\n * @see {@link concatAll}\n * @see {@link concatMap}\n * @see {@link concatMapTo}\n * @see {@link startWith}\n * @see {@link endWith}\n *\n * @param args Input Observables to concatenate.\n */\nexport function concat(...args: any[]): Observable<unknown> {\n  return concatAll()(from(args, popScheduler(args)));\n}\n", "import { Observable } from '../Observable';\nimport { ObservedValueOf, ObservableInput } from '../types';\nimport { innerFrom } from './innerFrom';\n\n/**\n * Creates an Observable that, on subscribe, calls an Observable factory to\n * make an Observable for each new Observer.\n *\n * <span class=\"informal\">Creates the Observable lazily, that is, only when it\n * is subscribed.\n * </span>\n *\n * ![](defer.png)\n *\n * `defer` allows you to create an Observable only when the Observer\n * subscribes. It waits until an Observer subscribes to it, calls the given\n * factory function to get an Observable -- where a factory function typically\n * generates a new Observable -- and subscribes the Observer to this Observable.\n * In case the factory function returns a falsy value, then EMPTY is used as\n * Observable instead. Last but not least, an exception during the factory\n * function call is transferred to the Observer by calling `error`.\n *\n * ## Example\n *\n * Subscribe to either an Observable of clicks or an Observable of interval, at random\n *\n * ```ts\n * import { defer, fromEvent, interval } from 'rxjs';\n *\n * const clicksOrInterval = defer(() => {\n *   return Math.random() > 0.5\n *     ? fromEvent(document, 'click')\n *     : interval(1000);\n * });\n * clicksOrInterval.subscribe(x => console.log(x));\n *\n * // Results in the following behavior:\n * // If the result of Math.random() is greater than 0.5 it will listen\n * // for clicks anywhere on the \"document\"; when document is clicked it\n * // will log a MouseEvent object to the console. If the result is less\n * // than 0.5 it will emit ascending numbers, one every second(1000ms).\n * ```\n *\n * @see {@link Observable}\n *\n * @param {function(): ObservableInput} observableFactory The Observable\n * factory function to invoke for each Observer that subscribes to the output\n * Observable. May also return a Promise, which will be converted on the fly\n * to an Observable.\n * @return {Observable} An Observable whose Observers' subscriptions trigger\n * an invocation of the given Observable factory function.\n */\nexport function defer<R extends ObservableInput<any>>(observableFactory: () => R): Observable<ObservedValueOf<R>> {\n  return new Observable<ObservedValueOf<R>>((subscriber) => {\n    innerFrom(observableFactory()).subscribe(subscriber);\n  });\n}\n", "import { innerFrom } from '../observable/innerFrom';\nimport { Observable } from '../Observable';\nimport { mergeMap } from '../operators/mergeMap';\nimport { isArrayLike } from '../util/isArrayLike';\nimport { isFunction } from '../util/isFunction';\nimport { mapOneOrManyArgs } from '../util/mapOneOrManyArgs';\n\n// These constants are used to create handler registry functions using array mapping below.\nconst nodeEventEmitterMethods = ['addListener', 'removeListener'] as const;\nconst eventTargetMethods = ['addEventListener', 'removeEventListener'] as const;\nconst jqueryMethods = ['on', 'off'] as const;\n\nexport interface NodeStyleEventEmitter {\n  addListener(eventName: string | symbol, handler: NodeEventHandler): this;\n  removeListener(eventName: string | symbol, handler: NodeEventHandler): this;\n}\n\nexport type NodeEventHandler = (...args: any[]) => void;\n\n// For APIs that implement `addListener` and `removeListener` methods that may\n// not use the same arguments or return EventEmitter values\n// such as React Native\nexport interface NodeCompatibleEventEmitter {\n  addListener(eventName: string, handler: NodeEventHandler): void | {};\n  removeListener(eventName: string, handler: NodeEventHandler): void | {};\n}\n\n// Use handler types like those in @types/jquery. See:\n// https://github.com/DefinitelyTyped/DefinitelyTyped/blob/847731ba1d7fa6db6b911c0e43aa0afe596e7723/types/jquery/misc.d.ts#L6395\nexport interface JQueryStyleEventEmitter<TContext, T> {\n  on(eventName: string, handler: (this: TContext, t: T, ...args: any[]) => any): void;\n  off(eventName: string, handler: (this: TContext, t: T, ...args: any[]) => any): void;\n}\n\nexport interface EventListenerObject<E> {\n  handleEvent(evt: E): void;\n}\n\nexport interface HasEventTargetAddRemove<E> {\n  addEventListener(\n    type: string,\n    listener: ((evt: E) => void) | EventListenerObject<E> | null,\n    options?: boolean | AddEventListenerOptions\n  ): void;\n  removeEventListener(\n    type: string,\n    listener: ((evt: E) => void) | EventListenerObject<E> | null,\n    options?: EventListenerOptions | boolean\n  ): void;\n}\n\nexport interface EventListenerOptions {\n  capture?: boolean;\n  passive?: boolean;\n  once?: boolean;\n}\n\nexport interface AddEventListenerOptions extends EventListenerOptions {\n  once?: boolean;\n  passive?: boolean;\n}\n\nexport function fromEvent<T>(target: HasEventTargetAddRemove<T> | ArrayLike<HasEventTargetAddRemove<T>>, eventName: string): Observable<T>;\nexport function fromEvent<T, R>(\n  target: HasEventTargetAddRemove<T> | ArrayLike<HasEventTargetAddRemove<T>>,\n  eventName: string,\n  resultSelector: (event: T) => R\n): Observable<R>;\nexport function fromEvent<T>(\n  target: HasEventTargetAddRemove<T> | ArrayLike<HasEventTargetAddRemove<T>>,\n  eventName: string,\n  options: EventListenerOptions\n): Observable<T>;\nexport function fromEvent<T, R>(\n  target: HasEventTargetAddRemove<T> | ArrayLike<HasEventTargetAddRemove<T>>,\n  eventName: string,\n  options: EventListenerOptions,\n  resultSelector: (event: T) => R\n): Observable<R>;\n\nexport function fromEvent(target: NodeStyleEventEmitter | ArrayLike<NodeStyleEventEmitter>, eventName: string): Observable<unknown>;\n/** @deprecated Do not specify explicit type parameters. Signatures with type parameters that cannot be inferred will be removed in v8. */\nexport function fromEvent<T>(target: NodeStyleEventEmitter | ArrayLike<NodeStyleEventEmitter>, eventName: string): Observable<T>;\nexport function fromEvent<R>(\n  target: NodeStyleEventEmitter | ArrayLike<NodeStyleEventEmitter>,\n  eventName: string,\n  resultSelector: (...args: any[]) => R\n): Observable<R>;\n\nexport function fromEvent(\n  target: NodeCompatibleEventEmitter | ArrayLike<NodeCompatibleEventEmitter>,\n  eventName: string\n): Observable<unknown>;\n/** @deprecated Do not specify explicit type parameters. Signatures with type parameters that cannot be inferred will be removed in v8. */\nexport function fromEvent<T>(target: NodeCompatibleEventEmitter | ArrayLike<NodeCompatibleEventEmitter>, eventName: string): Observable<T>;\nexport function fromEvent<R>(\n  target: NodeCompatibleEventEmitter | ArrayLike<NodeCompatibleEventEmitter>,\n  eventName: string,\n  resultSelector: (...args: any[]) => R\n): Observable<R>;\n\nexport function fromEvent<T>(\n  target: JQueryStyleEventEmitter<any, T> | ArrayLike<JQueryStyleEventEmitter<any, T>>,\n  eventName: string\n): Observable<T>;\nexport function fromEvent<T, R>(\n  target: JQueryStyleEventEmitter<any, T> | ArrayLike<JQueryStyleEventEmitter<any, T>>,\n  eventName: string,\n  resultSelector: (value: T, ...args: any[]) => R\n): Observable<R>;\n\n/**\n * Creates an Observable that emits events of a specific type coming from the\n * given event target.\n *\n * <span class=\"informal\">Creates an Observable from DOM events, or Node.js\n * EventEmitter events or others.</span>\n *\n * ![](fromEvent.png)\n *\n * `fromEvent` accepts as a first argument event target, which is an object with methods\n * for registering event handler functions. As a second argument it takes string that indicates\n * type of event we want to listen for. `fromEvent` supports selected types of event targets,\n * which are described in detail below. If your event target does not match any of the ones listed,\n * you should use {@link fromEventPattern}, which can be used on arbitrary APIs.\n * When it comes to APIs supported by `fromEvent`, their methods for adding and removing event\n * handler functions have different names, but they all accept a string describing event type\n * and function itself, which will be called whenever said event happens.\n *\n * Every time resulting Observable is subscribed, event handler function will be registered\n * to event target on given event type. When that event fires, value\n * passed as a first argument to registered function will be emitted by output Observable.\n * When Observable is unsubscribed, function will be unregistered from event target.\n *\n * Note that if event target calls registered function with more than one argument, second\n * and following arguments will not appear in resulting stream. In order to get access to them,\n * you can pass to `fromEvent` optional project function, which will be called with all arguments\n * passed to event handler. Output Observable will then emit value returned by project function,\n * instead of the usual value.\n *\n * Remember that event targets listed below are checked via duck typing. It means that\n * no matter what kind of object you have and no matter what environment you work in,\n * you can safely use `fromEvent` on that object if it exposes described methods (provided\n * of course they behave as was described above). So for example if Node.js library exposes\n * event target which has the same method names as DOM EventTarget, `fromEvent` is still\n * a good choice.\n *\n * If the API you use is more callback then event handler oriented (subscribed\n * callback function fires only once and thus there is no need to manually\n * unregister it), you should use {@link bindCallback} or {@link bindNodeCallback}\n * instead.\n *\n * `fromEvent` supports following types of event targets:\n *\n * **DOM EventTarget**\n *\n * This is an object with `addEventListener` and `removeEventListener` methods.\n *\n * In the browser, `addEventListener` accepts - apart from event type string and event\n * handler function arguments - optional third parameter, which is either an object or boolean,\n * both used for additional configuration how and when passed function will be called. When\n * `fromEvent` is used with event target of that type, you can provide this values\n * as third parameter as well.\n *\n * **Node.js EventEmitter**\n *\n * An object with `addListener` and `removeListener` methods.\n *\n * **JQuery-style event target**\n *\n * An object with `on` and `off` methods\n *\n * **DOM NodeList**\n *\n * List of DOM Nodes, returned for example by `document.querySelectorAll` or `Node.childNodes`.\n *\n * Although this collection is not event target in itself, `fromEvent` will iterate over all Nodes\n * it contains and install event handler function in every of them. When returned Observable\n * is unsubscribed, function will be removed from all Nodes.\n *\n * **DOM HtmlCollection**\n *\n * Just as in case of NodeList it is a collection of DOM nodes. Here as well event handler function is\n * installed and removed in each of elements.\n *\n *\n * ## Examples\n *\n * Emit clicks happening on the DOM document\n *\n * ```ts\n * import { fromEvent } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * clicks.subscribe(x => console.log(x));\n *\n * // Results in:\n * // MouseEvent object logged to console every time a click\n * // occurs on the document.\n * ```\n *\n * Use `addEventListener` with capture option\n *\n * ```ts\n * import { fromEvent } from 'rxjs';\n *\n * const div = document.createElement('div');\n * div.style.cssText = 'width: 200px; height: 200px; background: #09c;';\n * document.body.appendChild(div);\n *\n * // note optional configuration parameter which will be passed to addEventListener\n * const clicksInDocument = fromEvent(document, 'click', { capture: true });\n * const clicksInDiv = fromEvent(div, 'click');\n *\n * clicksInDocument.subscribe(() => console.log('document'));\n * clicksInDiv.subscribe(() => console.log('div'));\n *\n * // By default events bubble UP in DOM tree, so normally\n * // when we would click on div in document\n * // \"div\" would be logged first and then \"document\".\n * // Since we specified optional `capture` option, document\n * // will catch event when it goes DOWN DOM tree, so console\n * // will log \"document\" and then \"div\".\n * ```\n *\n * @see {@link bindCallback}\n * @see {@link bindNodeCallback}\n * @see {@link fromEventPattern}\n *\n * @param {FromEventTarget<T>} target The DOM EventTarget, Node.js\n * EventEmitter, JQuery-like event target, NodeList or HTMLCollection to attach the event handler to.\n * @param {string} eventName The event name of interest, being emitted by the\n * `target`.\n * @param {EventListenerOptions} [options] Options to pass through to addEventListener\n * @return {Observable<T>}\n */\nexport function fromEvent<T>(\n  target: any,\n  eventName: string,\n  options?: EventListenerOptions | ((...args: any[]) => T),\n  resultSelector?: (...args: any[]) => T\n): Observable<T> {\n  if (isFunction(options)) {\n    resultSelector = options;\n    options = undefined;\n  }\n  if (resultSelector) {\n    return fromEvent<T>(target, eventName, options as EventListenerOptions).pipe(mapOneOrManyArgs(resultSelector));\n  }\n\n  // Figure out our add and remove methods. In order to do this,\n  // we are going to analyze the target in a preferred order, if\n  // the target matches a given signature, we take the two \"add\" and \"remove\"\n  // method names and apply them to a map to create opposite versions of the\n  // same function. This is because they all operate in duplicate pairs,\n  // `addListener(name, handler)`, `removeListener(name, handler)`, for example.\n  // The call only differs by method name, as to whether or not you're adding or removing.\n  const [add, remove] =\n    // If it is an EventTarget, we need to use a slightly different method than the other two patterns.\n    isEventTarget(target)\n      ? eventTargetMethods.map((methodName) => (handler: any) => target[methodName](eventName, handler, options as EventListenerOptions))\n      : // In all other cases, the call pattern is identical with the exception of the method names.\n      isNodeStyleEventEmitter(target)\n      ? nodeEventEmitterMethods.map(toCommonHandlerRegistry(target, eventName))\n      : isJQueryStyleEventEmitter(target)\n      ? jqueryMethods.map(toCommonHandlerRegistry(target, eventName))\n      : [];\n\n  // If add is falsy, it's because we didn't match a pattern above.\n  // Check to see if it is an ArrayLike, because if it is, we want to\n  // try to apply fromEvent to all of it's items. We do this check last,\n  // because there are may be some types that are both ArrayLike *and* implement\n  // event registry points, and we'd rather delegate to that when possible.\n  if (!add) {\n    if (isArrayLike(target)) {\n      return mergeMap((subTarget: any) => fromEvent(subTarget, eventName, options as EventListenerOptions))(\n        innerFrom(target)\n      ) as Observable<T>;\n    }\n  }\n\n  // If add is falsy and we made it here, it's because we didn't\n  // match any valid target objects above.\n  if (!add) {\n    throw new TypeError('Invalid event target');\n  }\n\n  return new Observable<T>((subscriber) => {\n    // The handler we are going to register. Forwards the event object, by itself, or\n    // an array of arguments to the event handler, if there is more than one argument,\n    // to the consumer.\n    const handler = (...args: any[]) => subscriber.next(1 < args.length ? args : args[0]);\n    // Do the work of adding the handler to the target.\n    add(handler);\n    // When we finalize, we want to remove the handler and free up memory.\n    return () => remove!(handler);\n  });\n}\n\n/**\n * Used to create `add` and `remove` functions to register and unregister event handlers\n * from a target in the most common handler pattern, where there are only two arguments.\n * (e.g.  `on(name, fn)`, `off(name, fn)`, `addListener(name, fn)`, or `removeListener(name, fn)`)\n * @param target The target we're calling methods on\n * @param eventName The event name for the event we're creating register or unregister functions for\n */\nfunction toCommonHandlerRegistry(target: any, eventName: string) {\n  return (methodName: string) => (handler: any) => target[methodName](eventName, handler);\n}\n\n/**\n * Checks to see if the target implements the required node-style EventEmitter methods\n * for adding and removing event handlers.\n * @param target the object to check\n */\nfunction isNodeStyleEventEmitter(target: any): target is NodeStyleEventEmitter {\n  return isFunction(target.addListener) && isFunction(target.removeListener);\n}\n\n/**\n * Checks to see if the target implements the required jQuery-style EventEmitter methods\n * for adding and removing event handlers.\n * @param target the object to check\n */\nfunction isJQueryStyleEventEmitter(target: any): target is JQueryStyleEventEmitter<any, any> {\n  return isFunction(target.on) && isFunction(target.off);\n}\n\n/**\n * Checks to see if the target implements the required EventTarget methods\n * for adding and removing event handlers.\n * @param target the object to check\n */\nfunction isEventTarget(target: any): target is HasEventTargetAddRemove<any> {\n  return isFunction(target.addEventListener) && isFunction(target.removeEventListener);\n}\n", "import { Observable } from '../Observable';\nimport { isFunction } from '../util/isFunction';\nimport { NodeEventHandler } from './fromEvent';\nimport { mapOneOrManyArgs } from '../util/mapOneOrManyArgs';\n\n/* tslint:disable:max-line-length */\nexport function fromEventPattern<T>(\n  addHandler: (handler: NodeEventHandler) => any,\n  removeHandler?: (handler: NodeEventHandler, signal?: any) => void\n): Observable<T>;\nexport function fromEventPattern<T>(\n  addHandler: (handler: NodeEventHandler) => any,\n  removeHandler?: (handler: NodeEventHandler, signal?: any) => void,\n  resultSelector?: (...args: any[]) => T\n): Observable<T>;\n/* tslint:enable:max-line-length */\n\n/**\n * Creates an Observable from an arbitrary API for registering event handlers.\n *\n * <span class=\"informal\">When that method for adding event handler was something {@link fromEvent}\n * was not prepared for.</span>\n *\n * ![](fromEventPattern.png)\n *\n * `fromEventPattern` allows you to convert into an Observable any API that supports registering handler functions\n * for events. It is similar to {@link fromEvent}, but far\n * more flexible. In fact, all use cases of {@link fromEvent} could be easily handled by\n * `fromEventPattern` (although in slightly more verbose way).\n *\n * This operator accepts as a first argument an `addHandler` function, which will be injected with\n * handler parameter. That handler is actually an event handler function that you now can pass\n * to API expecting it. `addHandler` will be called whenever Observable\n * returned by the operator is subscribed, so registering handler in API will not\n * necessarily happen when `fromEventPattern` is called.\n *\n * After registration, every time an event that we listen to happens,\n * Observable returned by `fromEventPattern` will emit value that event handler\n * function was called with. Note that if event handler was called with more\n * than one argument, second and following arguments will not appear in the Observable.\n *\n * If API you are using allows to unregister event handlers as well, you can pass to `fromEventPattern`\n * another function - `removeHandler` - as a second parameter. It will be injected\n * with the same handler function as before, which now you can use to unregister\n * it from the API. `removeHandler` will be called when consumer of resulting Observable\n * unsubscribes from it.\n *\n * In some APIs unregistering is actually handled differently. Method registering an event handler\n * returns some kind of token, which is later used to identify which function should\n * be unregistered or it itself has method that unregisters event handler.\n * If that is the case with your API, make sure token returned\n * by registering method is returned by `addHandler`. Then it will be passed\n * as a second argument to `removeHandler`, where you will be able to use it.\n *\n * If you need access to all event handler parameters (not only the first one),\n * or you need to transform them in any way, you can call `fromEventPattern` with optional\n * third parameter - project function which will accept all arguments passed to\n * event handler when it is called. Whatever is returned from project function will appear on\n * resulting stream instead of usual event handlers first argument. This means\n * that default project can be thought of as function that takes its first parameter\n * and ignores the rest.\n *\n * ## Examples\n *\n * Emits clicks happening on the DOM document\n *\n * ```ts\n * import { fromEventPattern } from 'rxjs';\n *\n * function addClickHandler(handler) {\n *   document.addEventListener('click', handler);\n * }\n *\n * function removeClickHandler(handler) {\n *   document.removeEventListener('click', handler);\n * }\n *\n * const clicks = fromEventPattern(\n *   addClickHandler,\n *   removeClickHandler\n * );\n * clicks.subscribe(x => console.log(x));\n *\n * // Whenever you click anywhere in the browser, DOM MouseEvent\n * // object will be logged.\n * ```\n *\n * Use with API that returns cancellation token\n *\n * ```ts\n * import { fromEventPattern } from 'rxjs';\n *\n * const token = someAPI.registerEventHandler(function() {});\n * someAPI.unregisterEventHandler(token); // this APIs cancellation method accepts\n *                                        // not handler itself, but special token.\n *\n * const someAPIObservable = fromEventPattern(\n *   function(handler) { return someAPI.registerEventHandler(handler); }, // Note that we return the token here...\n *   function(handler, token) { someAPI.unregisterEventHandler(token); }  // ...to then use it here.\n * );\n * ```\n *\n * Use with project function\n *\n * ```ts\n * import { fromEventPattern } from 'rxjs';\n *\n * someAPI.registerEventHandler((eventType, eventMessage) => {\n *   console.log(eventType, eventMessage); // Logs 'EVENT_TYPE' 'EVENT_MESSAGE' to console.\n * });\n *\n * const someAPIObservable = fromEventPattern(\n *   handler => someAPI.registerEventHandler(handler),\n *   handler => someAPI.unregisterEventHandler(handler)\n *   (eventType, eventMessage) => eventType + ' --- ' + eventMessage // without that function only 'EVENT_TYPE'\n * );                                                                // would be emitted by the Observable\n *\n * someAPIObservable.subscribe(value => console.log(value));\n *\n * // Logs:\n * // 'EVENT_TYPE --- EVENT_MESSAGE'\n * ```\n *\n * @see {@link fromEvent}\n * @see {@link bindCallback}\n * @see {@link bindNodeCallback}\n *\n * @param {function(handler: Function): any} addHandler A function that takes\n * a `handler` function as argument and attaches it somehow to the actual\n * source of events.\n * @param {function(handler: Function, token?: any): void} [removeHandler] A function that\n * takes a `handler` function as an argument and removes it from the event source. If `addHandler`\n * returns some kind of token, `removeHandler` function will have it as a second parameter.\n * @param {function(...args: any): T} [project] A function to\n * transform results. It takes the arguments from the event handler and\n * should return a single value.\n * @return {Observable<T>} Observable which, when an event happens, emits first parameter\n * passed to registered event handler. Alternatively it emits whatever project function returns\n * at that moment.\n */\nexport function fromEventPattern<T>(\n  addHandler: (handler: NodeEventHandler) => any,\n  removeHandler?: (handler: NodeEventHandler, signal?: any) => void,\n  resultSelector?: (...args: any[]) => T\n): Observable<T | T[]> {\n  if (resultSelector) {\n    return fromEventPattern<T>(addHandler, removeHandler).pipe(mapOneOrManyArgs(resultSelector));\n  }\n\n  return new Observable<T | T[]>((subscriber) => {\n    const handler = (...e: T[]) => subscriber.next(e.length === 1 ? e[0] : e);\n    const retValue = addHandler(handler);\n    return isFunction(removeHandler) ? () => removeHandler(handler, retValue) : undefined;\n  });\n}\n", "import { Observable } from '../Observable';\nimport { SchedulerLike } from '../types';\nimport { async as asyncScheduler } from '../scheduler/async';\nimport { isScheduler } from '../util/isScheduler';\nimport { isValidDate } from '../util/isDate';\n\n/**\n * Creates an observable that will wait for a specified time period, or exact date, before\n * emitting the number 0.\n *\n * <span class=\"informal\">Used to emit a notification after a delay.</span>\n *\n * This observable is useful for creating delays in code, or racing against other values\n * for ad-hoc timeouts.\n *\n * The `delay` is specified by default in milliseconds, however providing a custom scheduler could\n * create a different behavior.\n *\n * ## Examples\n *\n * Wait 3 seconds and start another observable\n *\n * You might want to use `timer` to delay subscription to an\n * observable by a set amount of time. Here we use a timer with\n * {@link concatMapTo} or {@link concatMap} in order to wait\n * a few seconds and start a subscription to a source.\n *\n * ```ts\n * import { of, timer, concatMap } from 'rxjs';\n *\n * // This could be any observable\n * const source = of(1, 2, 3);\n *\n * timer(3000)\n *   .pipe(concatMap(() => source))\n *   .subscribe(console.log);\n * ```\n *\n * Take all values until the start of the next minute\n *\n * Using a `Date` as the trigger for the first emission, you can\n * do things like wait until midnight to fire an event, or in this case,\n * wait until a new minute starts (chosen so the example wouldn't take\n * too long to run) in order to stop watching a stream. Leveraging\n * {@link takeUntil}.\n *\n * ```ts\n * import { interval, takeUntil, timer } from 'rxjs';\n *\n * // Build a Date object that marks the\n * // next minute.\n * const currentDate = new Date();\n * const startOfNextMinute = new Date(\n *   currentDate.getFullYear(),\n *   currentDate.getMonth(),\n *   currentDate.getDate(),\n *   currentDate.getHours(),\n *   currentDate.getMinutes() + 1\n * );\n *\n * // This could be any observable stream\n * const source = interval(1000);\n *\n * const result = source.pipe(\n *   takeUntil(timer(startOfNextMinute))\n * );\n *\n * result.subscribe(console.log);\n * ```\n *\n * ### Known Limitations\n *\n * - The {@link asyncScheduler} uses `setTimeout` which has limitations for how far in the future it can be scheduled.\n *\n * - If a `scheduler` is provided that returns a timestamp other than an epoch from `now()`, and\n * a `Date` object is passed to the `dueTime` argument, the calculation for when the first emission\n * should occur will be incorrect. In this case, it would be best to do your own calculations\n * ahead of time, and pass a `number` in as the `dueTime`.\n *\n * @param due If a `number`, the amount of time in milliseconds to wait before emitting.\n * If a `Date`, the exact time at which to emit.\n * @param scheduler The scheduler to use to schedule the delay. Defaults to {@link asyncScheduler}.\n */\nexport function timer(due: number | Date, scheduler?: SchedulerLike): Observable<0>;\n\n/**\n * Creates an observable that starts an interval after a specified delay, emitting incrementing numbers -- starting at `0` --\n * on each interval after words.\n *\n * The `delay` and `intervalDuration` are specified by default in milliseconds, however providing a custom scheduler could\n * create a different behavior.\n *\n * ## Example\n *\n * ### Start an interval that starts right away\n *\n * Since {@link interval} waits for the passed delay before starting,\n * sometimes that's not ideal. You may want to start an interval immediately.\n * `timer` works well for this. Here we have both side-by-side so you can\n * see them in comparison.\n *\n * Note that this observable will never complete.\n *\n * ```ts\n * import { timer, interval } from 'rxjs';\n *\n * timer(0, 1000).subscribe(n => console.log('timer', n));\n * interval(1000).subscribe(n => console.log('interval', n));\n * ```\n *\n * ### Known Limitations\n *\n * - The {@link asyncScheduler} uses `setTimeout` which has limitations for how far in the future it can be scheduled.\n *\n * - If a `scheduler` is provided that returns a timestamp other than an epoch from `now()`, and\n * a `Date` object is passed to the `dueTime` argument, the calculation for when the first emission\n * should occur will be incorrect. In this case, it would be best to do your own calculations\n * ahead of time, and pass a `number` in as the `startDue`.\n * @param startDue If a `number`, is the time to wait before starting the interval.\n * If a `Date`, is the exact time at which to start the interval.\n * @param intervalDuration The delay between each value emitted in the interval. Passing a\n * negative number here will result in immediate completion after the first value is emitted, as though\n * no `intervalDuration` was passed at all.\n * @param scheduler The scheduler to use to schedule the delay. Defaults to {@link asyncScheduler}.\n */\nexport function timer(startDue: number | Date, intervalDuration: number, scheduler?: SchedulerLike): Observable<number>;\n\n/**\n * @deprecated The signature allowing `undefined` to be passed for `intervalDuration` will be removed in v8. Use the `timer(dueTime, scheduler?)` signature instead.\n */\nexport function timer(dueTime: number | Date, unused: undefined, scheduler?: SchedulerLike): Observable<0>;\n\nexport function timer(\n  dueTime: number | Date = 0,\n  intervalOrScheduler?: number | SchedulerLike,\n  scheduler: SchedulerLike = asyncScheduler\n): Observable<number> {\n  // Since negative intervalDuration is treated as though no\n  // interval was specified at all, we start with a negative number.\n  let intervalDuration = -1;\n\n  if (intervalOrScheduler != null) {\n    // If we have a second argument, and it's a scheduler,\n    // override the scheduler we had defaulted. Otherwise,\n    // it must be an interval.\n    if (isScheduler(intervalOrScheduler)) {\n      scheduler = intervalOrScheduler;\n    } else {\n      // Note that this *could* be negative, in which case\n      // it's like not passing an intervalDuration at all.\n      intervalDuration = intervalOrScheduler;\n    }\n  }\n\n  return new Observable((subscriber) => {\n    // If a valid date is passed, calculate how long to wait before\n    // executing the first value... otherwise, if it's a number just schedule\n    // that many milliseconds (or scheduler-specified unit size) in the future.\n    let due = isValidDate(dueTime) ? +dueTime - scheduler!.now() : dueTime;\n\n    if (due < 0) {\n      // Ensure we don't schedule in the future.\n      due = 0;\n    }\n\n    // The incrementing value we emit.\n    let n = 0;\n\n    // Start the timer.\n    return scheduler.schedule(function () {\n      if (!subscriber.closed) {\n        // Emit the next value and increment.\n        subscriber.next(n++);\n\n        if (0 <= intervalDuration) {\n          // If we have a interval after the initial timer,\n          // reschedule with the period.\n          this.schedule(undefined, intervalDuration);\n        } else {\n          // We didn't have an interval. So just complete.\n          subscriber.complete();\n        }\n      }\n    }, due);\n  });\n}\n", "import { Observable } from '../Observable';\nimport { ObservableInput, ObservableInputTuple, SchedulerLike } from '../types';\nimport { mergeAll } from '../operators/mergeAll';\nimport { innerFrom } from './innerFrom';\nimport { EMPTY } from './empty';\nimport { popNumber, popScheduler } from '../util/args';\nimport { from } from './from';\n\nexport function merge<A extends readonly unknown[]>(...sources: [...ObservableInputTuple<A>]): Observable<A[number]>;\nexport function merge<A extends readonly unknown[]>(...sourcesAndConcurrency: [...ObservableInputTuple<A>, number?]): Observable<A[number]>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `mergeAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function merge<A extends readonly unknown[]>(\n  ...sourcesAndScheduler: [...ObservableInputTuple<A>, SchedulerLike?]\n): Observable<A[number]>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `mergeAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function merge<A extends readonly unknown[]>(\n  ...sourcesAndConcurrencyAndScheduler: [...ObservableInputTuple<A>, number?, SchedulerLike?]\n): Observable<A[number]>;\n\n/**\n * Creates an output Observable which concurrently emits all values from every\n * given input Observable.\n *\n * <span class=\"informal\">Flattens multiple Observables together by blending\n * their values into one Observable.</span>\n *\n * ![](merge.png)\n *\n * `merge` subscribes to each given input Observable (as arguments), and simply\n * forwards (without doing any transformation) all the values from all the input\n * Observables to the output Observable. The output Observable only completes\n * once all input Observables have completed. Any error delivered by an input\n * Observable will be immediately emitted on the output Observable.\n *\n * ## Examples\n *\n * Merge together two Observables: 1s interval and clicks\n *\n * ```ts\n * import { merge, fromEvent, interval } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const timer = interval(1000);\n * const clicksOrTimer = merge(clicks, timer);\n * clicksOrTimer.subscribe(x => console.log(x));\n *\n * // Results in the following:\n * // timer will emit ascending values, one every second(1000ms) to console\n * // clicks logs MouseEvents to console every time the \"document\" is clicked\n * // Since the two streams are merged you see these happening\n * // as they occur.\n * ```\n *\n * Merge together 3 Observables, but run only 2 concurrently\n *\n * ```ts\n * import { interval, take, merge } from 'rxjs';\n *\n * const timer1 = interval(1000).pipe(take(10));\n * const timer2 = interval(2000).pipe(take(6));\n * const timer3 = interval(500).pipe(take(10));\n *\n * const concurrent = 2; // the argument\n * const merged = merge(timer1, timer2, timer3, concurrent);\n * merged.subscribe(x => console.log(x));\n *\n * // Results in the following:\n * // - First timer1 and timer2 will run concurrently\n * // - timer1 will emit a value every 1000ms for 10 iterations\n * // - timer2 will emit a value every 2000ms for 6 iterations\n * // - after timer1 hits its max iteration, timer2 will\n * //   continue, and timer3 will start to run concurrently with timer2\n * // - when timer2 hits its max iteration it terminates, and\n * //   timer3 will continue to emit a value every 500ms until it is complete\n * ```\n *\n * @see {@link mergeAll}\n * @see {@link mergeMap}\n * @see {@link mergeMapTo}\n * @see {@link mergeScan}\n *\n * @param {...ObservableInput} observables Input Observables to merge together.\n * @param {number} [concurrent=Infinity] Maximum number of input\n * Observables being subscribed to concurrently.\n * @param {SchedulerLike} [scheduler=null] The {@link SchedulerLike} to use for managing\n * concurrency of input Observables.\n * @return {Observable} an Observable that emits items that are the result of\n * every input Observable.\n */\nexport function merge(...args: (ObservableInput<unknown> | number | SchedulerLike)[]): Observable<unknown> {\n  const scheduler = popScheduler(args);\n  const concurrent = popNumber(args, Infinity);\n  const sources = args as ObservableInput<unknown>[];\n  return !sources.length\n    ? // No source provided\n      EMPTY\n    : sources.length === 1\n    ? // One source? Just return it.\n      innerFrom(sources[0])\n    : // Merge all sources\n      mergeAll(concurrent)(from(sources, scheduler));\n}\n", "import { Observable } from '../Observable';\nimport { noop } from '../util/noop';\n\n/**\n * An Observable that emits no items to the Observer and never completes.\n *\n * ![](never.png)\n *\n * A simple Observable that emits neither values nor errors nor the completion\n * notification. It can be used for testing purposes or for composing with other\n * Observables. Please note that by never emitting a complete notification, this\n * Observable keeps the subscription from being disposed automatically.\n * Subscriptions need to be manually disposed.\n *\n * ##  Example\n *\n * Emit the number 7, then never emit anything else (not even complete)\n *\n * ```ts\n * import { NEVER, startWith } from 'rxjs';\n *\n * const info = () => console.log('Will not be called');\n *\n * const result = NEVER.pipe(startWith(7));\n * result.subscribe({\n *   next: x => console.log(x),\n *   error: info,\n *   complete: info\n * });\n * ```\n *\n * @see {@link Observable}\n * @see {@link EMPTY}\n * @see {@link of}\n * @see {@link throwError}\n */\nexport const NEVER = new Observable<never>(noop);\n\n/**\n * @deprecated Replaced with the {@link NEVER} constant. Will be removed in v8.\n */\nexport function never() {\n  return NEVER;\n}\n", "const { isArray } = Array;\n\n/**\n * Used in operators and functions that accept either a list of arguments, or an array of arguments\n * as a single argument.\n */\nexport function argsOrArgArray<T>(args: (T | T[])[]): T[] {\n  return args.length === 1 && isArray(args[0]) ? args[0] : (args as T[]);\n}\n", "import { OperatorFunction, MonoTypeOperatorFunction, TruthyTypesOf } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/** @deprecated Use a closure instead of a `thisArg`. Signatures accepting a `thisArg` will be removed in v8. */\nexport function filter<T, S extends T, A>(predicate: (this: A, value: T, index: number) => value is S, thisArg: A): OperatorFunction<T, S>;\nexport function filter<T, S extends T>(predicate: (value: T, index: number) => value is S): OperatorFunction<T, S>;\nexport function filter<T>(predicate: BooleanConstructor): OperatorFunction<T, TruthyTypesOf<T>>;\n/** @deprecated Use a closure instead of a `thisArg`. Signatures accepting a `thisArg` will be removed in v8. */\nexport function filter<T, A>(predicate: (this: A, value: T, index: number) => boolean, thisArg: A): MonoTypeOperatorFunction<T>;\nexport function filter<T>(predicate: (value: T, index: number) => boolean): MonoTypeOperatorFunction<T>;\n\n/**\n * Filter items emitted by the source Observable by only emitting those that\n * satisfy a specified predicate.\n *\n * <span class=\"informal\">Like\n * [Array.prototype.filter()](https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array/filter),\n * it only emits a value from the source if it passes a criterion function.</span>\n *\n * ![](filter.png)\n *\n * Similar to the well-known `Array.prototype.filter` method, this operator\n * takes values from the source Observable, passes them through a `predicate`\n * function and only emits those values that yielded `true`.\n *\n * ## Example\n *\n * Emit only click events whose target was a DIV element\n *\n * ```ts\n * import { fromEvent, filter } from 'rxjs';\n *\n * const div = document.createElement('div');\n * div.style.cssText = 'width: 200px; height: 200px; background: #09c;';\n * document.body.appendChild(div);\n *\n * const clicks = fromEvent(document, 'click');\n * const clicksOnDivs = clicks.pipe(filter(ev => (<HTMLElement>ev.target).tagName === 'DIV'));\n * clicksOnDivs.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link distinct}\n * @see {@link distinctUntilChanged}\n * @see {@link distinctUntilKeyChanged}\n * @see {@link ignoreElements}\n * @see {@link partition}\n * @see {@link skip}\n *\n * @param predicate A function that\n * evaluates each value emitted by the source Observable. If it returns `true`,\n * the value is emitted, if `false` the value is not passed to the output\n * Observable. The `index` parameter is the number `i` for the i-th source\n * emission that has happened since the subscription, starting from the number\n * `0`.\n * @param thisArg An optional argument to determine the value of `this`\n * in the `predicate` function.\n * @return A function that returns an Observable that emits items from the\n * source Observable that satisfy the specified `predicate`.\n */\nexport function filter<T>(predicate: (value: T, index: number) => boolean, thisArg?: any): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    // An index passed to our predicate function on each call.\n    let index = 0;\n\n    // Subscribe to the source, all errors and completions are\n    // forwarded to the consumer.\n    source.subscribe(\n      // Call the predicate with the appropriate `this` context,\n      // if the predicate returns `true`, then send the value\n      // to the consumer.\n      createOperatorSubscriber(subscriber, (value) => predicate.call(thisArg, value, index++) && subscriber.next(value))\n    );\n  });\n}\n", "import { Observable } from '../Observable';\nimport { ObservableInputTuple } from '../types';\nimport { innerFrom } from './innerFrom';\nimport { argsOrArgArray } from '../util/argsOrArgArray';\nimport { EMPTY } from './empty';\nimport { createOperatorSubscriber } from '../operators/OperatorSubscriber';\nimport { popResultSelector } from '../util/args';\n\nexport function zip<A extends readonly unknown[]>(sources: [...ObservableInputTuple<A>]): Observable<A>;\nexport function zip<A extends readonly unknown[], R>(\n  sources: [...ObservableInputTuple<A>],\n  resultSelector: (...values: A) => R\n): Observable<R>;\nexport function zip<A extends readonly unknown[]>(...sources: [...ObservableInputTuple<A>]): Observable<A>;\nexport function zip<A extends readonly unknown[], R>(\n  ...sourcesAndResultSelector: [...ObservableInputTuple<A>, (...values: A) => R]\n): Observable<R>;\n\n/**\n * Combines multiple Observables to create an Observable whose values are calculated from the values, in order, of each\n * of its input Observables.\n *\n * If the last parameter is a function, this function is used to compute the created value from the input values.\n * Otherwise, an array of the input values is returned.\n *\n * ## Example\n *\n * Combine age and name from different sources\n *\n * ```ts\n * import { of, zip, map } from 'rxjs';\n *\n * const age$ = of(27, 25, 29);\n * const name$ = of('Foo', 'Bar', 'Beer');\n * const isDev$ = of(true, true, false);\n *\n * zip(age$, name$, isDev$).pipe(\n *   map(([age, name, isDev]) => ({ age, name, isDev }))\n * )\n * .subscribe(x => console.log(x));\n *\n * // Outputs\n * // { age: 27, name: 'Foo', isDev: true }\n * // { age: 25, name: 'Bar', isDev: true }\n * // { age: 29, name: 'Beer', isDev: false }\n * ```\n *\n * @param sources\n * @return {Observable<R>}\n */\nexport function zip(...args: unknown[]): Observable<unknown> {\n  const resultSelector = popResultSelector(args);\n\n  const sources = argsOrArgArray(args) as Observable<unknown>[];\n\n  return sources.length\n    ? new Observable<unknown[]>((subscriber) => {\n        // A collection of buffers of values from each source.\n        // Keyed by the same index with which the sources were passed in.\n        let buffers: unknown[][] = sources.map(() => []);\n\n        // An array of flags of whether or not the sources have completed.\n        // This is used to check to see if we should complete the result.\n        // Keyed by the same index with which the sources were passed in.\n        let completed = sources.map(() => false);\n\n        // When everything is done, release the arrays above.\n        subscriber.add(() => {\n          buffers = completed = null!;\n        });\n\n        // Loop over our sources and subscribe to each one. The index `i` is\n        // especially important here, because we use it in closures below to\n        // access the related buffers and completion properties\n        for (let sourceIndex = 0; !subscriber.closed && sourceIndex < sources.length; sourceIndex++) {\n          innerFrom(sources[sourceIndex]).subscribe(\n            createOperatorSubscriber(\n              subscriber,\n              (value) => {\n                buffers[sourceIndex].push(value);\n                // if every buffer has at least one value in it, then we\n                // can shift out the oldest value from each buffer and emit\n                // them as an array.\n                if (buffers.every((buffer) => buffer.length)) {\n                  const result: any = buffers.map((buffer) => buffer.shift()!);\n                  // Emit the array. If theres' a result selector, use that.\n                  subscriber.next(resultSelector ? resultSelector(...result) : result);\n                  // If any one of the sources is both complete and has an empty buffer\n                  // then we complete the result. This is because we cannot possibly have\n                  // any more values to zip together.\n                  if (buffers.some((buffer, i) => !buffer.length && completed[i])) {\n                    subscriber.complete();\n                  }\n                }\n              },\n              () => {\n                // This source completed. Mark it as complete so we can check it later\n                // if we have to.\n                completed[sourceIndex] = true;\n                // But, if this complete source has nothing in its buffer, then we\n                // can complete the result, because we can't possibly have any more\n                // values from this to zip together with the other values.\n                !buffers[sourceIndex].length && subscriber.complete();\n              }\n            )\n          );\n        }\n\n        // When everything is done, release the arrays above.\n        return () => {\n          buffers = completed = null!;\n        };\n      })\n    : EMPTY;\n}\n", "import { Subscriber } from '../Subscriber';\nimport { MonoTypeOperatorFunction, ObservableInput } from '../types';\n\nimport { operate } from '../util/lift';\nimport { innerFrom } from '../observable/innerFrom';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * Ignores source values for a duration determined by another Observable, then\n * emits the most recent value from the source Observable, then repeats this\n * process.\n *\n * <span class=\"informal\">It's like {@link auditTime}, but the silencing\n * duration is determined by a second Observable.</span>\n *\n * ![](audit.svg)\n *\n * `audit` is similar to `throttle`, but emits the last value from the silenced\n * time window, instead of the first value. `audit` emits the most recent value\n * from the source Observable on the output Observable as soon as its internal\n * timer becomes disabled, and ignores source values while the timer is enabled.\n * Initially, the timer is disabled. As soon as the first source value arrives,\n * the timer is enabled by calling the `durationSelector` function with the\n * source value, which returns the \"duration\" Observable. When the duration\n * Observable emits a value, the timer is disabled, then the most\n * recent source value is emitted on the output Observable, and this process\n * repeats for the next source value.\n *\n * ## Example\n *\n * Emit clicks at a rate of at most one click per second\n *\n * ```ts\n * import { fromEvent, audit, interval } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(audit(ev => interval(1000)));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link auditTime}\n * @see {@link debounce}\n * @see {@link delayWhen}\n * @see {@link sample}\n * @see {@link throttle}\n *\n * @param durationSelector A function\n * that receives a value from the source Observable, for computing the silencing\n * duration, returned as an Observable or a Promise.\n * @return A function that returns an Observable that performs rate-limiting of\n * emissions from the source Observable.\n */\nexport function audit<T>(durationSelector: (value: T) => ObservableInput<any>): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    let hasValue = false;\n    let lastValue: T | null = null;\n    let durationSubscriber: Subscriber<any> | null = null;\n    let isComplete = false;\n\n    const endDuration = () => {\n      durationSubscriber?.unsubscribe();\n      durationSubscriber = null;\n      if (hasValue) {\n        hasValue = false;\n        const value = lastValue!;\n        lastValue = null;\n        subscriber.next(value);\n      }\n      isComplete && subscriber.complete();\n    };\n\n    const cleanupDuration = () => {\n      durationSubscriber = null;\n      isComplete && subscriber.complete();\n    };\n\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value) => {\n          hasValue = true;\n          lastValue = value;\n          if (!durationSubscriber) {\n            innerFrom(durationSelector(value)).subscribe(\n              (durationSubscriber = createOperatorSubscriber(subscriber, endDuration, cleanupDuration))\n            );\n          }\n        },\n        () => {\n          isComplete = true;\n          (!hasValue || !durationSubscriber || durationSubscriber.closed) && subscriber.complete();\n        }\n      )\n    );\n  });\n}\n", "import { asyncScheduler } from '../scheduler/async';\nimport { audit } from './audit';\nimport { timer } from '../observable/timer';\nimport { MonoTypeOperatorFunction, SchedulerLike } from '../types';\n\n/**\n * Ignores source values for `duration` milliseconds, then emits the most recent\n * value from the source Observable, then repeats this process.\n *\n * <span class=\"informal\">When it sees a source value, it ignores that plus\n * the next ones for `duration` milliseconds, and then it emits the most recent\n * value from the source.</span>\n *\n * ![](auditTime.png)\n *\n * `auditTime` is similar to `throttleTime`, but emits the last value from the\n * silenced time window, instead of the first value. `auditTime` emits the most\n * recent value from the source Observable on the output Observable as soon as\n * its internal timer becomes disabled, and ignores source values while the\n * timer is enabled. Initially, the timer is disabled. As soon as the first\n * source value arrives, the timer is enabled. After `duration` milliseconds (or\n * the time unit determined internally by the optional `scheduler`) has passed,\n * the timer is disabled, then the most recent source value is emitted on the\n * output Observable, and this process repeats for the next source value.\n * Optionally takes a {@link SchedulerLike} for managing timers.\n *\n * ## Example\n *\n * Emit clicks at a rate of at most one click per second\n *\n * ```ts\n * import { fromEvent, auditTime } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(auditTime(1000));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link audit}\n * @see {@link debounceTime}\n * @see {@link delay}\n * @see {@link sampleTime}\n * @see {@link throttleTime}\n *\n * @param {number} duration Time to wait before emitting the most recent source\n * value, measured in milliseconds or the time unit determined internally\n * by the optional `scheduler`.\n * @param {SchedulerLike} [scheduler=async] The {@link SchedulerLike} to use for\n * managing the timers that handle the rate-limiting behavior.\n * @return A function that returns an Observable that performs rate-limiting of\n * emissions from the source Observable.\n */\nexport function auditTime<T>(duration: number, scheduler: SchedulerLike = asyncScheduler): MonoTypeOperatorFunction<T> {\n  return audit(() => timer(duration, scheduler));\n}\n", "import { OperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { arrRemove } from '../util/arrRemove';\n\n/**\n * Buffers the source Observable values until the size hits the maximum\n * `bufferSize` given.\n *\n * <span class=\"informal\">Collects values from the past as an array, and emits\n * that array only when its size reaches `bufferSize`.</span>\n *\n * ![](bufferCount.png)\n *\n * Buffers a number of values from the source Observable by `bufferSize` then\n * emits the buffer and clears it, and starts a new buffer each\n * `startBufferEvery` values. If `startBufferEvery` is not provided or is\n * `null`, then new buffers are started immediately at the start of the source\n * and when each buffer closes and is emitted.\n *\n * ## Examples\n *\n * Emit the last two click events as an array\n *\n * ```ts\n * import { fromEvent, bufferCount } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const buffered = clicks.pipe(bufferCount(2));\n * buffered.subscribe(x => console.log(x));\n * ```\n *\n * On every click, emit the last two click events as an array\n *\n * ```ts\n * import { fromEvent, bufferCount } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const buffered = clicks.pipe(bufferCount(2, 1));\n * buffered.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link buffer}\n * @see {@link bufferTime}\n * @see {@link bufferToggle}\n * @see {@link bufferWhen}\n * @see {@link pairwise}\n * @see {@link windowCount}\n *\n * @param {number} bufferSize The maximum size of the buffer emitted.\n * @param {number} [startBufferEvery] Interval at which to start a new buffer.\n * For example if `startBufferEvery` is `2`, then a new buffer will be started\n * on every other value from the source. A new buffer is started at the\n * beginning of the source by default.\n * @return A function that returns an Observable of arrays of buffered values.\n */\nexport function bufferCount<T>(bufferSize: number, startBufferEvery: number | null = null): OperatorFunction<T, T[]> {\n  // If no `startBufferEvery` value was supplied, then we're\n  // opening and closing on the bufferSize itself.\n  startBufferEvery = startBufferEvery ?? bufferSize;\n\n  return operate((source, subscriber) => {\n    let buffers: T[][] = [];\n    let count = 0;\n\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value) => {\n          let toEmit: T[][] | null = null;\n\n          // Check to see if we need to start a buffer.\n          // This will start one at the first value, and then\n          // a new one every N after that.\n          if (count++ % startBufferEvery! === 0) {\n            buffers.push([]);\n          }\n\n          // Push our value into our active buffers.\n          for (const buffer of buffers) {\n            buffer.push(value);\n            // Check to see if we're over the bufferSize\n            // if we are, record it so we can emit it later.\n            // If we emitted it now and removed it, it would\n            // mutate the `buffers` array while we're looping\n            // over it.\n            if (bufferSize <= buffer.length) {\n              toEmit = toEmit ?? [];\n              toEmit.push(buffer);\n            }\n          }\n\n          if (toEmit) {\n            // We have found some buffers that are over the\n            // `bufferSize`. Emit them, and remove them from our\n            // buffers list.\n            for (const buffer of toEmit) {\n              arrRemove(buffers, buffer);\n              subscriber.next(buffer);\n            }\n          }\n        },\n        () => {\n          // When the source completes, emit all of our\n          // active buffers.\n          for (const buffer of buffers) {\n            subscriber.next(buffer);\n          }\n          subscriber.complete();\n        },\n        // Pass all errors through to consumer.\n        undefined,\n        () => {\n          // Clean up our memory when we finalize\n          buffers = null!;\n        }\n      )\n    );\n  });\n}\n", "import { Observable } from '../Observable';\n\nimport { ObservableInput, OperatorFunction, ObservedValueOf } from '../types';\nimport { Subscription } from '../Subscription';\nimport { innerFrom } from '../observable/innerFrom';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { operate } from '../util/lift';\n\n/* tslint:disable:max-line-length */\nexport function catchError<T, O extends ObservableInput<any>>(\n  selector: (err: any, caught: Observable<T>) => O\n): OperatorFunction<T, T | ObservedValueOf<O>>;\n/* tslint:enable:max-line-length */\n\n/**\n * Catches errors on the observable to be handled by returning a new observable or throwing an error.\n *\n * <span class=\"informal\">\n * It only listens to the error channel and ignores notifications.\n * Handles errors from the source observable, and maps them to a new observable.\n * The error may also be rethrown, or a new error can be thrown to emit an error from the result.\n * </span>\n *\n * ![](catch.png)\n *\n * This operator handles errors, but forwards along all other events to the resulting observable.\n * If the source observable terminates with an error, it will map that error to a new observable,\n * subscribe to it, and forward all of its events to the resulting observable.\n *\n * ## Examples\n *\n * Continue with a different Observable when there's an error\n *\n * ```ts\n * import { of, map, catchError } from 'rxjs';\n *\n * of(1, 2, 3, 4, 5)\n *   .pipe(\n *     map(n => {\n *       if (n === 4) {\n *         throw 'four!';\n *       }\n *       return n;\n *     }),\n *     catchError(err => of('I', 'II', 'III', 'IV', 'V'))\n *   )\n *   .subscribe(x => console.log(x));\n *   // 1, 2, 3, I, II, III, IV, V\n * ```\n *\n * Retry the caught source Observable again in case of error, similar to `retry()` operator\n *\n * ```ts\n * import { of, map, catchError, take } from 'rxjs';\n *\n * of(1, 2, 3, 4, 5)\n *   .pipe(\n *     map(n => {\n *       if (n === 4) {\n *         throw 'four!';\n *       }\n *       return n;\n *     }),\n *     catchError((err, caught) => caught),\n *     take(30)\n *   )\n *   .subscribe(x => console.log(x));\n *   // 1, 2, 3, 1, 2, 3, ...\n * ```\n *\n * Throw a new error when the source Observable throws an error\n *\n * ```ts\n * import { of, map, catchError } from 'rxjs';\n *\n * of(1, 2, 3, 4, 5)\n *   .pipe(\n *     map(n => {\n *       if (n === 4) {\n *         throw 'four!';\n *       }\n *       return n;\n *     }),\n *     catchError(err => {\n *       throw 'error in source. Details: ' + err;\n *     })\n *   )\n *   .subscribe({\n *     next: x => console.log(x),\n *     error: err => console.log(err)\n *   });\n *   // 1, 2, 3, error in source. Details: four!\n * ```\n *\n * @see {@link onErrorResumeNext}\n * @see {@link repeat}\n * @see {@link repeatWhen}\n * @see {@link retry }\n * @see {@link retryWhen}\n *\n * @param {function} selector a function that takes as arguments `err`, which is the error, and `caught`, which\n * is the source observable, in case you'd like to \"retry\" that observable by returning it again. Whatever observable\n * is returned by the `selector` will be used to continue the observable chain.\n * @return A function that returns an Observable that originates from either\n * the source or the Observable returned by the `selector` function.\n */\nexport function catchError<T, O extends ObservableInput<any>>(\n  selector: (err: any, caught: Observable<T>) => O\n): OperatorFunction<T, T | ObservedValueOf<O>> {\n  return operate((source, subscriber) => {\n    let innerSub: Subscription | null = null;\n    let syncUnsub = false;\n    let handledResult: Observable<ObservedValueOf<O>>;\n\n    innerSub = source.subscribe(\n      createOperatorSubscriber(subscriber, undefined, undefined, (err) => {\n        handledResult = innerFrom(selector(err, catchError(selector)(source)));\n        if (innerSub) {\n          innerSub.unsubscribe();\n          innerSub = null;\n          handledResult.subscribe(subscriber);\n        } else {\n          // We don't have an innerSub yet, that means the error was synchronous\n          // because the subscribe call hasn't returned yet.\n          syncUnsub = true;\n        }\n      })\n    );\n\n    if (syncUnsub) {\n      // We have a synchronous error, we need to make sure to\n      // finalize right away. This ensures that callbacks in the `finalize` operator are called\n      // at the right time, and that finalization occurs at the expected\n      // time between the source error and the subscription to the\n      // next observable.\n      innerSub.unsubscribe();\n      innerSub = null;\n      handledResult!.subscribe(subscriber);\n    }\n  });\n}\n", "import { Observable } from '../Observable';\nimport { Subscriber } from '../Subscriber';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * A basic scan operation. This is used for `scan` and `reduce`.\n * @param accumulator The accumulator to use\n * @param seed The seed value for the state to accumulate\n * @param hasSeed Whether or not a seed was provided\n * @param emitOnNext Whether or not to emit the state on next\n * @param emitBeforeComplete Whether or not to emit the before completion\n */\n\nexport function scanInternals<V, A, S>(\n  accumulator: (acc: V | A | S, value: V, index: number) => A,\n  seed: S,\n  hasSeed: boolean,\n  emitOnNext: boolean,\n  emitBeforeComplete?: undefined | true\n) {\n  return (source: Observable<V>, subscriber: Subscriber<any>) => {\n    // Whether or not we have state yet. This will only be\n    // false before the first value arrives if we didn't get\n    // a seed value.\n    let hasState = hasSeed;\n    // The state that we're tracking, starting with the seed,\n    // if there is one, and then updated by the return value\n    // from the accumulator on each emission.\n    let state: any = seed;\n    // An index to pass to the accumulator function.\n    let index = 0;\n\n    // Subscribe to our source. All errors and completions are passed through.\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value) => {\n          // Always increment the index.\n          const i = index++;\n          // Set the state\n          state = hasState\n            ? // We already have state, so we can get the new state from the accumulator\n              accumulator(state, value, i)\n            : // We didn't have state yet, a seed value was not provided, so\n\n              // we set the state to the first value, and mark that we have state now\n              ((hasState = true), value);\n\n          // Maybe send it to the consumer.\n          emitOnNext && subscriber.next(state);\n        },\n        // If an onComplete was given, call it, otherwise\n        // just pass through the complete notification to the consumer.\n        emitBeforeComplete &&\n          (() => {\n            hasState && subscriber.next(state);\n            subscriber.complete();\n          })\n      )\n    );\n  };\n}\n", "import { combineLatestInit } from '../observable/combineLatest';\nimport { ObservableInput, ObservableInputTuple, OperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { argsOrArgArray } from '../util/argsOrArgArray';\nimport { mapOneOrManyArgs } from '../util/mapOneOrManyArgs';\nimport { pipe } from '../util/pipe';\nimport { popResultSelector } from '../util/args';\n\n/** @deprecated Replaced with {@link combineLatestWith}. Will be removed in v8. */\nexport function combineLatest<T, A extends readonly unknown[], R>(\n  sources: [...ObservableInputTuple<A>],\n  project: (...values: [T, ...A]) => R\n): OperatorFunction<T, R>;\n/** @deprecated Replaced with {@link combineLatestWith}. Will be removed in v8. */\nexport function combineLatest<T, A extends readonly unknown[], R>(sources: [...ObservableInputTuple<A>]): OperatorFunction<T, [T, ...A]>;\n\n/** @deprecated Replaced with {@link combineLatestWith}. Will be removed in v8. */\nexport function combineLatest<T, A extends readonly unknown[], R>(\n  ...sourcesAndProject: [...ObservableInputTuple<A>, (...values: [T, ...A]) => R]\n): OperatorFunction<T, R>;\n/** @deprecated Replaced with {@link combineLatestWith}. Will be removed in v8. */\nexport function combineLatest<T, A extends readonly unknown[], R>(...sources: [...ObservableInputTuple<A>]): OperatorFunction<T, [T, ...A]>;\n\n/**\n * @deprecated Replaced with {@link combineLatestWith}. Will be removed in v8.\n */\nexport function combineLatest<T, R>(...args: (ObservableInput<any> | ((...values: any[]) => R))[]): OperatorFunction<T, unknown> {\n  const resultSelector = popResultSelector(args);\n  return resultSelector\n    ? pipe(combineLatest(...(args as Array<ObservableInput<any>>)), mapOneOrManyArgs(resultSelector))\n    : operate((source, subscriber) => {\n        combineLatestInit([source, ...argsOrArgArray(args)])(subscriber);\n      });\n}\n", "import { ObservableInputTuple, OperatorFunction, Cons } from '../types';\nimport { combineLatest } from './combineLatest';\n\n/**\n * Create an observable that combines the latest values from all passed observables and the source\n * into arrays and emits them.\n *\n * Returns an observable, that when subscribed to, will subscribe to the source observable and all\n * sources provided as arguments. Once all sources emit at least one value, all of the latest values\n * will be emitted as an array. After that, every time any source emits a value, all of the latest values\n * will be emitted as an array.\n *\n * This is a useful operator for eagerly calculating values based off of changed inputs.\n *\n * ## Example\n *\n * Simple concatenation of values from two inputs\n *\n * ```ts\n * import { fromEvent, combineLatestWith, map } from 'rxjs';\n *\n * // Setup: Add two inputs to the page\n * const input1 = document.createElement('input');\n * document.body.appendChild(input1);\n * const input2 = document.createElement('input');\n * document.body.appendChild(input2);\n *\n * // Get streams of changes\n * const input1Changes$ = fromEvent(input1, 'change');\n * const input2Changes$ = fromEvent(input2, 'change');\n *\n * // Combine the changes by adding them together\n * input1Changes$.pipe(\n *   combineLatestWith(input2Changes$),\n *   map(([e1, e2]) => (<HTMLInputElement>e1.target).value + ' - ' + (<HTMLInputElement>e2.target).value)\n * )\n * .subscribe(x => console.log(x));\n * ```\n *\n * @param otherSources the other sources to subscribe to.\n * @return A function that returns an Observable that emits the latest\n * emissions from both source and provided Observables.\n */\nexport function combineLatestWith<T, A extends readonly unknown[]>(\n  ...otherSources: [...ObservableInputTuple<A>]\n): OperatorFunction<T, Cons<T, A>> {\n  return combineLatest(...otherSources);\n}\n", "import { asyncScheduler } from '../scheduler/async';\nimport { Subscription } from '../Subscription';\nimport { MonoTypeOperatorFunction, SchedulerAction, SchedulerLike } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * Emits a notification from the source Observable only after a particular time span\n * has passed without another source emission.\n *\n * <span class=\"informal\">It's like {@link delay}, but passes only the most\n * recent notification from each burst of emissions.</span>\n *\n * ![](debounceTime.png)\n *\n * `debounceTime` delays notifications emitted by the source Observable, but drops\n * previous pending delayed emissions if a new notification arrives on the source\n * Observable. This operator keeps track of the most recent notification from the\n * source Observable, and emits that only when `dueTime` has passed\n * without any other notification appearing on the source Observable. If a new value\n * appears before `dueTime` silence occurs, the previous notification will be dropped\n * and will not be emitted and a new `dueTime` is scheduled.\n * If the completing event happens during `dueTime` the last cached notification\n * is emitted before the completion event is forwarded to the output observable.\n * If the error event happens during `dueTime` or after it only the error event is\n * forwarded to the output observable. The cache notification is not emitted in this case.\n *\n * This is a rate-limiting operator, because it is impossible for more than one\n * notification to be emitted in any time window of duration `dueTime`, but it is also\n * a delay-like operator since output emissions do not occur at the same time as\n * they did on the source Observable. Optionally takes a {@link SchedulerLike} for\n * managing timers.\n *\n * ## Example\n *\n * Emit the most recent click after a burst of clicks\n *\n * ```ts\n * import { fromEvent, debounceTime } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(debounceTime(1000));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link audit}\n * @see {@link auditTime}\n * @see {@link debounce}\n * @see {@link sample}\n * @see {@link sampleTime}\n * @see {@link throttle}\n * @see {@link throttleTime}\n *\n * @param {number} dueTime The timeout duration in milliseconds (or the time\n * unit determined internally by the optional `scheduler`) for the window of\n * time required to wait for emission silence before emitting the most recent\n * source value.\n * @param {SchedulerLike} [scheduler=async] The {@link SchedulerLike} to use for\n * managing the timers that handle the timeout for each value.\n * @return A function that returns an Observable that delays the emissions of\n * the source Observable by the specified `dueTime`, and may drop some values\n * if they occur too frequently.\n */\nexport function debounceTime<T>(dueTime: number, scheduler: SchedulerLike = asyncScheduler): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    let activeTask: Subscription | null = null;\n    let lastValue: T | null = null;\n    let lastTime: number | null = null;\n\n    const emit = () => {\n      if (activeTask) {\n        // We have a value! Free up memory first, then emit the value.\n        activeTask.unsubscribe();\n        activeTask = null;\n        const value = lastValue!;\n        lastValue = null;\n        subscriber.next(value);\n      }\n    };\n    function emitWhenIdle(this: SchedulerAction<unknown>) {\n      // This is called `dueTime` after the first value\n      // but we might have received new values during this window!\n\n      const targetTime = lastTime! + dueTime;\n      const now = scheduler.now();\n      if (now < targetTime) {\n        // On that case, re-schedule to the new target\n        activeTask = this.schedule(undefined, targetTime - now);\n        subscriber.add(activeTask);\n        return;\n      }\n\n      emit();\n    }\n\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value: T) => {\n          lastValue = value;\n          lastTime = scheduler.now();\n\n          // Only set up a task if it's not already up\n          if (!activeTask) {\n            activeTask = scheduler.schedule(emitWhenIdle, dueTime);\n            subscriber.add(activeTask);\n          }\n        },\n        () => {\n          // Source completed.\n          // Emit any pending debounced values then complete\n          emit();\n          subscriber.complete();\n        },\n        // Pass all errors through to consumer.\n        undefined,\n        () => {\n          // Finalization.\n          lastValue = activeTask = null;\n        }\n      )\n    );\n  });\n}\n", "import { OperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * Emits a given value if the source Observable completes without emitting any\n * `next` value, otherwise mirrors the source Observable.\n *\n * <span class=\"informal\">If the source Observable turns out to be empty, then\n * this operator will emit a default value.</span>\n *\n * ![](defaultIfEmpty.png)\n *\n * `defaultIfEmpty` emits the values emitted by the source Observable or a\n * specified default value if the source Observable is empty (completes without\n * having emitted any `next` value).\n *\n * ## Example\n *\n * If no clicks happen in 5 seconds, then emit 'no clicks'\n *\n * ```ts\n * import { fromEvent, takeUntil, interval, defaultIfEmpty } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const clicksBeforeFive = clicks.pipe(takeUntil(interval(5000)));\n * const result = clicksBeforeFive.pipe(defaultIfEmpty('no clicks'));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link empty}\n * @see {@link last}\n *\n * @param defaultValue The default value used if the source\n * Observable is empty.\n * @return A function that returns an Observable that emits either the\n * specified `defaultValue` if the source Observable emits no items, or the\n * values emitted by the source Observable.\n */\nexport function defaultIfEmpty<T, R>(defaultValue: R): OperatorFunction<T, T | R> {\n  return operate((source, subscriber) => {\n    let hasValue = false;\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value) => {\n          hasValue = true;\n          subscriber.next(value);\n        },\n        () => {\n          if (!hasValue) {\n            subscriber.next(defaultValue!);\n          }\n          subscriber.complete();\n        }\n      )\n    );\n  });\n}\n", "import { MonoTypeOperatorFunction } from '../types';\nimport { EMPTY } from '../observable/empty';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * Emits only the first `count` values emitted by the source Observable.\n *\n * <span class=\"informal\">Takes the first `count` values from the source, then\n * completes.</span>\n *\n * ![](take.png)\n *\n * `take` returns an Observable that emits only the first `count` values emitted\n * by the source Observable. If the source emits fewer than `count` values then\n * all of its values are emitted. After that, it completes, regardless if the\n * source completes.\n *\n * ## Example\n *\n * Take the first 5 seconds of an infinite 1-second interval Observable\n *\n * ```ts\n * import { interval, take } from 'rxjs';\n *\n * const intervalCount = interval(1000);\n * const takeFive = intervalCount.pipe(take(5));\n * takeFive.subscribe(x => console.log(x));\n *\n * // Logs:\n * // 0\n * // 1\n * // 2\n * // 3\n * // 4\n * ```\n *\n * @see {@link takeLast}\n * @see {@link takeUntil}\n * @see {@link takeWhile}\n * @see {@link skip}\n *\n * @param count The maximum number of `next` values to emit.\n * @return A function that returns an Observable that emits only the first\n * `count` values emitted by the source Observable, or all of the values from\n * the source if the source emits fewer than `count` values.\n */\nexport function take<T>(count: number): MonoTypeOperatorFunction<T> {\n  return count <= 0\n    ? // If we are taking no values, that's empty.\n      () => EMPTY\n    : operate((source, subscriber) => {\n        let seen = 0;\n        source.subscribe(\n          createOperatorSubscriber(subscriber, (value) => {\n            // Increment the number of values we have seen,\n            // then check it against the allowed count to see\n            // if we are still letting values through.\n            if (++seen <= count) {\n              subscriber.next(value);\n              // If we have met or passed our allowed count,\n              // we need to complete. We have to do <= here,\n              // because re-entrant code will increment `seen` twice.\n              if (count <= seen) {\n                subscriber.complete();\n              }\n            }\n          })\n        );\n      });\n}\n", "import { OperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { noop } from '../util/noop';\n\n/**\n * Ignores all items emitted by the source Observable and only passes calls of `complete` or `error`.\n *\n * ![](ignoreElements.png)\n *\n * The `ignoreElements` operator suppresses all items emitted by the source Observable,\n * but allows its termination notification (either `error` or `complete`) to pass through unchanged.\n *\n * If you do not care about the items being emitted by an Observable, but you do want to be notified\n * when it completes or when it terminates with an error, you can apply the `ignoreElements` operator\n * to the Observable, which will ensure that it will never call its observers\u2019 `next` handlers.\n *\n * ## Example\n *\n * Ignore all `next` emissions from the source\n *\n * ```ts\n * import { of, ignoreElements } from 'rxjs';\n *\n * of('you', 'talking', 'to', 'me')\n *   .pipe(ignoreElements())\n *   .subscribe({\n *     next: word => console.log(word),\n *     error: err => console.log('error:', err),\n *     complete: () => console.log('the end'),\n *   });\n *\n * // result:\n * // 'the end'\n * ```\n *\n * @return A function that returns an empty Observable that only calls\n * `complete` or `error`, based on which one is called by the source\n * Observable.\n */\nexport function ignoreElements(): OperatorFunction<unknown, never> {\n  return operate((source, subscriber) => {\n    source.subscribe(createOperatorSubscriber(subscriber, noop));\n  });\n}\n", "import { OperatorFunction } from '../types';\nimport { map } from './map';\n\n/** @deprecated To be removed in v9. Use {@link map} instead: `map(() => value)`. */\nexport function mapTo<R>(value: R): OperatorFunction<unknown, R>;\n/**\n * @deprecated Do not specify explicit type parameters. Signatures with type parameters\n * that cannot be inferred will be removed in v8. `mapTo` itself will be removed in v9,\n * use {@link map} instead: `map(() => value)`.\n * */\nexport function mapTo<T, R>(value: R): OperatorFunction<T, R>;\n\n/**\n * Emits the given constant value on the output Observable every time the source\n * Observable emits a value.\n *\n * <span class=\"informal\">Like {@link map}, but it maps every source value to\n * the same output value every time.</span>\n *\n * ![](mapTo.png)\n *\n * Takes a constant `value` as argument, and emits that whenever the source\n * Observable emits a value. In other words, ignores the actual source value,\n * and simply uses the emission moment to know when to emit the given `value`.\n *\n * ## Example\n *\n * Map every click to the string `'Hi'`\n *\n * ```ts\n * import { fromEvent, mapTo } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const greetings = clicks.pipe(mapTo('Hi'));\n *\n * greetings.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link map}\n *\n * @param value The value to map each source value to.\n * @return A function that returns an Observable that emits the given `value`\n * every time the source Observable emits.\n * @deprecated To be removed in v9. Use {@link map} instead: `map(() => value)`.\n */\nexport function mapTo<R>(value: R): OperatorFunction<unknown, R> {\n  return map(() => value);\n}\n", "import { Observable } from '../Observable';\nimport { MonoTypeOperatorFunction, ObservableInput } from '../types';\nimport { concat } from '../observable/concat';\nimport { take } from './take';\nimport { ignoreElements } from './ignoreElements';\nimport { mapTo } from './mapTo';\nimport { mergeMap } from './mergeMap';\nimport { innerFrom } from '../observable/innerFrom';\n\n/** @deprecated The `subscriptionDelay` parameter will be removed in v8. */\nexport function delayWhen<T>(\n  delayDurationSelector: (value: T, index: number) => ObservableInput<any>,\n  subscriptionDelay: Observable<any>\n): MonoTypeOperatorFunction<T>;\nexport function delayWhen<T>(delayDurationSelector: (value: T, index: number) => ObservableInput<any>): MonoTypeOperatorFunction<T>;\n\n/**\n * Delays the emission of items from the source Observable by a given time span\n * determined by the emissions of another Observable.\n *\n * <span class=\"informal\">It's like {@link delay}, but the time span of the\n * delay duration is determined by a second Observable.</span>\n *\n * ![](delayWhen.png)\n *\n * `delayWhen` operator shifts each emitted value from the source Observable by\n * a time span determined by another Observable. When the source emits a value,\n * the `delayDurationSelector` function is called with the value emitted from\n * the source Observable as the first argument to the `delayDurationSelector`.\n * The `delayDurationSelector` function should return an {@link ObservableInput},\n * that is internally converted to an Observable that is called the \"duration\"\n * Observable.\n *\n * The source value is emitted on the output Observable only when the \"duration\"\n * Observable emits ({@link guide/glossary-and-semantics#next next}s) any value.\n * Upon that, the \"duration\" Observable gets unsubscribed.\n *\n * Before RxJS V7, the {@link guide/glossary-and-semantics#complete completion}\n * of the \"duration\" Observable would have been triggering the emission of the\n * source value to the output Observable, but with RxJS V7, this is not the case\n * anymore.\n *\n * Only next notifications (from the \"duration\" Observable) trigger values from\n * the source Observable to be passed to the output Observable. If the \"duration\"\n * Observable only emits the complete notification (without next), the value\n * emitted by the source Observable will never get to the output Observable - it\n * will be swallowed. If the \"duration\" Observable errors, the error will be\n * propagated to the output Observable.\n *\n * Optionally, `delayWhen` takes a second argument, `subscriptionDelay`, which\n * is an Observable. When `subscriptionDelay` emits its first value or\n * completes, the source Observable is subscribed to and starts behaving like\n * described in the previous paragraph. If `subscriptionDelay` is not provided,\n * `delayWhen` will subscribe to the source Observable as soon as the output\n * Observable is subscribed.\n *\n * ## Example\n *\n * Delay each click by a random amount of time, between 0 and 5 seconds\n *\n * ```ts\n * import { fromEvent, delayWhen, interval } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const delayedClicks = clicks.pipe(\n *   delayWhen(() => interval(Math.random() * 5000))\n * );\n * delayedClicks.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link delay}\n * @see {@link throttle}\n * @see {@link throttleTime}\n * @see {@link debounce}\n * @see {@link debounceTime}\n * @see {@link sample}\n * @see {@link sampleTime}\n * @see {@link audit}\n * @see {@link auditTime}\n *\n * @param delayDurationSelector A function that returns an `ObservableInput` for\n * each `value` emitted by the source Observable, which is then used to delay the\n * emission of that `value` on the output Observable until the `ObservableInput`\n * returned from this function emits a next value. When called, beside `value`,\n * this function receives a zero-based `index` of the emission order.\n * @param subscriptionDelay An Observable that triggers the subscription to the\n * source Observable once it emits any value.\n * @return A function that returns an Observable that delays the emissions of\n * the source Observable by an amount of time specified by the Observable\n * returned by `delayDurationSelector`.\n */\nexport function delayWhen<T>(\n  delayDurationSelector: (value: T, index: number) => ObservableInput<any>,\n  subscriptionDelay?: Observable<any>\n): MonoTypeOperatorFunction<T> {\n  if (subscriptionDelay) {\n    // DEPRECATED PATH\n    return (source: Observable<T>) =>\n      concat(subscriptionDelay.pipe(take(1), ignoreElements()), source.pipe(delayWhen(delayDurationSelector)));\n  }\n\n  return mergeMap((value, index) => innerFrom(delayDurationSelector(value, index)).pipe(take(1), mapTo(value)));\n}\n", "import { asyncScheduler } from '../scheduler/async';\nimport { MonoTypeOperatorFunction, SchedulerLike } from '../types';\nimport { delayWhen } from './delayWhen';\nimport { timer } from '../observable/timer';\n\n/**\n * Delays the emission of items from the source Observable by a given timeout or\n * until a given Date.\n *\n * <span class=\"informal\">Time shifts each item by some specified amount of\n * milliseconds.</span>\n *\n * ![](delay.svg)\n *\n * If the delay argument is a Number, this operator time shifts the source\n * Observable by that amount of time expressed in milliseconds. The relative\n * time intervals between the values are preserved.\n *\n * If the delay argument is a Date, this operator time shifts the start of the\n * Observable execution until the given date occurs.\n *\n * ## Examples\n *\n * Delay each click by one second\n *\n * ```ts\n * import { fromEvent, delay } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const delayedClicks = clicks.pipe(delay(1000)); // each click emitted after 1 second\n * delayedClicks.subscribe(x => console.log(x));\n * ```\n *\n * Delay all clicks until a future date happens\n *\n * ```ts\n * import { fromEvent, delay } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const date = new Date('March 15, 2050 12:00:00'); // in the future\n * const delayedClicks = clicks.pipe(delay(date)); // click emitted only after that date\n * delayedClicks.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link delayWhen}\n * @see {@link throttle}\n * @see {@link throttleTime}\n * @see {@link debounce}\n * @see {@link debounceTime}\n * @see {@link sample}\n * @see {@link sampleTime}\n * @see {@link audit}\n * @see {@link auditTime}\n *\n * @param {number|Date} due The delay duration in milliseconds (a `number`) or\n * a `Date` until which the emission of the source items is delayed.\n * @param {SchedulerLike} [scheduler=async] The {@link SchedulerLike} to use for\n * managing the timers that handle the time-shift for each item.\n * @return A function that returns an Observable that delays the emissions of\n * the source Observable by the specified timeout or Date.\n */\nexport function delay<T>(due: number | Date, scheduler: SchedulerLike = asyncScheduler): MonoTypeOperatorFunction<T> {\n  const duration = timer(due, scheduler);\n  return delayWhen(() => duration);\n}\n", "import { MonoTypeOperatorFunction } from '../types';\nimport { identity } from '../util/identity';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\nexport function distinctUntilChanged<T>(comparator?: (previous: T, current: T) => boolean): MonoTypeOperatorFunction<T>;\nexport function distinctUntilChanged<T, K>(\n  comparator: (previous: K, current: K) => boolean,\n  keySelector: (value: T) => K\n): MonoTypeOperatorFunction<T>;\n\n/**\n * Returns a result {@link Observable} that emits all values pushed by the source observable if they\n * are distinct in comparison to the last value the result observable emitted.\n *\n * When provided without parameters or with the first parameter (`{@link distinctUntilChanged#comparator comparator}`),\n * it behaves like this:\n *\n * 1. It will always emit the first value from the source.\n * 2. For all subsequent values pushed by the source, they will be compared to the previously emitted values\n *    using the provided `comparator` or an `===` equality check.\n * 3. If the value pushed by the source is determined to be unequal by this check, that value is emitted and\n *    becomes the new \"previously emitted value\" internally.\n *\n * When the second parameter (`{@link distinctUntilChanged#keySelector keySelector}`) is provided, the behavior\n * changes:\n *\n * 1. It will always emit the first value from the source.\n * 2. The `keySelector` will be run against all values, including the first value.\n * 3. For all values after the first, the selected key will be compared against the key selected from\n *    the previously emitted value using the `comparator`.\n * 4. If the keys are determined to be unequal by this check, the value (not the key), is emitted\n *    and the selected key from that value is saved for future comparisons against other keys.\n *\n * ## Examples\n *\n * A very basic example with no `{@link distinctUntilChanged#comparator comparator}`. Note that `1` is emitted more than once,\n * because it's distinct in comparison to the _previously emitted_ value,\n * not in comparison to _all other emitted values_.\n *\n * ```ts\n * import { of, distinctUntilChanged } from 'rxjs';\n *\n * of(1, 1, 1, 2, 2, 2, 1, 1, 3, 3)\n *   .pipe(distinctUntilChanged())\n *   .subscribe(console.log);\n * // Logs: 1, 2, 1, 3\n * ```\n *\n * With a `{@link distinctUntilChanged#comparator comparator}`, you can do custom comparisons. Let's say\n * you only want to emit a value when all of its components have\n * changed:\n *\n * ```ts\n * import { of, distinctUntilChanged } from 'rxjs';\n *\n * const totallyDifferentBuilds$ = of(\n *   { engineVersion: '1.1.0', transmissionVersion: '1.2.0' },\n *   { engineVersion: '1.1.0', transmissionVersion: '1.4.0' },\n *   { engineVersion: '1.3.0', transmissionVersion: '1.4.0' },\n *   { engineVersion: '1.3.0', transmissionVersion: '1.5.0' },\n *   { engineVersion: '2.0.0', transmissionVersion: '1.5.0' }\n * ).pipe(\n *   distinctUntilChanged((prev, curr) => {\n *     return (\n *       prev.engineVersion === curr.engineVersion ||\n *       prev.transmissionVersion === curr.transmissionVersion\n *     );\n *   })\n * );\n *\n * totallyDifferentBuilds$.subscribe(console.log);\n *\n * // Logs:\n * // { engineVersion: '1.1.0', transmissionVersion: '1.2.0' }\n * // { engineVersion: '1.3.0', transmissionVersion: '1.4.0' }\n * // { engineVersion: '2.0.0', transmissionVersion: '1.5.0' }\n * ```\n *\n * You can also provide a custom `{@link distinctUntilChanged#comparator comparator}` to check that emitted\n * changes are only in one direction. Let's say you only want to get\n * the next record temperature:\n *\n * ```ts\n * import { of, distinctUntilChanged } from 'rxjs';\n *\n * const temps$ = of(30, 31, 20, 34, 33, 29, 35, 20);\n *\n * const recordHighs$ = temps$.pipe(\n *   distinctUntilChanged((prevHigh, temp) => {\n *     // If the current temp is less than\n *     // or the same as the previous record,\n *     // the record hasn't changed.\n *     return temp <= prevHigh;\n *   })\n * );\n *\n * recordHighs$.subscribe(console.log);\n * // Logs: 30, 31, 34, 35\n * ```\n *\n * Selecting update events only when the `updatedBy` field shows\n * the account changed hands.\n *\n * ```ts\n * import { of, distinctUntilChanged } from 'rxjs';\n *\n * // A stream of updates to a given account\n * const accountUpdates$ = of(\n *   { updatedBy: 'blesh', data: [] },\n *   { updatedBy: 'blesh', data: [] },\n *   { updatedBy: 'ncjamieson', data: [] },\n *   { updatedBy: 'ncjamieson', data: [] },\n *   { updatedBy: 'blesh', data: [] }\n * );\n *\n * // We only want the events where it changed hands\n * const changedHands$ = accountUpdates$.pipe(\n *   distinctUntilChanged(undefined, update => update.updatedBy)\n * );\n *\n * changedHands$.subscribe(console.log);\n * // Logs:\n * // { updatedBy: 'blesh', data: Array[0] }\n * // { updatedBy: 'ncjamieson', data: Array[0] }\n * // { updatedBy: 'blesh', data: Array[0] }\n * ```\n *\n * @see {@link distinct}\n * @see {@link distinctUntilKeyChanged}\n *\n * @param comparator A function used to compare the previous and current keys for\n * equality. Defaults to a `===` check.\n * @param keySelector Used to select a key value to be passed to the `comparator`.\n *\n * @return A function that returns an Observable that emits items from the\n * source Observable with distinct values.\n */\nexport function distinctUntilChanged<T, K>(\n  comparator?: (previous: K, current: K) => boolean,\n  keySelector: (value: T) => K = identity as (value: T) => K\n): MonoTypeOperatorFunction<T> {\n  // We've been allowing `null` do be passed as the `compare`, so we can't do\n  // a default value for the parameter, because that will only work\n  // for `undefined`.\n  comparator = comparator ?? defaultCompare;\n\n  return operate((source, subscriber) => {\n    // The previous key, used to compare against keys selected\n    // from new arrivals to determine \"distinctiveness\".\n    let previousKey: K;\n    // Whether or not this is the first value we've gotten.\n    let first = true;\n\n    source.subscribe(\n      createOperatorSubscriber(subscriber, (value) => {\n        // We always call the key selector.\n        const currentKey = keySelector(value);\n\n        // If it's the first value, we always emit it.\n        // Otherwise, we compare this key to the previous key, and\n        // if the comparer returns false, we emit.\n        if (first || !comparator!(previousKey, currentKey)) {\n          // Update our state *before* we emit the value\n          // as emission can be the source of re-entrant code\n          // in functional libraries like this. We only really\n          // need to do this if it's the first value, or if the\n          // key we're tracking in previous needs to change.\n          first = false;\n          previousKey = currentKey;\n\n          // Emit the value!\n          subscriber.next(value);\n        }\n      })\n    );\n  });\n}\n\nfunction defaultCompare(a: any, b: any) {\n  return a === b;\n}\n", "import { distinctUntilChanged } from './distinctUntilChanged';\nimport { MonoTypeOperatorFunction } from '../types';\n\n/* tslint:disable:max-line-length */\nexport function distinctUntilKeyChanged<T>(key: keyof T): MonoTypeOperatorFunction<T>;\nexport function distinctUntilKeyChanged<T, K extends keyof T>(key: K, compare: (x: T[K], y: T[K]) => boolean): MonoTypeOperatorFunction<T>;\n/* tslint:enable:max-line-length */\n\n/**\n * Returns an Observable that emits all items emitted by the source Observable that are distinct by comparison from the previous item,\n * using a property accessed by using the key provided to check if the two items are distinct.\n *\n * If a comparator function is provided, then it will be called for each item to test for whether or not that value should be emitted.\n *\n * If a comparator function is not provided, an equality check is used by default.\n *\n * ## Examples\n *\n * An example comparing the name of persons\n *\n * ```ts\n * import { of, distinctUntilKeyChanged } from 'rxjs';\n *\n * of(\n *   { age: 4, name: 'Foo' },\n *   { age: 7, name: 'Bar' },\n *   { age: 5, name: 'Foo' },\n *   { age: 6, name: 'Foo' }\n * ).pipe(\n *   distinctUntilKeyChanged('name')\n * )\n * .subscribe(x => console.log(x));\n *\n * // displays:\n * // { age: 4, name: 'Foo' }\n * // { age: 7, name: 'Bar' }\n * // { age: 5, name: 'Foo' }\n * ```\n *\n * An example comparing the first letters of the name\n *\n * ```ts\n * import { of, distinctUntilKeyChanged } from 'rxjs';\n *\n * of(\n *   { age: 4, name: 'Foo1' },\n *   { age: 7, name: 'Bar' },\n *   { age: 5, name: 'Foo2' },\n *   { age: 6, name: 'Foo3' }\n * ).pipe(\n *   distinctUntilKeyChanged('name', (x, y) => x.substring(0, 3) === y.substring(0, 3))\n * )\n * .subscribe(x => console.log(x));\n *\n * // displays:\n * // { age: 4, name: 'Foo1' }\n * // { age: 7, name: 'Bar' }\n * // { age: 5, name: 'Foo2' }\n * ```\n *\n * @see {@link distinct}\n * @see {@link distinctUntilChanged}\n *\n * @param {string} key String key for object property lookup on each item.\n * @param {function} [compare] Optional comparison function called to test if an item is distinct from the previous item in the source.\n * @return A function that returns an Observable that emits items from the\n * source Observable with distinct values based on the key specified.\n */\nexport function distinctUntilKeyChanged<T, K extends keyof T>(key: K, compare?: (x: T[K], y: T[K]) => boolean): MonoTypeOperatorFunction<T> {\n  return distinctUntilChanged((x: T, y: T) => compare ? compare(x[key], y[key]) : x[key] === y[key]);\n}\n", "import { EmptyError } from '../util/EmptyError';\nimport { MonoTypeOperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * If the source observable completes without emitting a value, it will emit\n * an error. The error will be created at that time by the optional\n * `errorFactory` argument, otherwise, the error will be {@link EmptyError}.\n *\n * ![](throwIfEmpty.png)\n *\n * ## Example\n *\n * Throw an error if the document wasn't clicked within 1 second\n *\n * ```ts\n * import { fromEvent, takeUntil, timer, throwIfEmpty } from 'rxjs';\n *\n * const click$ = fromEvent(document, 'click');\n *\n * click$.pipe(\n *   takeUntil(timer(1000)),\n *   throwIfEmpty(() => new Error('The document was not clicked within 1 second'))\n * )\n * .subscribe({\n *   next() {\n *    console.log('The document was clicked');\n *   },\n *   error(err) {\n *     console.error(err.message);\n *   }\n * });\n * ```\n *\n * @param errorFactory A factory function called to produce the\n * error to be thrown when the source observable completes without emitting a\n * value.\n * @return A function that returns an Observable that throws an error if the\n * source Observable completed without emitting.\n */\nexport function throwIfEmpty<T>(errorFactory: () => any = defaultErrorFactory): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    let hasValue = false;\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value) => {\n          hasValue = true;\n          subscriber.next(value);\n        },\n        () => (hasValue ? subscriber.complete() : subscriber.error(errorFactory()))\n      )\n    );\n  });\n}\n\nfunction defaultErrorFactory() {\n  return new EmptyError();\n}\n", "/** prettier */\nimport { Observable } from '../Observable';\nimport { concat } from '../observable/concat';\nimport { of } from '../observable/of';\nimport { MonoTypeOperatorFunction, SchedulerLike, OperatorFunction, ValueFromArray } from '../types';\n\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `concatAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function endWith<T>(scheduler: SchedulerLike): MonoTypeOperatorFunction<T>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `concatAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function endWith<T, A extends unknown[] = T[]>(\n  ...valuesAndScheduler: [...A, SchedulerLike]\n): OperatorFunction<T, T | ValueFromArray<A>>;\n\nexport function endWith<T, A extends unknown[] = T[]>(...values: A): OperatorFunction<T, T | ValueFromArray<A>>;\n\n/**\n * Returns an observable that will emit all values from the source, then synchronously emit\n * the provided value(s) immediately after the source completes.\n *\n * NOTE: Passing a last argument of a Scheduler is _deprecated_, and may result in incorrect\n * types in TypeScript.\n *\n * This is useful for knowing when an observable ends. Particularly when paired with an\n * operator like {@link takeUntil}\n *\n * ![](endWith.png)\n *\n * ## Example\n *\n * Emit values to know when an interval starts and stops. The interval will\n * stop when a user clicks anywhere on the document.\n *\n * ```ts\n * import { interval, map, fromEvent, startWith, takeUntil, endWith } from 'rxjs';\n *\n * const ticker$ = interval(5000).pipe(\n *   map(() => 'tick')\n * );\n *\n * const documentClicks$ = fromEvent(document, 'click');\n *\n * ticker$.pipe(\n *   startWith('interval started'),\n *   takeUntil(documentClicks$),\n *   endWith('interval ended by click')\n * )\n * .subscribe(x => console.log(x));\n *\n * // Result (assuming a user clicks after 15 seconds)\n * // 'interval started'\n * // 'tick'\n * // 'tick'\n * // 'tick'\n * // 'interval ended by click'\n * ```\n *\n * @see {@link startWith}\n * @see {@link concat}\n * @see {@link takeUntil}\n *\n * @param values Items you want the modified Observable to emit last.\n * @return A function that returns an Observable that emits all values from the\n * source, then synchronously emits the provided value(s) immediately after the\n * source completes.\n */\nexport function endWith<T>(...values: Array<T | SchedulerLike>): MonoTypeOperatorFunction<T> {\n  return (source: Observable<T>) => concat(source, of(...values)) as Observable<T>;\n}\n", "import { MonoTypeOperatorFunction } from '../types';\nimport { operate } from '../util/lift';\n\n/**\n * Returns an Observable that mirrors the source Observable, but will call a specified function when\n * the source terminates on complete or error.\n * The specified function will also be called when the subscriber explicitly unsubscribes.\n *\n * ## Examples\n *\n * Execute callback function when the observable completes\n *\n * ```ts\n * import { interval, take, finalize } from 'rxjs';\n *\n * // emit value in sequence every 1 second\n * const source = interval(1000);\n * const example = source.pipe(\n *   take(5), //take only the first 5 values\n *   finalize(() => console.log('Sequence complete')) // Execute when the observable completes\n * );\n * const subscribe = example.subscribe(val => console.log(val));\n *\n * // results:\n * // 0\n * // 1\n * // 2\n * // 3\n * // 4\n * // 'Sequence complete'\n * ```\n *\n * Execute callback function when the subscriber explicitly unsubscribes\n *\n * ```ts\n * import { interval, finalize, tap, noop, timer } from 'rxjs';\n *\n * const source = interval(100).pipe(\n *   finalize(() => console.log('[finalize] Called')),\n *   tap({\n *     next: () => console.log('[next] Called'),\n *     error: () => console.log('[error] Not called'),\n *     complete: () => console.log('[tap complete] Not called')\n *   })\n * );\n *\n * const sub = source.subscribe({\n *   next: x => console.log(x),\n *   error: noop,\n *   complete: () => console.log('[complete] Not called')\n * });\n *\n * timer(150).subscribe(() => sub.unsubscribe());\n *\n * // results:\n * // '[next] Called'\n * // 0\n * // '[finalize] Called'\n * ```\n *\n * @param {function} callback Function to be called when source terminates.\n * @return A function that returns an Observable that mirrors the source, but\n * will call the specified function on termination.\n */\nexport function finalize<T>(callback: () => void): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    // TODO: This try/finally was only added for `useDeprecatedSynchronousErrorHandling`.\n    // REMOVE THIS WHEN THAT HOT GARBAGE IS REMOVED IN V8.\n    try {\n      source.subscribe(subscriber);\n    } finally {\n      subscriber.add(callback);\n    }\n  });\n}\n", "import { Observable } from '../Observable';\nimport { EmptyError } from '../util/EmptyError';\nimport { OperatorFunction, TruthyTypesOf } from '../types';\nimport { filter } from './filter';\nimport { take } from './take';\nimport { defaultIfEmpty } from './defaultIfEmpty';\nimport { throwIfEmpty } from './throwIfEmpty';\nimport { identity } from '../util/identity';\n\nexport function first<T, D = T>(predicate?: null, defaultValue?: D): OperatorFunction<T, T | D>;\nexport function first<T>(predicate: BooleanConstructor): OperatorFunction<T, TruthyTypesOf<T>>;\nexport function first<T, D>(predicate: BooleanConstructor, defaultValue: D): OperatorFunction<T, TruthyTypesOf<T> | D>;\nexport function first<T, S extends T>(\n  predicate: (value: T, index: number, source: Observable<T>) => value is S,\n  defaultValue?: S\n): OperatorFunction<T, S>;\nexport function first<T, S extends T, D>(\n  predicate: (value: T, index: number, source: Observable<T>) => value is S,\n  defaultValue: D\n): OperatorFunction<T, S | D>;\nexport function first<T, D = T>(\n  predicate: (value: T, index: number, source: Observable<T>) => boolean,\n  defaultValue?: D\n): OperatorFunction<T, T | D>;\n\n/**\n * Emits only the first value (or the first value that meets some condition)\n * emitted by the source Observable.\n *\n * <span class=\"informal\">Emits only the first value. Or emits only the first\n * value that passes some test.</span>\n *\n * ![](first.png)\n *\n * If called with no arguments, `first` emits the first value of the source\n * Observable, then completes. If called with a `predicate` function, `first`\n * emits the first value of the source that matches the specified condition. Throws an error if\n * `defaultValue` was not provided and a matching element is not found.\n *\n * ## Examples\n *\n * Emit only the first click that happens on the DOM\n *\n * ```ts\n * import { fromEvent, first } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(first());\n * result.subscribe(x => console.log(x));\n * ```\n *\n * Emits the first click that happens on a DIV\n *\n * ```ts\n * import { fromEvent, first } from 'rxjs';\n *\n * const div = document.createElement('div');\n * div.style.cssText = 'width: 200px; height: 200px; background: #09c;';\n * document.body.appendChild(div);\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(first(ev => (<HTMLElement>ev.target).tagName === 'DIV'));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link filter}\n * @see {@link find}\n * @see {@link take}\n *\n * @throws {EmptyError} Delivers an EmptyError to the Observer's `error`\n * callback if the Observable completes before any `next` notification was sent.\n * This is how `first()` is different from {@link take}(1) which completes instead.\n *\n * @param {function(value: T, index: number, source: Observable<T>): boolean} [predicate]\n * An optional function called with each item to test for condition matching.\n * @param {D} [defaultValue] The default value emitted in case no valid value\n * was found on the source.\n * @return A function that returns an Observable that emits the first item that\n * matches the condition.\n */\nexport function first<T, D>(\n  predicate?: ((value: T, index: number, source: Observable<T>) => boolean) | null,\n  defaultValue?: D\n): OperatorFunction<T, T | D> {\n  const hasDefaultValue = arguments.length >= 2;\n  return (source: Observable<T>) =>\n    source.pipe(\n      predicate ? filter((v, i) => predicate(v, i, source)) : identity,\n      take(1),\n      hasDefaultValue ? defaultIfEmpty(defaultValue!) : throwIfEmpty(() => new EmptyError())\n    );\n}\n", "import { EMPTY } from '../observable/empty';\nimport { MonoTypeOperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * Waits for the source to complete, then emits the last N values from the source,\n * as specified by the `count` argument.\n *\n * ![](takeLast.png)\n *\n * `takeLast` results in an observable that will hold values up to `count` values in memory,\n * until the source completes. It then pushes all values in memory to the consumer, in the\n * order they were received from the source, then notifies the consumer that it is\n * complete.\n *\n * If for some reason the source completes before the `count` supplied to `takeLast` is reached,\n * all values received until that point are emitted, and then completion is notified.\n *\n * **Warning**: Using `takeLast` with an observable that never completes will result\n * in an observable that never emits a value.\n *\n * ## Example\n *\n * Take the last 3 values of an Observable with many values\n *\n * ```ts\n * import { range, takeLast } from 'rxjs';\n *\n * const many = range(1, 100);\n * const lastThree = many.pipe(takeLast(3));\n * lastThree.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link take}\n * @see {@link takeUntil}\n * @see {@link takeWhile}\n * @see {@link skip}\n *\n * @param count The maximum number of values to emit from the end of\n * the sequence of values emitted by the source Observable.\n * @return A function that returns an Observable that emits at most the last\n * `count` values emitted by the source Observable.\n */\nexport function takeLast<T>(count: number): MonoTypeOperatorFunction<T> {\n  return count <= 0\n    ? () => EMPTY\n    : operate((source, subscriber) => {\n        // This buffer will hold the values we are going to emit\n        // when the source completes. Since we only want to take the\n        // last N values, we can't emit until we're sure we're not getting\n        // any more values.\n        let buffer: T[] = [];\n        source.subscribe(\n          createOperatorSubscriber(\n            subscriber,\n            (value) => {\n              // Add the most recent value onto the end of our buffer.\n              buffer.push(value);\n              // If our buffer is now larger than the number of values we\n              // want to take, we remove the oldest value from the buffer.\n              count < buffer.length && buffer.shift();\n            },\n            () => {\n              // The source completed, we now know what are last values\n              // are, emit them in the order they were received.\n              for (const value of buffer) {\n                subscriber.next(value);\n              }\n              subscriber.complete();\n            },\n            // Errors are passed through to the consumer\n            undefined,\n            () => {\n              // During finalization release the values in our buffer.\n              buffer = null!;\n            }\n          )\n        );\n      });\n}\n", "import { ObservableInput, ObservableInputTuple, OperatorFunction, SchedulerLike } from '../types';\nimport { operate } from '../util/lift';\nimport { argsOrArgArray } from '../util/argsOrArgArray';\nimport { mergeAll } from './mergeAll';\nimport { popNumber, popScheduler } from '../util/args';\nimport { from } from '../observable/from';\n\n/** @deprecated Replaced with {@link mergeWith}. Will be removed in v8. */\nexport function merge<T, A extends readonly unknown[]>(...sources: [...ObservableInputTuple<A>]): OperatorFunction<T, T | A[number]>;\n/** @deprecated Replaced with {@link mergeWith}. Will be removed in v8. */\nexport function merge<T, A extends readonly unknown[]>(\n  ...sourcesAndConcurrency: [...ObservableInputTuple<A>, number]\n): OperatorFunction<T, T | A[number]>;\n/** @deprecated Replaced with {@link mergeWith}. Will be removed in v8. */\nexport function merge<T, A extends readonly unknown[]>(\n  ...sourcesAndScheduler: [...ObservableInputTuple<A>, SchedulerLike]\n): OperatorFunction<T, T | A[number]>;\n/** @deprecated Replaced with {@link mergeWith}. Will be removed in v8. */\nexport function merge<T, A extends readonly unknown[]>(\n  ...sourcesAndConcurrencyAndScheduler: [...ObservableInputTuple<A>, number, SchedulerLike]\n): OperatorFunction<T, T | A[number]>;\n\nexport function merge<T>(...args: unknown[]): OperatorFunction<T, unknown> {\n  const scheduler = popScheduler(args);\n  const concurrent = popNumber(args, Infinity);\n  args = argsOrArgArray(args);\n\n  return operate((source, subscriber) => {\n    mergeAll(concurrent)(from([source, ...(args as ObservableInput<T>[])], scheduler)).subscribe(subscriber);\n  });\n}\n", "import { ObservableInputTuple, OperatorFunction } from '../types';\nimport { merge } from './merge';\n\n/**\n * Merge the values from all observables to a single observable result.\n *\n * Creates an observable, that when subscribed to, subscribes to the source\n * observable, and all other sources provided as arguments. All values from\n * every source are emitted from the resulting subscription.\n *\n * When all sources complete, the resulting observable will complete.\n *\n * When any source errors, the resulting observable will error.\n *\n * ## Example\n *\n * Joining all outputs from multiple user input event streams\n *\n * ```ts\n * import { fromEvent, map, mergeWith } from 'rxjs';\n *\n * const clicks$ = fromEvent(document, 'click').pipe(map(() => 'click'));\n * const mousemoves$ = fromEvent(document, 'mousemove').pipe(map(() => 'mousemove'));\n * const dblclicks$ = fromEvent(document, 'dblclick').pipe(map(() => 'dblclick'));\n *\n * mousemoves$\n *   .pipe(mergeWith(clicks$, dblclicks$))\n *   .subscribe(x => console.log(x));\n *\n * // result (assuming user interactions)\n * // 'mousemove'\n * // 'mousemove'\n * // 'mousemove'\n * // 'click'\n * // 'click'\n * // 'dblclick'\n * ```\n *\n * @see {@link merge}\n *\n * @param otherSources the sources to combine the current source with.\n * @return A function that returns an Observable that merges the values from\n * all given Observables.\n */\nexport function mergeWith<T, A extends readonly unknown[]>(\n  ...otherSources: [...ObservableInputTuple<A>]\n): OperatorFunction<T, T | A[number]> {\n  return merge(...otherSources);\n}\n", "import { Subscription } from '../Subscription';\nimport { EMPTY } from '../observable/empty';\nimport { operate } from '../util/lift';\nimport { MonoTypeOperatorFunction, ObservableInput } from '../types';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { innerFrom } from '../observable/innerFrom';\nimport { timer } from '../observable/timer';\n\nexport interface RepeatConfig {\n  /**\n   * The number of times to repeat the source. Defaults to `Infinity`.\n   */\n  count?: number;\n\n  /**\n   * If a `number`, will delay the repeat of the source by that number of milliseconds.\n   * If a function, it will provide the number of times the source has been subscribed to,\n   * and the return value should be a valid observable input that will notify when the source\n   * should be repeated. If the notifier observable is empty, the result will complete.\n   */\n  delay?: number | ((count: number) => ObservableInput<any>);\n}\n\n/**\n * Returns an Observable that will resubscribe to the source stream when the source stream completes.\n *\n * <span class=\"informal\">Repeats all values emitted on the source. It's like {@link retry}, but for non error cases.</span>\n *\n * ![](repeat.png)\n *\n * Repeat will output values from a source until the source completes, then it will resubscribe to the\n * source a specified number of times, with a specified delay. Repeat can be particularly useful in\n * combination with closing operators like {@link take}, {@link takeUntil}, {@link first}, or {@link takeWhile},\n * as it can be used to restart a source again from scratch.\n *\n * Repeat is very similar to {@link retry}, where {@link retry} will resubscribe to the source in the error case, but\n * `repeat` will resubscribe if the source completes.\n *\n * Note that `repeat` will _not_ catch errors. Use {@link retry} for that.\n *\n * - `repeat(0)` returns an empty observable\n * - `repeat()` will repeat forever\n * - `repeat({ delay: 200 })` will repeat forever, with a delay of 200ms between repetitions.\n * - `repeat({ count: 2, delay: 400 })` will repeat twice, with a delay of 400ms between repetitions.\n * - `repeat({ delay: (count) => timer(count * 1000) })` will repeat forever, but will have a delay that grows by one second for each repetition.\n *\n * ## Example\n *\n * Repeat a message stream\n *\n * ```ts\n * import { of, repeat } from 'rxjs';\n *\n * const source = of('Repeat message');\n * const result = source.pipe(repeat(3));\n *\n * result.subscribe(x => console.log(x));\n *\n * // Results\n * // 'Repeat message'\n * // 'Repeat message'\n * // 'Repeat message'\n * ```\n *\n * Repeat 3 values, 2 times\n *\n * ```ts\n * import { interval, take, repeat } from 'rxjs';\n *\n * const source = interval(1000);\n * const result = source.pipe(take(3), repeat(2));\n *\n * result.subscribe(x => console.log(x));\n *\n * // Results every second\n * // 0\n * // 1\n * // 2\n * // 0\n * // 1\n * // 2\n * ```\n *\n * Defining two complex repeats with delays on the same source.\n * Note that the second repeat cannot be called until the first\n * repeat as exhausted it's count.\n *\n * ```ts\n * import { defer, of, repeat } from 'rxjs';\n *\n * const source = defer(() => {\n *    return of(`Hello, it is ${new Date()}`)\n * });\n *\n * source.pipe(\n *    // Repeat 3 times with a delay of 1 second between repetitions\n *    repeat({\n *      count: 3,\n *      delay: 1000,\n *    }),\n *\n *    // *Then* repeat forever, but with an exponential step-back\n *    // maxing out at 1 minute.\n *    repeat({\n *      delay: (count) => timer(Math.min(60000, 2 ^ count * 1000))\n *    })\n * )\n * ```\n *\n * @see {@link repeatWhen}\n * @see {@link retry}\n *\n * @param count The number of times the source Observable items are repeated, a count of 0 will yield\n * an empty Observable.\n */\nexport function repeat<T>(countOrConfig?: number | RepeatConfig): MonoTypeOperatorFunction<T> {\n  let count = Infinity;\n  let delay: RepeatConfig['delay'];\n\n  if (countOrConfig != null) {\n    if (typeof countOrConfig === 'object') {\n      ({ count = Infinity, delay } = countOrConfig);\n    } else {\n      count = countOrConfig;\n    }\n  }\n\n  return count <= 0\n    ? () => EMPTY\n    : operate((source, subscriber) => {\n        let soFar = 0;\n        let sourceSub: Subscription | null;\n\n        const resubscribe = () => {\n          sourceSub?.unsubscribe();\n          sourceSub = null;\n          if (delay != null) {\n            const notifier = typeof delay === 'number' ? timer(delay) : innerFrom(delay(soFar));\n            const notifierSubscriber = createOperatorSubscriber(subscriber, () => {\n              notifierSubscriber.unsubscribe();\n              subscribeToSource();\n            });\n            notifier.subscribe(notifierSubscriber);\n          } else {\n            subscribeToSource();\n          }\n        };\n\n        const subscribeToSource = () => {\n          let syncUnsub = false;\n          sourceSub = source.subscribe(\n            createOperatorSubscriber(subscriber, undefined, () => {\n              if (++soFar < count) {\n                if (sourceSub) {\n                  resubscribe();\n                } else {\n                  syncUnsub = true;\n                }\n              } else {\n                subscriber.complete();\n              }\n            })\n          );\n\n          if (syncUnsub) {\n            resubscribe();\n          }\n        };\n\n        subscribeToSource();\n      });\n}\n", "import { OperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { scanInternals } from './scanInternals';\n\nexport function scan<V, A = V>(accumulator: (acc: A | V, value: V, index: number) => A): OperatorFunction<V, V | A>;\nexport function scan<V, A>(accumulator: (acc: A, value: V, index: number) => A, seed: A): OperatorFunction<V, A>;\nexport function scan<V, A, S>(accumulator: (acc: A | S, value: V, index: number) => A, seed: S): OperatorFunction<V, A>;\n\n// TODO: link to a \"redux pattern\" section in the guide (location TBD)\n\n/**\n * Useful for encapsulating and managing state. Applies an accumulator (or \"reducer function\")\n * to each value from the source after an initial state is established -- either via\n * a `seed` value (second argument), or from the first value from the source.\n *\n * <span class=\"informal\">It's like {@link reduce}, but emits the current\n * accumulation state after each update</span>\n *\n * ![](scan.png)\n *\n * This operator maintains an internal state and emits it after processing each value as follows:\n *\n * 1. First value arrives\n *   - If a `seed` value was supplied (as the second argument to `scan`), let `state = seed` and `value = firstValue`.\n *   - If NO `seed` value was supplied (no second argument), let `state = firstValue` and go to 3.\n * 2. Let `state = accumulator(state, value)`.\n *   - If an error is thrown by `accumulator`, notify the consumer of an error. The process ends.\n * 3. Emit `state`.\n * 4. Next value arrives, let `value = nextValue`, go to 2.\n *\n * ## Examples\n *\n * An average of previous numbers. This example shows how\n * not providing a `seed` can prime the stream with the\n * first value from the source.\n *\n * ```ts\n * import { of, scan, map } from 'rxjs';\n *\n * const numbers$ = of(1, 2, 3);\n *\n * numbers$\n *   .pipe(\n *     // Get the sum of the numbers coming in.\n *     scan((total, n) => total + n),\n *     // Get the average by dividing the sum by the total number\n *     // received so far (which is 1 more than the zero-based index).\n *     map((sum, index) => sum / (index + 1))\n *   )\n *   .subscribe(console.log);\n * ```\n *\n * The Fibonacci sequence. This example shows how you can use\n * a seed to prime accumulation process. Also... you know... Fibonacci.\n * So important to like, computers and stuff that its whiteboarded\n * in job interviews. Now you can show them the Rx version! (Please don't, haha)\n *\n * ```ts\n * import { interval, scan, map, startWith } from 'rxjs';\n *\n * const firstTwoFibs = [0, 1];\n * // An endless stream of Fibonacci numbers.\n * const fibonacci$ = interval(1000).pipe(\n *   // Scan to get the fibonacci numbers (after 0, 1)\n *   scan(([a, b]) => [b, a + b], firstTwoFibs),\n *   // Get the second number in the tuple, it's the one you calculated\n *   map(([, n]) => n),\n *   // Start with our first two digits :)\n *   startWith(...firstTwoFibs)\n * );\n *\n * fibonacci$.subscribe(console.log);\n * ```\n *\n * @see {@link expand}\n * @see {@link mergeScan}\n * @see {@link reduce}\n * @see {@link switchScan}\n *\n * @param accumulator A \"reducer function\". This will be called for each value after an initial state is\n * acquired.\n * @param seed The initial state. If this is not provided, the first value from the source will\n * be used as the initial state, and emitted without going through the accumulator. All subsequent values\n * will be processed by the accumulator function. If this is provided, all values will go through\n * the accumulator function.\n * @return A function that returns an Observable of the accumulated values.\n */\nexport function scan<V, A, S>(accumulator: (acc: V | A | S, value: V, index: number) => A, seed?: S): OperatorFunction<V, V | A> {\n  // providing a seed of `undefined` *should* be valid and trigger\n  // hasSeed! so don't use `seed !== undefined` checks!\n  // For this reason, we have to check it here at the original call site\n  // otherwise inside Operator/Subscriber we won't know if `undefined`\n  // means they didn't provide anything or if they literally provided `undefined`\n  return operate(scanInternals(accumulator, seed as S, arguments.length >= 2, true));\n}\n", "import { innerFrom } from '../observable/innerFrom';\nimport { Subject } from '../Subject';\nimport { SafeSubscriber } from '../Subscriber';\nimport { Subscription } from '../Subscription';\nimport { MonoTypeOperatorFunction, SubjectLike, ObservableInput } from '../types';\nimport { operate } from '../util/lift';\n\nexport interface ShareConfig<T> {\n  /**\n   * The factory used to create the subject that will connect the source observable to\n   * multicast consumers.\n   */\n  connector?: () => SubjectLike<T>;\n  /**\n   * If `true`, the resulting observable will reset internal state on error from source and return to a \"cold\" state. This\n   * allows the resulting observable to be \"retried\" in the event of an error.\n   * If `false`, when an error comes from the source it will push the error into the connecting subject, and the subject\n   * will remain the connecting subject, meaning the resulting observable will not go \"cold\" again, and subsequent retries\n   * or resubscriptions will resubscribe to that same subject. In all cases, RxJS subjects will emit the same error again, however\n   * {@link ReplaySubject} will also push its buffered values before pushing the error.\n   * It is also possible to pass a notifier factory returning an `ObservableInput` instead which grants more fine-grained\n   * control over how and when the reset should happen. This allows behaviors like conditional or delayed resets.\n   */\n  resetOnError?: boolean | ((error: any) => ObservableInput<any>);\n  /**\n   * If `true`, the resulting observable will reset internal state on completion from source and return to a \"cold\" state. This\n   * allows the resulting observable to be \"repeated\" after it is done.\n   * If `false`, when the source completes, it will push the completion through the connecting subject, and the subject\n   * will remain the connecting subject, meaning the resulting observable will not go \"cold\" again, and subsequent repeats\n   * or resubscriptions will resubscribe to that same subject.\n   * It is also possible to pass a notifier factory returning an `ObservableInput` instead which grants more fine-grained\n   * control over how and when the reset should happen. This allows behaviors like conditional or delayed resets.\n   */\n  resetOnComplete?: boolean | (() => ObservableInput<any>);\n  /**\n   * If `true`, when the number of subscribers to the resulting observable reaches zero due to those subscribers unsubscribing, the\n   * internal state will be reset and the resulting observable will return to a \"cold\" state. This means that the next\n   * time the resulting observable is subscribed to, a new subject will be created and the source will be subscribed to\n   * again.\n   * If `false`, when the number of subscribers to the resulting observable reaches zero due to unsubscription, the subject\n   * will remain connected to the source, and new subscriptions to the result will be connected through that same subject.\n   * It is also possible to pass a notifier factory returning an `ObservableInput` instead which grants more fine-grained\n   * control over how and when the reset should happen. This allows behaviors like conditional or delayed resets.\n   */\n  resetOnRefCountZero?: boolean | (() => ObservableInput<any>);\n}\n\nexport function share<T>(): MonoTypeOperatorFunction<T>;\n\nexport function share<T>(options: ShareConfig<T>): MonoTypeOperatorFunction<T>;\n\n/**\n * Returns a new Observable that multicasts (shares) the original Observable. As long as there is at least one\n * Subscriber this Observable will be subscribed and emitting data. When all subscribers have unsubscribed it will\n * unsubscribe from the source Observable. Because the Observable is multicasting it makes the stream `hot`.\n * This is an alias for `multicast(() => new Subject()), refCount()`.\n *\n * The subscription to the underlying source Observable can be reset (unsubscribe and resubscribe for new subscribers),\n * if the subscriber count to the shared observable drops to 0, or if the source Observable errors or completes. It is\n * possible to use notifier factories for the resets to allow for behaviors like conditional or delayed resets. Please\n * note that resetting on error or complete of the source Observable does not behave like a transparent retry or restart\n * of the source because the error or complete will be forwarded to all subscribers and their subscription will be\n * closed. Only new subscribers after a reset on error or complete happened will cause a fresh subscription to the\n * source. To achieve transparent retries or restarts pipe the source through appropriate operators before sharing.\n *\n * ![](share.png)\n *\n * ## Example\n *\n * Generate new multicast Observable from the `source` Observable value\n *\n * ```ts\n * import { interval, tap, map, take, share } from 'rxjs';\n *\n * const source = interval(1000).pipe(\n *   tap(x => console.log('Processing: ', x)),\n *   map(x => x * x),\n *   take(6),\n *   share()\n * );\n *\n * source.subscribe(x => console.log('subscription 1: ', x));\n * source.subscribe(x => console.log('subscription 2: ', x));\n *\n * // Logs:\n * // Processing: 0\n * // subscription 1: 0\n * // subscription 2: 0\n * // Processing: 1\n * // subscription 1: 1\n * // subscription 2: 1\n * // Processing: 2\n * // subscription 1: 4\n * // subscription 2: 4\n * // Processing: 3\n * // subscription 1: 9\n * // subscription 2: 9\n * // Processing: 4\n * // subscription 1: 16\n * // subscription 2: 16\n * // Processing: 5\n * // subscription 1: 25\n * // subscription 2: 25\n * ```\n *\n * ## Example with notifier factory: Delayed reset\n *\n * ```ts\n * import { interval, take, share, timer } from 'rxjs';\n *\n * const source = interval(1000).pipe(\n *   take(3),\n *   share({\n *     resetOnRefCountZero: () => timer(1000)\n *   })\n * );\n *\n * const subscriptionOne = source.subscribe(x => console.log('subscription 1: ', x));\n * setTimeout(() => subscriptionOne.unsubscribe(), 1300);\n *\n * setTimeout(() => source.subscribe(x => console.log('subscription 2: ', x)), 1700);\n *\n * setTimeout(() => source.subscribe(x => console.log('subscription 3: ', x)), 5000);\n *\n * // Logs:\n * // subscription 1:  0\n * // (subscription 1 unsubscribes here)\n * // (subscription 2 subscribes here ~400ms later, source was not reset)\n * // subscription 2:  1\n * // subscription 2:  2\n * // (subscription 2 unsubscribes here)\n * // (subscription 3 subscribes here ~2000ms later, source did reset before)\n * // subscription 3:  0\n * // subscription 3:  1\n * // subscription 3:  2\n * ```\n *\n * @see {@link shareReplay}\n *\n * @return A function that returns an Observable that mirrors the source.\n */\nexport function share<T>(options: ShareConfig<T> = {}): MonoTypeOperatorFunction<T> {\n  const { connector = () => new Subject<T>(), resetOnError = true, resetOnComplete = true, resetOnRefCountZero = true } = options;\n  // It's necessary to use a wrapper here, as the _operator_ must be\n  // referentially transparent. Otherwise, it cannot be used in calls to the\n  // static `pipe` function - to create a partial pipeline.\n  //\n  // The _operator function_ - the function returned by the _operator_ - will\n  // not be referentially transparent - as it shares its source - but the\n  // _operator function_ is called when the complete pipeline is composed via a\n  // call to a source observable's `pipe` method - not when the static `pipe`\n  // function is called.\n  return (wrapperSource) => {\n    let connection: SafeSubscriber<T> | undefined;\n    let resetConnection: Subscription | undefined;\n    let subject: SubjectLike<T> | undefined;\n    let refCount = 0;\n    let hasCompleted = false;\n    let hasErrored = false;\n\n    const cancelReset = () => {\n      resetConnection?.unsubscribe();\n      resetConnection = undefined;\n    };\n    // Used to reset the internal state to a \"cold\"\n    // state, as though it had never been subscribed to.\n    const reset = () => {\n      cancelReset();\n      connection = subject = undefined;\n      hasCompleted = hasErrored = false;\n    };\n    const resetAndUnsubscribe = () => {\n      // We need to capture the connection before\n      // we reset (if we need to reset).\n      const conn = connection;\n      reset();\n      conn?.unsubscribe();\n    };\n\n    return operate<T, T>((source, subscriber) => {\n      refCount++;\n      if (!hasErrored && !hasCompleted) {\n        cancelReset();\n      }\n\n      // Create the subject if we don't have one yet. Grab a local reference to\n      // it as well, which avoids non-null assertions when using it and, if we\n      // connect to it now, then error/complete need a reference after it was\n      // reset.\n      const dest = (subject = subject ?? connector());\n\n      // Add the finalization directly to the subscriber - instead of returning it -\n      // so that the handling of the subscriber's unsubscription will be wired\n      // up _before_ the subscription to the source occurs. This is done so that\n      // the assignment to the source connection's `closed` property will be seen\n      // by synchronous firehose sources.\n      subscriber.add(() => {\n        refCount--;\n\n        // If we're resetting on refCount === 0, and it's 0, we only want to do\n        // that on \"unsubscribe\", really. Resetting on error or completion is a different\n        // configuration.\n        if (refCount === 0 && !hasErrored && !hasCompleted) {\n          resetConnection = handleReset(resetAndUnsubscribe, resetOnRefCountZero);\n        }\n      });\n\n      // The following line adds the subscription to the subscriber passed.\n      // Basically, `subscriber === dest.subscribe(subscriber)` is `true`.\n      dest.subscribe(subscriber);\n\n      if (\n        !connection &&\n        // Check this shareReplay is still activate - it can be reset to 0\n        // and be \"unsubscribed\" _before_ it actually subscribes.\n        // If we were to subscribe then, it'd leak and get stuck.\n        refCount > 0\n      ) {\n        // We need to create a subscriber here - rather than pass an observer and\n        // assign the returned subscription to connection - because it's possible\n        // for reentrant subscriptions to the shared observable to occur and in\n        // those situations we want connection to be already-assigned so that we\n        // don't create another connection to the source.\n        connection = new SafeSubscriber({\n          next: (value) => dest.next(value),\n          error: (err) => {\n            hasErrored = true;\n            cancelReset();\n            resetConnection = handleReset(reset, resetOnError, err);\n            dest.error(err);\n          },\n          complete: () => {\n            hasCompleted = true;\n            cancelReset();\n            resetConnection = handleReset(reset, resetOnComplete);\n            dest.complete();\n          },\n        });\n        innerFrom(source).subscribe(connection);\n      }\n    })(wrapperSource);\n  };\n}\n\nfunction handleReset<T extends unknown[] = never[]>(\n  reset: () => void,\n  on: boolean | ((...args: T) => ObservableInput<any>),\n  ...args: T\n): Subscription | undefined {\n  if (on === true) {\n    reset();\n    return;\n  }\n\n  if (on === false) {\n    return;\n  }\n\n  const onSubscriber = new SafeSubscriber({\n    next: () => {\n      onSubscriber.unsubscribe();\n      reset();\n    },\n  });\n\n  return innerFrom(on(...args)).subscribe(onSubscriber);\n}\n", "import { ReplaySubject } from '../ReplaySubject';\nimport { MonoTypeOperatorFunction, SchedulerLike } from '../types';\nimport { share } from './share';\n\nexport interface ShareReplayConfig {\n  bufferSize?: number;\n  windowTime?: number;\n  refCount: boolean;\n  scheduler?: SchedulerLike;\n}\n\nexport function shareReplay<T>(config: ShareReplayConfig): MonoTypeOperatorFunction<T>;\nexport function shareReplay<T>(bufferSize?: number, windowTime?: number, scheduler?: SchedulerLike): MonoTypeOperatorFunction<T>;\n\n/**\n * Share source and replay specified number of emissions on subscription.\n *\n * This operator is a specialization of `replay` that connects to a source observable\n * and multicasts through a `ReplaySubject` constructed with the specified arguments.\n * A successfully completed source will stay cached in the `shareReplay`ed observable forever,\n * but an errored source can be retried.\n *\n * ## Why use `shareReplay`?\n *\n * You generally want to use `shareReplay` when you have side-effects or taxing computations\n * that you do not wish to be executed amongst multiple subscribers.\n * It may also be valuable in situations where you know you will have late subscribers to\n * a stream that need access to previously emitted values.\n * This ability to replay values on subscription is what differentiates {@link share} and `shareReplay`.\n *\n * ## Reference counting\n *\n * By default `shareReplay` will use `refCount` of false, meaning that it will _not_ unsubscribe the\n * source when the reference counter drops to zero, i.e. the inner `ReplaySubject` will _not_ be unsubscribed\n * (and potentially run for ever).\n * This is the default as it is expected that `shareReplay` is often used to keep around expensive to setup\n * observables which we want to keep running instead of having to do the expensive setup again.\n *\n * As of RXJS version 6.4.0 a new overload signature was added to allow for manual control over what\n * happens when the operators internal reference counter drops to zero.\n * If `refCount` is true, the source will be unsubscribed from once the reference count drops to zero, i.e.\n * the inner `ReplaySubject` will be unsubscribed. All new subscribers will receive value emissions from a\n * new `ReplaySubject` which in turn will cause a new subscription to the source observable.\n *\n * ## Examples\n *\n * Example with a third subscriber coming late to the party\n *\n * ```ts\n * import { interval, take, shareReplay } from 'rxjs';\n *\n * const shared$ = interval(2000).pipe(\n *   take(6),\n *   shareReplay(3)\n * );\n *\n * shared$.subscribe(x => console.log('sub A: ', x));\n * shared$.subscribe(y => console.log('sub B: ', y));\n *\n * setTimeout(() => {\n *   shared$.subscribe(y => console.log('sub C: ', y));\n * }, 11000);\n *\n * // Logs:\n * // (after ~2000 ms)\n * // sub A: 0\n * // sub B: 0\n * // (after ~4000 ms)\n * // sub A: 1\n * // sub B: 1\n * // (after ~6000 ms)\n * // sub A: 2\n * // sub B: 2\n * // (after ~8000 ms)\n * // sub A: 3\n * // sub B: 3\n * // (after ~10000 ms)\n * // sub A: 4\n * // sub B: 4\n * // (after ~11000 ms, sub C gets the last 3 values)\n * // sub C: 2\n * // sub C: 3\n * // sub C: 4\n * // (after ~12000 ms)\n * // sub A: 5\n * // sub B: 5\n * // sub C: 5\n * ```\n *\n * Example for `refCount` usage\n *\n * ```ts\n * import { Observable, tap, interval, shareReplay, take } from 'rxjs';\n *\n * const log = <T>(name: string, source: Observable<T>) => source.pipe(\n *   tap({\n *     subscribe: () => console.log(`${ name }: subscribed`),\n *     next: value => console.log(`${ name }: ${ value }`),\n *     complete: () => console.log(`${ name }: completed`),\n *     finalize: () => console.log(`${ name }: unsubscribed`)\n *   })\n * );\n *\n * const obs$ = log('source', interval(1000));\n *\n * const shared$ = log('shared', obs$.pipe(\n *   shareReplay({ bufferSize: 1, refCount: true }),\n *   take(2)\n * ));\n *\n * shared$.subscribe(x => console.log('sub A: ', x));\n * shared$.subscribe(y => console.log('sub B: ', y));\n *\n * // PRINTS:\n * // shared: subscribed <-- reference count = 1\n * // source: subscribed\n * // shared: subscribed <-- reference count = 2\n * // source: 0\n * // shared: 0\n * // sub A: 0\n * // shared: 0\n * // sub B: 0\n * // source: 1\n * // shared: 1\n * // sub A: 1\n * // shared: completed <-- take(2) completes the subscription for sub A\n * // shared: unsubscribed <-- reference count = 1\n * // shared: 1\n * // sub B: 1\n * // shared: completed <-- take(2) completes the subscription for sub B\n * // shared: unsubscribed <-- reference count = 0\n * // source: unsubscribed <-- replaySubject unsubscribes from source observable because the reference count dropped to 0 and refCount is true\n *\n * // In case of refCount being false, the unsubscribe is never called on the source and the source would keep on emitting, even if no subscribers\n * // are listening.\n * // source: 2\n * // source: 3\n * // source: 4\n * // ...\n * ```\n *\n * @see {@link publish}\n * @see {@link share}\n * @see {@link publishReplay}\n *\n * @param configOrBufferSize Maximum element count of the replay buffer or {@link ShareReplayConfig configuration}\n * object.\n * @param windowTime Maximum time length of the replay buffer in milliseconds.\n * @param scheduler Scheduler where connected observers within the selector function\n * will be invoked on.\n * @return A function that returns an Observable sequence that contains the\n * elements of a sequence produced by multicasting the source sequence within a\n * selector function.\n */\nexport function shareReplay<T>(\n  configOrBufferSize?: ShareReplayConfig | number,\n  windowTime?: number,\n  scheduler?: SchedulerLike\n): MonoTypeOperatorFunction<T> {\n  let bufferSize: number;\n  let refCount = false;\n  if (configOrBufferSize && typeof configOrBufferSize === 'object') {\n    ({ bufferSize = Infinity, windowTime = Infinity, refCount = false, scheduler } = configOrBufferSize);\n  } else {\n    bufferSize = (configOrBufferSize ?? Infinity) as number;\n  }\n  return share<T>({\n    connector: () => new ReplaySubject(bufferSize, windowTime, scheduler),\n    resetOnError: true,\n    resetOnComplete: false,\n    resetOnRefCountZero: refCount,\n  });\n}\n", "import { MonoTypeOperatorFunction } from '../types';\nimport { filter } from './filter';\n\n/**\n * Returns an Observable that skips the first `count` items emitted by the source Observable.\n *\n * ![](skip.png)\n *\n * Skips the values until the sent notifications are equal or less than provided skip count. It raises\n * an error if skip count is equal or more than the actual number of emits and source raises an error.\n *\n * ## Example\n *\n * Skip the values before the emission\n *\n * ```ts\n * import { interval, skip } from 'rxjs';\n *\n * // emit every half second\n * const source = interval(500);\n * // skip the first 10 emitted values\n * const result = source.pipe(skip(10));\n *\n * result.subscribe(value => console.log(value));\n * // output: 10...11...12...13...\n * ```\n *\n * @see {@link last}\n * @see {@link skipWhile}\n * @see {@link skipUntil}\n * @see {@link skipLast}\n *\n * @param {Number} count - The number of times, items emitted by source Observable should be skipped.\n * @return A function that returns an Observable that skips the first `count`\n * values emitted by the source Observable.\n */\nexport function skip<T>(count: number): MonoTypeOperatorFunction<T> {\n  return filter((_, index) => count <= index);\n}\n", "import { MonoTypeOperatorFunction, ObservableInput } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { innerFrom } from '../observable/innerFrom';\nimport { noop } from '../util/noop';\n\n/**\n * Returns an Observable that skips items emitted by the source Observable until a second Observable emits an item.\n *\n * The `skipUntil` operator causes the observable stream to skip the emission of values until the passed in observable\n * emits the first value. This can be particularly useful in combination with user interactions, responses of HTTP\n * requests or waiting for specific times to pass by.\n *\n * ![](skipUntil.png)\n *\n * Internally, the `skipUntil` operator subscribes to the passed in `notifier` `ObservableInput` (which gets converted\n * to an Observable) in order to recognize the emission of its first value. When `notifier` emits next, the operator\n * unsubscribes from it and starts emitting the values of the *source* observable until it completes or errors. It\n * will never let the *source* observable emit any values if the `notifier` completes or throws an error without\n * emitting a value before.\n *\n * ## Example\n *\n * In the following example, all emitted values of the interval observable are skipped until the user clicks anywhere\n * within the page\n *\n * ```ts\n * import { interval, fromEvent, skipUntil } from 'rxjs';\n *\n * const intervalObservable = interval(1000);\n * const click = fromEvent(document, 'click');\n *\n * const emitAfterClick = intervalObservable.pipe(\n *   skipUntil(click)\n * );\n * // clicked at 4.6s. output: 5...6...7...8........ or\n * // clicked at 7.3s. output: 8...9...10..11.......\n * emitAfterClick.subscribe(value => console.log(value));\n * ```\n *\n * @see {@link last}\n * @see {@link skip}\n * @see {@link skipWhile}\n * @see {@link skipLast}\n *\n * @param notifier An `ObservableInput` that has to emit an item before the source Observable elements begin to\n * be mirrored by the resulting Observable.\n * @return A function that returns an Observable that skips items from the\n * source Observable until the `notifier` Observable emits an item, then emits the\n * remaining items.\n */\nexport function skipUntil<T>(notifier: ObservableInput<any>): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    let taking = false;\n\n    const skipSubscriber = createOperatorSubscriber(\n      subscriber,\n      () => {\n        skipSubscriber?.unsubscribe();\n        taking = true;\n      },\n      noop\n    );\n\n    innerFrom(notifier).subscribe(skipSubscriber);\n\n    source.subscribe(createOperatorSubscriber(subscriber, (value) => taking && subscriber.next(value)));\n  });\n}\n", "import { concat } from '../observable/concat';\nimport { OperatorFunction, SchedulerLike, ValueFromArray } from '../types';\nimport { popScheduler } from '../util/args';\nimport { operate } from '../util/lift';\n\n// Devs are more likely to pass null or undefined than they are a scheduler\n// without accompanying values. To make things easier for (naughty) devs who\n// use the `strictNullChecks: false` TypeScript compiler option, these\n// overloads with explicit null and undefined values are included.\n\nexport function startWith<T>(value: null): OperatorFunction<T, T | null>;\nexport function startWith<T>(value: undefined): OperatorFunction<T, T | undefined>;\n\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `concatAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function startWith<T, A extends readonly unknown[] = T[]>(\n  ...valuesAndScheduler: [...A, SchedulerLike]\n): OperatorFunction<T, T | ValueFromArray<A>>;\nexport function startWith<T, A extends readonly unknown[] = T[]>(...values: A): OperatorFunction<T, T | ValueFromArray<A>>;\n\n/**\n * Returns an observable that, at the moment of subscription, will synchronously emit all\n * values provided to this operator, then subscribe to the source and mirror all of its emissions\n * to subscribers.\n *\n * This is a useful way to know when subscription has occurred on an existing observable.\n *\n * <span class=\"informal\">First emits its arguments in order, and then any\n * emissions from the source.</span>\n *\n * ![](startWith.png)\n *\n * ## Examples\n *\n * Emit a value when a timer starts.\n *\n * ```ts\n * import { timer, map, startWith } from 'rxjs';\n *\n * timer(1000)\n *   .pipe(\n *     map(() => 'timer emit'),\n *     startWith('timer start')\n *   )\n *   .subscribe(x => console.log(x));\n *\n * // results:\n * // 'timer start'\n * // 'timer emit'\n * ```\n *\n * @param values Items you want the modified Observable to emit first.\n * @return A function that returns an Observable that synchronously emits\n * provided values before subscribing to the source Observable.\n *\n * @see {@link endWith}\n * @see {@link finalize}\n * @see {@link concat}\n */\nexport function startWith<T, D>(...values: D[]): OperatorFunction<T, T | D> {\n  const scheduler = popScheduler(values);\n  return operate((source, subscriber) => {\n    // Here we can't pass `undefined` as a scheduler, because if we did, the\n    // code inside of `concat` would be confused by the `undefined`, and treat it\n    // like an invalid observable. So we have to split it two different ways.\n    (scheduler ? concat(values, source, scheduler) : concat(values, source)).subscribe(subscriber);\n  });\n}\n", "import { Subscriber } from '../Subscriber';\nimport { ObservableInput, OperatorFunction, ObservedValueOf } from '../types';\nimport { innerFrom } from '../observable/innerFrom';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/* tslint:disable:max-line-length */\nexport function switchMap<T, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O\n): OperatorFunction<T, ObservedValueOf<O>>;\n/** @deprecated The `resultSelector` parameter will be removed in v8. Use an inner `map` instead. Details: https://rxjs.dev/deprecations/resultSelector */\nexport function switchMap<T, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O,\n  resultSelector: undefined\n): OperatorFunction<T, ObservedValueOf<O>>;\n/** @deprecated The `resultSelector` parameter will be removed in v8. Use an inner `map` instead. Details: https://rxjs.dev/deprecations/resultSelector */\nexport function switchMap<T, R, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O,\n  resultSelector: (outerValue: T, innerValue: ObservedValueOf<O>, outerIndex: number, innerIndex: number) => R\n): OperatorFunction<T, R>;\n/* tslint:enable:max-line-length */\n\n/**\n * Projects each source value to an Observable which is merged in the output\n * Observable, emitting values only from the most recently projected Observable.\n *\n * <span class=\"informal\">Maps each value to an Observable, then flattens all of\n * these inner Observables using {@link switchAll}.</span>\n *\n * ![](switchMap.png)\n *\n * Returns an Observable that emits items based on applying a function that you\n * supply to each item emitted by the source Observable, where that function\n * returns an (so-called \"inner\") Observable. Each time it observes one of these\n * inner Observables, the output Observable begins emitting the items emitted by\n * that inner Observable. When a new inner Observable is emitted, `switchMap`\n * stops emitting items from the earlier-emitted inner Observable and begins\n * emitting items from the new one. It continues to behave like this for\n * subsequent inner Observables.\n *\n * ## Example\n *\n * Generate new Observable according to source Observable values\n *\n * ```ts\n * import { of, switchMap } from 'rxjs';\n *\n * const switched = of(1, 2, 3).pipe(switchMap(x => of(x, x ** 2, x ** 3)));\n * switched.subscribe(x => console.log(x));\n * // outputs\n * // 1\n * // 1\n * // 1\n * // 2\n * // 4\n * // 8\n * // 3\n * // 9\n * // 27\n * ```\n *\n * Restart an interval Observable on every click event\n *\n * ```ts\n * import { fromEvent, switchMap, interval } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(switchMap(() => interval(1000)));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link concatMap}\n * @see {@link exhaustMap}\n * @see {@link mergeMap}\n * @see {@link switchAll}\n * @see {@link switchMapTo}\n *\n * @param {function(value: T, index: number): ObservableInput} project A function\n * that, when applied to an item emitted by the source Observable, returns an\n * Observable.\n * @return A function that returns an Observable that emits the result of\n * applying the projection function (and the optional deprecated\n * `resultSelector`) to each item emitted by the source Observable and taking\n * only the values from the most recently projected inner Observable.\n */\nexport function switchMap<T, R, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O,\n  resultSelector?: (outerValue: T, innerValue: ObservedValueOf<O>, outerIndex: number, innerIndex: number) => R\n): OperatorFunction<T, ObservedValueOf<O> | R> {\n  return operate((source, subscriber) => {\n    let innerSubscriber: Subscriber<ObservedValueOf<O>> | null = null;\n    let index = 0;\n    // Whether or not the source subscription has completed\n    let isComplete = false;\n\n    // We only complete the result if the source is complete AND we don't have an active inner subscription.\n    // This is called both when the source completes and when the inners complete.\n    const checkComplete = () => isComplete && !innerSubscriber && subscriber.complete();\n\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value) => {\n          // Cancel the previous inner subscription if there was one\n          innerSubscriber?.unsubscribe();\n          let innerIndex = 0;\n          const outerIndex = index++;\n          // Start the next inner subscription\n          innerFrom(project(value, outerIndex)).subscribe(\n            (innerSubscriber = createOperatorSubscriber(\n              subscriber,\n              // When we get a new inner value, next it through. Note that this is\n              // handling the deprecate result selector here. This is because with this architecture\n              // it ends up being smaller than using the map operator.\n              (innerValue) => subscriber.next(resultSelector ? resultSelector(value, innerValue, outerIndex, innerIndex++) : innerValue),\n              () => {\n                // The inner has completed. Null out the inner subscriber to\n                // free up memory and to signal that we have no inner subscription\n                // currently.\n                innerSubscriber = null!;\n                checkComplete();\n              }\n            ))\n          );\n        },\n        () => {\n          isComplete = true;\n          checkComplete();\n        }\n      )\n    );\n  });\n}\n", "import { MonoTypeOperatorFunction, ObservableInput } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { innerFrom } from '../observable/innerFrom';\nimport { noop } from '../util/noop';\n\n/**\n * Emits the values emitted by the source Observable until a `notifier`\n * Observable emits a value.\n *\n * <span class=\"informal\">Lets values pass until a second Observable,\n * `notifier`, emits a value. Then, it completes.</span>\n *\n * ![](takeUntil.png)\n *\n * `takeUntil` subscribes and begins mirroring the source Observable. It also\n * monitors a second Observable, `notifier` that you provide. If the `notifier`\n * emits a value, the output Observable stops mirroring the source Observable\n * and completes. If the `notifier` doesn't emit any value and completes\n * then `takeUntil` will pass all values.\n *\n * ## Example\n *\n * Tick every second until the first click happens\n *\n * ```ts\n * import { interval, fromEvent, takeUntil } from 'rxjs';\n *\n * const source = interval(1000);\n * const clicks = fromEvent(document, 'click');\n * const result = source.pipe(takeUntil(clicks));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link take}\n * @see {@link takeLast}\n * @see {@link takeWhile}\n * @see {@link skip}\n *\n * @param {Observable} notifier The Observable whose first emitted value will\n * cause the output Observable of `takeUntil` to stop emitting values from the\n * source Observable.\n * @return A function that returns an Observable that emits the values from the\n * source Observable until `notifier` emits its first value.\n */\nexport function takeUntil<T>(notifier: ObservableInput<any>): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    innerFrom(notifier).subscribe(createOperatorSubscriber(subscriber, () => subscriber.complete(), noop));\n    !subscriber.closed && source.subscribe(subscriber);\n  });\n}\n", "import { OperatorFunction, MonoTypeOperatorFunction, TruthyTypesOf } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\nexport function takeWhile<T>(predicate: BooleanConstructor, inclusive: true): MonoTypeOperatorFunction<T>;\nexport function takeWhile<T>(predicate: BooleanConstructor, inclusive: false): OperatorFunction<T, TruthyTypesOf<T>>;\nexport function takeWhile<T>(predicate: BooleanConstructor): OperatorFunction<T, TruthyTypesOf<T>>;\nexport function takeWhile<T, S extends T>(predicate: (value: T, index: number) => value is S): OperatorFunction<T, S>;\nexport function takeWhile<T, S extends T>(predicate: (value: T, index: number) => value is S, inclusive: false): OperatorFunction<T, S>;\nexport function takeWhile<T>(predicate: (value: T, index: number) => boolean, inclusive?: boolean): MonoTypeOperatorFunction<T>;\n\n/**\n * Emits values emitted by the source Observable so long as each value satisfies\n * the given `predicate`, and then completes as soon as this `predicate` is not\n * satisfied.\n *\n * <span class=\"informal\">Takes values from the source only while they pass the\n * condition given. When the first value does not satisfy, it completes.</span>\n *\n * ![](takeWhile.png)\n *\n * `takeWhile` subscribes and begins mirroring the source Observable. Each value\n * emitted on the source is given to the `predicate` function which returns a\n * boolean, representing a condition to be satisfied by the source values. The\n * output Observable emits the source values until such time as the `predicate`\n * returns false, at which point `takeWhile` stops mirroring the source\n * Observable and completes the output Observable.\n *\n * ## Example\n *\n * Emit click events only while the clientX property is greater than 200\n *\n * ```ts\n * import { fromEvent, takeWhile } from 'rxjs';\n *\n * const clicks = fromEvent<PointerEvent>(document, 'click');\n * const result = clicks.pipe(takeWhile(ev => ev.clientX > 200));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link take}\n * @see {@link takeLast}\n * @see {@link takeUntil}\n * @see {@link skip}\n *\n * @param {function(value: T, index: number): boolean} predicate A function that\n * evaluates a value emitted by the source Observable and returns a boolean.\n * Also takes the (zero-based) index as the second argument.\n * @param {boolean} inclusive When set to `true` the value that caused\n * `predicate` to return `false` will also be emitted.\n * @return A function that returns an Observable that emits values from the\n * source Observable so long as each value satisfies the condition defined by\n * the `predicate`, then completes.\n */\nexport function takeWhile<T>(predicate: (value: T, index: number) => boolean, inclusive = false): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    let index = 0;\n    source.subscribe(\n      createOperatorSubscriber(subscriber, (value) => {\n        const result = predicate(value, index++);\n        (result || inclusive) && subscriber.next(value);\n        !result && subscriber.complete();\n      })\n    );\n  });\n}\n", "import { MonoTypeOperatorFunction, Observer } from '../types';\nimport { isFunction } from '../util/isFunction';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { identity } from '../util/identity';\n\n/**\n * An extension to the {@link Observer} interface used only by the {@link tap} operator.\n *\n * It provides a useful set of callbacks a user can register to do side-effects in\n * cases other than what the usual {@link Observer} callbacks are\n * ({@link guide/glossary-and-semantics#next next},\n * {@link guide/glossary-and-semantics#error error} and/or\n * {@link guide/glossary-and-semantics#complete complete}).\n *\n * ## Example\n *\n * ```ts\n * import { fromEvent, switchMap, tap, interval, take } from 'rxjs';\n *\n * const source$ = fromEvent(document, 'click');\n * const result$ = source$.pipe(\n *   switchMap((_, i) => i % 2 === 0\n *     ? fromEvent(document, 'mousemove').pipe(\n *         tap({\n *           subscribe: () => console.log('Subscribed to the mouse move events after click #' + i),\n *           unsubscribe: () => console.log('Mouse move events #' + i + ' unsubscribed'),\n *           finalize: () => console.log('Mouse move events #' + i + ' finalized')\n *         })\n *       )\n *     : interval(1_000).pipe(\n *         take(5),\n *         tap({\n *           subscribe: () => console.log('Subscribed to the 1-second interval events after click #' + i),\n *           unsubscribe: () => console.log('1-second interval events #' + i + ' unsubscribed'),\n *           finalize: () => console.log('1-second interval events #' + i + ' finalized')\n *         })\n *       )\n *   )\n * );\n *\n * const subscription = result$.subscribe({\n *   next: console.log\n * });\n *\n * setTimeout(() => {\n *   console.log('Unsubscribe after 60 seconds');\n *   subscription.unsubscribe();\n * }, 60_000);\n * ```\n */\nexport interface TapObserver<T> extends Observer<T> {\n  /**\n   * The callback that `tap` operator invokes at the moment when the source Observable\n   * gets subscribed to.\n   */\n  subscribe: () => void;\n  /**\n   * The callback that `tap` operator invokes when an explicit\n   * {@link guide/glossary-and-semantics#unsubscription unsubscribe} happens. It won't get invoked on\n   * `error` or `complete` events.\n   */\n  unsubscribe: () => void;\n  /**\n   * The callback that `tap` operator invokes when any kind of\n   * {@link guide/glossary-and-semantics#finalization finalization} happens - either when\n   * the source Observable `error`s or `complete`s or when it gets explicitly unsubscribed\n   * by the user. There is no difference in using this callback or the {@link finalize}\n   * operator, but if you're already using `tap` operator, you can use this callback\n   * instead. You'd get the same result in either case.\n   */\n  finalize: () => void;\n}\nexport function tap<T>(observerOrNext?: Partial<TapObserver<T>> | ((value: T) => void)): MonoTypeOperatorFunction<T>;\n/** @deprecated Instead of passing separate callback arguments, use an observer argument. Signatures taking separate callback arguments will be removed in v8. Details: https://rxjs.dev/deprecations/subscribe-arguments */\nexport function tap<T>(\n  next?: ((value: T) => void) | null,\n  error?: ((error: any) => void) | null,\n  complete?: (() => void) | null\n): MonoTypeOperatorFunction<T>;\n\n/**\n * Used to perform side-effects for notifications from the source observable\n *\n * <span class=\"informal\">Used when you want to affect outside state with a notification without altering the notification</span>\n *\n * ![](tap.png)\n *\n * Tap is designed to allow the developer a designated place to perform side effects. While you _could_ perform side-effects\n * inside of a `map` or a `mergeMap`, that would make their mapping functions impure, which isn't always a big deal, but will\n * make it so you can't do things like memoize those functions. The `tap` operator is designed solely for such side-effects to\n * help you remove side-effects from other operations.\n *\n * For any notification, next, error, or complete, `tap` will call the appropriate callback you have provided to it, via a function\n * reference, or a partial observer, then pass that notification down the stream.\n *\n * The observable returned by `tap` is an exact mirror of the source, with one exception: Any error that occurs -- synchronously -- in a handler\n * provided to `tap` will be emitted as an error from the returned observable.\n *\n * > Be careful! You can mutate objects as they pass through the `tap` operator's handlers.\n *\n * The most common use of `tap` is actually for debugging. You can place a `tap(console.log)` anywhere\n * in your observable `pipe`, log out the notifications as they are emitted by the source returned by the previous\n * operation.\n *\n * ## Examples\n *\n * Check a random number before it is handled. Below is an observable that will use a random number between 0 and 1,\n * and emit `'big'` or `'small'` depending on the size of that number. But we wanted to log what the original number\n * was, so we have added a `tap(console.log)`.\n *\n * ```ts\n * import { of, tap, map } from 'rxjs';\n *\n * of(Math.random()).pipe(\n *   tap(console.log),\n *   map(n => n > 0.5 ? 'big' : 'small')\n * ).subscribe(console.log);\n * ```\n *\n * Using `tap` to analyze a value and force an error. Below is an observable where in our system we only\n * want to emit numbers 3 or less we get from another source. We can force our observable to error\n * using `tap`.\n *\n * ```ts\n * import { of, tap } from 'rxjs';\n *\n * const source = of(1, 2, 3, 4, 5);\n *\n * source.pipe(\n *   tap(n => {\n *     if (n > 3) {\n *       throw new TypeError(`Value ${ n } is greater than 3`);\n *     }\n *   })\n * )\n * .subscribe({ next: console.log, error: err => console.log(err.message) });\n * ```\n *\n * We want to know when an observable completes before moving on to the next observable. The system\n * below will emit a random series of `'X'` characters from 3 different observables in sequence. The\n * only way we know when one observable completes and moves to the next one, in this case, is because\n * we have added a `tap` with the side effect of logging to console.\n *\n * ```ts\n * import { of, concatMap, interval, take, map, tap } from 'rxjs';\n *\n * of(1, 2, 3).pipe(\n *   concatMap(n => interval(1000).pipe(\n *     take(Math.round(Math.random() * 10)),\n *     map(() => 'X'),\n *     tap({ complete: () => console.log(`Done with ${ n }`) })\n *   ))\n * )\n * .subscribe(console.log);\n * ```\n *\n * @see {@link finalize}\n * @see {@link TapObserver}\n *\n * @param observerOrNext A next handler or partial observer\n * @param error An error handler\n * @param complete A completion handler\n * @return A function that returns an Observable identical to the source, but\n * runs the specified Observer or callback(s) for each item.\n */\nexport function tap<T>(\n  observerOrNext?: Partial<TapObserver<T>> | ((value: T) => void) | null,\n  error?: ((e: any) => void) | null,\n  complete?: (() => void) | null\n): MonoTypeOperatorFunction<T> {\n  // We have to check to see not only if next is a function,\n  // but if error or complete were passed. This is because someone\n  // could technically call tap like `tap(null, fn)` or `tap(null, null, fn)`.\n  const tapObserver =\n    isFunction(observerOrNext) || error || complete\n      ? // tslint:disable-next-line: no-object-literal-type-assertion\n        ({ next: observerOrNext as Exclude<typeof observerOrNext, Partial<TapObserver<T>>>, error, complete } as Partial<TapObserver<T>>)\n      : observerOrNext;\n\n  return tapObserver\n    ? operate((source, subscriber) => {\n        tapObserver.subscribe?.();\n        let isUnsub = true;\n        source.subscribe(\n          createOperatorSubscriber(\n            subscriber,\n            (value) => {\n              tapObserver.next?.(value);\n              subscriber.next(value);\n            },\n            () => {\n              isUnsub = false;\n              tapObserver.complete?.();\n              subscriber.complete();\n            },\n            (err) => {\n              isUnsub = false;\n              tapObserver.error?.(err);\n              subscriber.error(err);\n            },\n            () => {\n              if (isUnsub) {\n                tapObserver.unsubscribe?.();\n              }\n              tapObserver.finalize?.();\n            }\n          )\n        );\n      })\n    : // Tap was called with no valid tap observer or handler\n      // (e.g. `tap(null, null, null)` or `tap(null)` or `tap()`)\n      // so we're going to just mirror the source.\n      identity;\n}\n", "import { Subscription } from '../Subscription';\n\nimport { MonoTypeOperatorFunction, ObservableInput } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { innerFrom } from '../observable/innerFrom';\n\n/**\n * An object interface used by {@link throttle} or {@link throttleTime} that ensure\n * configuration options of these operators.\n *\n * @see {@link throttle}\n * @see {@link throttleTime}\n */\nexport interface ThrottleConfig {\n  /**\n   * If `true`, the resulting Observable will emit the first value from the source\n   * Observable at the **start** of the \"throttling\" process (when starting an\n   * internal timer that prevents other emissions from the source to pass through).\n   * If `false`, it will not emit the first value from the source Observable at the\n   * start of the \"throttling\" process.\n   *\n   * If not provided, defaults to: `true`.\n   */\n  leading?: boolean;\n  /**\n   * If `true`, the resulting Observable will emit the last value from the source\n   * Observable at the **end** of the \"throttling\" process (when ending an internal\n   * timer that prevents other emissions from the source to pass through).\n   * If `false`, it will not emit the last value from the source Observable at the\n   * end of the \"throttling\" process.\n   *\n   * If not provided, defaults to: `false`.\n   */\n  trailing?: boolean;\n}\n\n/**\n * Emits a value from the source Observable, then ignores subsequent source\n * values for a duration determined by another Observable, then repeats this\n * process.\n *\n * <span class=\"informal\">It's like {@link throttleTime}, but the silencing\n * duration is determined by a second Observable.</span>\n *\n * ![](throttle.svg)\n *\n * `throttle` emits the source Observable values on the output Observable\n * when its internal timer is disabled, and ignores source values when the timer\n * is enabled. Initially, the timer is disabled. As soon as the first source\n * value arrives, it is forwarded to the output Observable, and then the timer\n * is enabled by calling the `durationSelector` function with the source value,\n * which returns the \"duration\" Observable. When the duration Observable emits a\n * value, the timer is disabled, and this process repeats for the\n * next source value.\n *\n * ## Example\n *\n * Emit clicks at a rate of at most one click per second\n *\n * ```ts\n * import { fromEvent, throttle, interval } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(throttle(() => interval(1000)));\n *\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link audit}\n * @see {@link debounce}\n * @see {@link delayWhen}\n * @see {@link sample}\n * @see {@link throttleTime}\n *\n * @param durationSelector A function that receives a value from the source\n * Observable, for computing the silencing duration for each source value,\n * returned as an `ObservableInput`.\n * @param config A configuration object to define `leading` and `trailing`\n * behavior. Defaults to `{ leading: true, trailing: false }`.\n * @return A function that returns an Observable that performs the throttle\n * operation to limit the rate of emissions from the source.\n */\nexport function throttle<T>(durationSelector: (value: T) => ObservableInput<any>, config?: ThrottleConfig): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    const { leading = true, trailing = false } = config ?? {};\n    let hasValue = false;\n    let sendValue: T | null = null;\n    let throttled: Subscription | null = null;\n    let isComplete = false;\n\n    const endThrottling = () => {\n      throttled?.unsubscribe();\n      throttled = null;\n      if (trailing) {\n        send();\n        isComplete && subscriber.complete();\n      }\n    };\n\n    const cleanupThrottling = () => {\n      throttled = null;\n      isComplete && subscriber.complete();\n    };\n\n    const startThrottle = (value: T) =>\n      (throttled = innerFrom(durationSelector(value)).subscribe(createOperatorSubscriber(subscriber, endThrottling, cleanupThrottling)));\n\n    const send = () => {\n      if (hasValue) {\n        // Ensure we clear out our value and hasValue flag\n        // before we emit, otherwise reentrant code can cause\n        // issues here.\n        hasValue = false;\n        const value = sendValue!;\n        sendValue = null;\n        // Emit the value.\n        subscriber.next(value);\n        !isComplete && startThrottle(value);\n      }\n    };\n\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        // Regarding the presence of throttled.closed in the following\n        // conditions, if a synchronous duration selector is specified - weird,\n        // but legal - an already-closed subscription will be assigned to\n        // throttled, so the subscription's closed property needs to be checked,\n        // too.\n        (value) => {\n          hasValue = true;\n          sendValue = value;\n          !(throttled && !throttled.closed) && (leading ? send() : startThrottle(value));\n        },\n        () => {\n          isComplete = true;\n          !(trailing && hasValue && throttled && !throttled.closed) && subscriber.complete();\n        }\n      )\n    );\n  });\n}\n", "import { asyncScheduler } from '../scheduler/async';\nimport { throttle, ThrottleConfig } from './throttle';\nimport { MonoTypeOperatorFunction, SchedulerLike } from '../types';\nimport { timer } from '../observable/timer';\n\n/**\n * Emits a value from the source Observable, then ignores subsequent source\n * values for `duration` milliseconds, then repeats this process.\n *\n * <span class=\"informal\">Lets a value pass, then ignores source values for the\n * next `duration` milliseconds.</span>\n *\n * ![](throttleTime.png)\n *\n * `throttleTime` emits the source Observable values on the output Observable\n * when its internal timer is disabled, and ignores source values when the timer\n * is enabled. Initially, the timer is disabled. As soon as the first source\n * value arrives, it is forwarded to the output Observable, and then the timer\n * is enabled. After `duration` milliseconds (or the time unit determined\n * internally by the optional `scheduler`) has passed, the timer is disabled,\n * and this process repeats for the next source value. Optionally takes a\n * {@link SchedulerLike} for managing timers.\n *\n * ## Examples\n *\n * ### Limit click rate\n *\n * Emit clicks at a rate of at most one click per second\n *\n * ```ts\n * import { fromEvent, throttleTime } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(throttleTime(1000));\n *\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link auditTime}\n * @see {@link debounceTime}\n * @see {@link delay}\n * @see {@link sampleTime}\n * @see {@link throttle}\n *\n * @param duration Time to wait before emitting another value after\n * emitting the last value, measured in milliseconds or the time unit determined\n * internally by the optional `scheduler`.\n * @param scheduler The {@link SchedulerLike} to use for\n * managing the timers that handle the throttling. Defaults to {@link asyncScheduler}.\n * @param config A configuration object to define `leading` and\n * `trailing` behavior. Defaults to `{ leading: true, trailing: false }`.\n * @return A function that returns an Observable that performs the throttle\n * operation to limit the rate of emissions from the source.\n */\nexport function throttleTime<T>(\n  duration: number,\n  scheduler: SchedulerLike = asyncScheduler,\n  config?: ThrottleConfig\n): MonoTypeOperatorFunction<T> {\n  const duration$ = timer(duration, scheduler);\n  return throttle(() => duration$, config);\n}\n", "import { OperatorFunction, ObservableInputTuple } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { innerFrom } from '../observable/innerFrom';\nimport { identity } from '../util/identity';\nimport { noop } from '../util/noop';\nimport { popResultSelector } from '../util/args';\n\nexport function withLatestFrom<T, O extends unknown[]>(...inputs: [...ObservableInputTuple<O>]): OperatorFunction<T, [T, ...O]>;\n\nexport function withLatestFrom<T, O extends unknown[], R>(\n  ...inputs: [...ObservableInputTuple<O>, (...value: [T, ...O]) => R]\n): OperatorFunction<T, R>;\n\n/**\n * Combines the source Observable with other Observables to create an Observable\n * whose values are calculated from the latest values of each, only when the\n * source emits.\n *\n * <span class=\"informal\">Whenever the source Observable emits a value, it\n * computes a formula using that value plus the latest values from other input\n * Observables, then emits the output of that formula.</span>\n *\n * ![](withLatestFrom.png)\n *\n * `withLatestFrom` combines each value from the source Observable (the\n * instance) with the latest values from the other input Observables only when\n * the source emits a value, optionally using a `project` function to determine\n * the value to be emitted on the output Observable. All input Observables must\n * emit at least one value before the output Observable will emit a value.\n *\n * ## Example\n *\n * On every click event, emit an array with the latest timer event plus the click event\n *\n * ```ts\n * import { fromEvent, interval, withLatestFrom } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const timer = interval(1000);\n * const result = clicks.pipe(withLatestFrom(timer));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link combineLatest}\n *\n * @param {ObservableInput} other An input Observable to combine with the source\n * Observable. More than one input Observables may be given as argument.\n * @param {Function} [project] Projection function for combining values\n * together. Receives all values in order of the Observables passed, where the\n * first parameter is a value from the source Observable. (e.g.\n * `a.pipe(withLatestFrom(b, c), map(([a1, b1, c1]) => a1 + b1 + c1))`). If this is not\n * passed, arrays will be emitted on the output Observable.\n * @return A function that returns an Observable of projected values from the\n * most recent values from each input Observable, or an array of the most\n * recent values from each input Observable.\n */\nexport function withLatestFrom<T, R>(...inputs: any[]): OperatorFunction<T, R | any[]> {\n  const project = popResultSelector(inputs) as ((...args: any[]) => R) | undefined;\n\n  return operate((source, subscriber) => {\n    const len = inputs.length;\n    const otherValues = new Array(len);\n    // An array of whether or not the other sources have emitted. Matched with them by index.\n    // TODO: At somepoint, we should investigate the performance implications here, and look\n    // into using a `Set()` and checking the `size` to see if we're ready.\n    let hasValue = inputs.map(() => false);\n    // Flipped true when we have at least one value from all other sources and\n    // we are ready to start emitting values.\n    let ready = false;\n\n    // Other sources. Note that here we are not checking `subscriber.closed`,\n    // this causes all inputs to be subscribed to, even if nothing can be emitted\n    // from them. This is an important distinction because subscription constitutes\n    // a side-effect.\n    for (let i = 0; i < len; i++) {\n      innerFrom(inputs[i]).subscribe(\n        createOperatorSubscriber(\n          subscriber,\n          (value) => {\n            otherValues[i] = value;\n            if (!ready && !hasValue[i]) {\n              // If we're not ready yet, flag to show this observable has emitted.\n              hasValue[i] = true;\n              // Intentionally terse code.\n              // If all of our other observables have emitted, set `ready` to `true`,\n              // so we know we can start emitting values, then clean up the `hasValue` array,\n              // because we don't need it anymore.\n              (ready = hasValue.every(identity)) && (hasValue = null!);\n            }\n          },\n          // Completing one of the other sources has\n          // no bearing on the completion of our result.\n          noop\n        )\n      );\n    }\n\n    // Source subscription\n    source.subscribe(\n      createOperatorSubscriber(subscriber, (value) => {\n        if (ready) {\n          // We have at least one value from the other sources. Go ahead and emit.\n          const values = [value, ...otherValues];\n          subscriber.next(project ? project(...values) : values);\n        }\n      })\n    );\n  });\n}\n", "import { zip as zipStatic } from '../observable/zip';\nimport { ObservableInput, ObservableInputTuple, OperatorFunction, Cons } from '../types';\nimport { operate } from '../util/lift';\n\n/** @deprecated Replaced with {@link zipWith}. Will be removed in v8. */\nexport function zip<T, A extends readonly unknown[]>(otherInputs: [...ObservableInputTuple<A>]): OperatorFunction<T, Cons<T, A>>;\n/** @deprecated Replaced with {@link zipWith}. Will be removed in v8. */\nexport function zip<T, A extends readonly unknown[], R>(\n  otherInputsAndProject: [...ObservableInputTuple<A>],\n  project: (...values: Cons<T, A>) => R\n): OperatorFunction<T, R>;\n/** @deprecated Replaced with {@link zipWith}. Will be removed in v8. */\nexport function zip<T, A extends readonly unknown[]>(...otherInputs: [...ObservableInputTuple<A>]): OperatorFunction<T, Cons<T, A>>;\n/** @deprecated Replaced with {@link zipWith}. Will be removed in v8. */\nexport function zip<T, A extends readonly unknown[], R>(\n  ...otherInputsAndProject: [...ObservableInputTuple<A>, (...values: Cons<T, A>) => R]\n): OperatorFunction<T, R>;\n\n/**\n * @deprecated Replaced with {@link zipWith}. Will be removed in v8.\n */\nexport function zip<T, R>(...sources: Array<ObservableInput<any> | ((...values: Array<any>) => R)>): OperatorFunction<T, any> {\n  return operate((source, subscriber) => {\n    zipStatic(source as ObservableInput<any>, ...(sources as Array<ObservableInput<any>>)).subscribe(subscriber);\n  });\n}\n", "import { ObservableInputTuple, OperatorFunction, Cons } from '../types';\nimport { zip } from './zip';\n\n/**\n * Subscribes to the source, and the observable inputs provided as arguments, and combines their values, by index, into arrays.\n *\n * What is meant by \"combine by index\": The first value from each will be made into a single array, then emitted,\n * then the second value from each will be combined into a single array and emitted, then the third value\n * from each will be combined into a single array and emitted, and so on.\n *\n * This will continue until it is no longer able to combine values of the same index into an array.\n *\n * After the last value from any one completed source is emitted in an array, the resulting observable will complete,\n * as there is no way to continue \"zipping\" values together by index.\n *\n * Use-cases for this operator are limited. There are memory concerns if one of the streams is emitting\n * values at a much faster rate than the others. Usage should likely be limited to streams that emit\n * at a similar pace, or finite streams of known length.\n *\n * In many cases, authors want `combineLatestWith` and not `zipWith`.\n *\n * @param otherInputs other observable inputs to collate values from.\n * @return A function that returns an Observable that emits items by index\n * combined from the source Observable and provided Observables, in form of an\n * array.\n */\nexport function zipWith<T, A extends readonly unknown[]>(...otherInputs: [...ObservableInputTuple<A>]): OperatorFunction<T, Cons<T, A>> {\n  return zip(...otherInputs);\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  ReplaySubject,\n  Subject,\n  fromEvent\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch document\n *\n * Documents are implemented as subjects, so all downstream observables are\n * automatically updated when a new document is emitted.\n *\n * @returns Document subject\n */\nexport function watchDocument(): Subject<Document> {\n  const document$ = new ReplaySubject<Document>(1)\n  fromEvent(document, \"DOMContentLoaded\", { once: true })\n    .subscribe(() => document$.next(document))\n\n  /* Return document */\n  return document$\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve all elements matching the query selector\n *\n * @template T - Element type\n *\n * @param selector - Query selector\n * @param node - Node of reference\n *\n * @returns Elements\n */\nexport function getElements<T extends keyof HTMLElementTagNameMap>(\n  selector: T, node?: ParentNode\n): HTMLElementTagNameMap[T][]\n\nexport function getElements<T extends HTMLElement>(\n  selector: string, node?: ParentNode\n): T[]\n\nexport function getElements<T extends HTMLElement>(\n  selector: string, node: ParentNode = document\n): T[] {\n  return Array.from(node.querySelectorAll<T>(selector))\n}\n\n/**\n * Retrieve an element matching a query selector or throw a reference error\n *\n * Note that this function assumes that the element is present. If unsure if an\n * element is existent, use the `getOptionalElement` function instead.\n *\n * @template T - Element type\n *\n * @param selector - Query selector\n * @param node - Node of reference\n *\n * @returns Element\n */\nexport function getElement<T extends keyof HTMLElementTagNameMap>(\n  selector: T, node?: ParentNode\n): HTMLElementTagNameMap[T]\n\nexport function getElement<T extends HTMLElement>(\n  selector: string, node?: ParentNode\n): T\n\nexport function getElement<T extends HTMLElement>(\n  selector: string, node: ParentNode = document\n): T {\n  const el = getOptionalElement<T>(selector, node)\n  if (typeof el === \"undefined\")\n    throw new ReferenceError(\n      `Missing element: expected \"${selector}\" to be present`\n    )\n\n  /* Return element */\n  return el\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Retrieve an optional element matching the query selector\n *\n * @template T - Element type\n *\n * @param selector - Query selector\n * @param node - Node of reference\n *\n * @returns Element or nothing\n */\nexport function getOptionalElement<T extends keyof HTMLElementTagNameMap>(\n  selector: T, node?: ParentNode\n): HTMLElementTagNameMap[T] | undefined\n\nexport function getOptionalElement<T extends HTMLElement>(\n  selector: string, node?: ParentNode\n): T | undefined\n\nexport function getOptionalElement<T extends HTMLElement>(\n  selector: string, node: ParentNode = document\n): T | undefined {\n  return node.querySelector<T>(selector) || undefined\n}\n\n/**\n * Retrieve the currently active element\n *\n * @returns Element or nothing\n */\nexport function getActiveElement(): HTMLElement | undefined {\n  return (\n    document.activeElement?.shadowRoot?.activeElement as HTMLElement ??\n    document.activeElement as HTMLElement ??\n    undefined\n  )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  debounceTime,\n  distinctUntilChanged,\n  fromEvent,\n  map,\n  merge,\n  shareReplay,\n  startWith\n} from \"rxjs\"\n\nimport { getActiveElement } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Focus observable\n *\n * Previously, this observer used `focus` and `blur` events to determine whether\n * an element is focused, but this doesn't work if there are focusable elements\n * within the elements itself. A better solutions are `focusin` and `focusout`\n * events, which bubble up the tree and allow for more fine-grained control.\n *\n * `debounceTime` is necessary, because when a focus change happens inside an\n * element, the observable would first emit `false` and then `true` again.\n */\nconst observer$ = merge(\n  fromEvent(document.body, \"focusin\"),\n  fromEvent(document.body, \"focusout\")\n)\n  .pipe(\n    debounceTime(1),\n    startWith(undefined),\n    map(() => getActiveElement() || document.body),\n    shareReplay(1)\n  )\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch element focus\n *\n * @param el - Element\n *\n * @returns Element focus observable\n */\nexport function watchElementFocus(\n  el: HTMLElement\n): Observable<boolean> {\n  return observer$\n    .pipe(\n      map(active => el.contains(active)),\n      distinctUntilChanged()\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  debounceTime,\n  fromEvent,\n  identity,\n  map,\n  merge,\n  startWith\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch element hover\n *\n * @param el - Element\n * @param duration - Debounce duration\n *\n * @returns Element hover observable\n */\nexport function watchElementHover(\n  el: HTMLElement, duration?: number\n): Observable<boolean> {\n  return merge(\n    fromEvent(el, \"mouseenter\").pipe(map(() => true)),\n    fromEvent(el, \"mouseleave\").pipe(map(() => false))\n  )\n    .pipe(\n      duration ? debounceTime(duration) : identity,\n      startWith(false)\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  animationFrameScheduler,\n  auditTime,\n  fromEvent,\n  map,\n  merge,\n  startWith\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Element offset\n */\nexport interface ElementOffset {\n  x: number                            /* Horizontal offset */\n  y: number                            /* Vertical offset */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve element offset\n *\n * @param el - Element\n *\n * @returns Element offset\n */\nexport function getElementOffset(\n  el: HTMLElement\n): ElementOffset {\n  return {\n    x: el.offsetLeft,\n    y: el.offsetTop\n  }\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch element offset\n *\n * @param el - Element\n *\n * @returns Element offset observable\n */\nexport function watchElementOffset(\n  el: HTMLElement\n): Observable<ElementOffset> {\n  return merge(\n    fromEvent(window, \"load\"),\n    fromEvent(window, \"resize\")\n  )\n    .pipe(\n      auditTime(0, animationFrameScheduler),\n      map(() => getElementOffset(el)),\n      startWith(getElementOffset(el))\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  animationFrameScheduler,\n  auditTime,\n  fromEvent,\n  map,\n  merge,\n  startWith\n} from \"rxjs\"\n\nimport { ElementOffset } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve element content offset (= scroll offset)\n *\n * @param el - Element\n *\n * @returns Element content offset\n */\nexport function getElementContentOffset(\n  el: HTMLElement\n): ElementOffset {\n  return {\n    x: el.scrollLeft,\n    y: el.scrollTop\n  }\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch element content offset\n *\n * @param el - Element\n *\n * @returns Element content offset observable\n */\nexport function watchElementContentOffset(\n  el: HTMLElement\n): Observable<ElementOffset> {\n  return merge(\n    fromEvent(el, \"scroll\"),\n    fromEvent(window, \"resize\")\n  )\n    .pipe(\n      auditTime(0, animationFrameScheduler),\n      map(() => getElementContentOffset(el)),\n      startWith(getElementContentOffset(el))\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { JSX as JSXInternal } from \"preact\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * HTML attributes\n */\ntype Attributes =\n  & JSXInternal.HTMLAttributes\n  & JSXInternal.SVGAttributes\n  & Record<string, any>\n\n/**\n * Child element\n */\ntype Child =\n  | ChildNode\n  | HTMLElement\n  | Text\n  | string\n  | number\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Append a child node to an element\n *\n * @param el - Element\n * @param child - Child node(s)\n */\nfunction appendChild(el: HTMLElement, child: Child | Child[]): void {\n\n  /* Handle primitive types (including raw HTML) */\n  if (typeof child === \"string\" || typeof child === \"number\") {\n    el.innerHTML += child.toString()\n\n  /* Handle nodes */\n  } else if (child instanceof Node) {\n    el.appendChild(child)\n\n  /* Handle nested children */\n  } else if (Array.isArray(child)) {\n    for (const node of child)\n      appendChild(el, node)\n  }\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * JSX factory\n *\n * @template T - Element type\n *\n * @param tag - HTML tag\n * @param attributes - HTML attributes\n * @param children - Child elements\n *\n * @returns Element\n */\nexport function h<T extends keyof HTMLElementTagNameMap>(\n  tag: T, attributes?: Attributes | null, ...children: Child[]\n): HTMLElementTagNameMap[T]\n\nexport function h<T extends h.JSX.Element>(\n  tag: string, attributes?: Attributes | null, ...children: Child[]\n): T\n\nexport function h<T extends h.JSX.Element>(\n  tag: string, attributes?: Attributes | null, ...children: Child[]\n): T {\n  const el = document.createElement(tag)\n\n  /* Set attributes, if any */\n  if (attributes)\n    for (const attr of Object.keys(attributes)) {\n      if (typeof attributes[attr] === \"undefined\")\n        continue\n\n      /* Set default attribute or boolean */\n      if (typeof attributes[attr] !== \"boolean\")\n        el.setAttribute(attr, attributes[attr])\n      else\n        el.setAttribute(attr, \"\")\n    }\n\n  /* Append child nodes */\n  for (const child of children)\n    appendChild(el, child)\n\n  /* Return element */\n  return el as T\n}\n\n/* ----------------------------------------------------------------------------\n * Namespace\n * ------------------------------------------------------------------------- */\n\nexport declare namespace h {\n  namespace JSX {\n    type Element = HTMLElement\n    type IntrinsicElements = JSXInternal.IntrinsicElements\n  }\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Round a number for display with repository facts\n *\n * This is a reverse-engineered version of GitHub's weird rounding algorithm\n * for stars, forks and all other numbers. While all numbers below `1,000` are\n * returned as-is, bigger numbers are converted to fixed numbers:\n *\n * - `1,049` => `1k`\n * - `1,050` => `1.1k`\n * - `1,949` => `1.9k`\n * - `1,950` => `2k`\n *\n * @param value - Original value\n *\n * @returns Rounded value\n */\nexport function round(value: number): string {\n  if (value > 999) {\n    const digits = +((value - 950) % 1000 > 99)\n    return `${((value + 0.000001) / 1000).toFixed(digits)}k`\n  } else {\n    return value.toString()\n  }\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  defer,\n  finalize,\n  fromEvent,\n  map,\n  merge,\n  switchMap,\n  take,\n  throwError\n} from \"rxjs\"\n\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Create and load a `script` element\n *\n * This function returns an observable that will emit when the script was\n * successfully loaded, or throw an error if it wasn't.\n *\n * @param src - Script URL\n *\n * @returns Script observable\n */\nexport function watchScript(src: string): Observable<void> {\n  const script = h(\"script\", { src })\n  return defer(() => {\n    document.head.appendChild(script)\n    return merge(\n      fromEvent(script, \"load\"),\n      fromEvent(script, \"error\")\n        .pipe(\n          switchMap(() => (\n            throwError(() => new ReferenceError(`Invalid script: ${src}`))\n          ))\n        )\n    )\n      .pipe(\n        map(() => undefined),\n        finalize(() => document.head.removeChild(script)),\n        take(1)\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  NEVER,\n  Observable,\n  Subject,\n  defer,\n  filter,\n  finalize,\n  map,\n  merge,\n  of,\n  shareReplay,\n  startWith,\n  switchMap,\n  tap\n} from \"rxjs\"\n\nimport { watchScript } from \"../../../script\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Element offset\n */\nexport interface ElementSize {\n  width: number                        /* Element width */\n  height: number                       /* Element height */\n}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Resize observer entry subject\n */\nconst entry$ = new Subject<ResizeObserverEntry>()\n\n/**\n * Resize observer observable\n *\n * This observable will create a `ResizeObserver` on the first subscription\n * and will automatically terminate it when there are no more subscribers.\n * It's quite important to centralize observation in a single `ResizeObserver`,\n * as the performance difference can be quite dramatic, as the link shows.\n *\n * If the browser doesn't have a `ResizeObserver` implementation available, a\n * polyfill is automatically downloaded from unpkg.com. This is also compatible\n * with the built-in privacy plugin, which will download the polyfill and put\n * it alongside the built site for self-hosting.\n *\n * @see https://bit.ly/3iIYfEm - Google Groups on performance\n */\nconst observer$ = defer(() => (\n  typeof ResizeObserver === \"undefined\"\n    ? watchScript(\"https://unpkg.com/resize-observer-polyfill\")\n    : of(undefined)\n))\n  .pipe(\n    map(() => new ResizeObserver(entries => {\n      for (const entry of entries)\n        entry$.next(entry)\n    })),\n    switchMap(observer => merge(NEVER, of(observer))\n      .pipe(\n        finalize(() => observer.disconnect())\n      )\n    ),\n    shareReplay(1)\n  )\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve element size\n *\n * @param el - Element\n *\n * @returns Element size\n */\nexport function getElementSize(\n  el: HTMLElement\n): ElementSize {\n  return {\n    width:  el.offsetWidth,\n    height: el.offsetHeight\n  }\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch element size\n *\n * This function returns an observable that subscribes to a single internal\n * instance of `ResizeObserver` upon subscription, and emit resize events until\n * termination. Note that this function should not be called with the same\n * element twice, as the first unsubscription will terminate observation.\n *\n * Sadly, we can't use the `DOMRect` objects returned by the observer, because\n * we need the emitted values to be consistent with `getElementSize`, which will\n * return the used values (rounded) and not actual values (unrounded). Thus, we\n * use the `offset*` properties. See the linked GitHub issue.\n *\n * @see https://bit.ly/3m0k3he - GitHub issue\n *\n * @param el - Element\n *\n * @returns Element size observable\n */\nexport function watchElementSize(\n  el: HTMLElement\n): Observable<ElementSize> {\n  return observer$\n    .pipe(\n      tap(observer => observer.observe(el)),\n      switchMap(observer => entry$\n        .pipe(\n          filter(({ target }) => target === el),\n          finalize(() => observer.unobserve(el)),\n          map(() => getElementSize(el))\n        )\n      ),\n      startWith(getElementSize(el))\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { ElementSize } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve element content size (= scroll width and height)\n *\n * @param el - Element\n *\n * @returns Element content size\n */\nexport function getElementContentSize(\n  el: HTMLElement\n): ElementSize {\n  return {\n    width:  el.scrollWidth,\n    height: el.scrollHeight\n  }\n}\n\n/**\n * Retrieve the overflowing container of an element, if any\n *\n * @param el - Element\n *\n * @returns Overflowing container or nothing\n */\nexport function getElementContainer(\n  el: HTMLElement\n): HTMLElement | undefined {\n  let parent = el.parentElement\n  while (parent)\n    if (\n      el.scrollWidth  <= parent.scrollWidth &&\n      el.scrollHeight <= parent.scrollHeight\n    )\n      parent = (el = parent).parentElement\n    else\n      break\n\n  /* Return overflowing container */\n  return parent ? el : undefined\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  NEVER,\n  Observable,\n  Subject,\n  defer,\n  distinctUntilChanged,\n  filter,\n  finalize,\n  map,\n  merge,\n  of,\n  shareReplay,\n  switchMap,\n  tap\n} from \"rxjs\"\n\nimport {\n  getElementContentSize,\n  getElementSize,\n  watchElementContentOffset\n} from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Intersection observer entry subject\n */\nconst entry$ = new Subject<IntersectionObserverEntry>()\n\n/**\n * Intersection observer observable\n *\n * This observable will create an `IntersectionObserver` on first subscription\n * and will automatically terminate it when there are no more subscribers.\n *\n * @see https://bit.ly/3iIYfEm - Google Groups on performance\n */\nconst observer$ = defer(() => of(\n  new IntersectionObserver(entries => {\n    for (const entry of entries)\n      entry$.next(entry)\n  }, {\n    threshold: 0\n  })\n))\n  .pipe(\n    switchMap(observer => merge(NEVER, of(observer))\n      .pipe(\n        finalize(() => observer.disconnect())\n      )\n    ),\n    shareReplay(1)\n  )\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch element visibility\n *\n * @param el - Element\n *\n * @returns Element visibility observable\n */\nexport function watchElementVisibility(\n  el: HTMLElement\n): Observable<boolean> {\n  return observer$\n    .pipe(\n      tap(observer => observer.observe(el)),\n      switchMap(observer => entry$\n        .pipe(\n          filter(({ target }) => target === el),\n          finalize(() => observer.unobserve(el)),\n          map(({ isIntersecting }) => isIntersecting)\n        )\n      )\n    )\n}\n\n/**\n * Watch element boundary\n *\n * This function returns an observable which emits whether the bottom content\n * boundary (= scroll offset) of an element is within a certain threshold.\n *\n * @param el - Element\n * @param threshold - Threshold\n *\n * @returns Element boundary observable\n */\nexport function watchElementBoundary(\n  el: HTMLElement, threshold = 16\n): Observable<boolean> {\n  return watchElementContentOffset(el)\n    .pipe(\n      map(({ y }) => {\n        const visible = getElementSize(el)\n        const content = getElementContentSize(el)\n        return y >= (\n          content.height - visible.height - threshold\n        )\n      }),\n      distinctUntilChanged()\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  fromEvent,\n  map,\n  startWith\n} from \"rxjs\"\n\nimport { getElement } from \"../element\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Toggle\n */\nexport type Toggle =\n  | \"drawer\"                           /* Toggle for drawer */\n  | \"search\"                           /* Toggle for search */\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Toggle map\n */\nconst toggles: Record<Toggle, HTMLInputElement> = {\n  drawer: getElement(\"[data-md-toggle=drawer]\"),\n  search: getElement(\"[data-md-toggle=search]\")\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve the value of a toggle\n *\n * @param name - Toggle\n *\n * @returns Toggle value\n */\nexport function getToggle(name: Toggle): boolean {\n  return toggles[name].checked\n}\n\n/**\n * Set toggle\n *\n * Simulating a click event seems to be the most cross-browser compatible way\n * of changing the value while also emitting a `change` event. Before, Material\n * used `CustomEvent` to programmatically change the value of a toggle, but this\n * is a much simpler and cleaner solution which doesn't require a polyfill.\n *\n * @param name - Toggle\n * @param value - Toggle value\n */\nexport function setToggle(name: Toggle, value: boolean): void {\n  if (toggles[name].checked !== value)\n    toggles[name].click()\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch toggle\n *\n * @param name - Toggle\n *\n * @returns Toggle value observable\n */\nexport function watchToggle(name: Toggle): Observable<boolean> {\n  const el = toggles[name]\n  return fromEvent(el, \"change\")\n    .pipe(\n      map(() => el.checked),\n      startWith(el.checked)\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  filter,\n  fromEvent,\n  map,\n  merge,\n  share,\n  startWith,\n  switchMap\n} from \"rxjs\"\n\nimport { getActiveElement } from \"../element\"\nimport { getToggle } from \"../toggle\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Keyboard mode\n */\nexport type KeyboardMode =\n  | \"global\"                           /* Global */\n  | \"search\"                           /* Search is open */\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Keyboard\n */\nexport interface Keyboard {\n  mode: KeyboardMode                   /* Keyboard mode */\n  type: string                         /* Key type */\n  claim(): void                        /* Key claim */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Check whether an element may receive keyboard input\n *\n * @param el - Element\n * @param type - Key type\n *\n * @returns Test result\n */\nfunction isSusceptibleToKeyboard(\n  el: HTMLElement, type: string\n): boolean {\n  switch (el.constructor) {\n\n    /* Input elements */\n    case HTMLInputElement:\n      /* @ts-expect-error - omit unnecessary type cast */\n      if (el.type === \"radio\")\n        return /^Arrow/.test(type)\n      else\n        return true\n\n    /* Select element and textarea */\n    case HTMLSelectElement:\n    case HTMLTextAreaElement:\n      return true\n\n    /* Everything else */\n    default:\n      return el.isContentEditable\n  }\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch composition events\n *\n * @returns Composition observable\n */\nexport function watchComposition(): Observable<boolean> {\n  return merge(\n    fromEvent(window, \"compositionstart\").pipe(map(() => true)),\n    fromEvent(window, \"compositionend\").pipe(map(() => false))\n  )\n    .pipe(\n      startWith(false)\n    )\n}\n\n/**\n * Watch keyboard\n *\n * @returns Keyboard observable\n */\nexport function watchKeyboard(): Observable<Keyboard> {\n  const keyboard$ = fromEvent<KeyboardEvent>(window, \"keydown\")\n    .pipe(\n      filter(ev => !(ev.metaKey || ev.ctrlKey)),\n      map(ev => ({\n        mode: getToggle(\"search\") ? \"search\" : \"global\",\n        type: ev.key,\n        claim() {\n          ev.preventDefault()\n          ev.stopPropagation()\n        }\n      } as Keyboard)),\n      filter(({ mode, type }) => {\n        if (mode === \"global\") {\n          const active = getActiveElement()\n          if (typeof active !== \"undefined\")\n            return !isSusceptibleToKeyboard(active, type)\n        }\n        return true\n      }),\n      share()\n    )\n\n  /* Don't emit during composition events - see https://bit.ly/3te3Wl8 */\n  return watchComposition()\n    .pipe(\n      switchMap(active => !active ? keyboard$ : EMPTY)\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { Subject } from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve location\n *\n * This function returns a `URL` object (and not `Location`) to normalize the\n * typings across the application. Furthermore, locations need to be tracked\n * without setting them and `Location` is a singleton which represents the\n * current location.\n *\n * @returns URL\n */\nexport function getLocation(): URL {\n  return new URL(location.href)\n}\n\n/**\n * Set location\n *\n * If instant navigation is enabled, this function creates a temporary anchor\n * element, sets the `href` attribute, appends it to the body, clicks it, and\n * then removes it again. The event will bubble up the DOM and trigger be\n * intercepted by the instant loading business logic.\n *\n * Note that we must append and remove the anchor element, or the event will\n * not bubble up the DOM, making it impossible to intercept it.\n *\n * @param url - URL to navigate to\n * @param navigate - Force navigation\n */\nexport function setLocation(\n  url: URL | HTMLLinkElement, navigate = false\n): void {\n  if (feature(\"navigation.instant\") && !navigate) {\n    const el = h(\"a\", { href: url.href })\n    document.body.appendChild(el)\n    el.click()\n    el.remove()\n\n  // If we're not using instant navigation, and the page should not be reloaded\n  // just instruct the browser to navigate to the given URL\n  } else {\n    location.href = url.href\n  }\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch location\n *\n * @returns Location subject\n */\nexport function watchLocation(): Subject<URL> {\n  return new Subject<URL>()\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  filter,\n  fromEvent,\n  map,\n  merge,\n  shareReplay,\n  startWith\n} from \"rxjs\"\n\nimport { getOptionalElement } from \"~/browser\"\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve location hash\n *\n * @returns Location hash\n */\nexport function getLocationHash(): string {\n  return location.hash.slice(1)\n}\n\n/**\n * Set location hash\n *\n * Setting a new fragment identifier via `location.hash` will have no effect\n * if the value doesn't change. When a new fragment identifier is set, we want\n * the browser to target the respective element at all times, which is why we\n * use this dirty little trick.\n *\n * @param hash - Location hash\n */\nexport function setLocationHash(hash: string): void {\n  const el = h(\"a\", { href: hash })\n  el.addEventListener(\"click\", ev => ev.stopPropagation())\n  el.click()\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch location hash\n *\n * @param location$ - Location observable\n *\n * @returns Location hash observable\n */\nexport function watchLocationHash(\n  location$: Observable<URL>\n): Observable<string> {\n  return merge(\n    fromEvent<HashChangeEvent>(window, \"hashchange\"),\n    location$\n  )\n    .pipe(\n      map(getLocationHash),\n      startWith(getLocationHash()),\n      filter(hash => hash.length > 0),\n      shareReplay(1)\n    )\n}\n\n/**\n * Watch location target\n *\n * @param location$ - Location observable\n *\n * @returns Location target observable\n */\nexport function watchLocationTarget(\n  location$: Observable<URL>\n): Observable<HTMLElement> {\n  return watchLocationHash(location$)\n    .pipe(\n      map(id => getOptionalElement(`[id=\"${id}\"]`)!),\n      filter(el => typeof el !== \"undefined\")\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  fromEvent,\n  fromEventPattern,\n  map,\n  merge,\n  startWith,\n  switchMap\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch media query\n *\n * Note that although `MediaQueryList.addListener` is deprecated we have to\n * use it, because it's the only way to ensure proper downward compatibility.\n *\n * @see https://bit.ly/3dUBH2m - GitHub issue\n *\n * @param query - Media query\n *\n * @returns Media observable\n */\nexport function watchMedia(query: string): Observable<boolean> {\n  const media = matchMedia(query)\n  return fromEventPattern<boolean>(next => (\n    media.addListener(() => next(media.matches))\n  ))\n    .pipe(\n      startWith(media.matches)\n    )\n}\n\n/**\n * Watch print mode\n *\n * @returns Print observable\n */\nexport function watchPrint(): Observable<boolean> {\n  const media = matchMedia(\"print\")\n  return merge(\n    fromEvent(window, \"beforeprint\").pipe(map(() => true)),\n    fromEvent(window, \"afterprint\").pipe(map(() => false))\n  )\n    .pipe(\n      startWith(media.matches)\n    )\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Toggle an observable with a media observable\n *\n * @template T - Data type\n *\n * @param query$ - Media observable\n * @param factory - Observable factory\n *\n * @returns Toggled observable\n */\nexport function at<T>(\n  query$: Observable<boolean>, factory: () => Observable<T>\n): Observable<T> {\n  return query$\n    .pipe(\n      switchMap(active => active ? factory() : EMPTY)\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  map,\n  shareReplay,\n  switchMap\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Options\n */\ninterface Options {\n  progress$?: Subject<number>          // Progress subject\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch the given URL\n *\n * This function returns an observable that emits the response as a blob and\n * completes, or emits an error if the request failed. The caller can cancel\n * the request by unsubscribing at any time, which will automatically abort\n * the inflight request and complete the observable.\n *\n * Note that we use `XMLHTTPRequest` not because we're nostalgic, but because\n * it's the only way to get progress events for downloads and also allow for\n * cancellation of requests, as the official Fetch API does not support this\n * yet, even though we're already in 2024.\n *\n * @param url - Request URL\n * @param options - Options\n *\n * @returns Data observable\n */\nexport function request(\n  url: URL | string, options?: Options\n): Observable<Blob> {\n  return new Observable<Blob>(observer => {\n    const req = new XMLHttpRequest()\n    req.open(\"GET\", `${url}`)\n    req.responseType = \"blob\"\n\n    // Handle response\n    req.addEventListener(\"load\", () => {\n      if (req.status >= 200 && req.status < 300) {\n        observer.next(req.response)\n        observer.complete()\n\n      // Every response that is not in the 2xx range is considered an error\n      } else {\n        observer.error(new Error(req.statusText))\n      }\n    })\n\n    // Handle network errors\n    req.addEventListener(\"error\", () => {\n      observer.error(new Error(\"Network error\"))\n    })\n\n    // Handle aborted requests\n    req.addEventListener(\"abort\", () => {\n      observer.complete()\n    })\n\n    // Handle download progress\n    if (typeof options?.progress$ !== \"undefined\") {\n      req.addEventListener(\"progress\", event => {\n        if (event.lengthComputable) {\n          options.progress$!.next((event.loaded / event.total) * 100)\n\n        // Hack: Chromium doesn't report the total number of bytes if content\n        // is compressed, so we need this fallback - see https://t.ly/ZXofI\n        } else {\n          const length = req.getResponseHeader(\"Content-Length\") ?? 0\n          options.progress$!.next((event.loaded / +length) * 100)\n        }\n      })\n\n      // Immediately set progress to 5% to indicate that we're loading\n      options.progress$.next(5)\n    }\n\n    // Send request and automatically abort request upon unsubscription\n    req.send()\n    return () => req.abort()\n  })\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Fetch JSON from the given URL\n *\n * @template T - Data type\n *\n * @param url - Request URL\n * @param options - Options\n *\n * @returns Data observable\n */\nexport function requestJSON<T>(\n  url: URL | string, options?: Options\n): Observable<T> {\n  return request(url, options)\n    .pipe(\n      switchMap(res => res.text()),\n      map(body => JSON.parse(body) as T),\n      shareReplay(1)\n    )\n}\n\n/**\n * Fetch HTML from the given URL\n *\n * @param url - Request URL\n * @param options - Options\n *\n * @returns Data observable\n */\nexport function requestHTML(\n  url: URL | string, options?: Options\n): Observable<Document> {\n  const dom = new DOMParser()\n  return request(url, options)\n    .pipe(\n      switchMap(res => res.text()),\n      map(res => dom.parseFromString(res, \"text/html\")),\n      shareReplay(1)\n    )\n}\n\n/**\n * Fetch XML from the given URL\n *\n * @param url - Request URL\n * @param options - Options\n *\n * @returns Data observable\n */\nexport function requestXML(\n  url: URL | string, options?: Options\n): Observable<Document> {\n  const dom = new DOMParser()\n  return request(url, options)\n    .pipe(\n      switchMap(res => res.text()),\n      map(res => dom.parseFromString(res, \"text/xml\")),\n      shareReplay(1)\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  fromEvent,\n  map,\n  merge,\n  startWith\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Viewport offset\n */\nexport interface ViewportOffset {\n  x: number                            /* Horizontal offset */\n  y: number                            /* Vertical offset */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve viewport offset\n *\n * On iOS Safari, viewport offset can be negative due to overflow scrolling.\n * As this may induce strange behaviors downstream, we'll just limit it to 0.\n *\n * @returns Viewport offset\n */\nexport function getViewportOffset(): ViewportOffset {\n  return {\n    x: Math.max(0, scrollX),\n    y: Math.max(0, scrollY)\n  }\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch viewport offset\n *\n * @returns Viewport offset observable\n */\nexport function watchViewportOffset(): Observable<ViewportOffset> {\n  return merge(\n    fromEvent(window, \"scroll\", { passive: true }),\n    fromEvent(window, \"resize\", { passive: true })\n  )\n    .pipe(\n      map(getViewportOffset),\n      startWith(getViewportOffset())\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  fromEvent,\n  map,\n  startWith\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Viewport size\n */\nexport interface ViewportSize {\n  width: number                        /* Viewport width */\n  height: number                       /* Viewport height */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve viewport size\n *\n * @returns Viewport size\n */\nexport function getViewportSize(): ViewportSize {\n  return {\n    width:  innerWidth,\n    height: innerHeight\n  }\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch viewport size\n *\n * @returns Viewport size observable\n */\nexport function watchViewportSize(): Observable<ViewportSize> {\n  return fromEvent(window, \"resize\", { passive: true })\n    .pipe(\n      map(getViewportSize),\n      startWith(getViewportSize())\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  combineLatest,\n  map,\n  shareReplay\n} from \"rxjs\"\n\nimport {\n  ViewportOffset,\n  watchViewportOffset\n} from \"../offset\"\nimport {\n  ViewportSize,\n  watchViewportSize\n} from \"../size\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Viewport\n */\nexport interface Viewport {\n  offset: ViewportOffset               /* Viewport offset */\n  size: ViewportSize                   /* Viewport size */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch viewport\n *\n * @returns Viewport observable\n */\nexport function watchViewport(): Observable<Viewport> {\n  return combineLatest([\n    watchViewportOffset(),\n    watchViewportSize()\n  ])\n    .pipe(\n      map(([offset, size]) => ({ offset, size })),\n      shareReplay(1)\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  combineLatest,\n  distinctUntilKeyChanged,\n  map\n} from \"rxjs\"\n\nimport { Header } from \"~/components\"\n\nimport { getElementOffset } from \"../../element\"\nimport { Viewport } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch viewport relative to element\n *\n * @param el - Element\n * @param options - Options\n *\n * @returns Viewport observable\n */\nexport function watchViewportAt(\n  el: HTMLElement, { viewport$, header$ }: WatchOptions\n): Observable<Viewport> {\n  const size$ = viewport$\n    .pipe(\n      distinctUntilKeyChanged(\"size\")\n    )\n\n  /* Compute element offset */\n  const offset$ = combineLatest([size$, header$])\n    .pipe(\n      map(() => getElementOffset(el))\n    )\n\n  /* Compute relative viewport, return hot observable */\n  return combineLatest([header$, viewport$, offset$])\n    .pipe(\n      map(([{ height }, { offset, size }, { x, y }]) => ({\n        offset: {\n          x: offset.x - x,\n          y: offset.y - y + height\n        },\n        size\n      }))\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  endWith,\n  fromEvent,\n  ignoreElements,\n  mergeWith,\n  share,\n  takeUntil\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Worker message\n */\nexport interface WorkerMessage {\n  type: unknown                        /* Message type */\n  data?: unknown                       /* Message data */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Create an observable for receiving from a web worker\n *\n * @template T - Data type\n *\n * @param worker - Web worker\n *\n * @returns Message observable\n */\nfunction recv<T>(worker: Worker): Observable<T> {\n  return fromEvent<MessageEvent<T>, T>(worker, \"message\", ev => ev.data)\n}\n\n/**\n * Create a subject for sending to a web worker\n *\n * @template T - Data type\n *\n * @param worker - Web worker\n *\n * @returns Message subject\n */\nfunction send<T>(worker: Worker): Subject<T> {\n  const send$ = new Subject<T>()\n  send$.subscribe(data => worker.postMessage(data))\n\n  /* Return message subject */\n  return send$\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Create a bidirectional communication channel to a web worker\n *\n * @template T - Data type\n *\n * @param url - Worker URL\n * @param worker - Worker\n *\n * @returns Worker subject\n */\nexport function watchWorker<T extends WorkerMessage>(\n  url: string, worker = new Worker(url)\n): Subject<T> {\n  const recv$ = recv<T>(worker)\n  const send$ = send<T>(worker)\n\n  /* Create worker subject and forward messages */\n  const worker$ = new Subject<T>()\n  worker$.subscribe(send$)\n\n  /* Return worker subject */\n  const done$ = send$.pipe(ignoreElements(), endWith(true))\n  return worker$\n    .pipe(\n      ignoreElements(),\n      mergeWith(recv$.pipe(takeUntil(done$))),\n      share()\n    ) as Subject<T>\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { getElement, getLocation } from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Feature flag\n */\nexport type Flag =\n  | \"announce.dismiss\"                 /* Dismissable announcement bar */\n  | \"content.code.annotate\"            /* Code annotations */\n  | \"content.code.copy\"                /* Code copy button */\n  | \"content.lazy\"                     /* Lazy content elements */\n  | \"content.tabs.link\"                /* Link content tabs */\n  | \"content.tooltips\"                 /* Tooltips */\n  | \"header.autohide\"                  /* Hide header */\n  | \"navigation.expand\"                /* Automatic expansion */\n  | \"navigation.indexes\"               /* Section pages */\n  | \"navigation.instant\"               /* Instant navigation */\n  | \"navigation.instant.progress\"      /* Instant navigation progress */\n  | \"navigation.sections\"              /* Section navigation */\n  | \"navigation.tabs\"                  /* Tabs navigation */\n  | \"navigation.tabs.sticky\"           /* Tabs navigation (sticky) */\n  | \"navigation.top\"                   /* Back-to-top button */\n  | \"navigation.tracking\"              /* Anchor tracking */\n  | \"search.highlight\"                 /* Search highlighting */\n  | \"search.share\"                     /* Search sharing */\n  | \"search.suggest\"                   /* Search suggestions */\n  | \"toc.follow\"                       /* Following table of contents */\n  | \"toc.integrate\"                    /* Integrated table of contents */\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Translation\n */\nexport type Translation =\n  | \"clipboard.copy\"                   /* Copy to clipboard */\n  | \"clipboard.copied\"                 /* Copied to clipboard */\n  | \"search.result.placeholder\"        /* Type to start searching */\n  | \"search.result.none\"               /* No matching documents */\n  | \"search.result.one\"                /* 1 matching document */\n  | \"search.result.other\"              /* # matching documents */\n  | \"search.result.more.one\"           /* 1 more on this page */\n  | \"search.result.more.other\"         /* # more on this page */\n  | \"search.result.term.missing\"       /* Missing */\n  | \"select.version\"                   /* Version selector */\n\n/**\n * Translations\n */\nexport type Translations =\n  Record<Translation, string>\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Versioning\n */\nexport interface Versioning {\n  provider: \"mike\"                     /* Version provider */\n  default?: string | string[]          /* Default version */\n}\n\n/**\n * Configuration\n */\nexport interface Config {\n  base: string                         /* Base URL */\n  features: Flag[]                     /* Feature flags */\n  translations: Translations           /* Translations */\n  search: string                       /* Search worker URL */\n  tags?: Record<string, string>        /* Tags mapping */\n  version?: Versioning                 /* Versioning */\n}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve global configuration and make base URL absolute\n */\nconst script = getElement(\"#__config\")\nconst config: Config = JSON.parse(script.textContent!)\nconfig.base = `${new URL(config.base, getLocation())}`\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve global configuration\n *\n * @returns Global configuration\n */\nexport function configuration(): Config {\n  return config\n}\n\n/**\n * Check whether a feature flag is enabled\n *\n * @param flag - Feature flag\n *\n * @returns Test result\n */\nexport function feature(flag: Flag): boolean {\n  return config.features.includes(flag)\n}\n\n/**\n * Retrieve the translation for the given key\n *\n * @param key - Key to be translated\n * @param value - Positional value, if any\n *\n * @returns Translation\n */\nexport function translation(\n  key: Translation, value?: string | number\n): string {\n  return typeof value !== \"undefined\"\n    ? config.translations[key].replace(\"#\", value.toString())\n    : config.translations[key]\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { getElement, getElements } from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Component type\n */\nexport type ComponentType =\n  | \"announce\"                         /* Announcement bar */\n  | \"container\"                        /* Container */\n  | \"consent\"                          /* Consent */\n  | \"content\"                          /* Content */\n  | \"dialog\"                           /* Dialog */\n  | \"header\"                           /* Header */\n  | \"header-title\"                     /* Header title */\n  | \"header-topic\"                     /* Header topic */\n  | \"main\"                             /* Main area */\n  | \"outdated\"                         /* Version warning */\n  | \"palette\"                          /* Color palette */\n  | \"progress\"                         /* Progress indicator */\n  | \"search\"                           /* Search */\n  | \"search-query\"                     /* Search input */\n  | \"search-result\"                    /* Search results */\n  | \"search-share\"                     /* Search sharing */\n  | \"search-suggest\"                   /* Search suggestions */\n  | \"sidebar\"                          /* Sidebar */\n  | \"skip\"                             /* Skip link */\n  | \"source\"                           /* Repository information */\n  | \"tabs\"                             /* Navigation tabs */\n  | \"toc\"                              /* Table of contents */\n  | \"top\"                              /* Back-to-top button */\n\n/**\n * Component\n *\n * @template T - Component type\n * @template U - Reference type\n */\nexport type Component<\n  T extends {} = {},\n  U extends HTMLElement = HTMLElement\n> =\n  T & {\n    ref: U                             /* Component reference */\n  }\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Component type map\n */\ninterface ComponentTypeMap {\n  \"announce\": HTMLElement              /* Announcement bar */\n  \"container\": HTMLElement             /* Container */\n  \"consent\": HTMLElement               /* Consent */\n  \"content\": HTMLElement               /* Content */\n  \"dialog\": HTMLElement                /* Dialog */\n  \"header\": HTMLElement                /* Header */\n  \"header-title\": HTMLElement          /* Header title */\n  \"header-topic\": HTMLElement          /* Header topic */\n  \"main\": HTMLElement                  /* Main area */\n  \"outdated\": HTMLElement              /* Version warning */\n  \"palette\": HTMLElement               /* Color palette */\n  \"progress\": HTMLElement              /* Progress indicator */\n  \"search\": HTMLElement                /* Search */\n  \"search-query\": HTMLInputElement     /* Search input */\n  \"search-result\": HTMLElement         /* Search results */\n  \"search-share\": HTMLAnchorElement    /* Search sharing */\n  \"search-suggest\": HTMLElement        /* Search suggestions */\n  \"sidebar\": HTMLElement               /* Sidebar */\n  \"skip\": HTMLAnchorElement            /* Skip link */\n  \"source\": HTMLAnchorElement          /* Repository information */\n  \"tabs\": HTMLElement                  /* Navigation tabs */\n  \"toc\": HTMLElement                   /* Table of contents */\n  \"top\": HTMLAnchorElement             /* Back-to-top button */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve the element for a given component or throw a reference error\n *\n * @template T - Component type\n *\n * @param type - Component type\n * @param node - Node of reference\n *\n * @returns Element\n */\nexport function getComponentElement<T extends ComponentType>(\n  type: T, node: ParentNode = document\n): ComponentTypeMap[T] {\n  return getElement(`[data-md-component=${type}]`, node)\n}\n\n/**\n * Retrieve all elements for a given component\n *\n * @template T - Component type\n *\n * @param type - Component type\n * @param node - Node of reference\n *\n * @returns Elements\n */\nexport function getComponentElements<T extends ComponentType>(\n  type: T, node: ParentNode = document\n): ComponentTypeMap[T][] {\n  return getElements(`[data-md-component=${type}]`, node)\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  defer,\n  finalize,\n  fromEvent,\n  map,\n  tap\n} from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport { getElement } from \"~/browser\"\n\nimport { Component } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Announcement bar\n */\nexport interface Announce {\n  hash: number                        /* Content hash */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch announcement bar\n *\n * @param el - Announcement bar element\n *\n * @returns Announcement bar observable\n */\nexport function watchAnnounce(\n  el: HTMLElement\n): Observable<Announce> {\n  const button = getElement(\".md-typeset > :first-child\", el)\n  return fromEvent(button, \"click\", { once: true })\n    .pipe(\n      map(() => getElement(\".md-typeset\", el)),\n      map(content => ({ hash: __md_hash(content.innerHTML) }))\n    )\n}\n\n/**\n * Mount announcement bar\n *\n * @param el - Announcement bar element\n *\n * @returns Announcement bar component observable\n */\nexport function mountAnnounce(\n  el: HTMLElement\n): Observable<Component<Announce>> {\n  if (!feature(\"announce.dismiss\") || !el.childElementCount)\n    return EMPTY\n\n  /* Support instant navigation - see https://t.ly/3FTme */\n  if (!el.hidden) {\n    const content = getElement(\".md-typeset\", el)\n    if (__md_hash(content.innerHTML) === __md_get(\"__announce\"))\n      el.hidden = true\n  }\n\n  /* Mount component on subscription */\n  return defer(() => {\n    const push$ = new Subject<Announce>()\n    push$.subscribe(({ hash }) => {\n      el.hidden = true\n\n      /* Persist preference in local storage */\n      __md_set<number>(\"__announce\", hash)\n    })\n\n    /* Create and return component */\n    return watchAnnounce(el)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  finalize,\n  map,\n  tap\n} from \"rxjs\"\n\nimport { Component } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Consent\n */\nexport interface Consent {\n  hidden: boolean                      /* Consent is hidden */\n}\n\n/**\n * Consent defaults\n */\nexport interface ConsentDefaults {\n  analytics?: boolean                  /* Consent for Analytics */\n  github?: boolean                     /* Consent for GitHub */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  target$: Observable<HTMLElement>     /* Target observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  target$: Observable<HTMLElement>     /* Target observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch consent\n *\n * @param el - Consent element\n * @param options - Options\n *\n * @returns Consent observable\n */\nexport function watchConsent(\n  el: HTMLElement, { target$ }: WatchOptions\n): Observable<Consent> {\n  return target$\n    .pipe(\n      map(target => ({ hidden: target !== el }))\n    )\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Mount consent\n *\n * @param el - Consent element\n * @param options - Options\n *\n * @returns Consent component observable\n */\nexport function mountConsent(\n  el: HTMLElement, options: MountOptions\n): Observable<Component<Consent>> {\n  const internal$ = new Subject<Consent>()\n  internal$.subscribe(({ hidden }) => {\n    el.hidden = hidden\n  })\n\n  /* Create and return component */\n  return watchConsent(el, options)\n    .pipe(\n      tap(state => internal$.next(state)),\n      finalize(() => internal$.complete()),\n      map(state => ({ ref: el, ...state }))\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Tooltip style\n */\nexport type TooltipStyle =\n  | \"inline\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render a tooltip\n *\n * @param id - Tooltip identifier\n * @param style - Tooltip style\n *\n * @returns Element\n */\nexport function renderTooltip(\n  id?: string, style?: TooltipStyle\n): HTMLElement {\n  if (style === \"inline\") { // @todo refactor control flow\n    return (\n      <div class=\"md-tooltip md-tooltip--inline\" id={id} role=\"tooltip\">\n        <div class=\"md-tooltip__inner md-typeset\"></div>\n      </div>\n    )\n  } else {\n    return (\n      <div class=\"md-tooltip\" id={id} role=\"tooltip\">\n        <div class=\"md-tooltip__inner md-typeset\"></div>\n      </div>\n    )\n  }\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { h } from \"~/utilities\"\n\nimport { renderTooltip } from \"../tooltip\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render an annotation\n *\n * @param id - Annotation identifier\n * @param prefix - Tooltip identifier prefix\n *\n * @returns Element\n */\nexport function renderAnnotation(\n  id: string | number, prefix?: string\n): HTMLElement {\n  prefix = prefix ? `${prefix}_annotation_${id}` : undefined\n\n  /* Render tooltip with anchor, if given */\n  if (prefix) {\n    const anchor = prefix ? `#${prefix}` : undefined\n    return (\n      <aside class=\"md-annotation\" tabIndex={0}>\n        {renderTooltip(prefix)}\n        <a href={anchor} class=\"md-annotation__index\" tabIndex={-1}>\n          <span data-md-annotation-id={id}></span>\n        </a>\n      </aside>\n    )\n  } else {\n    return (\n      <aside class=\"md-annotation\" tabIndex={0}>\n        {renderTooltip(prefix)}\n        <span class=\"md-annotation__index\" tabIndex={-1}>\n          <span data-md-annotation-id={id}></span>\n        </span>\n      </aside>\n    )\n  }\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { translation } from \"~/_\"\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render a 'copy-to-clipboard' button\n *\n * @param id - Unique identifier\n *\n * @returns Element\n */\nexport function renderClipboardButton(id: string): HTMLElement {\n  return (\n    <button\n      class=\"md-clipboard md-icon\"\n      title={translation(\"clipboard.copy\")}\n      data-clipboard-target={`#${id} > code`}\n    ></button>\n  )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { ComponentChild } from \"preact\"\n\nimport { configuration, feature, translation } from \"~/_\"\nimport { SearchItem } from \"~/integrations/search\"\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Render flag\n */\nconst enum Flag {\n  TEASER = 1,                          /* Render teaser */\n  PARENT = 2                           /* Render as parent */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper function\n * ------------------------------------------------------------------------- */\n\n/**\n * Render a search document\n *\n * @param document - Search document\n * @param flag - Render flags\n *\n * @returns Element\n */\nfunction renderSearchDocument(\n  document: SearchItem, flag: Flag\n): HTMLElement {\n  const parent = flag & Flag.PARENT\n  const teaser = flag & Flag.TEASER\n\n  /* Render missing query terms */\n  const missing = Object.keys(document.terms)\n    .filter(key => !document.terms[key])\n    .reduce<ComponentChild[]>((list, key) => [\n      ...list, <del>{key}</del>, \" \"\n    ], [])\n    .slice(0, -1)\n\n  /* Assemble query string for highlighting */\n  const config = configuration()\n  const url = new URL(document.location, config.base)\n  if (feature(\"search.highlight\"))\n    url.searchParams.set(\"h\", Object.entries(document.terms)\n      .filter(([, match]) => match)\n      .reduce((highlight, [value]) => `${highlight} ${value}`.trim(), \"\")\n    )\n\n  /* Render article or section, depending on flags */\n  const { tags } = configuration()\n  return (\n    <a href={`${url}`} class=\"md-search-result__link\" tabIndex={-1}>\n      <article\n        class=\"md-search-result__article md-typeset\"\n        data-md-score={document.score.toFixed(2)}\n      >\n        {parent > 0 && <div class=\"md-search-result__icon md-icon\"></div>}\n        {parent > 0 && <h1>{document.title}</h1>}\n        {parent <= 0 && <h2>{document.title}</h2>}\n        {teaser > 0 && document.text.length > 0 &&\n          document.text\n        }\n        {document.tags && document.tags.map(tag => {\n          const type = tags\n            ? tag in tags\n              ? `md-tag-icon md-tag--${tags[tag]}`\n              : \"md-tag-icon\"\n            : \"\"\n          return (\n            <span class={`md-tag ${type}`}>{tag}</span>\n          )\n        })}\n        {teaser > 0 && missing.length > 0 &&\n          <p class=\"md-search-result__terms\">\n            {translation(\"search.result.term.missing\")}: {...missing}\n          </p>\n        }\n      </article>\n    </a>\n  )\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render a search result\n *\n * @param result - Search result\n *\n * @returns Element\n */\nexport function renderSearchResultItem(\n  result: SearchItem[]\n): HTMLElement {\n  const threshold = result[0].score\n  const docs = [...result]\n\n  const config = configuration()\n\n  /* Find and extract parent article */\n  const parent = docs.findIndex(doc => {\n    const l = `${new URL(doc.location, config.base)}` // @todo hacky\n    return !l.includes(\"#\")\n  })\n  const [article] = docs.splice(parent, 1)\n\n  /* Determine last index above threshold */\n  let index = docs.findIndex(doc => doc.score < threshold)\n  if (index === -1)\n    index = docs.length\n\n  /* Partition sections */\n  const best = docs.slice(0, index)\n  const more = docs.slice(index)\n\n  /* Render children */\n  const children = [\n    renderSearchDocument(article, Flag.PARENT | +(!parent && index === 0)),\n    ...best.map(section => renderSearchDocument(section, Flag.TEASER)),\n    ...more.length ? [\n      <details class=\"md-search-result__more\">\n        <summary tabIndex={-1}>\n          <div>\n            {more.length > 0 && more.length === 1\n              ? translation(\"search.result.more.one\")\n              : translation(\"search.result.more.other\", more.length)\n            }\n          </div>\n        </summary>\n        {...more.map(section => renderSearchDocument(section, Flag.TEASER))}\n      </details>\n    ] : []\n  ]\n\n  /* Render search result */\n  return (\n    <li class=\"md-search-result__item\">\n      {children}\n    </li>\n  )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { SourceFacts } from \"~/components\"\nimport { h, round } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render repository facts\n *\n * @param facts - Repository facts\n *\n * @returns Element\n */\nexport function renderSourceFacts(facts: SourceFacts): HTMLElement {\n  return (\n    <ul class=\"md-source__facts\">\n      {Object.entries(facts).map(([key, value]) => (\n        <li class={`md-source__fact md-source__fact--${key}`}>\n          {typeof value === \"number\" ? round(value) : value}\n        </li>\n      ))}\n    </ul>\n  )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Tabbed control type\n */\ntype TabbedControlType =\n  | \"prev\"\n  | \"next\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render control for content tabs\n *\n * @param type - Control type\n *\n * @returns Element\n */\nexport function renderTabbedControl(\n  type: TabbedControlType\n): HTMLElement {\n  const classes = `tabbed-control tabbed-control--${type}`\n  return (\n    <div class={classes} hidden>\n      <button class=\"tabbed-button\" tabIndex={-1} aria-hidden=\"true\"></button>\n    </div>\n  )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render a table inside a wrapper to improve scrolling on mobile\n *\n * @param table - Table element\n *\n * @returns Element\n */\nexport function renderTable(table: HTMLElement): HTMLElement {\n  return (\n    <div class=\"md-typeset__scrollwrap\">\n      <div class=\"md-typeset__table\">\n        {table}\n      </div>\n    </div>\n  )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { configuration, translation } from \"~/_\"\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Version properties\n */\nexport interface VersionProperties {\n  hidden?: boolean                     /* Version is hidden */\n}\n\n/**\n * Version\n */\nexport interface Version {\n  version: string                      /* Version identifier */\n  title: string                        /* Version title */\n  aliases: string[]                    /* Version aliases */\n  properties?: VersionProperties       /* Version properties */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render a version\n *\n * @param version - Version\n *\n * @returns Element\n */\nfunction renderVersion(version: Version): HTMLElement {\n  const config = configuration()\n\n  /* Ensure trailing slash - see https://bit.ly/3rL5u3f */\n  const url = new URL(`../${version.version}/`, config.base)\n  return (\n    <li class=\"md-version__item\">\n      <a href={`${url}`} class=\"md-version__link\">\n        {version.title}\n      </a>\n    </li>\n  )\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render a version selector\n *\n * @param versions - Versions\n * @param active - Active version\n *\n * @returns Element\n */\nexport function renderVersionSelector(\n  versions: Version[], active: Version\n): HTMLElement {\n  versions = versions.filter(version => !version.properties?.hidden)\n  return (\n    <div class=\"md-version\">\n      <button\n        class=\"md-version__current\"\n        aria-label={translation(\"select.version\")}\n      >\n        {active.title}\n      </button>\n      <ul class=\"md-version__list\">\n        {versions.map(renderVersion)}\n      </ul>\n    </div>\n  )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  animationFrameScheduler,\n  asyncScheduler,\n  auditTime,\n  combineLatest,\n  debounceTime,\n  defer,\n  distinctUntilChanged,\n  filter,\n  finalize,\n  map,\n  merge,\n  of,\n  subscribeOn,\n  tap,\n  throttleTime\n} from \"rxjs\"\n\nimport {\n  ElementOffset,\n  getElement,\n  getElementContainer,\n  getElementOffset,\n  getElementSize,\n  watchElementContentOffset,\n  watchElementFocus,\n  watchElementHover\n} from \"~/browser\"\nimport { renderTooltip } from \"~/templates\"\n\nimport { Component } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Tooltip\n */\nexport interface Tooltip {\n  active: boolean                      /* Tooltip is active */\n  offset: ElementOffset                /* Tooltip offset */\n}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Global sequence number for tooltips\n */\nlet sequence = 0\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch tooltip\n *\n * This function will append the tooltip temporarily to compute its width,\n * which is necessary for correct centering, and then removing it again.\n *\n * @param el - Tooltip element\n * @param host - Host element\n *\n * @returns Tooltip observable\n */\nexport function watchTooltip(\n  el: HTMLElement, host: HTMLElement\n): Observable<Tooltip> {\n  document.body.append(el)\n\n  /* Compute width and remove tooltip immediately */\n  const { width } = getElementSize(el)\n  el.style.setProperty(\"--md-tooltip-width\", `${width}px`)\n  el.remove()\n\n  /* Retrieve and watch containing element */\n  const container = getElementContainer(host)\n  const scroll$ =\n    typeof container !== \"undefined\"\n      ? watchElementContentOffset(container)\n      : of({ x: 0, y: 0 })\n\n  /* Compute tooltip visibility */\n  const active$ = merge(\n    watchElementFocus(host),\n    watchElementHover(host)\n  )\n    .pipe(\n      distinctUntilChanged()\n    )\n\n  /* Compute tooltip offset */\n  return combineLatest([active$, scroll$])\n    .pipe(\n      map(([active, scroll]) => {\n        let { x, y } = getElementOffset(host)\n        const size = getElementSize(host)\n\n        /**\n         * Experimental: fix handling of tables - see https://bit.ly/3TQEj5O\n         *\n         * If this proves to be a viable fix, we should refactor tooltip\n         * positioning and somehow streamline the current process. This might\n         * also fix positioning for annotations inside tables, which is another\n         * limitation.\n         */\n        const table = host.closest(\"table\")\n        if (table && host.parentElement) {\n          x += table.offsetLeft + host.parentElement.offsetLeft\n          y += table.offsetTop  + host.parentElement.offsetTop\n        }\n        return {\n          active,\n          offset: {\n            x: x - scroll.x + size.width  / 2 - width / 2,\n            y: y - scroll.y + size.height + 8\n          }\n        }\n      })\n    )\n}\n\n/**\n * Mount tooltip\n *\n * @param el - Host element\n *\n * @returns Tooltip component observable\n */\nexport function mountTooltip(\n  el: HTMLElement\n): Observable<Component<Tooltip>> {\n  const title = el.title\n  if (!title.length)\n    return EMPTY\n\n  /* Render tooltip and set title from host element */\n  const id = `__tooltip_${sequence++}`\n  const tooltip = renderTooltip(id, \"inline\")\n  const typeset = getElement(\".md-typeset\", tooltip)\n  typeset.innerHTML = title\n\n  /* Mount component on subscription */\n  return defer(() => {\n    const push$ = new Subject<Tooltip>()\n    push$.subscribe({\n\n      /* Handle emission */\n      next({ offset }) {\n        tooltip.style.setProperty(\"--md-tooltip-x\", `${offset.x}px`)\n        tooltip.style.setProperty(\"--md-tooltip-y\", `${offset.y}px`)\n      },\n\n      /* Handle complete */\n      complete() {\n        tooltip.style.removeProperty(\"--md-tooltip-x\")\n        tooltip.style.removeProperty(\"--md-tooltip-y\")\n      }\n    })\n\n    /* Toggle tooltip presence to mitigate empty lines when copying */\n    merge(\n      push$.pipe(filter(({ active }) => active)),\n      push$.pipe(debounceTime(250), filter(({ active }) => !active))\n    )\n      .subscribe({\n\n        /* Handle emission */\n        next({ active }) {\n          if (active) {\n            el.insertAdjacentElement(\"afterend\", tooltip)\n            el.setAttribute(\"aria-describedby\", id)\n            el.removeAttribute(\"title\")\n          } else {\n            tooltip.remove()\n            el.removeAttribute(\"aria-describedby\")\n            el.setAttribute(\"title\", title)\n          }\n        },\n\n        /* Handle complete */\n        complete() {\n          tooltip.remove()\n          el.removeAttribute(\"aria-describedby\")\n          el.setAttribute(\"title\", title)\n        }\n      })\n\n    /* Toggle tooltip visibility */\n    push$\n      .pipe(\n        auditTime(16, animationFrameScheduler)\n      )\n        .subscribe(({ active }) => {\n          tooltip.classList.toggle(\"md-tooltip--active\", active)\n        })\n\n    // @todo - refactor positioning together with annotations \u2013 there are\n    // several things that overlap and are identical in handling\n\n    /* Track relative origin of tooltip */\n    push$\n      .pipe(\n        throttleTime(125, animationFrameScheduler),\n        filter(() => !!el.offsetParent),\n        map(() => el.offsetParent!.getBoundingClientRect()),\n        map(({ x }) => x)\n      )\n      .subscribe({\n\n        /* Handle emission */\n        next(origin) {\n          if (origin)\n            tooltip.style.setProperty(\"--md-tooltip-0\", `${-origin}px`)\n          else\n            tooltip.style.removeProperty(\"--md-tooltip-0\")\n        },\n\n        /* Handle complete */\n        complete() {\n          tooltip.style.removeProperty(\"--md-tooltip-0\")\n        }\n      })\n\n    /* Create and return component */\n    return watchTooltip(tooltip, el)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n    .pipe(\n      subscribeOn(asyncScheduler)\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  animationFrameScheduler,\n  auditTime,\n  combineLatest,\n  debounceTime,\n  defer,\n  delay,\n  endWith,\n  filter,\n  finalize,\n  fromEvent,\n  ignoreElements,\n  map,\n  merge,\n  switchMap,\n  take,\n  takeUntil,\n  tap,\n  throttleTime,\n  withLatestFrom\n} from \"rxjs\"\n\nimport {\n  ElementOffset,\n  getActiveElement,\n  getElementSize,\n  watchElementContentOffset,\n  watchElementFocus,\n  watchElementOffset,\n  watchElementVisibility\n} from \"~/browser\"\n\nimport { Component } from \"../../../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Annotation\n */\nexport interface Annotation {\n  active: boolean                      /* Annotation is active */\n  offset: ElementOffset                /* Annotation offset */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  target$: Observable<HTMLElement>     /* Location target observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch annotation\n *\n * @param el - Annotation element\n * @param container - Containing element\n *\n * @returns Annotation observable\n */\nexport function watchAnnotation(\n  el: HTMLElement, container: HTMLElement\n): Observable<Annotation> {\n  const offset$ = defer(() => combineLatest([\n    watchElementOffset(el),\n    watchElementContentOffset(container)\n  ]))\n    .pipe(\n      map(([{ x, y }, scroll]): ElementOffset => {\n        const { width, height } = getElementSize(el)\n        return ({\n          x: x - scroll.x + width  / 2,\n          y: y - scroll.y + height / 2\n        })\n      })\n    )\n\n  /* Actively watch annotation on focus */\n  return watchElementFocus(el)\n    .pipe(\n      switchMap(active => offset$\n        .pipe(\n          map(offset => ({ active, offset })),\n          take(+!active || Infinity)\n        )\n      )\n    )\n}\n\n/**\n * Mount annotation\n *\n * @param el - Annotation element\n * @param container - Containing element\n * @param options - Options\n *\n * @returns Annotation component observable\n */\nexport function mountAnnotation(\n  el: HTMLElement, container: HTMLElement, { target$ }: MountOptions\n): Observable<Component<Annotation>> {\n  const [tooltip, index] = Array.from(el.children)\n\n  /* Mount component on subscription */\n  return defer(() => {\n    const push$ = new Subject<Annotation>()\n    const done$ = push$.pipe(ignoreElements(), endWith(true))\n    push$.subscribe({\n\n      /* Handle emission */\n      next({ offset }) {\n        el.style.setProperty(\"--md-tooltip-x\", `${offset.x}px`)\n        el.style.setProperty(\"--md-tooltip-y\", `${offset.y}px`)\n      },\n\n      /* Handle complete */\n      complete() {\n        el.style.removeProperty(\"--md-tooltip-x\")\n        el.style.removeProperty(\"--md-tooltip-y\")\n      }\n    })\n\n    /* Start animation only when annotation is visible */\n    watchElementVisibility(el)\n      .pipe(\n        takeUntil(done$)\n      )\n        .subscribe(visible => {\n          el.toggleAttribute(\"data-md-visible\", visible)\n        })\n\n    /* Toggle tooltip presence to mitigate empty lines when copying */\n    merge(\n      push$.pipe(filter(({ active }) => active)),\n      push$.pipe(debounceTime(250), filter(({ active }) => !active))\n    )\n      .subscribe({\n\n        /* Handle emission */\n        next({ active }) {\n          if (active)\n            el.prepend(tooltip)\n          else\n            tooltip.remove()\n        },\n\n        /* Handle complete */\n        complete() {\n          el.prepend(tooltip)\n        }\n      })\n\n    /* Toggle tooltip visibility */\n    push$\n      .pipe(\n        auditTime(16, animationFrameScheduler)\n      )\n        .subscribe(({ active }) => {\n          tooltip.classList.toggle(\"md-tooltip--active\", active)\n        })\n\n    /* Track relative origin of tooltip */\n    push$\n      .pipe(\n        throttleTime(125, animationFrameScheduler),\n        filter(() => !!el.offsetParent),\n        map(() => el.offsetParent!.getBoundingClientRect()),\n        map(({ x }) => x)\n      )\n        .subscribe({\n\n          /* Handle emission */\n          next(origin) {\n            if (origin)\n              el.style.setProperty(\"--md-tooltip-0\", `${-origin}px`)\n            else\n              el.style.removeProperty(\"--md-tooltip-0\")\n          },\n\n          /* Handle complete */\n          complete() {\n            el.style.removeProperty(\"--md-tooltip-0\")\n          }\n        })\n\n    /* Allow to copy link without scrolling to anchor */\n    fromEvent<MouseEvent>(index, \"click\")\n      .pipe(\n        takeUntil(done$),\n        filter(ev => !(ev.metaKey || ev.ctrlKey))\n      )\n        .subscribe(ev => {\n          ev.stopPropagation()\n          ev.preventDefault()\n        })\n\n    /* Allow to open link in new tab or blur on close */\n    fromEvent<MouseEvent>(index, \"mousedown\")\n      .pipe(\n        takeUntil(done$),\n        withLatestFrom(push$)\n      )\n        .subscribe(([ev, { active }]) => {\n\n          /* Open in new tab */\n          if (ev.button !== 0 || ev.metaKey || ev.ctrlKey) {\n            ev.preventDefault()\n\n          /* Close annotation */\n          } else if (active) {\n            ev.preventDefault()\n\n            /* Focus parent annotation, if any */\n            const parent = el.parentElement!.closest(\".md-annotation\")\n            if (parent instanceof HTMLElement)\n              parent.focus()\n            else\n              getActiveElement()?.blur()\n          }\n        })\n\n    /* Open and focus annotation on location target */\n    target$\n      .pipe(\n        takeUntil(done$),\n        filter(target => target === tooltip),\n        delay(125)\n      )\n        .subscribe(() => el.focus())\n\n    /* Create and return component */\n    return watchAnnotation(el, container)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  defer,\n  endWith,\n  finalize,\n  ignoreElements,\n  merge,\n  share,\n  takeUntil\n} from \"rxjs\"\n\nimport {\n  getElement,\n  getElements,\n  getOptionalElement\n} from \"~/browser\"\nimport { renderAnnotation } from \"~/templates\"\n\nimport { Component } from \"../../../_\"\nimport {\n  Annotation,\n  mountAnnotation\n} from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  target$: Observable<HTMLElement>     /* Location target observable */\n  print$: Observable<boolean>          /* Media print observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Find all annotation hosts in the containing element\n *\n * @param container - Containing element\n *\n * @returns Annotation hosts\n */\nfunction findHosts(container: HTMLElement): HTMLElement[] {\n  return container.tagName === \"CODE\"\n    ? getElements(\".c, .c1, .cm\", container)\n    : [container]\n}\n\n/**\n * Find all annotation markers in the containing element\n *\n * @param container - Containing element\n *\n * @returns Annotation markers\n */\nfunction findMarkers(container: HTMLElement): Text[] {\n  const markers: Text[] = []\n  for (const el of findHosts(container)) {\n    const nodes: Text[] = []\n\n    /* Find all text nodes in current element */\n    const it = document.createNodeIterator(el, NodeFilter.SHOW_TEXT)\n    for (let node = it.nextNode(); node; node = it.nextNode())\n      nodes.push(node as Text)\n\n    /* Find all markers in each text node */\n    for (let text of nodes) {\n      let match: RegExpExecArray | null\n\n      /* Split text at marker and add to list */\n      while ((match = /(\\(\\d+\\))(!)?/.exec(text.textContent!))) {\n        const [, id, force] = match\n        if (typeof force === \"undefined\") {\n          const marker = text.splitText(match.index)\n          text = marker.splitText(id.length)\n          markers.push(marker)\n\n        /* Replace entire text with marker */\n        } else {\n          text.textContent = id\n          markers.push(text)\n          break\n        }\n      }\n    }\n  }\n  return markers\n}\n\n/**\n * Swap the child nodes of two elements\n *\n * @param source - Source element\n * @param target - Target element\n */\nfunction swap(source: HTMLElement, target: HTMLElement): void {\n  target.append(...Array.from(source.childNodes))\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount annotation list\n *\n * This function analyzes the containing code block and checks for markers\n * referring to elements in the given annotation list. If no markers are found,\n * the list is left untouched. Otherwise, list elements are rendered as\n * annotations inside the code block.\n *\n * @param el - Annotation list element\n * @param container - Containing element\n * @param options - Options\n *\n * @returns Annotation component observable\n */\nexport function mountAnnotationList(\n  el: HTMLElement, container: HTMLElement, { target$, print$ }: MountOptions\n): Observable<Component<Annotation>> {\n\n  /* Compute prefix for tooltip anchors */\n  const parent = container.closest(\"[id]\")\n  const prefix = parent?.id\n\n  /* Find and replace all markers with empty annotations */\n  const annotations = new Map<string, HTMLElement>()\n  for (const marker of findMarkers(container)) {\n    const [, id] = marker.textContent!.match(/\\((\\d+)\\)/)!\n    if (getOptionalElement(`:scope > li:nth-child(${id})`, el)) {\n      annotations.set(id, renderAnnotation(id, prefix))\n      marker.replaceWith(annotations.get(id)!)\n    }\n  }\n\n  /* Keep list if there are no annotations to render */\n  if (annotations.size === 0)\n    return EMPTY\n\n  /* Mount component on subscription */\n  return defer(() => {\n    const push$ = new Subject()\n    const done$ = push$.pipe(ignoreElements(), endWith(true))\n\n    /* Retrieve container pairs for swapping */\n    const pairs: [HTMLElement, HTMLElement][] = []\n    for (const [id, annotation] of annotations)\n      pairs.push([\n        getElement(\".md-typeset\", annotation),\n        getElement(`:scope > li:nth-child(${id})`, el)\n      ])\n\n    /* Handle print mode - see https://bit.ly/3rgPdpt */\n    print$.pipe(takeUntil(done$))\n      .subscribe(active => {\n        el.hidden = !active\n\n        /* Add class to discern list element */\n        el.classList.toggle(\"md-annotation-list\", active)\n\n        /* Show annotations in code block or list (print) */\n        for (const [inner, child] of pairs)\n          if (!active)\n            swap(child, inner)\n          else\n            swap(inner, child)\n      })\n\n    /* Create and return component */\n    return merge(...[...annotations]\n      .map(([, annotation]) => (\n        mountAnnotation(annotation, container, { target$ })\n      ))\n    )\n      .pipe(\n        finalize(() => push$.complete()),\n        share()\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { EMPTY, Observable, defer } from \"rxjs\"\n\nimport { Component } from \"../../../_\"\nimport { Annotation } from \"../_\"\nimport { mountAnnotationList } from \"../list\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  target$: Observable<HTMLElement>     /* Location target observable */\n  print$: Observable<boolean>          /* Media print observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Find list element directly following a block\n *\n * @param el - Annotation block element\n *\n * @returns List element or nothing\n */\nfunction findList(el: HTMLElement): HTMLElement | undefined {\n  if (el.nextElementSibling) {\n    const sibling = el.nextElementSibling as HTMLElement\n    if (sibling.tagName === \"OL\")\n      return sibling\n\n    /* Skip empty paragraphs - see https://bit.ly/3r4ZJ2O */\n    else if (sibling.tagName === \"P\" && !sibling.children.length)\n      return findList(sibling)\n  }\n\n  /* Everything else */\n  return undefined\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount annotation block\n *\n * @param el - Annotation block element\n * @param options - Options\n *\n * @returns Annotation component observable\n */\nexport function mountAnnotationBlock(\n  el: HTMLElement, options: MountOptions\n): Observable<Component<Annotation>> {\n  return defer(() => {\n    const list = findList(el)\n    return typeof list !== \"undefined\"\n      ? mountAnnotationList(list, el, options)\n      : EMPTY\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport ClipboardJS from \"clipboard\"\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  defer,\n  distinctUntilChanged,\n  distinctUntilKeyChanged,\n  filter,\n  finalize,\n  map,\n  mergeWith,\n  switchMap,\n  take,\n  takeLast,\n  takeUntil,\n  tap\n} from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport {\n  getElementContentSize,\n  watchElementSize,\n  watchElementVisibility\n} from \"~/browser\"\nimport { renderClipboardButton } from \"~/templates\"\n\nimport { Component } from \"../../../_\"\nimport {\n  Tooltip,\n  mountTooltip\n} from \"../../../tooltip\"\nimport {\n  Annotation,\n  mountAnnotationList\n} from \"../../annotation\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Code block overflow\n */\nexport interface Overflow {\n  scrollable: boolean                  /* Code block overflows */\n}\n\n/**\n * Code block\n */\nexport type CodeBlock =\n  | Overflow\n  | Annotation\n  | Tooltip\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  target$: Observable<HTMLElement>     /* Location target observable */\n  print$: Observable<boolean>          /* Media print observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Global sequence number for code blocks\n */\nlet sequence = 0\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Find candidate list element directly following a code block\n *\n * @param el - Code block element\n *\n * @returns List element or nothing\n */\nfunction findCandidateList(el: HTMLElement): HTMLElement | undefined {\n  if (el.nextElementSibling) {\n    const sibling = el.nextElementSibling as HTMLElement\n    if (sibling.tagName === \"OL\")\n      return sibling\n\n    /* Skip empty paragraphs - see https://bit.ly/3r4ZJ2O */\n    else if (sibling.tagName === \"P\" && !sibling.children.length)\n      return findCandidateList(sibling)\n  }\n\n  /* Everything else */\n  return undefined\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch code block\n *\n * This function monitors size changes of the viewport, as well as switches of\n * content tabs with embedded code blocks, as both may trigger overflow.\n *\n * @param el - Code block element\n *\n * @returns Code block observable\n */\nexport function watchCodeBlock(\n  el: HTMLElement\n): Observable<Overflow> {\n  return watchElementSize(el)\n    .pipe(\n      map(({ width }) => {\n        const content = getElementContentSize(el)\n        return {\n          scrollable: content.width > width\n        }\n      }),\n      distinctUntilKeyChanged(\"scrollable\")\n    )\n}\n\n/**\n * Mount code block\n *\n * This function ensures that an overflowing code block is focusable through\n * keyboard, so it can be scrolled without a mouse to improve on accessibility.\n * Furthermore, if code annotations are enabled, they are mounted if and only\n * if the code block is currently visible, e.g., not in a hidden content tab.\n *\n * Note that code blocks may be mounted eagerly or lazily. If they're mounted\n * lazily (on first visibility), code annotation anchor links will not work,\n * as they are evaluated on initial page load, and code annotations in general\n * might feel a little bumpier.\n *\n * @param el - Code block element\n * @param options - Options\n *\n * @returns Code block and annotation component observable\n */\nexport function mountCodeBlock(\n  el: HTMLElement, options: MountOptions\n): Observable<Component<CodeBlock>> {\n  const { matches: hover } = matchMedia(\"(hover)\")\n\n  /* Defer mounting of code block - see https://bit.ly/3vHVoVD */\n  const factory$ = defer(() => {\n    const push$ = new Subject<Overflow>()\n    const done$ = push$.pipe(takeLast(1))\n    push$.subscribe(({ scrollable }) => {\n      if (scrollable && hover)\n        el.setAttribute(\"tabindex\", \"0\")\n      else\n        el.removeAttribute(\"tabindex\")\n    })\n\n    /* Render button for Clipboard.js integration */\n    const content$: Array<Observable<Component<CodeBlock>>> = []\n    if (ClipboardJS.isSupported()) {\n      if (el.closest(\".copy\") || (\n        feature(\"content.code.copy\") && !el.closest(\".no-copy\")\n      )) {\n        const parent = el.closest(\"pre\")!\n        parent.id = `__code_${sequence++}`\n\n        /* Mount tooltip, if enabled */\n        const button = renderClipboardButton(parent.id)\n        parent.insertBefore(button, el)\n        if (feature(\"content.tooltips\"))\n          content$.push(mountTooltip(button))\n      }\n    }\n\n    /* Handle code annotations */\n    const container = el.closest(\".highlight\")\n    if (container instanceof HTMLElement) {\n      const list = findCandidateList(container)\n\n      /* Mount code annotations, if enabled */\n      if (typeof list !== \"undefined\" && (\n        container.classList.contains(\"annotate\") ||\n        feature(\"content.code.annotate\")\n      )) {\n        const annotations$ = mountAnnotationList(list, el, options)\n        content$.push(\n          watchElementSize(container)\n            .pipe(\n              takeUntil(done$),\n              map(({ width, height }) => width && height),\n              distinctUntilChanged(),\n              switchMap(active => active ? annotations$ : EMPTY)\n            )\n        )\n      }\n    }\n\n    /* Create and return component */\n    return watchCodeBlock(el)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state })),\n        mergeWith(...content$)\n      )\n  })\n\n  /* Mount code block lazily */\n  if (feature(\"content.lazy\"))\n    return watchElementVisibility(el)\n      .pipe(\n        filter(visible => visible),\n        take(1),\n        switchMap(() => factory$)\n      )\n\n  /* Mount code block */\n  return factory$\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  defer,\n  filter,\n  finalize,\n  map,\n  merge,\n  tap\n} from \"rxjs\"\n\nimport { Component } from \"../../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Details\n */\nexport interface Details {\n  action: \"open\" | \"close\"             /* Details state */\n  reveal?: boolean                     /* Details is revealed */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  target$: Observable<HTMLElement>     /* Location target observable */\n  print$: Observable<boolean>          /* Media print observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  target$: Observable<HTMLElement>     /* Location target observable */\n  print$: Observable<boolean>          /* Media print observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch details\n *\n * @param el - Details element\n * @param options - Options\n *\n * @returns Details observable\n */\nexport function watchDetails(\n  el: HTMLDetailsElement, { target$, print$ }: WatchOptions\n): Observable<Details> {\n  let open = true\n  return merge(\n\n    /* Open and focus details on location target */\n    target$\n      .pipe(\n        map(target => target.closest(\"details:not([open])\")!),\n        filter(details => el === details),\n        map(() => ({\n          action: \"open\", reveal: true\n        }) as Details)\n      ),\n\n    /* Open details on print and close afterwards */\n    print$\n      .pipe(\n        filter(active => active || !open),\n        tap(() => open = el.open),\n        map(active => ({\n          action: active ? \"open\" : \"close\"\n        }) as Details)\n      )\n  )\n}\n\n/**\n * Mount details\n *\n * This function ensures that `details` tags are opened on anchor jumps and\n * prior to printing, so the whole content of the page is visible.\n *\n * @param el - Details element\n * @param options - Options\n *\n * @returns Details component observable\n */\nexport function mountDetails(\n  el: HTMLDetailsElement, options: MountOptions\n): Observable<Component<Details>> {\n  return defer(() => {\n    const push$ = new Subject<Details>()\n    push$.subscribe(({ action, reveal }) => {\n      el.toggleAttribute(\"open\", action === \"open\")\n      if (reveal)\n        el.scrollIntoView()\n    })\n\n    /* Create and return component */\n    return watchDetails(el, options)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", ".node circle,.node ellipse,.node path,.node polygon,.node rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}marker{fill:var(--md-mermaid-edge-color)!important}.edgeLabel .label rect{fill:#0000}.label{color:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.label foreignObject{line-height:normal;overflow:visible}.label div .edgeLabel{color:var(--md-mermaid-label-fg-color)}.edgeLabel,.edgeLabel rect,.label div .edgeLabel{background-color:var(--md-mermaid-label-bg-color)}.edgeLabel,.edgeLabel rect{fill:var(--md-mermaid-label-bg-color);color:var(--md-mermaid-edge-color)}.edgePath .path,.flowchart-link{stroke:var(--md-mermaid-edge-color);stroke-width:.05rem}.edgePath .arrowheadPath{fill:var(--md-mermaid-edge-color);stroke:none}.cluster rect{fill:var(--md-default-fg-color--lightest);stroke:var(--md-default-fg-color--lighter)}.cluster span{color:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}g #flowchart-circleEnd,g #flowchart-circleStart,g #flowchart-crossEnd,g #flowchart-crossStart,g #flowchart-pointEnd,g #flowchart-pointStart{stroke:none}g.classGroup line,g.classGroup rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}g.classGroup text{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.classLabel .box{fill:var(--md-mermaid-label-bg-color);background-color:var(--md-mermaid-label-bg-color);opacity:1}.classLabel .label{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.node .divider{stroke:var(--md-mermaid-node-fg-color)}.relation{stroke:var(--md-mermaid-edge-color)}.cardinality{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.cardinality text{fill:inherit!important}defs #classDiagram-compositionEnd,defs #classDiagram-compositionStart,defs #classDiagram-dependencyEnd,defs #classDiagram-dependencyStart,defs #classDiagram-extensionEnd,defs #classDiagram-extensionStart{fill:var(--md-mermaid-edge-color)!important;stroke:var(--md-mermaid-edge-color)!important}defs #classDiagram-aggregationEnd,defs #classDiagram-aggregationStart{fill:var(--md-mermaid-label-bg-color)!important;stroke:var(--md-mermaid-edge-color)!important}g.stateGroup rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}g.stateGroup .state-title{fill:var(--md-mermaid-label-fg-color)!important;font-family:var(--md-mermaid-font-family)}g.stateGroup .composit{fill:var(--md-mermaid-label-bg-color)}.nodeLabel,.nodeLabel p{color:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.node circle.state-end,.node circle.state-start,.start-state{fill:var(--md-mermaid-edge-color);stroke:none}.end-state-inner,.end-state-outer{fill:var(--md-mermaid-edge-color)}.end-state-inner,.node circle.state-end{stroke:var(--md-mermaid-label-bg-color)}.transition{stroke:var(--md-mermaid-edge-color)}[id^=state-fork] rect,[id^=state-join] rect{fill:var(--md-mermaid-edge-color)!important;stroke:none!important}.statediagram-cluster.statediagram-cluster .inner{fill:var(--md-default-bg-color)}.statediagram-cluster rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}.statediagram-state rect.divider{fill:var(--md-default-fg-color--lightest);stroke:var(--md-default-fg-color--lighter)}defs #statediagram-barbEnd{stroke:var(--md-mermaid-edge-color)}.attributeBoxEven,.attributeBoxOdd{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}.entityBox{fill:var(--md-mermaid-label-bg-color);stroke:var(--md-mermaid-node-fg-color)}.entityLabel{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.relationshipLabelBox{fill:var(--md-mermaid-label-bg-color);fill-opacity:1;background-color:var(--md-mermaid-label-bg-color);opacity:1}.relationshipLabel{fill:var(--md-mermaid-label-fg-color)}.relationshipLine{stroke:var(--md-mermaid-edge-color)}defs #ONE_OR_MORE_END *,defs #ONE_OR_MORE_START *,defs #ONLY_ONE_END *,defs #ONLY_ONE_START *,defs #ZERO_OR_MORE_END *,defs #ZERO_OR_MORE_START *,defs #ZERO_OR_ONE_END *,defs #ZERO_OR_ONE_START *{stroke:var(--md-mermaid-edge-color)!important}defs #ZERO_OR_MORE_END circle,defs #ZERO_OR_MORE_START circle{fill:var(--md-mermaid-label-bg-color)}.actor{fill:var(--md-mermaid-sequence-actor-bg-color);stroke:var(--md-mermaid-sequence-actor-border-color)}text.actor>tspan{fill:var(--md-mermaid-sequence-actor-fg-color);font-family:var(--md-mermaid-font-family)}line{stroke:var(--md-mermaid-sequence-actor-line-color)}.actor-man circle,.actor-man line{fill:var(--md-mermaid-sequence-actorman-bg-color);stroke:var(--md-mermaid-sequence-actorman-line-color)}.messageLine0,.messageLine1{stroke:var(--md-mermaid-sequence-message-line-color)}.note{fill:var(--md-mermaid-sequence-note-bg-color);stroke:var(--md-mermaid-sequence-note-border-color)}.loopText,.loopText>tspan,.messageText,.noteText>tspan{stroke:none;font-family:var(--md-mermaid-font-family)!important}.messageText{fill:var(--md-mermaid-sequence-message-fg-color)}.loopText,.loopText>tspan{fill:var(--md-mermaid-sequence-loop-fg-color)}.noteText>tspan{fill:var(--md-mermaid-sequence-note-fg-color)}#arrowhead path{fill:var(--md-mermaid-sequence-message-line-color);stroke:none}.loopLine{fill:var(--md-mermaid-sequence-loop-bg-color);stroke:var(--md-mermaid-sequence-loop-border-color)}.labelBox{fill:var(--md-mermaid-sequence-label-bg-color);stroke:none}.labelText,.labelText>span{fill:var(--md-mermaid-sequence-label-fg-color);font-family:var(--md-mermaid-font-family)}.sequenceNumber{fill:var(--md-mermaid-sequence-number-fg-color)}rect.rect{fill:var(--md-mermaid-sequence-box-bg-color);stroke:none}rect.rect+text.text{fill:var(--md-mermaid-sequence-box-fg-color)}defs #sequencenumber{fill:var(--md-mermaid-sequence-number-bg-color)!important}", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  map,\n  of,\n  shareReplay,\n  tap\n} from \"rxjs\"\n\nimport { watchScript } from \"~/browser\"\nimport { h } from \"~/utilities\"\n\nimport { Component } from \"../../_\"\n\nimport themeCSS from \"./index.css\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mermaid diagram\n */\nexport interface Mermaid {}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Mermaid instance observable\n */\nlet mermaid$: Observable<void>\n\n/**\n * Global sequence number for diagrams\n */\nlet sequence = 0\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch Mermaid script\n *\n * @returns Mermaid scripts observable\n */\nfunction fetchScripts(): Observable<void> {\n  return typeof mermaid === \"undefined\" || mermaid instanceof Element\n    ? watchScript(\"https://unpkg.com/mermaid@10.7.0/dist/mermaid.min.js\")\n    : of(undefined)\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount Mermaid diagram\n *\n * @param el - Code block element\n *\n * @returns Mermaid diagram component observable\n */\nexport function mountMermaid(\n  el: HTMLElement\n): Observable<Component<Mermaid>> {\n  el.classList.remove(\"mermaid\") // Hack: mitigate https://bit.ly/3CiN6Du\n  mermaid$ ||= fetchScripts()\n    .pipe(\n      tap(() => mermaid.initialize({\n        startOnLoad: false,\n        themeCSS,\n        sequence: {\n          actorFontSize: \"16px\", // Hack: mitigate https://bit.ly/3y0NEi3\n          messageFontSize: \"16px\",\n          noteFontSize: \"16px\"\n        }\n      })),\n      map(() => undefined),\n      shareReplay(1)\n    )\n\n  /* Render diagram */\n  mermaid$.subscribe(async () => {\n    el.classList.add(\"mermaid\") // Hack: mitigate https://bit.ly/3CiN6Du\n    const id = `__mermaid_${sequence++}`\n\n    /* Create host element to replace code block */\n    const host = h(\"div\", { class: \"mermaid\" })\n    const text = el.textContent\n\n    /* Render and inject diagram */\n    const { svg, fn } = await mermaid.render(id, text)\n\n    /* Create a shadow root and inject diagram */\n    const shadow = host.attachShadow({ mode: \"closed\" })\n    shadow.innerHTML = svg\n\n    /* Replace code block with diagram and bind functions */\n    el.replaceWith(host)\n    fn?.(shadow)\n  })\n\n  /* Create and return component */\n  return mermaid$\n    .pipe(\n      map(() => ({ ref: el }))\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { Observable, of } from \"rxjs\"\n\nimport { renderTable } from \"~/templates\"\nimport { h } from \"~/utilities\"\n\nimport { Component } from \"../../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Data table\n */\nexport interface DataTable {}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Sentinel for replacement\n */\nconst sentinel = h(\"table\")\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount data table\n *\n * This function wraps a data table in another scrollable container, so it can\n * be smoothly scrolled on smaller screen sizes and won't break the layout.\n *\n * @param el - Data table element\n *\n * @returns Data table component observable\n */\nexport function mountDataTable(\n  el: HTMLElement\n): Observable<Component<DataTable>> {\n  el.replaceWith(sentinel)\n  sentinel.replaceWith(renderTable(el))\n\n  /* Create and return component */\n  return of({ ref: el })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  animationFrameScheduler,\n  asyncScheduler,\n  auditTime,\n  combineLatest,\n  defer,\n  endWith,\n  filter,\n  finalize,\n  fromEvent,\n  ignoreElements,\n  map,\n  merge,\n  skip,\n  startWith,\n  subscribeOn,\n  takeUntil,\n  tap,\n  withLatestFrom\n} from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport {\n  Viewport,\n  getElement,\n  getElementContentOffset,\n  getElementContentSize,\n  getElementOffset,\n  getElementSize,\n  getElements,\n  watchElementContentOffset,\n  watchElementSize\n} from \"~/browser\"\nimport { renderTabbedControl } from \"~/templates\"\nimport { h } from \"~/utilities\"\n\nimport { Component } from \"../../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Content tabs\n */\nexport interface ContentTabs {\n  active: HTMLLabelElement             /* Active tab label */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  target$: Observable<HTMLElement>     /* Location target observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch content tabs\n *\n * @param inputs - Content tabs input elements\n *\n * @returns Content tabs observable\n */\nexport function watchContentTabs(\n  inputs: HTMLInputElement[]\n): Observable<ContentTabs> {\n  const initial = inputs.find(input => input.checked) || inputs[0]\n  return merge(...inputs.map(input => fromEvent(input, \"change\")\n    .pipe(\n      map(() => getElement<HTMLLabelElement>(`label[for=\"${input.id}\"]`))\n    )\n  ))\n    .pipe(\n      startWith(getElement<HTMLLabelElement>(`label[for=\"${initial.id}\"]`)),\n      map(active => ({ active }))\n    )\n}\n\n/**\n * Mount content tabs\n *\n * @param el - Content tabs element\n * @param options - Options\n *\n * @returns Content tabs component observable\n */\nexport function mountContentTabs(\n  el: HTMLElement, { viewport$, target$ }: MountOptions\n): Observable<Component<ContentTabs>> {\n  const container = getElement(\".tabbed-labels\", el)\n  const inputs = getElements<HTMLInputElement>(\":scope > input\", el)\n\n  /* Render content tab previous button for pagination */\n  const prev = renderTabbedControl(\"prev\")\n  el.append(prev)\n\n  /* Render content tab next button for pagination */\n  const next = renderTabbedControl(\"next\")\n  el.append(next)\n\n  /* Mount component on subscription */\n  return defer(() => {\n    const push$ = new Subject<ContentTabs>()\n    const done$ = push$.pipe(ignoreElements(), endWith(true))\n    combineLatest([push$, watchElementSize(el)])\n      .pipe(\n        takeUntil(done$),\n        auditTime(1, animationFrameScheduler)\n      )\n        .subscribe({\n\n          /* Handle emission */\n          next([{ active }, size]) {\n            const offset = getElementOffset(active)\n            const { width } = getElementSize(active)\n\n            /* Set tab indicator offset and width */\n            el.style.setProperty(\"--md-indicator-x\", `${offset.x}px`)\n            el.style.setProperty(\"--md-indicator-width\", `${width}px`)\n\n            /* Scroll container to active content tab */\n            const content = getElementContentOffset(container)\n            if (\n              offset.x         < content.x              ||\n              offset.x + width > content.x + size.width\n            )\n              container.scrollTo({\n                left: Math.max(0, offset.x - 16),\n                behavior: \"smooth\"\n              })\n          },\n\n          /* Handle complete */\n          complete() {\n            el.style.removeProperty(\"--md-indicator-x\")\n            el.style.removeProperty(\"--md-indicator-width\")\n          }\n        })\n\n    /* Hide content tab buttons on borders */\n    combineLatest([\n      watchElementContentOffset(container),\n      watchElementSize(container)\n    ])\n      .pipe(\n        takeUntil(done$)\n      )\n        .subscribe(([offset, size]) => {\n          const content = getElementContentSize(container)\n          prev.hidden = offset.x < 16\n          next.hidden = offset.x > content.width - size.width - 16\n        })\n\n    /* Paginate content tab container on click */\n    merge(\n      fromEvent(prev, \"click\").pipe(map(() => -1)),\n      fromEvent(next, \"click\").pipe(map(() => +1))\n    )\n      .pipe(\n        takeUntil(done$)\n      )\n        .subscribe(direction => {\n          const { width } = getElementSize(container)\n          container.scrollBy({\n            left: width * direction,\n            behavior: \"smooth\"\n          })\n        })\n\n    /* Switch to content tab target */\n    target$\n      .pipe(\n        takeUntil(done$),\n        filter(input => inputs.includes(input as HTMLInputElement))\n      )\n        .subscribe(input => input.click())\n\n    /* Add link to each content tab label */\n    container.classList.add(\"tabbed-labels--linked\")\n    for (const input of inputs) {\n      const label = getElement<HTMLLabelElement>(`label[for=\"${input.id}\"]`)\n      label.replaceChildren(h(\"a\", {\n        href: `#${label.htmlFor}`,\n        tabIndex: -1\n      }, ...Array.from(label.childNodes)))\n\n      /* Allow to copy link without scrolling to anchor */\n      fromEvent<MouseEvent>(label.firstElementChild!, \"click\")\n        .pipe(\n          takeUntil(done$),\n          filter(ev => !(ev.metaKey || ev.ctrlKey)),\n          tap(ev => {\n            ev.preventDefault()\n            ev.stopPropagation()\n          })\n        )\n          // @todo we might need to remove the anchor link on complete\n          .subscribe(() => {\n            history.replaceState({}, \"\", `#${label.htmlFor}`)\n            label.click()\n          })\n    }\n\n    /* Set up linking of content tabs, if enabled */\n    if (feature(\"content.tabs.link\"))\n      push$.pipe(\n        skip(1),\n        withLatestFrom(viewport$)\n      )\n        .subscribe(([{ active }, { offset }]) => {\n          const tab = active.innerText.trim()\n          if (active.hasAttribute(\"data-md-switching\")) {\n            active.removeAttribute(\"data-md-switching\")\n\n          /* Determine viewport offset of active tab */\n          } else {\n            const y = el.offsetTop - offset.y\n\n            /* Passively activate other tabs */\n            for (const set of getElements(\"[data-tabs]\"))\n              for (const input of getElements<HTMLInputElement>(\n                \":scope > input\", set\n              )) {\n                const label = getElement(`label[for=\"${input.id}\"]`)\n                if (\n                  label !== active &&\n                  label.innerText.trim() === tab\n                ) {\n                  label.setAttribute(\"data-md-switching\", \"\")\n                  input.click()\n                  break\n                }\n              }\n\n            /* Bring active tab into view */\n            window.scrollTo({\n              top: el.offsetTop - y\n            })\n\n            /* Persist active tabs in local storage */\n            const tabs = __md_get<string[]>(\"__tabs\") || []\n            __md_set(\"__tabs\", [...new Set([tab, ...tabs])])\n          }\n        })\n\n    /* Pause media (audio, video) on switch - see https://bit.ly/3Bk6cel */\n    push$.pipe(takeUntil(done$))\n      .subscribe(() => {\n        for (const media of getElements<HTMLAudioElement>(\"audio, video\", el))\n          media.pause()\n      })\n\n    /* Create and return component */\n    return watchContentTabs(inputs)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n    .pipe(\n      subscribeOn(asyncScheduler)\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { Observable, merge } from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport { Viewport, getElements } from \"~/browser\"\n\nimport { Component } from \"../../_\"\nimport {\n  Tooltip,\n  mountTooltip\n} from \"../../tooltip\"\nimport {\n  Annotation,\n  mountAnnotationBlock\n} from \"../annotation\"\nimport {\n  CodeBlock,\n  mountCodeBlock\n} from \"../code\"\nimport {\n  Details,\n  mountDetails\n} from \"../details\"\nimport {\n  Mermaid,\n  mountMermaid\n} from \"../mermaid\"\nimport {\n  DataTable,\n  mountDataTable\n} from \"../table\"\nimport {\n  ContentTabs,\n  mountContentTabs\n} from \"../tabs\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Content\n */\nexport type Content =\n  | Annotation\n  | CodeBlock\n  | ContentTabs\n  | DataTable\n  | Details\n  | Mermaid\n  | Tooltip\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  target$: Observable<HTMLElement>     /* Location target observable */\n  print$: Observable<boolean>          /* Media print observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount content\n *\n * This function mounts all components that are found in the content of the\n * actual article, including code blocks, data tables and details.\n *\n * @param el - Content element\n * @param options - Options\n *\n * @returns Content component observable\n */\nexport function mountContent(\n  el: HTMLElement, { viewport$, target$, print$ }: MountOptions\n): Observable<Component<Content>> {\n  return merge(\n\n    /* Annotations */\n    ...getElements(\".annotate:not(.highlight)\", el)\n      .map(child => mountAnnotationBlock(child, { target$, print$ })),\n\n    /* Code blocks */\n    ...getElements(\"pre:not(.mermaid) > code\", el)\n      .map(child => mountCodeBlock(child, { target$, print$ })),\n\n    /* Mermaid diagrams */\n    ...getElements(\"pre.mermaid\", el)\n      .map(child => mountMermaid(child)),\n\n    /* Data tables */\n    ...getElements(\"table:not([class])\", el)\n      .map(child => mountDataTable(child)),\n\n    /* Details */\n    ...getElements(\"details\", el)\n      .map(child => mountDetails(child, { target$, print$ })),\n\n    /* Content tabs */\n    ...getElements(\"[data-tabs]\", el)\n      .map(child => mountContentTabs(child, { viewport$, target$ })),\n\n    /* Tooltips */\n    ...getElements(\"[title]\", el)\n      .filter(() => feature(\"content.tooltips\"))\n      .map(child => mountTooltip(child))\n  )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  defer,\n  delay,\n  finalize,\n  map,\n  merge,\n  of,\n  switchMap,\n  tap\n} from \"rxjs\"\n\nimport { getElement } from \"~/browser\"\n\nimport { Component } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Dialog\n */\nexport interface Dialog {\n  message: string                      /* Dialog message */\n  active: boolean                      /* Dialog is active */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  alert$: Subject<string>              /* Alert subject */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  alert$: Subject<string>              /* Alert subject */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch dialog\n *\n * @param _el - Dialog element\n * @param options - Options\n *\n * @returns Dialog observable\n */\nexport function watchDialog(\n  _el: HTMLElement, { alert$ }: WatchOptions\n): Observable<Dialog> {\n  return alert$\n    .pipe(\n      switchMap(message => merge(\n        of(true),\n        of(false).pipe(delay(2000))\n      )\n        .pipe(\n          map(active => ({ message, active }))\n        )\n      )\n    )\n}\n\n/**\n * Mount dialog\n *\n * This function reveals the dialog in the right corner when a new alert is\n * emitted through the subject that is passed as part of the options.\n *\n * @param el - Dialog element\n * @param options - Options\n *\n * @returns Dialog component observable\n */\nexport function mountDialog(\n  el: HTMLElement, options: MountOptions\n): Observable<Component<Dialog>> {\n  const inner = getElement(\".md-typeset\", el)\n  return defer(() => {\n    const push$ = new Subject<Dialog>()\n    push$.subscribe(({ message, active }) => {\n      el.classList.toggle(\"md-dialog--active\", active)\n      inner.textContent = message\n    })\n\n    /* Create and return component */\n    return watchDialog(el, options)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  bufferCount,\n  combineLatest,\n  combineLatestWith,\n  defer,\n  distinctUntilChanged,\n  distinctUntilKeyChanged,\n  endWith,\n  filter,\n  from,\n  ignoreElements,\n  map,\n  mergeMap,\n  mergeWith,\n  of,\n  shareReplay,\n  startWith,\n  switchMap,\n  takeUntil\n} from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport {\n  Viewport,\n  getElements,\n  watchElementSize,\n  watchToggle\n} from \"~/browser\"\n\nimport { Component } from \"../../_\"\nimport { Main } from \"../../main\"\nimport {\n  Tooltip,\n  mountTooltip\n} from \"../../tooltip\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Header\n */\nexport interface Header {\n  height: number                       /* Header visible height */\n  hidden: boolean                      /* Header is hidden */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n  main$: Observable<Main>              /* Main area observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Compute whether the header is hidden\n *\n * If the user scrolls past a certain threshold, the header can be hidden when\n * scrolling down, and shown when scrolling up.\n *\n * @param options - Options\n *\n * @returns Toggle observable\n */\nfunction isHidden({ viewport$ }: WatchOptions): Observable<boolean> {\n  if (!feature(\"header.autohide\"))\n    return of(false)\n\n  /* Compute direction and turning point */\n  const direction$ = viewport$\n    .pipe(\n      map(({ offset: { y } }) => y),\n      bufferCount(2, 1),\n      map(([a, b]) => [a < b, b] as const),\n      distinctUntilKeyChanged(0)\n    )\n\n  /* Compute whether header should be hidden */\n  const hidden$ = combineLatest([viewport$, direction$])\n    .pipe(\n      filter(([{ offset }, [, y]]) => Math.abs(y - offset.y) > 100),\n      map(([, [direction]]) => direction),\n      distinctUntilChanged()\n    )\n\n  /* Compute threshold for hiding */\n  const search$ = watchToggle(\"search\")\n  return combineLatest([viewport$, search$])\n    .pipe(\n      map(([{ offset }, search]) => offset.y > 400 && !search),\n      distinctUntilChanged(),\n      switchMap(active => active ? hidden$ : of(false)),\n      startWith(false)\n    )\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch header\n *\n * @param el - Header element\n * @param options - Options\n *\n * @returns Header observable\n */\nexport function watchHeader(\n  el: HTMLElement, options: WatchOptions\n): Observable<Header> {\n  return defer(() => combineLatest([\n    watchElementSize(el),\n    isHidden(options)\n  ]))\n    .pipe(\n      map(([{ height }, hidden]) => ({\n        height,\n        hidden\n      })),\n      distinctUntilChanged((a, b) => (\n        a.height === b.height &&\n        a.hidden === b.hidden\n      )),\n      shareReplay(1)\n    )\n}\n\n/**\n * Mount header\n *\n * This function manages the different states of the header, i.e. whether it's\n * hidden or rendered with a shadow. This depends heavily on the main area.\n *\n * @param el - Header element\n * @param options - Options\n *\n * @returns Header component observable\n */\nexport function mountHeader(\n  el: HTMLElement, { header$, main$ }: MountOptions\n): Observable<Component<Header | Tooltip>> {\n  return defer(() => {\n    const push$ = new Subject<Main>()\n    const done$ = push$.pipe(ignoreElements(), endWith(true))\n    push$\n      .pipe(\n        distinctUntilKeyChanged(\"active\"),\n        combineLatestWith(header$)\n      )\n        .subscribe(([{ active }, { hidden }]) => {\n          el.classList.toggle(\"md-header--shadow\", active && !hidden)\n          el.hidden = hidden\n        })\n\n    /* Mount tooltips, if enabled */\n    const tooltips = from(getElements(\"[title]\", el))\n      .pipe(\n        filter(() => feature(\"content.tooltips\")),\n        mergeMap(child => mountTooltip(child))\n      )\n\n    /* Link to main area */\n    main$.subscribe(push$)\n\n    /* Create and return component */\n    return header$\n      .pipe(\n        takeUntil(done$),\n        map(state => ({ ref: el, ...state })),\n        mergeWith(tooltips.pipe(takeUntil(done$)))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  defer,\n  distinctUntilKeyChanged,\n  finalize,\n  map,\n  tap\n} from \"rxjs\"\n\nimport {\n  Viewport,\n  getElementSize,\n  getOptionalElement,\n  watchViewportAt\n} from \"~/browser\"\n\nimport { Component } from \"../../_\"\nimport { Header } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Header\n */\nexport interface HeaderTitle {\n  active: boolean                      /* Header title is active */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch header title\n *\n * @param el - Heading element\n * @param options - Options\n *\n * @returns Header title observable\n */\nexport function watchHeaderTitle(\n  el: HTMLElement, { viewport$, header$ }: WatchOptions\n): Observable<HeaderTitle> {\n  return watchViewportAt(el, { viewport$, header$ })\n    .pipe(\n      map(({ offset: { y } }) => {\n        const { height } = getElementSize(el)\n        return {\n          active: y >= height\n        }\n      }),\n      distinctUntilKeyChanged(\"active\")\n    )\n}\n\n/**\n * Mount header title\n *\n * This function swaps the header title from the site title to the title of the\n * current page when the user scrolls past the first headline.\n *\n * @param el - Header title element\n * @param options - Options\n *\n * @returns Header title component observable\n */\nexport function mountHeaderTitle(\n  el: HTMLElement, options: MountOptions\n): Observable<Component<HeaderTitle>> {\n  return defer(() => {\n    const push$ = new Subject<HeaderTitle>()\n    push$.subscribe({\n\n      /* Handle emission */\n      next({ active }) {\n        el.classList.toggle(\"md-header__title--active\", active)\n      },\n\n      /* Handle complete */\n      complete() {\n        el.classList.remove(\"md-header__title--active\")\n      }\n    })\n\n    /* Obtain headline, if any */\n    const heading = getOptionalElement(\".md-content h1\")\n    if (typeof heading === \"undefined\")\n      return EMPTY\n\n    /* Create and return component */\n    return watchHeaderTitle(heading, options)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  combineLatest,\n  distinctUntilChanged,\n  distinctUntilKeyChanged,\n  map,\n  switchMap\n} from \"rxjs\"\n\nimport {\n  Viewport,\n  watchElementSize\n} from \"~/browser\"\n\nimport { Header } from \"../header\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Main area\n */\nexport interface Main {\n  offset: number                       /* Main area top offset */\n  height: number                       /* Main area visible height */\n  active: boolean                      /* Main area is active */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch main area\n *\n * This function returns an observable that computes the visual parameters of\n * the main area which depends on the viewport vertical offset and height, as\n * well as the height of the header element, if the header is fixed.\n *\n * @param el - Main area element\n * @param options - Options\n *\n * @returns Main area observable\n */\nexport function watchMain(\n  el: HTMLElement, { viewport$, header$ }: WatchOptions\n): Observable<Main> {\n\n  /* Compute necessary adjustment for header */\n  const adjust$ = header$\n    .pipe(\n      map(({ height }) => height),\n      distinctUntilChanged()\n    )\n\n  /* Compute the main area's top and bottom borders */\n  const border$ = adjust$\n    .pipe(\n      switchMap(() => watchElementSize(el)\n        .pipe(\n          map(({ height }) => ({\n            top:    el.offsetTop,\n            bottom: el.offsetTop + height\n          })),\n          distinctUntilKeyChanged(\"bottom\")\n        )\n      )\n    )\n\n  /* Compute the main area's offset, visible height and if we scrolled past */\n  return combineLatest([adjust$, border$, viewport$])\n    .pipe(\n      map(([header, { top, bottom }, { offset: { y }, size: { height } }]) => {\n        height = Math.max(0, height\n          - Math.max(0, top    - y,  header)\n          - Math.max(0, height + y - bottom)\n        )\n        return {\n          offset: top - header,\n          height,\n          active: top - header <= y\n        }\n      }),\n      distinctUntilChanged((a, b) => (\n        a.offset === b.offset &&\n        a.height === b.height &&\n        a.active === b.active\n      ))\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  asyncScheduler,\n  defer,\n  filter,\n  finalize,\n  fromEvent,\n  map,\n  mergeMap,\n  observeOn,\n  of,\n  repeat,\n  shareReplay,\n  skip,\n  startWith,\n  takeUntil,\n  tap,\n  withLatestFrom\n} from \"rxjs\"\n\nimport { getElements, watchMedia } from \"~/browser\"\nimport { h } from \"~/utilities\"\n\nimport {\n  Component,\n  getComponentElement\n} from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Palette colors\n */\nexport interface PaletteColor {\n  media?: string                       /* Media query */\n  scheme?: string                      /* Color scheme */\n  primary?: string                     /* Primary color */\n  accent?: string                      /* Accent color */\n}\n\n/**\n * Palette\n */\nexport interface Palette {\n  index: number                        /* Palette index */\n  color: PaletteColor                  /* Palette colors */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch color palette\n *\n * @param inputs - Color palette element\n *\n * @returns Color palette observable\n */\nexport function watchPalette(\n  inputs: HTMLInputElement[]\n): Observable<Palette> {\n  const current = __md_get<Palette>(\"__palette\") || {\n    index: inputs.findIndex(input => matchMedia(\n      input.getAttribute(\"data-md-color-media\")!\n    ).matches)\n  }\n\n  /* Emit changes in color palette */\n  const index = Math.max(0, Math.min(current.index, inputs.length - 1))\n  return of(...inputs)\n    .pipe(\n      mergeMap(input => fromEvent(input, \"change\").pipe(map(() => input))),\n      startWith(inputs[index]),\n      map(input => ({\n        index: inputs.indexOf(input),\n        color: {\n          media:   input.getAttribute(\"data-md-color-media\"),\n          scheme:  input.getAttribute(\"data-md-color-scheme\"),\n          primary: input.getAttribute(\"data-md-color-primary\"),\n          accent:  input.getAttribute(\"data-md-color-accent\")\n        }\n      } as Palette)),\n      shareReplay(1)\n    )\n}\n\n/**\n * Mount color palette\n *\n * @param el - Color palette element\n *\n * @returns Color palette component observable\n */\nexport function mountPalette(\n  el: HTMLElement\n): Observable<Component<Palette>> {\n  const inputs = getElements<HTMLInputElement>(\"input\", el)\n  const meta = h(\"meta\", { name: \"theme-color\" })\n  document.head.appendChild(meta)\n\n  // Add color scheme meta tag\n  const scheme = h(\"meta\", { name: \"color-scheme\" })\n  document.head.appendChild(scheme)\n\n  /* Mount component on subscription */\n  const media$ = watchMedia(\"(prefers-color-scheme: light)\")\n  return defer(() => {\n    const push$ = new Subject<Palette>()\n    push$.subscribe(palette => {\n      document.body.setAttribute(\"data-md-color-switching\", \"\")\n\n      /* Retrieve color palette for system preference */\n      if (palette.color.media === \"(prefers-color-scheme)\") {\n        const media = matchMedia(\"(prefers-color-scheme: light)\")\n        const input = document.querySelector(media.matches\n          ? \"[data-md-color-media='(prefers-color-scheme: light)']\"\n          : \"[data-md-color-media='(prefers-color-scheme: dark)']\"\n        )!\n\n        /* Retrieve colors for system preference */\n        palette.color.scheme  = input.getAttribute(\"data-md-color-scheme\")!\n        palette.color.primary = input.getAttribute(\"data-md-color-primary\")!\n        palette.color.accent  = input.getAttribute(\"data-md-color-accent\")!\n      }\n\n      /* Set color palette */\n      for (const [key, value] of Object.entries(palette.color))\n        document.body.setAttribute(`data-md-color-${key}`, value)\n\n      /* Set toggle visibility */\n      for (let index = 0; index < inputs.length; index++) {\n        const label = inputs[index].nextElementSibling\n        if (label instanceof HTMLElement)\n          label.hidden = palette.index !== index\n      }\n\n      /* Persist preference in local storage */\n      __md_set(\"__palette\", palette)\n    })\n\n    // Handle color switch on Enter or Space - see https://t.ly/YIhVj\n    fromEvent<KeyboardEvent>(el, \"keydown\").pipe(\n      filter(ev => ev.key === \"Enter\"),\n      withLatestFrom(push$, (_, palette) => palette)\n    )\n      .subscribe(({ index }) => {\n        index = (index + 1) % inputs.length\n        inputs[index].click()\n        inputs[index].focus()\n      })\n\n    /* Update theme-color meta tag */\n    push$\n      .pipe(\n        map(() => {\n          const header = getComponentElement(\"header\")\n          const style  = window.getComputedStyle(header)\n\n          // Set color scheme\n          scheme.content = style.colorScheme\n\n          /* Return color in hexadecimal format */\n          return style.backgroundColor.match(/\\d+/g)!\n            .map(value => (+value).toString(16).padStart(2, \"0\"))\n            .join(\"\")\n        })\n      )\n        .subscribe(color => meta.content = `#${color}`)\n\n    /* Revert transition durations after color switch */\n    push$.pipe(observeOn(asyncScheduler))\n      .subscribe(() => {\n        document.body.removeAttribute(\"data-md-color-switching\")\n      })\n\n    /* Create and return component */\n    return watchPalette(inputs)\n      .pipe(\n        takeUntil(media$.pipe(skip(1))),\n        repeat(),\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  defer,\n  finalize,\n  map,\n  tap\n} from \"rxjs\"\n\nimport { Component } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Progress indicator\n */\nexport interface Progress {\n  value: number                        // Progress value\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  progress$: Subject<number>           // Progress subject\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount progress indicator\n *\n * @param el - Progress indicator element\n * @param options - Options\n *\n * @returns Progress indicator component observable\n */\nexport function mountProgress(\n  el: HTMLElement, { progress$ }: MountOptions\n): Observable<Component<Progress>> {\n\n  // Mount component on subscription\n  return defer(() => {\n    const push$ = new Subject<Progress>()\n    push$.subscribe(({ value }) => {\n      el.style.setProperty(\"--md-progress-value\", `${value}`)\n    })\n\n    // Create and return component\n    return progress$\n      .pipe(\n        tap(value => push$.next({ value })),\n        finalize(() => push$.complete()),\n        map(value => ({ ref: el, value }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport ClipboardJS from \"clipboard\"\nimport {\n  Observable,\n  Subject,\n  map,\n  tap\n} from \"rxjs\"\n\nimport { translation } from \"~/_\"\nimport { getElement } from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Setup options\n */\ninterface SetupOptions {\n  alert$: Subject<string>              /* Alert subject */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Extract text to copy\n *\n * @param el - HTML element\n *\n * @returns Extracted text\n */\nfunction extract(el: HTMLElement): string {\n  el.setAttribute(\"data-md-copying\", \"\")\n  const copy = el.closest(\"[data-copy]\")\n  const text = copy\n    ? copy.getAttribute(\"data-copy\")!\n    : el.innerText\n  el.removeAttribute(\"data-md-copying\")\n  return text.trimEnd()\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Set up Clipboard.js integration\n *\n * @param options - Options\n */\nexport function setupClipboardJS(\n  { alert$ }: SetupOptions\n): void {\n  if (ClipboardJS.isSupported()) {\n    new Observable<ClipboardJS.Event>(subscriber => {\n      new ClipboardJS(\"[data-clipboard-target], [data-clipboard-text]\", {\n        text: el => (\n          el.getAttribute(\"data-clipboard-text\")! ||\n          extract(getElement(\n            el.getAttribute(\"data-clipboard-target\")!\n          ))\n        )\n      })\n        .on(\"success\", ev => subscriber.next(ev))\n    })\n      .pipe(\n        tap(ev => {\n          const trigger = ev.trigger as HTMLElement\n          trigger.focus()\n        }),\n        map(() => translation(\"clipboard.copied\"))\n      )\n        .subscribe(alert$)\n  }\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  catchError,\n  map,\n  of\n} from \"rxjs\"\n\nimport {\n  getElement,\n  getElements,\n  requestXML\n} from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Sitemap, i.e. a list of URLs\n */\nexport type Sitemap = Map<string, URL[]>\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Resolve URL to the given base URL\n *\n * When serving the site with instant navigation, MkDocs will set the hostname\n * to the value as specified in `dev_addr`, but the browser allows for several\n * hostnames to be used: `localhost`, `127.0.0.1` or even `0.0.0.0`, depending\n * on configuration. This function resolves the URL to the given hostname.\n *\n * @param url - URL\n * @param base - Base URL\n *\n * @returns Resolved URL\n */\nfunction resolve(url: URL, base: URL) {\n  url.protocol = base.protocol\n  url.hostname = base.hostname\n  return url\n}\n\n/**\n * Extract sitemap from document\n *\n * This function extracts the URLs and alternate links from the document, and\n * associates alternate links to the original URL as found in `loc`, allowing\n * the browser to navigate to the correct page when switching languages. The\n * format of the sitemap is expected to adhere to:\n *\n * ``` xml\n * <urlset>\n *   <url>\n *     <loc>...</loc>\n *     <xhtml:link rel=\"alternate\" hreflang=\"en\" href=\"...\"/>\n *     <xhtml:link rel=\"alternate\" hreflang=\"de\" href=\"...\"/>\n *     ...\n *   </url>\n *   ...\n * </urlset>\n * ```\n *\n * @param document - Document\n * @param base - Base URL\n *\n * @returns Sitemap\n */\nfunction extract(document: Document, base: URL): Sitemap {\n  const sitemap: Sitemap = new Map()\n  for (const el of getElements(\"url\", document)) {\n    const url = getElement(\"loc\", el)\n\n    // Create entry for location and add it to the list of links\n    const links = [resolve(new URL(url.textContent!), base)]\n    sitemap.set(`${links[0]}`, links)\n\n    // Attach alternate links to current entry\n    for (const link of getElements(\"[rel=alternate]\", el)) {\n      const href = link.getAttribute(\"href\")\n      if (href != null)\n        links.push(resolve(new URL(href), base))\n    }\n  }\n\n  // Return sitemap\n  return sitemap\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch the sitemap for the given base URL\n *\n * If a network or parsing error occurs, we just default to an empty sitemap,\n * which means the caller should fall back to regular navigation.\n *\n * @param base - Base URL\n *\n * @returns Sitemap observable\n */\nexport function fetchSitemap(base: URL | string): Observable<Sitemap> {\n  return requestXML(new URL(\"sitemap.xml\", base))\n    .pipe(\n      map(document => extract(document, new URL(base))),\n      catchError(() => of(new Map())),\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  catchError,\n  combineLatestWith,\n  concat,\n  debounceTime,\n  distinctUntilChanged,\n  distinctUntilKeyChanged,\n  endWith,\n  fromEvent,\n  ignoreElements,\n  map,\n  merge,\n  of,\n  share,\n  switchMap,\n  tap,\n  withLatestFrom\n} from \"rxjs\"\n\nimport { configuration, feature } from \"~/_\"\nimport {\n  Viewport,\n  getElements,\n  getLocation,\n  getOptionalElement,\n  requestHTML,\n  setLocation,\n  setLocationHash\n} from \"~/browser\"\nimport { getComponentElement } from \"~/components\"\n\nimport { Sitemap, fetchSitemap } from \"../sitemap\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Setup options\n */\ninterface SetupOptions {\n  location$: Subject<URL>              // Location subject\n  viewport$: Observable<Viewport>      // Viewport observable\n  progress$: Subject<number>           // Progress subject\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Handle clicks on internal URLs while skipping external URLs\n *\n * @param ev - Mouse event\n * @param sitemap - Sitemap\n *\n * @returns URL observable\n */\nfunction handle(\n  ev: MouseEvent, sitemap: Sitemap\n): Observable<URL> {\n  if (!(ev.target instanceof Element))\n    return EMPTY\n\n  // Skip, as target is not within a link - clicks on non-link elements are\n  // also captured, which we need to exclude from processing\n  const el = ev.target.closest(\"a\")\n  if (el === null)\n    return EMPTY\n\n  // Skip, as link opens in new window - we now know we have captured a click\n  // on a link, but the link either has a `target` property defined, or the\n  // user pressed the `meta` or `ctrl` key to open it in a new window. Thus,\n  // we need to filter this event as well.\n  if (el.target || ev.metaKey || ev.ctrlKey)\n    return EMPTY\n\n  // Next, we must check if the URL is relevant for us, i.e., if it's an\n  // internal link to a page that is managed by MkDocs. Only then we can be\n  // sure that the structure of the page to be loaded adheres to the current\n  // document structure and can subsequently be injected into it without doing\n  // a full reload. For this reason, we must canonicalize the URL by removing\n  // all search parameters and hash fragments.\n  const url = new URL(el.href)\n  url.search = url.hash = \"\"\n\n  // Skip, if URL is not included in the sitemap - this could be the case when\n  // linking between versions or languages, or to another page that the author\n  // included as part of the build, but that is not managed by MkDocs. In that\n  // case we must not continue with instant navigation.\n  if (!sitemap.has(`${url}`))\n    return EMPTY\n\n  // We now know that we have a link to an internal page, so we prevent the\n  // browser from navigation and emit the URL for instant navigation. Note that\n  // this also includes anchor links, which means we need to implement anchor\n  // positioning ourselves. The reason for this is that if we wouldn't manage\n  // anchor links as well, scroll restoration will not work correctly (e.g.\n  // following an anchor link and scrolling).\n  ev.preventDefault()\n  return of(new URL(el.href))\n}\n\n/**\n * Create a map of head elements for lookup and replacement\n *\n * @param document - Document\n *\n * @returns Tag map\n */\nfunction head(document: Document): Map<string, HTMLElement> {\n  const tags = new Map<string, HTMLElement>()\n  for (const el of getElements(\":scope > *\", document.head))\n    tags.set(el.outerHTML, el)\n\n  // Return tag map\n  return tags\n}\n\n/**\n * Resolve relative URLs in the given document\n *\n * This function resolves relative `href` and `src` attributes, which can belong\n * to all sorts of tags, like meta tags, links, images, scripts and more.\n *\n * @param document - Document\n *\n * @returns Document observable\n */\nfunction resolve(document: Document): Observable<Document> {\n  for (const el of getElements(\"[href], [src]\", document))\n    for (const key of [\"href\", \"src\"]) {\n      const value = el.getAttribute(key)\n      if (value && !/^(?:[a-z]+:)?\\/\\//i.test(value)) {\n        // @ts-expect-error - trick: self-assign to resolve URL\n        el[key] = el[key]\n        break\n      }\n    }\n\n  // Return document observable\n  return of(document)\n}\n\n/**\n * Inject the contents of a document into the current one\n *\n * @param next - Next document\n *\n * @returns Document observable\n */\nfunction inject(next: Document): Observable<Document> {\n  for (const selector of [\n    \"[data-md-component=announce]\",\n    \"[data-md-component=container]\",\n    \"[data-md-component=header-topic]\",\n    \"[data-md-component=outdated]\",\n    \"[data-md-component=logo]\",\n    \"[data-md-component=skip]\",\n    ...feature(\"navigation.tabs.sticky\")\n      ? [\"[data-md-component=tabs]\"]\n      : []\n  ]) {\n    const source = getOptionalElement(selector)\n    const target = getOptionalElement(selector, next)\n    if (\n      typeof source !== \"undefined\" &&\n      typeof target !== \"undefined\"\n    ) {\n      source.replaceWith(target)\n    }\n  }\n\n  // Update meta tags\n  const tags = head(document)\n  for (const [html, el] of head(next))\n    if (tags.has(html))\n      tags.delete(html)\n    else\n      document.head.appendChild(el)\n\n  // Remove meta tags that are not present in the new document\n  for (const el of tags.values()) {\n    const name = el.getAttribute(\"name\")\n    // @todo - find a better way to handle attributes we add dynamically in\n    // other components without mounting components on every navigation, as\n    // this might impact overall performance - see https://t.ly/ehp_O\n    if (name !== \"theme-color\" && name !== \"color-scheme\")\n      el.remove()\n  }\n\n  // After components and meta tags were replaced, re-evaluate scripts\n  // that were provided by the author as part of Markdown files\n  const container = getComponentElement(\"container\")\n  return concat(getElements(\"script\", container))\n    .pipe(\n      switchMap(el => {\n        const script = next.createElement(\"script\")\n        if (el.src) {\n          for (const name of el.getAttributeNames())\n            script.setAttribute(name, el.getAttribute(name)!)\n          el.replaceWith(script)\n\n          // Complete when script is loaded\n          return new Observable(observer => {\n            script.onload = () => observer.complete()\n          })\n\n        // Complete immediately\n        } else {\n          script.textContent = el.textContent\n          el.replaceWith(script)\n          return EMPTY\n        }\n      }),\n      ignoreElements(),\n      endWith(document)\n    )\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Set up instant navigation\n *\n * This is a heavily orchestrated operation - see inline comments to learn how\n * this works with Material for MkDocs, and how you can hook into it.\n *\n * @param options - Options\n *\n * @returns Document observable\n */\nexport function setupInstantNavigation(\n  { location$, viewport$, progress$ }: SetupOptions\n): Observable<Document> {\n  const config = configuration()\n  if (location.protocol === \"file:\")\n    return EMPTY\n\n  // Load sitemap immediately, so we have it available when the user initiates\n  // the first navigation request without any perceivable delay\n  const sitemap$ = fetchSitemap(config.base)\n\n  // Since we might be on a slow connection, the user might trigger multiple\n  // instant navigation events that overlap. MkDocs produces relative URLs for\n  // all internal links, which becomes a problem in this case, because we need\n  // to change the base URL the moment the user clicks a link that should be\n  // intercepted in order to be consistent with popstate, which means that the\n  // base URL would now be incorrect when resolving another relative link from\n  // the same site. For this reason we always resolve all relative links to\n  // absolute links, so we can be sure this never happens.\n  of(document)\n    .subscribe(resolve)\n\n  // --------------------------------------------------------------------------\n  // Navigation interception\n  // --------------------------------------------------------------------------\n\n  // Intercept navigation - to keep the number of event listeners down we use\n  // the fact that uncaptured events bubble up to the body. This has the nice\n  // property that we don't need to detach and then re-attach event listeners\n  // when the document is replaced after a navigation event.\n  const instant$ =\n    fromEvent<MouseEvent>(document.body, \"click\")\n      .pipe(\n        combineLatestWith(sitemap$),\n        switchMap(([ev, sitemap]) => handle(ev, sitemap)),\n        share()\n      )\n\n  // Intercept history change events, e.g. when the user uses the browser's\n  // back or forward buttons, and emit new location for fetching and parsing\n  const history$ =\n    fromEvent<PopStateEvent>(window, \"popstate\")\n      .pipe(\n        map(getLocation),\n        share()\n      )\n\n  // While it would be better UX to defer navigation events until the document\n  // is fully fetched and parsed, we must schedule it here to synchronize with\n  // popstate events, as they are emitted immediately. Moreover we need to\n  // store the current viewport offset for scroll restoration later on.\n  instant$.pipe(withLatestFrom(viewport$))\n    .subscribe(([url, { offset }]) => {\n      history.replaceState(offset, \"\")\n      history.pushState(null, \"\", url)\n    })\n\n  // Emit URLs that should be fetched via instant navigation on location subject\n  // which was passed into this function. The state of instant navigation can be\n  // intercepted by other parts of the application, which can synchronously back\n  // up or restore state before or after instant navigation happens.\n  merge(instant$, history$)\n    .subscribe(location$)\n\n  // --------------------------------------------------------------------------\n  // Fetching and parsing\n  // --------------------------------------------------------------------------\n\n  // Fetch document - we deduplicate requests to the same location, so we don't\n  // end up with multiple requests for the same page. We use `switchMap`, since\n  // we want to cancel the previous request when a new one is triggered, which\n  // is automatically handled by the observable returned by `request`. This is\n  // essential to ensure a good user experience, as we don't want to load pages\n  // that are not needed anymore, e.g., when the user clicks multiple links in\n  // quick succession or on slow connections. If the request fails for some\n  // reason, we fall back and use regular navigation, forcing a reload.\n  const document$ =\n    location$.pipe(\n      distinctUntilKeyChanged(\"pathname\"),\n      switchMap(url => requestHTML(url, { progress$ })\n        .pipe(\n          catchError(() => {\n            setLocation(url, true)\n            return EMPTY\n          })\n        )\n      ),\n\n      // The document was successfully fetched and parsed, so we can inject its\n      // contents into the currently active document\n      switchMap(resolve),\n      switchMap(inject),\n      share()\n    )\n\n  // --------------------------------------------------------------------------\n  // Scroll restoration\n  // --------------------------------------------------------------------------\n\n  // Handle scroll restoration - we must restore the viewport offset after the\n  // document has been fetched and injected, and every time the user clicks an\n  // anchor that leads to an element on the same page, which might also happen\n  // when the user uses the back or forward button.\n  merge(\n    document$.pipe(withLatestFrom(location$, (_, url) => url)),\n\n    // Handle instant navigation events that are triggered by the user clicking\n    // on an anchor link with a hash fragment different from the current one, as\n    // well as from popstate events, which are emitted when the user navigates\n    // back and forth between pages. We use a two-layered subscription to scope\n    // the scroll restoration to the current page, as we don't need to restore\n    // the viewport offset when the user navigates to a different page, as this\n    // is already handled by the previous observable.\n    location$.pipe(\n      distinctUntilKeyChanged(\"pathname\"),\n      switchMap(() => location$),\n      distinctUntilKeyChanged(\"hash\"),\n    ),\n\n    // Handle instant navigation events that are triggered by the user clicking\n    // on an anchor link with the same hash fragment as the current one in the\n    // URL. Is is essential that we only intercept those from instant navigation\n    // events and not from history change events, or we'll end up in and endless\n    // loop. The top-level history entry must be removed, as it will be replaced\n    // with a new one, which would otherwise lead to a duplicate entry.\n    location$.pipe(\n      distinctUntilChanged((a, b) => (\n        a.pathname === b.pathname &&\n        a.hash     === b.hash\n      )),\n      switchMap(() => instant$),\n      tap(() => history.back())\n    )\n  )\n    .subscribe(url => {\n\n      // Check if the current history entry has a state, which happens when the\n      // user presses the back or forward button to visit a page we've already\n      // seen. If there's no state, it means a new page was visited and we must\n      // scroll to the top, unless an anchor is given.\n      if (history.state !== null || !url.hash) {\n        window.scrollTo(0, history.state?.y ?? 0)\n      } else {\n        history.scrollRestoration = \"auto\"\n        setLocationHash(url.hash)\n        history.scrollRestoration = \"manual\"\n      }\n    })\n\n  // Disable scroll restoration when an instant navigation event occurs, so the\n  // browser does not immediately set the viewport offset to the prior history\n  // entry, scrolling to the position on the same page, which would look odd.\n  // Instead, we manually restore the position once the page has loaded.\n  location$.subscribe(() => {\n    history.scrollRestoration = \"manual\"\n  })\n\n  // Enable scroll restoration before window unloads - this is essential to\n  // ensure that full reloads (F5) restore the viewport offset correctly. If\n  // only popstate events wouldn't reset the viewport offset prior to their\n  // emission, we could just reset this in popstate. Meh.\n  fromEvent(window, \"beforeunload\")\n    .subscribe(() => {\n      history.scrollRestoration = \"auto\"\n    })\n\n  // Track viewport offset, so we can restore it when the user navigates back\n  // and forth between pages. Note that this must be debounced and cannot be\n  // done in popstate, as popstate has already removed the entry from the\n  // history, which means it is too late.\n  viewport$.pipe(\n    distinctUntilKeyChanged(\"offset\"),\n    debounceTime(100)\n  )\n    .subscribe(({ offset }) => {\n      history.replaceState(offset, \"\")\n    })\n\n  // Return document observable\n  return document$\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport escapeHTML from \"escape-html\"\n\nimport { SearchConfig } from \"../config\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search highlight function\n *\n * @param value - Value\n *\n * @returns Highlighted value\n */\nexport type SearchHighlightFn = (value: string) => string\n\n/**\n * Search highlight factory function\n *\n * @param query - Query value\n *\n * @returns Search highlight function\n */\nexport type SearchHighlightFactoryFn = (query: string) => SearchHighlightFn\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Create a search highlighter\n *\n * @param config - Search configuration\n *\n * @returns Search highlight factory function\n */\nexport function setupSearchHighlighter(\n  config: SearchConfig\n): SearchHighlightFactoryFn {\n  // Hack: temporarily remove pure lookaheads and lookbehinds\n  const regex = config.separator.split(\"|\").map(term => {\n    const temp = term.replace(/(\\(\\?[!=<][^)]+\\))/g, \"\")\n    return temp.length === 0 ? \"\uFFFD\" : term\n  })\n    .join(\"|\")\n\n  const separator = new RegExp(regex, \"img\")\n  const highlight = (_: unknown, data: string, term: string) => {\n    return `${data}<mark data-md-highlight>${term}</mark>`\n  }\n\n  /* Return factory function */\n  return (query: string) => {\n    query = query\n      .replace(/[\\s*+\\-:~^]+/g, \" \")\n      .trim()\n\n    /* Create search term match expression */\n    const match = new RegExp(`(^|${config.separator}|)(${\n      query\n        .replace(/[|\\\\{}()[\\]^$+*?.-]/g, \"\\\\$&\")\n        .replace(separator, \"|\")\n    })`, \"img\")\n\n    /* Highlight string value */\n    return value => escapeHTML(value)\n      .replace(match, highlight)\n      .replace(/<\\/mark>(\\s+)<mark[^>]*>/img, \"$1\")\n  }\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A RTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { SearchResult } from \"../../_\"\nimport { SearchIndex } from \"../../config\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search message type\n */\nexport const enum SearchMessageType {\n  SETUP,                               /* Search index setup */\n  READY,                               /* Search index ready */\n  QUERY,                               /* Search query */\n  RESULT                               /* Search results */\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Message containing the data necessary to setup the search index\n */\nexport interface SearchSetupMessage {\n  type: SearchMessageType.SETUP        /* Message type */\n  data: SearchIndex                    /* Message data */\n}\n\n/**\n * Message indicating the search index is ready\n */\nexport interface SearchReadyMessage {\n  type: SearchMessageType.READY        /* Message type */\n}\n\n/**\n * Message containing a search query\n */\nexport interface SearchQueryMessage {\n  type: SearchMessageType.QUERY        /* Message type */\n  data: string                         /* Message data */\n}\n\n/**\n * Message containing results for a search query\n */\nexport interface SearchResultMessage {\n  type: SearchMessageType.RESULT       /* Message type */\n  data: SearchResult                   /* Message data */\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Message exchanged with the search worker\n */\nexport type SearchMessage =\n  | SearchSetupMessage\n  | SearchReadyMessage\n  | SearchQueryMessage\n  | SearchResultMessage\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Type guard for search ready messages\n *\n * @param message - Search worker message\n *\n * @returns Test result\n */\nexport function isSearchReadyMessage(\n  message: SearchMessage\n): message is SearchReadyMessage {\n  return message.type === SearchMessageType.READY\n}\n\n/**\n * Type guard for search result messages\n *\n * @param message - Search worker message\n *\n * @returns Test result\n */\nexport function isSearchResultMessage(\n  message: SearchMessage\n): message is SearchResultMessage {\n  return message.type === SearchMessageType.RESULT\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A RTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  ObservableInput,\n  Subject,\n  first,\n  merge,\n  of,\n  switchMap\n} from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport { watchToggle, watchWorker } from \"~/browser\"\n\nimport { SearchIndex } from \"../../config\"\nimport {\n  SearchMessage,\n  SearchMessageType\n} from \"../message\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Set up search worker\n *\n * This function creates and initializes a web worker that is used for search,\n * so that the user interface doesn't freeze. In general, the application does\n * not care how search is implemented, as long as the web worker conforms to\n * the format expected by the application as defined in `SearchMessage`. This\n * allows the author to implement custom search functionality, by providing a\n * custom web worker via configuration.\n *\n * Material for MkDocs' built-in search implementation makes use of Lunr.js, an\n * efficient and fast implementation for client-side search. Leveraging a tiny\n * iframe-based web worker shim, search is even supported for the `file://`\n * protocol, enabling search for local non-hosted builds.\n *\n * If the protocol is `file://`, search initialization is deferred to mitigate\n * freezing, as it's now synchronous by design - see https://bit.ly/3C521EO\n *\n * @see https://bit.ly/3igvtQv - How to implement custom search\n *\n * @param url - Worker URL\n * @param index$ - Search index observable input\n *\n * @returns Search worker\n */\nexport function setupSearchWorker(\n  url: string, index$: ObservableInput<SearchIndex>\n): Subject<SearchMessage> {\n  const worker$ = watchWorker<SearchMessage>(url)\n  merge(\n    of(location.protocol !== \"file:\"),\n    watchToggle(\"search\")\n  )\n    .pipe(\n      first(active => active),\n      switchMap(() => index$)\n    )\n      .subscribe(({ config, docs }) => worker$.next({\n        type: SearchMessageType.SETUP,\n        data: {\n          config,\n          docs,\n          options: {\n            suggest: feature(\"search.suggest\")\n          }\n        }\n      }))\n\n  /* Return search worker */\n  return worker$\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Subject,\n  catchError,\n  combineLatest,\n  filter,\n  fromEvent,\n  map,\n  of,\n  switchMap,\n  withLatestFrom\n} from \"rxjs\"\n\nimport { configuration } from \"~/_\"\nimport {\n  getElement,\n  getLocation,\n  requestJSON,\n  setLocation\n} from \"~/browser\"\nimport { getComponentElements } from \"~/components\"\nimport {\n  Version,\n  renderVersionSelector\n} from \"~/templates\"\n\nimport { fetchSitemap } from \"../sitemap\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Setup options\n */\ninterface SetupOptions {\n  document$: Subject<Document>         /* Document subject */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Set up version selector\n *\n * @param options - Options\n */\nexport function setupVersionSelector(\n  { document$ }: SetupOptions\n): void {\n  const config = configuration()\n  const versions$ = requestJSON<Version[]>(\n    new URL(\"../versions.json\", config.base)\n  )\n    .pipe(\n      catchError(() => EMPTY) // @todo refactor instant loading\n    )\n\n  /* Determine current version */\n  const current$ = versions$\n    .pipe(\n      map(versions => {\n        const [, current] = config.base.match(/([^/]+)\\/?$/)!\n        return versions.find(({ version, aliases }) => (\n          version === current || aliases.includes(current)\n        )) || versions[0]\n      })\n    )\n\n  /* Intercept inter-version navigation */\n  versions$\n    .pipe(\n      map(versions => new Map(versions.map(version => [\n        `${new URL(`../${version.version}/`, config.base)}`,\n        version\n      ]))),\n      switchMap(urls => fromEvent<MouseEvent>(document.body, \"click\")\n        .pipe(\n          filter(ev => !ev.metaKey && !ev.ctrlKey),\n          withLatestFrom(current$),\n          switchMap(([ev, current]) => {\n            if (ev.target instanceof Element) {\n              const el = ev.target.closest(\"a\")\n              if (el && !el.target && urls.has(el.href)) {\n                const url = el.href\n                // This is a temporary hack to detect if a version inside the\n                // version selector or on another part of the site was clicked.\n                // If we're inside the version selector, we definitely want to\n                // find the same page, as we might have different deployments\n                // due to aliases. However, if we're outside the version\n                // selector, we must abort here, because we might otherwise\n                // interfere with instant navigation. We need to refactor this\n                // at some point together with instant navigation.\n                //\n                // See https://github.com/squidfunk/mkdocs-material/issues/4012\n                if (!ev.target.closest(\".md-version\")) {\n                  const version = urls.get(url)!\n                  if (version === current)\n                    return EMPTY\n                }\n                ev.preventDefault()\n                return of(url)\n              }\n            }\n            return EMPTY\n          }),\n          switchMap(url => {\n            const { version } = urls.get(url)!\n            return fetchSitemap(new URL(url))\n              .pipe(\n                map(sitemap => {\n                  const location = getLocation()\n                  const path = location.href.replace(config.base, \"\")\n                  return sitemap.has(path.split(\"#\")[0])\n                    ? new URL(`../${version}/${path}`, config.base)\n                    : new URL(url)\n                })\n              )\n          })\n        )\n      )\n    )\n      .subscribe(url => setLocation(url, true))\n\n  /* Render version selector and warning */\n  combineLatest([versions$, current$])\n    .subscribe(([versions, current]) => {\n      const topic = getElement(\".md-header__topic\")\n      topic.appendChild(renderVersionSelector(versions, current))\n    })\n\n  /* Integrate outdated version banner with instant navigation */\n  document$.pipe(switchMap(() => current$))\n    .subscribe(current => {\n\n      /* Check if version state was already determined */\n      let outdated = __md_get(\"__outdated\", sessionStorage)\n      if (outdated === null) {\n        outdated = true\n\n        /* Obtain and normalize default versions */\n        let ignored = config.version?.default || \"latest\"\n        if (!Array.isArray(ignored))\n          ignored = [ignored]\n\n        /* Check if version is considered a default */\n        main: for (const ignore of ignored)\n          for (const version of current.aliases.concat(current.version))\n            if (new RegExp(ignore, \"i\").test(version)) {\n              outdated = false\n              break main\n            }\n\n        /* Persist version state in session storage */\n        __md_set(\"__outdated\", outdated, sessionStorage)\n      }\n\n      /* Unhide outdated version banner */\n      if (outdated)\n        for (const warning of getComponentElements(\"outdated\"))\n          warning.hidden = false\n    })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  combineLatest,\n  distinctUntilChanged,\n  distinctUntilKeyChanged,\n  endWith,\n  finalize,\n  first,\n  fromEvent,\n  ignoreElements,\n  map,\n  merge,\n  shareReplay,\n  takeUntil,\n  tap\n} from \"rxjs\"\n\nimport {\n  getElement,\n  getLocation,\n  setToggle,\n  watchElementFocus,\n  watchToggle\n} from \"~/browser\"\nimport {\n  SearchMessage,\n  SearchMessageType,\n  isSearchReadyMessage\n} from \"~/integrations\"\n\nimport { Component } from \"../../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search query\n */\nexport interface SearchQuery {\n  value: string                        /* Query value */\n  focus: boolean                       /* Query focus */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  worker$: Subject<SearchMessage>      /* Search worker */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  worker$: Subject<SearchMessage>      /* Search worker */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch search query\n *\n * Note that the focus event which triggers re-reading the current query value\n * is delayed by `1ms` so the input's empty state is allowed to propagate.\n *\n * @param el - Search query element\n * @param options - Options\n *\n * @returns Search query observable\n */\nexport function watchSearchQuery(\n  el: HTMLInputElement, { worker$ }: WatchOptions\n): Observable<SearchQuery> {\n\n  /* Support search deep linking */\n  const { searchParams } = getLocation()\n  if (searchParams.has(\"q\")) {\n    setToggle(\"search\", true)\n\n    /* Set query from parameter */\n    el.value = searchParams.get(\"q\")!\n    el.focus()\n\n    /* Remove query parameter on close */\n    watchToggle(\"search\")\n      .pipe(\n        first(active => !active)\n      )\n        .subscribe(() => {\n          const url = getLocation()\n          url.searchParams.delete(\"q\")\n          history.replaceState({}, \"\", `${url}`)\n        })\n  }\n\n  /* Intercept focus and input events */\n  const focus$ = watchElementFocus(el)\n  const value$ = merge(\n    worker$.pipe(first(isSearchReadyMessage)),\n    fromEvent(el, \"keyup\"),\n    focus$\n  )\n    .pipe(\n      map(() => el.value),\n      distinctUntilChanged()\n    )\n\n  /* Combine into single observable */\n  return combineLatest([value$, focus$])\n    .pipe(\n      map(([value, focus]) => ({ value, focus })),\n      shareReplay(1)\n    )\n}\n\n/**\n * Mount search query\n *\n * @param el - Search query element\n * @param options - Options\n *\n * @returns Search query component observable\n */\nexport function mountSearchQuery(\n  el: HTMLInputElement, { worker$ }: MountOptions\n): Observable<Component<SearchQuery, HTMLInputElement>> {\n  const push$ = new Subject<SearchQuery>()\n  const done$ = push$.pipe(ignoreElements(), endWith(true))\n\n  /* Handle value change */\n  combineLatest([\n    worker$.pipe(first(isSearchReadyMessage)),\n    push$\n  ], (_, query) => query)\n    .pipe(\n      distinctUntilKeyChanged(\"value\")\n    )\n      .subscribe(({ value }) => worker$.next({\n        type: SearchMessageType.QUERY,\n        data: value\n      }))\n\n  /* Handle focus change */\n  push$\n    .pipe(\n      distinctUntilKeyChanged(\"focus\")\n    )\n      .subscribe(({ focus }) => {\n        if (focus)\n          setToggle(\"search\", focus)\n      })\n\n  /* Handle reset */\n  fromEvent(el.form!, \"reset\")\n    .pipe(\n      takeUntil(done$)\n    )\n      .subscribe(() => el.focus())\n\n  // Focus search query on label click - note that this is necessary to bring\n  // up the keyboard on iOS and other mobile platforms, as the search dialog is\n  // not visible at first, and programatically focusing an input element must\n  // be triggered by a user interaction - see https://t.ly/Cb30n\n  const label = getElement(\"header [for=__search]\")\n  fromEvent(label, \"click\")\n    .subscribe(() => el.focus())\n\n  /* Create and return component */\n  return watchSearchQuery(el, { worker$ })\n    .pipe(\n      tap(state => push$.next(state)),\n      finalize(() => push$.complete()),\n      map(state => ({ ref: el, ...state })),\n      shareReplay(1)\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  bufferCount,\n  filter,\n  finalize,\n  first,\n  fromEvent,\n  map,\n  merge,\n  mergeMap,\n  of,\n  share,\n  skipUntil,\n  switchMap,\n  takeUntil,\n  tap,\n  withLatestFrom,\n  zipWith\n} from \"rxjs\"\n\nimport { translation } from \"~/_\"\nimport {\n  getElement,\n  getOptionalElement,\n  watchElementBoundary,\n  watchToggle\n} from \"~/browser\"\nimport {\n  SearchMessage,\n  SearchResult,\n  isSearchReadyMessage,\n  isSearchResultMessage\n} from \"~/integrations\"\nimport { renderSearchResultItem } from \"~/templates\"\nimport { round } from \"~/utilities\"\n\nimport { Component } from \"../../_\"\nimport { SearchQuery } from \"../query\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  query$: Observable<SearchQuery>      /* Search query observable */\n  worker$: Subject<SearchMessage>      /* Search worker */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount search result list\n *\n * This function performs a lazy rendering of the search results, depending on\n * the vertical offset of the search result container.\n *\n * @param el - Search result list element\n * @param options - Options\n *\n * @returns Search result list component observable\n */\nexport function mountSearchResult(\n  el: HTMLElement, { worker$, query$ }: MountOptions\n): Observable<Component<SearchResult>> {\n  const push$ = new Subject<SearchResult>()\n  const boundary$ = watchElementBoundary(el.parentElement!)\n    .pipe(\n      filter(Boolean)\n    )\n\n  /* Retrieve container */\n  const container = el.parentElement!\n\n  /* Retrieve nested components */\n  const meta = getElement(\":scope > :first-child\", el)\n  const list = getElement(\":scope > :last-child\", el)\n\n  /* Reveal to accessibility tree \u2013 see https://bit.ly/3iAA7t8 */\n  watchToggle(\"search\")\n    .subscribe(active => list.setAttribute(\n      \"role\", active ? \"list\" : \"presentation\"\n    ))\n\n  /* Update search result metadata */\n  push$\n    .pipe(\n      withLatestFrom(query$),\n      skipUntil(worker$.pipe(first(isSearchReadyMessage)))\n    )\n      .subscribe(([{ items }, { value }]) => {\n        switch (items.length) {\n\n          /* No results */\n          case 0:\n            meta.textContent = value.length\n              ? translation(\"search.result.none\")\n              : translation(\"search.result.placeholder\")\n            break\n\n          /* One result */\n          case 1:\n            meta.textContent = translation(\"search.result.one\")\n            break\n\n          /* Multiple result */\n          default:\n            const count = round(items.length)\n            meta.textContent = translation(\"search.result.other\", count)\n        }\n      })\n\n  /* Render search result item */\n  const render$ = push$\n    .pipe(\n      tap(() => list.innerHTML = \"\"),\n      switchMap(({ items }) => merge(\n        of(...items.slice(0, 10)),\n        of(...items.slice(10))\n          .pipe(\n            bufferCount(4),\n            zipWith(boundary$),\n            switchMap(([chunk]) => chunk)\n          )\n      )),\n      map(renderSearchResultItem),\n      share()\n    )\n\n  /* Update search result list */\n  render$.subscribe(item => list.appendChild(item))\n  render$\n    .pipe(\n      mergeMap(item => {\n        const details = getOptionalElement(\"details\", item)\n        if (typeof details === \"undefined\")\n          return EMPTY\n\n        /* Keep position of details element stable */\n        return fromEvent(details, \"toggle\")\n          .pipe(\n            takeUntil(push$),\n            map(() => details)\n          )\n      })\n    )\n      .subscribe(details => {\n        if (\n          details.open === false &&\n          details.offsetTop <= container.scrollTop\n        )\n          container.scrollTo({ top: details.offsetTop })\n      })\n\n  /* Filter search result message */\n  const result$ = worker$\n    .pipe(\n      filter(isSearchResultMessage),\n      map(({ data }) => data)\n    )\n\n  /* Create and return component */\n  return result$\n    .pipe(\n      tap(state => push$.next(state)),\n      finalize(() => push$.complete()),\n      map(state => ({ ref: el, ...state }))\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  endWith,\n  finalize,\n  fromEvent,\n  ignoreElements,\n  map,\n  takeUntil,\n  tap\n} from \"rxjs\"\n\nimport { getLocation } from \"~/browser\"\n\nimport { Component } from \"../../_\"\nimport { SearchQuery } from \"../query\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search sharing\n */\nexport interface SearchShare {\n  url: URL                             /* Deep link for sharing */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  query$: Observable<SearchQuery>      /* Search query observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  query$: Observable<SearchQuery>      /* Search query observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount search sharing\n *\n * @param _el - Search sharing element\n * @param options - Options\n *\n * @returns Search sharing observable\n */\nexport function watchSearchShare(\n  _el: HTMLElement, { query$ }: WatchOptions\n): Observable<SearchShare> {\n  return query$\n    .pipe(\n      map(({ value }) => {\n        const url = getLocation()\n        url.hash = \"\"\n\n        /* Compute readable query strings */\n        value = value\n          .replace(/\\s+/g, \"+\")        /* Collapse whitespace */\n          .replace(/&/g, \"%26\")        /* Escape '&' character */\n          .replace(/=/g, \"%3D\")        /* Escape '=' character */\n\n        /* Replace query string */\n        url.search = `q=${value}`\n        return { url }\n      })\n    )\n}\n\n/**\n * Mount search sharing\n *\n * @param el - Search sharing element\n * @param options - Options\n *\n * @returns Search sharing component observable\n */\nexport function mountSearchShare(\n  el: HTMLAnchorElement, options: MountOptions\n): Observable<Component<SearchShare>> {\n  const push$ = new Subject<SearchShare>()\n  const done$ = push$.pipe(ignoreElements(), endWith(true))\n  push$.subscribe(({ url }) => {\n    el.setAttribute(\"data-clipboard-text\", el.href)\n    el.href = `${url}`\n  })\n\n  /* Prevent following of link */\n  fromEvent(el, \"click\")\n    .pipe(\n      takeUntil(done$)\n    )\n      .subscribe(ev => ev.preventDefault())\n\n  /* Create and return component */\n  return watchSearchShare(el, options)\n    .pipe(\n      tap(state => push$.next(state)),\n      finalize(() => push$.complete()),\n      map(state => ({ ref: el, ...state }))\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  asyncScheduler,\n  combineLatestWith,\n  distinctUntilChanged,\n  filter,\n  finalize,\n  fromEvent,\n  map,\n  merge,\n  observeOn,\n  tap\n} from \"rxjs\"\n\nimport { Keyboard } from \"~/browser\"\nimport {\n  SearchMessage,\n  SearchResult,\n  isSearchResultMessage\n} from \"~/integrations\"\n\nimport { Component, getComponentElement } from \"../../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search suggestions\n */\nexport interface SearchSuggest {}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  keyboard$: Observable<Keyboard>      /* Keyboard observable */\n  worker$: Subject<SearchMessage>      /* Search worker */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount search suggestions\n *\n * This function will perform a lazy rendering of the search results, depending\n * on the vertical offset of the search result container.\n *\n * @param el - Search result list element\n * @param options - Options\n *\n * @returns Search result list component observable\n */\nexport function mountSearchSuggest(\n  el: HTMLElement, { worker$, keyboard$ }: MountOptions\n): Observable<Component<SearchSuggest>> {\n  const push$ = new Subject<SearchResult>()\n\n  /* Retrieve query component and track all changes */\n  const query  = getComponentElement(\"search-query\")\n  const query$ = merge(\n    fromEvent(query, \"keydown\"),\n    fromEvent(query, \"focus\")\n  )\n    .pipe(\n      observeOn(asyncScheduler),\n      map(() => query.value),\n      distinctUntilChanged(),\n    )\n\n  /* Update search suggestions */\n  push$\n    .pipe(\n      combineLatestWith(query$),\n      map(([{ suggest }, value]) => {\n        const words = value.split(/([\\s-]+)/)\n        if (suggest?.length && words[words.length - 1]) {\n          const last = suggest[suggest.length - 1]\n          if (last.startsWith(words[words.length - 1]))\n            words[words.length - 1] = last\n        } else {\n          words.length = 0\n        }\n        return words\n      })\n    )\n      .subscribe(words => el.innerHTML = words\n        .join(\"\")\n        .replace(/\\s/g, \"&nbsp;\")\n      )\n\n  /* Set up search keyboard handlers */\n  keyboard$\n    .pipe(\n      filter(({ mode }) => mode === \"search\")\n    )\n      .subscribe(key => {\n        switch (key.type) {\n\n          /* Right arrow: accept current suggestion */\n          case \"ArrowRight\":\n            if (\n              el.innerText.length &&\n              query.selectionStart === query.value.length\n            )\n              query.value = el.innerText\n            break\n        }\n      })\n\n  /* Filter search result message */\n  const result$ = worker$\n    .pipe(\n      filter(isSearchResultMessage),\n      map(({ data }) => data)\n    )\n\n  /* Create and return component */\n  return result$\n    .pipe(\n      tap(state => push$.next(state)),\n      finalize(() => push$.complete()),\n      map(() => ({ ref: el }))\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  NEVER,\n  Observable,\n  ObservableInput,\n  filter,\n  fromEvent,\n  merge,\n  mergeWith\n} from \"rxjs\"\n\nimport { configuration } from \"~/_\"\nimport {\n  Keyboard,\n  getActiveElement,\n  getElements,\n  setToggle\n} from \"~/browser\"\nimport {\n  SearchIndex,\n  SearchResult,\n  setupSearchWorker\n} from \"~/integrations\"\n\nimport {\n  Component,\n  getComponentElement,\n  getComponentElements\n} from \"../../_\"\nimport {\n  SearchQuery,\n  mountSearchQuery\n} from \"../query\"\nimport { mountSearchResult } from \"../result\"\nimport {\n  SearchShare,\n  mountSearchShare\n} from \"../share\"\nimport {\n  SearchSuggest,\n  mountSearchSuggest\n} from \"../suggest\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search\n */\nexport type Search =\n  | SearchQuery\n  | SearchResult\n  | SearchShare\n  | SearchSuggest\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  index$: ObservableInput<SearchIndex> /* Search index observable */\n  keyboard$: Observable<Keyboard>      /* Keyboard observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount search\n *\n * This function sets up the search functionality, including the underlying\n * web worker and all keyboard bindings.\n *\n * @param el - Search element\n * @param options - Options\n *\n * @returns Search component observable\n */\nexport function mountSearch(\n  el: HTMLElement, { index$, keyboard$ }: MountOptions\n): Observable<Component<Search>> {\n  const config = configuration()\n  try {\n    const worker$ = setupSearchWorker(config.search, index$)\n\n    /* Retrieve query and result components */\n    const query  = getComponentElement(\"search-query\", el)\n    const result = getComponentElement(\"search-result\", el)\n\n    /* Always close search on result selection */\n    fromEvent<PointerEvent>(el, \"click\")\n      .pipe(\n        filter(({ target }) => (\n          target instanceof Element && !!target.closest(\"a\")\n        ))\n      )\n        .subscribe(() => setToggle(\"search\", false))\n\n    /* Set up search keyboard handlers */\n    keyboard$\n      .pipe(\n        filter(({ mode }) => mode === \"search\")\n      )\n        .subscribe(key => {\n          const active = getActiveElement()\n          switch (key.type) {\n\n            /* Enter: go to first (best) result */\n            case \"Enter\":\n              if (active === query) {\n                const anchors = new Map<HTMLAnchorElement, number>()\n                for (const anchor of getElements<HTMLAnchorElement>(\n                  \":first-child [href]\", result\n                )) {\n                  const article = anchor.firstElementChild!\n                  anchors.set(anchor, parseFloat(\n                    article.getAttribute(\"data-md-score\")!\n                  ))\n                }\n\n                /* Go to result with highest score, if any */\n                if (anchors.size) {\n                  const [[best]] = [...anchors].sort(([, a], [, b]) => b - a)\n                  best.click()\n                }\n\n                /* Otherwise omit form submission */\n                key.claim()\n              }\n              break\n\n            /* Escape or Tab: close search */\n            case \"Escape\":\n            case \"Tab\":\n              setToggle(\"search\", false)\n              query.blur()\n              break\n\n            /* Vertical arrows: select previous or next search result */\n            case \"ArrowUp\":\n            case \"ArrowDown\":\n              if (typeof active === \"undefined\") {\n                query.focus()\n              } else {\n                const els = [query, ...getElements(\n                  \":not(details) > [href], summary, details[open] [href]\",\n                  result\n                )]\n                const i = Math.max(0, (\n                  Math.max(0, els.indexOf(active)) + els.length + (\n                    key.type === \"ArrowUp\" ? -1 : +1\n                  )\n                ) % els.length)\n                els[i].focus()\n              }\n\n              /* Prevent scrolling of page */\n              key.claim()\n              break\n\n            /* All other keys: hand to search query */\n            default:\n              if (query !== getActiveElement())\n                query.focus()\n          }\n        })\n\n    /* Set up global keyboard handlers */\n    keyboard$\n      .pipe(\n        filter(({ mode }) => mode === \"global\")\n      )\n        .subscribe(key => {\n          switch (key.type) {\n\n            /* Open search and select query */\n            case \"f\":\n            case \"s\":\n            case \"/\":\n              query.focus()\n              query.select()\n\n              /* Prevent scrolling of page */\n              key.claim()\n              break\n          }\n        })\n\n    /* Create and return component */\n    const query$ = mountSearchQuery(query, { worker$ })\n    return merge(\n      query$,\n      mountSearchResult(result, { worker$, query$ })\n    )\n      .pipe(\n        mergeWith(\n\n          /* Search sharing */\n          ...getComponentElements(\"search-share\", el)\n            .map(child => mountSearchShare(child, { query$ })),\n\n          /* Search suggestions */\n          ...getComponentElements(\"search-suggest\", el)\n            .map(child => mountSearchSuggest(child, { worker$, keyboard$ }))\n        )\n      )\n\n  /* Gracefully handle broken search */\n  } catch (err) {\n    el.hidden = true\n    return NEVER\n  }\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  ObservableInput,\n  combineLatest,\n  filter,\n  map,\n  startWith\n} from \"rxjs\"\n\nimport { getLocation } from \"~/browser\"\nimport {\n  SearchIndex,\n  setupSearchHighlighter\n} from \"~/integrations\"\nimport { h } from \"~/utilities\"\n\nimport { Component } from \"../../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search highlighting\n */\nexport interface SearchHighlight {\n  nodes: Map<ChildNode, string>        /* Map of replacements */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  index$: ObservableInput<SearchIndex> /* Search index observable */\n  location$: Observable<URL>           /* Location observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount search highlighting\n *\n * @param el - Content element\n * @param options - Options\n *\n * @returns Search highlighting component observable\n */\nexport function mountSearchHiglight(\n  el: HTMLElement, { index$, location$ }: MountOptions\n): Observable<Component<SearchHighlight>> {\n  return combineLatest([\n    index$,\n    location$\n      .pipe(\n        startWith(getLocation()),\n        filter(url => !!url.searchParams.get(\"h\"))\n      )\n  ])\n    .pipe(\n      map(([index, url]) => setupSearchHighlighter(index.config)(\n        url.searchParams.get(\"h\")!\n      )),\n      map(fn => {\n        const nodes = new Map<ChildNode, string>()\n\n        /* Traverse text nodes and collect matches */\n        const it = document.createNodeIterator(el, NodeFilter.SHOW_TEXT)\n        for (let node = it.nextNode(); node; node = it.nextNode()) {\n          if (node.parentElement?.offsetHeight) {\n            const original = node.textContent!\n            const replaced = fn(original)\n            if (replaced.length > original.length)\n              nodes.set(node as ChildNode, replaced)\n          }\n        }\n\n        /* Replace original nodes with matches */\n        for (const [node, text] of nodes) {\n          const { childNodes } = h(\"span\", null, text)\n          node.replaceWith(...Array.from(childNodes))\n        }\n\n        /* Return component */\n        return { ref: el, nodes }\n      })\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  animationFrameScheduler,\n  asyncScheduler,\n  auditTime,\n  combineLatest,\n  defer,\n  distinctUntilChanged,\n  endWith,\n  finalize,\n  first,\n  from,\n  fromEvent,\n  ignoreElements,\n  map,\n  mergeMap,\n  observeOn,\n  takeUntil,\n  tap,\n  withLatestFrom\n} from \"rxjs\"\n\nimport {\n  Viewport,\n  getElement,\n  getElementOffset,\n  getElementSize,\n  getElements\n} from \"~/browser\"\n\nimport { Component } from \"../_\"\nimport { Header } from \"../header\"\nimport { Main } from \"../main\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Sidebar\n */\nexport interface Sidebar {\n  height: number                       /* Sidebar height */\n  locked: boolean                      /* Sidebar is locked */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  main$: Observable<Main>              /* Main area observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n  main$: Observable<Main>              /* Main area observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch sidebar\n *\n * This function returns an observable that computes the visual parameters of\n * the sidebar which depends on the vertical viewport offset, as well as the\n * height of the main area. When the page is scrolled beyond the header, the\n * sidebar is locked and fills the remaining space.\n *\n * @param el - Sidebar element\n * @param options - Options\n *\n * @returns Sidebar observable\n */\nexport function watchSidebar(\n  el: HTMLElement, { viewport$, main$ }: WatchOptions\n): Observable<Sidebar> {\n  const parent = el.closest<HTMLElement>(\".md-grid\")!\n  const adjust =\n    parent.offsetTop -\n    parent.parentElement!.offsetTop\n\n  /* Compute the sidebar's available height and if it should be locked */\n  return combineLatest([main$, viewport$])\n    .pipe(\n      map(([{ offset, height }, { offset: { y } }]) => {\n        height = height\n          + Math.min(adjust, Math.max(0, y - offset))\n          - adjust\n        return {\n          height,\n          locked: y >= offset + adjust\n        }\n      }),\n      distinctUntilChanged((a, b) => (\n        a.height === b.height &&\n        a.locked === b.locked\n      ))\n    )\n}\n\n/**\n * Mount sidebar\n *\n * This function doesn't set the height of the actual sidebar, but of its first\n * child \u2013 the `.md-sidebar__scrollwrap` element in order to mitigiate jittery\n * sidebars when the footer is scrolled into view. At some point we switched\n * from `absolute` / `fixed` positioning to `sticky` positioning, significantly\n * reducing jitter in some browsers (respectively Firefox and Safari) when\n * scrolling from the top. However, top-aligned sticky positioning means that\n * the sidebar snaps to the bottom when the end of the container is reached.\n * This is what leads to the mentioned jitter, as the sidebar's height may be\n * updated too slowly.\n *\n * This behaviour can be mitigiated by setting the height of the sidebar to `0`\n * while preserving the padding, and the height on its first element.\n *\n * @param el - Sidebar element\n * @param options - Options\n *\n * @returns Sidebar component observable\n */\nexport function mountSidebar(\n  el: HTMLElement, { header$, ...options }: MountOptions\n): Observable<Component<Sidebar>> {\n  const inner = getElement(\".md-sidebar__scrollwrap\", el)\n  const { y } = getElementOffset(inner)\n  return defer(() => {\n    const push$ = new Subject<Sidebar>()\n    const done$ = push$.pipe(ignoreElements(), endWith(true))\n    const next$ = push$\n      .pipe(\n        auditTime(0, animationFrameScheduler)\n      )\n\n    /* Update sidebar height and offset */\n    next$.pipe(withLatestFrom(header$))\n      .subscribe({\n\n        /* Handle emission */\n        next([{ height }, { height: offset }]) {\n          inner.style.height = `${height - 2 * y}px`\n          el.style.top       = `${offset}px`\n        },\n\n        /* Handle complete */\n        complete() {\n          inner.style.height = \"\"\n          el.style.top       = \"\"\n        }\n      })\n\n    /* Bring active item into view on initial load */\n    next$.pipe(first())\n      .subscribe(() => {\n        for (const item of getElements(\".md-nav__link--active[href]\", el)) {\n          if (!item.clientHeight) // skip invisible toc in left sidebar\n            continue\n          const container = item.closest<HTMLElement>(\".md-sidebar__scrollwrap\")!\n          if (typeof container !== \"undefined\") {\n            const offset = item.offsetTop - container.offsetTop\n            const { height } = getElementSize(container)\n            container.scrollTo({\n              top: offset - height / 2\n            })\n          }\n        }\n      })\n\n    /* Handle accessibility for expandable items, see https://bit.ly/3jaod9p */\n    from(getElements<HTMLLabelElement>(\"label[tabindex]\", el))\n      .pipe(\n        mergeMap(label => fromEvent(label, \"click\")\n          .pipe(\n            observeOn(asyncScheduler),\n            map(() => label),\n            takeUntil(done$)\n          )\n        )\n      )\n        .subscribe(label => {\n          const input = getElement<HTMLInputElement>(`[id=\"${label.htmlFor}\"]`)\n          const nav = getElement(`[aria-labelledby=\"${label.id}\"]`)\n          nav.setAttribute(\"aria-expanded\", `${input.checked}`)\n        })\n\n    /* Create and return component */\n    return watchSidebar(el, options)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { Repo, User } from \"github-types\"\nimport {\n  EMPTY,\n  Observable,\n  catchError,\n  defaultIfEmpty,\n  map,\n  zip\n} from \"rxjs\"\n\nimport { requestJSON } from \"~/browser\"\n\nimport { SourceFacts } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * GitHub release (partial)\n */\ninterface Release {\n  tag_name: string                     /* Tag name */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch GitHub repository facts\n *\n * @param user - GitHub user or organization\n * @param repo - GitHub repository\n *\n * @returns Repository facts observable\n */\nexport function fetchSourceFactsFromGitHub(\n  user: string, repo?: string\n): Observable<SourceFacts> {\n  if (typeof repo !== \"undefined\") {\n    const url = `https://api.github.com/repos/${user}/${repo}`\n    return zip(\n\n      /* Fetch version */\n      requestJSON<Release>(`${url}/releases/latest`)\n        .pipe(\n          catchError(() => EMPTY), // @todo refactor instant loading\n          map(release => ({\n            version: release.tag_name\n          })),\n          defaultIfEmpty({})\n        ),\n\n      /* Fetch stars and forks */\n      requestJSON<Repo>(url)\n        .pipe(\n          catchError(() => EMPTY), // @todo refactor instant loading\n          map(info => ({\n            stars: info.stargazers_count,\n            forks: info.forks_count\n          })),\n          defaultIfEmpty({})\n        )\n    )\n      .pipe(\n        map(([release, info]) => ({ ...release, ...info }))\n      )\n\n  /* User or organization */\n  } else {\n    const url = `https://api.github.com/users/${user}`\n    return requestJSON<User>(url)\n      .pipe(\n        map(info => ({\n          repositories: info.public_repos\n        })),\n        defaultIfEmpty({})\n      )\n  }\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { ProjectSchema } from \"gitlab\"\nimport {\n  EMPTY,\n  Observable,\n  catchError,\n  defaultIfEmpty,\n  map\n} from \"rxjs\"\n\nimport { requestJSON } from \"~/browser\"\n\nimport { SourceFacts } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch GitLab repository facts\n *\n * @param base - GitLab base\n * @param project - GitLab project\n *\n * @returns Repository facts observable\n */\nexport function fetchSourceFactsFromGitLab(\n  base: string, project: string\n): Observable<SourceFacts> {\n  const url = `https://${base}/api/v4/projects/${encodeURIComponent(project)}`\n  return requestJSON<ProjectSchema>(url)\n    .pipe(\n      catchError(() => EMPTY), // @todo refactor instant loading\n      map(({ star_count, forks_count }) => ({\n        stars: star_count,\n        forks: forks_count\n      })),\n      defaultIfEmpty({})\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { EMPTY, Observable } from \"rxjs\"\n\nimport { fetchSourceFactsFromGitHub } from \"../github\"\nimport { fetchSourceFactsFromGitLab } from \"../gitlab\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Repository facts for repositories\n */\nexport interface RepositoryFacts {\n  stars?: number                       /* Number of stars */\n  forks?: number                       /* Number of forks */\n  version?: string                     /* Latest version */\n}\n\n/**\n * Repository facts for organizations\n */\nexport interface OrganizationFacts {\n  repositories?: number                /* Number of repositories */\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Repository facts\n */\nexport type SourceFacts =\n  | RepositoryFacts\n  | OrganizationFacts\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch repository facts\n *\n * @param url - Repository URL\n *\n * @returns Repository facts observable\n */\nexport function fetchSourceFacts(\n  url: string\n): Observable<SourceFacts> {\n\n  /* Try to match GitHub repository */\n  let match = url.match(/^.+github\\.com\\/([^/]+)\\/?([^/]+)?/i)\n  if (match) {\n    const [, user, repo] = match\n    return fetchSourceFactsFromGitHub(user, repo)\n  }\n\n  /* Try to match GitLab repository */\n  match = url.match(/^.+?([^/]*gitlab[^/]+)\\/(.+?)\\/?$/i)\n  if (match) {\n    const [, base, slug] = match\n    return fetchSourceFactsFromGitLab(base, slug)\n  }\n\n  /* Fallback */\n  return EMPTY\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  catchError,\n  defer,\n  filter,\n  finalize,\n  map,\n  of,\n  shareReplay,\n  tap\n} from \"rxjs\"\n\nimport { getElement } from \"~/browser\"\nimport { ConsentDefaults } from \"~/components/consent\"\nimport { renderSourceFacts } from \"~/templates\"\n\nimport {\n  Component,\n  getComponentElements\n} from \"../../_\"\nimport {\n  SourceFacts,\n  fetchSourceFacts\n} from \"../facts\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Repository information\n */\nexport interface Source {\n  facts: SourceFacts                   /* Repository facts */\n}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Repository information observable\n */\nlet fetch$: Observable<Source>\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch repository information\n *\n * This function tries to read the repository facts from session storage, and\n * if unsuccessful, fetches them from the underlying provider.\n *\n * @param el - Repository information element\n *\n * @returns Repository information observable\n */\nexport function watchSource(\n  el: HTMLAnchorElement\n): Observable<Source> {\n  return fetch$ ||= defer(() => {\n    const cached = __md_get<SourceFacts>(\"__source\", sessionStorage)\n    if (cached) {\n      return of(cached)\n    } else {\n\n      /* Check if consent is configured and was given */\n      const els = getComponentElements(\"consent\")\n      if (els.length) {\n        const consent = __md_get<ConsentDefaults>(\"__consent\")\n        if (!(consent && consent.github))\n          return EMPTY\n      }\n\n      /* Fetch repository facts */\n      return fetchSourceFacts(el.href)\n        .pipe(\n          tap(facts => __md_set(\"__source\", facts, sessionStorage))\n        )\n    }\n  })\n    .pipe(\n      catchError(() => EMPTY),\n      filter(facts => Object.keys(facts).length > 0),\n      map(facts => ({ facts })),\n      shareReplay(1)\n    )\n}\n\n/**\n * Mount repository information\n *\n * @param el - Repository information element\n *\n * @returns Repository information component observable\n */\nexport function mountSource(\n  el: HTMLAnchorElement\n): Observable<Component<Source>> {\n  const inner = getElement(\":scope > :last-child\", el)\n  return defer(() => {\n    const push$ = new Subject<Source>()\n    push$.subscribe(({ facts }) => {\n      inner.appendChild(renderSourceFacts(facts))\n      inner.classList.add(\"md-source__repository--active\")\n    })\n\n    /* Create and return component */\n    return watchSource(el)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  defer,\n  distinctUntilKeyChanged,\n  finalize,\n  map,\n  of,\n  switchMap,\n  tap\n} from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport {\n  Viewport,\n  watchElementSize,\n  watchViewportAt\n} from \"~/browser\"\n\nimport { Component } from \"../_\"\nimport { Header } from \"../header\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Navigation tabs\n */\nexport interface Tabs {\n  hidden: boolean                      /* Navigation tabs are hidden */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch navigation tabs\n *\n * @param el - Navigation tabs element\n * @param options - Options\n *\n * @returns Navigation tabs observable\n */\nexport function watchTabs(\n  el: HTMLElement, { viewport$, header$ }: WatchOptions\n): Observable<Tabs> {\n  return watchElementSize(document.body)\n    .pipe(\n      switchMap(() => watchViewportAt(el, { header$, viewport$ })),\n      map(({ offset: { y } }) => {\n        return {\n          hidden: y >= 10\n        }\n      }),\n      distinctUntilKeyChanged(\"hidden\")\n    )\n}\n\n/**\n * Mount navigation tabs\n *\n * This function hides the navigation tabs when scrolling past the threshold\n * and makes them reappear in a nice CSS animation when scrolling back up.\n *\n * @param el - Navigation tabs element\n * @param options - Options\n *\n * @returns Navigation tabs component observable\n */\nexport function mountTabs(\n  el: HTMLElement, options: MountOptions\n): Observable<Component<Tabs>> {\n  return defer(() => {\n    const push$ = new Subject<Tabs>()\n    push$.subscribe({\n\n      /* Handle emission */\n      next({ hidden }) {\n        el.hidden = hidden\n      },\n\n      /* Handle complete */\n      complete() {\n        el.hidden = false\n      }\n    })\n\n    /* Create and return component */\n    return (\n      feature(\"navigation.tabs.sticky\")\n        ? of({ hidden: false })\n        : watchTabs(el, options)\n    )\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  asyncScheduler,\n  bufferCount,\n  combineLatestWith,\n  debounceTime,\n  defer,\n  distinctUntilChanged,\n  distinctUntilKeyChanged,\n  endWith,\n  filter,\n  finalize,\n  ignoreElements,\n  map,\n  merge,\n  observeOn,\n  of,\n  repeat,\n  scan,\n  share,\n  skip,\n  startWith,\n  switchMap,\n  takeUntil,\n  tap,\n  withLatestFrom\n} from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport {\n  Viewport,\n  getElement,\n  getElementContainer,\n  getElementSize,\n  getElements,\n  getLocation,\n  getOptionalElement,\n  watchElementSize\n} from \"~/browser\"\n\nimport {\n  Component,\n  getComponentElement\n} from \"../_\"\nimport { Header } from \"../header\"\nimport { Main } from \"../main\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Table of contents\n */\nexport interface TableOfContents {\n  prev: HTMLAnchorElement[][]          /* Anchors (previous) */\n  next: HTMLAnchorElement[][]          /* Anchors (next) */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n  main$: Observable<Main>              /* Main area observable */\n  target$: Observable<HTMLElement>     /* Location target observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch table of contents\n *\n * This is effectively a scroll spy implementation which will account for the\n * fixed header and automatically re-calculate anchor offsets when the viewport\n * is resized. The returned observable will only emit if the table of contents\n * needs to be repainted.\n *\n * This implementation tracks an anchor element's entire path starting from its\n * level up to the top-most anchor element, e.g. `[h3, h2, h1]`. Although the\n * Material theme currently doesn't make use of this information, it enables\n * the styling of the entire hierarchy through customization.\n *\n * Note that the current anchor is the last item of the `prev` anchor list.\n *\n * @param el - Table of contents element\n * @param options - Options\n *\n * @returns Table of contents observable\n */\nexport function watchTableOfContents(\n  el: HTMLElement, { viewport$, header$ }: WatchOptions\n): Observable<TableOfContents> {\n  const table = new Map<HTMLAnchorElement, HTMLElement>()\n\n  /* Compute anchor-to-target mapping */\n  const anchors = getElements<HTMLAnchorElement>(\".md-nav__link\", el)\n  for (const anchor of anchors) {\n    const id = decodeURIComponent(anchor.hash.substring(1))\n    const target = getOptionalElement(`[id=\"${id}\"]`)\n    if (typeof target !== \"undefined\")\n      table.set(anchor, target)\n  }\n\n  /* Compute necessary adjustment for header */\n  const adjust$ = header$\n    .pipe(\n      distinctUntilKeyChanged(\"height\"),\n      map(({ height }) => {\n        const main = getComponentElement(\"main\")\n        const grid = getElement(\":scope > :first-child\", main)\n        return height + 0.8 * (\n          grid.offsetTop -\n          main.offsetTop\n        )\n      }),\n      share()\n    )\n\n  /* Compute partition of previous and next anchors */\n  const partition$ = watchElementSize(document.body)\n    .pipe(\n      distinctUntilKeyChanged(\"height\"),\n\n      /* Build index to map anchor paths to vertical offsets */\n      switchMap(body => defer(() => {\n        let path: HTMLAnchorElement[] = []\n        return of([...table].reduce((index, [anchor, target]) => {\n          while (path.length) {\n            const last = table.get(path[path.length - 1])!\n            if (last.tagName >= target.tagName) {\n              path.pop()\n            } else {\n              break\n            }\n          }\n\n          /* If the current anchor is hidden, continue with its parent */\n          let offset = target.offsetTop\n          while (!offset && target.parentElement) {\n            target = target.parentElement\n            offset = target.offsetTop\n          }\n\n          /* Fix anchor offsets in tables - see https://bit.ly/3CUFOcn */\n          let parent = target.offsetParent as HTMLElement\n          for (; parent; parent = parent.offsetParent as HTMLElement)\n            offset += parent.offsetTop\n\n          /* Map reversed anchor path to vertical offset */\n          return index.set(\n            [...path = [...path, anchor]].reverse(),\n            offset\n          )\n        }, new Map<HTMLAnchorElement[], number>()))\n      })\n        .pipe(\n\n          /* Sort index by vertical offset (see https://bit.ly/30z6QSO) */\n          map(index => new Map([...index].sort(([, a], [, b]) => a - b))),\n          combineLatestWith(adjust$),\n\n          /* Re-compute partition when viewport offset changes */\n          switchMap(([index, adjust]) => viewport$\n            .pipe(\n              scan(([prev, next], { offset: { y }, size }) => {\n                const last = y + size.height >= Math.floor(body.height)\n\n                /* Look forward */\n                while (next.length) {\n                  const [, offset] = next[0]\n                  if (offset - adjust < y || last) {\n                    prev = [...prev, next.shift()!]\n                  } else {\n                    break\n                  }\n                }\n\n                /* Look backward */\n                while (prev.length) {\n                  const [, offset] = prev[prev.length - 1]\n                  if (offset - adjust >= y && !last) {\n                    next = [prev.pop()!, ...next]\n                  } else {\n                    break\n                  }\n                }\n\n                /* Return partition */\n                return [prev, next]\n              }, [[], [...index]]),\n              distinctUntilChanged((a, b) => (\n                a[0] === b[0] &&\n                a[1] === b[1]\n              ))\n            )\n          )\n        )\n      )\n    )\n\n  /* Compute and return anchor list migrations */\n  return partition$\n    .pipe(\n      map(([prev, next]) => ({\n        prev: prev.map(([path]) => path),\n        next: next.map(([path]) => path)\n      })),\n\n      /* Extract anchor list migrations */\n      startWith({ prev: [], next: [] }),\n      bufferCount(2, 1),\n      map(([a, b]) => {\n\n        /* Moving down */\n        if (a.prev.length < b.prev.length) {\n          return {\n            prev: b.prev.slice(Math.max(0, a.prev.length - 1), b.prev.length),\n            next: []\n          }\n\n        /* Moving up */\n        } else {\n          return {\n            prev: b.prev.slice(-1),\n            next: b.next.slice(0, b.next.length - a.next.length)\n          }\n        }\n      })\n    )\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Mount table of contents\n *\n * @param el - Table of contents element\n * @param options - Options\n *\n * @returns Table of contents component observable\n */\nexport function mountTableOfContents(\n  el: HTMLElement, { viewport$, header$, main$, target$ }: MountOptions\n): Observable<Component<TableOfContents>> {\n  return defer(() => {\n    const push$ = new Subject<TableOfContents>()\n    const done$ = push$.pipe(ignoreElements(), endWith(true))\n    push$.subscribe(({ prev, next }) => {\n\n      /* Look forward */\n      for (const [anchor] of next) {\n        anchor.classList.remove(\"md-nav__link--passed\")\n        anchor.classList.remove(\"md-nav__link--active\")\n      }\n\n      /* Look backward */\n      for (const [index, [anchor]] of prev.entries()) {\n        anchor.classList.add(\"md-nav__link--passed\")\n        anchor.classList.toggle(\n          \"md-nav__link--active\",\n          index === prev.length - 1\n        )\n      }\n    })\n\n    /* Set up following, if enabled */\n    if (feature(\"toc.follow\")) {\n\n      /* Toggle smooth scrolling only for anchor clicks */\n      const smooth$ = merge(\n        viewport$.pipe(debounceTime(1), map(() => undefined)),\n        viewport$.pipe(debounceTime(250), map(() => \"smooth\" as const))\n      )\n\n      /* Bring active anchor into view */ // @todo: refactor\n      push$\n        .pipe(\n          filter(({ prev }) => prev.length > 0),\n          combineLatestWith(main$.pipe(observeOn(asyncScheduler))),\n          withLatestFrom(smooth$)\n        )\n          .subscribe(([[{ prev }], behavior]) => {\n            const [anchor] = prev[prev.length - 1]\n            if (anchor.offsetHeight) {\n\n              /* Retrieve overflowing container and scroll */\n              const container = getElementContainer(anchor)\n              if (typeof container !== \"undefined\") {\n                const offset = anchor.offsetTop - container.offsetTop\n                const { height } = getElementSize(container)\n                container.scrollTo({\n                  top: offset - height / 2,\n                  behavior\n                })\n              }\n            }\n          })\n    }\n\n    /* Set up anchor tracking, if enabled */\n    if (feature(\"navigation.tracking\"))\n      viewport$\n        .pipe(\n          takeUntil(done$),\n          distinctUntilKeyChanged(\"offset\"),\n          debounceTime(250),\n          skip(1),\n          takeUntil(target$.pipe(skip(1))),\n          repeat({ delay: 250 }),\n          withLatestFrom(push$)\n        )\n          .subscribe(([, { prev }]) => {\n            const url = getLocation()\n\n            /* Set hash fragment to active anchor */\n            const anchor = prev[prev.length - 1]\n            if (anchor && anchor.length) {\n              const [active] = anchor\n              const { hash } = new URL(active.href)\n              if (url.hash !== hash) {\n                url.hash = hash\n                history.replaceState({}, \"\", `${url}`)\n              }\n\n            /* Reset anchor when at the top */\n            } else {\n              url.hash = \"\"\n              history.replaceState({}, \"\", `${url}`)\n            }\n          })\n\n    /* Create and return component */\n    return watchTableOfContents(el, { viewport$, header$ })\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  bufferCount,\n  combineLatest,\n  distinctUntilChanged,\n  distinctUntilKeyChanged,\n  endWith,\n  finalize,\n  fromEvent,\n  ignoreElements,\n  map,\n  repeat,\n  skip,\n  takeUntil,\n  tap\n} from \"rxjs\"\n\nimport { Viewport } from \"~/browser\"\n\nimport { Component } from \"../_\"\nimport { Header } from \"../header\"\nimport { Main } from \"../main\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Back-to-top button\n */\nexport interface BackToTop {\n  hidden: boolean                      /* Back-to-top button is hidden */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  main$: Observable<Main>              /* Main area observable */\n  target$: Observable<HTMLElement>     /* Location target observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n  main$: Observable<Main>              /* Main area observable */\n  target$: Observable<HTMLElement>     /* Location target observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch back-to-top\n *\n * @param _el - Back-to-top element\n * @param options - Options\n *\n * @returns Back-to-top observable\n */\nexport function watchBackToTop(\n  _el: HTMLElement, { viewport$, main$, target$ }: WatchOptions\n): Observable<BackToTop> {\n\n  /* Compute direction */\n  const direction$ = viewport$\n    .pipe(\n      map(({ offset: { y } }) => y),\n      bufferCount(2, 1),\n      map(([a, b]) => a > b && b > 0),\n      distinctUntilChanged()\n    )\n\n  /* Compute whether main area is active */\n  const active$ = main$\n    .pipe(\n      map(({ active }) => active)\n    )\n\n  /* Compute threshold for hiding */\n  return combineLatest([active$, direction$])\n    .pipe(\n      map(([active, direction]) => !(active && direction)),\n      distinctUntilChanged(),\n      takeUntil(target$.pipe(skip(1))),\n      endWith(true),\n      repeat({ delay: 250 }),\n      map(hidden => ({ hidden }))\n    )\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Mount back-to-top\n *\n * @param el - Back-to-top element\n * @param options - Options\n *\n * @returns Back-to-top component observable\n */\nexport function mountBackToTop(\n  el: HTMLElement, { viewport$, header$, main$, target$ }: MountOptions\n): Observable<Component<BackToTop>> {\n  const push$ = new Subject<BackToTop>()\n  const done$ = push$.pipe(ignoreElements(), endWith(true))\n  push$.subscribe({\n\n    /* Handle emission */\n    next({ hidden }) {\n      el.hidden = hidden\n      if (hidden) {\n        el.setAttribute(\"tabindex\", \"-1\")\n        el.blur()\n      } else {\n        el.removeAttribute(\"tabindex\")\n      }\n    },\n\n    /* Handle complete */\n    complete() {\n      el.style.top = \"\"\n      el.hidden = true\n      el.removeAttribute(\"tabindex\")\n    }\n  })\n\n  /* Watch header height */\n  header$\n    .pipe(\n      takeUntil(done$),\n      distinctUntilKeyChanged(\"height\")\n    )\n      .subscribe(({ height }) => {\n        el.style.top = `${height + 16}px`\n      })\n\n  /* Go back to top */\n  fromEvent(el, \"click\")\n    .subscribe(ev => {\n      ev.preventDefault()\n      window.scrollTo({ top: 0 })\n    })\n\n  /* Create and return component */\n  return watchBackToTop(el, { viewport$, main$, target$ })\n    .pipe(\n      tap(state => push$.next(state)),\n      finalize(() => push$.complete()),\n      map(state => ({ ref: el, ...state }))\n    )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  filter,\n  finalize,\n  map,\n  mergeMap,\n  skip,\n  switchMap,\n  take,\n  takeUntil\n} from \"rxjs\"\n\nimport {\n  getElements,\n  watchElementVisibility\n} from \"~/browser\"\nimport { mountTooltip } from \"~/components\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Patch options\n */\ninterface PatchOptions {\n  document$: Observable<Document>      /* Document observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Patch ellipsis\n *\n * This function will fetch all elements that are shortened with ellipsis, and\n * filter those which are visible. Once they become visible, they stay in that\n * state, even though they may be hidden again. This optimization is necessary\n * to reduce pressure on the browser, with elements fading in and out of view.\n *\n * @param options - Options\n */\nexport function patchEllipsis(\n  { document$ }: PatchOptions\n): void {\n  document$\n    .pipe(\n      switchMap(() => getElements(\".md-ellipsis\")),\n      mergeMap(el => watchElementVisibility(el)\n        .pipe(\n          takeUntil(document$.pipe(skip(1))),\n          filter(visible => visible),\n          map(() => el),\n          take(1)\n        )\n      ),\n      filter(el => el.offsetWidth < el.scrollWidth),\n      mergeMap(el => {\n        const text = el.innerText\n        const host = el.closest(\"a\") || el\n        host.title = text\n\n        /* Mount tooltip */\n        return mountTooltip(host)\n          .pipe(\n            takeUntil(document$.pipe(skip(1))),\n            finalize(() => host.removeAttribute(\"title\"))\n          )\n      })\n    )\n      .subscribe()\n\n  // @todo move this outside of here and fix memleaks\n  document$\n    .pipe(\n      switchMap(() => getElements(\".md-status\")),\n      mergeMap(el => mountTooltip(el))\n    )\n      .subscribe()\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  fromEvent,\n  map,\n  mergeMap,\n  switchMap,\n  takeWhile,\n  tap,\n  withLatestFrom\n} from \"rxjs\"\n\nimport { getElements } from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Patch options\n */\ninterface PatchOptions {\n  document$: Observable<Document>      /* Document observable */\n  tablet$: Observable<boolean>         /* Media tablet observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Patch indeterminate checkboxes\n *\n * This function replaces the indeterminate \"pseudo state\" with the actual\n * indeterminate state, which is used to keep navigation always expanded.\n *\n * @param options - Options\n */\nexport function patchIndeterminate(\n  { document$, tablet$ }: PatchOptions\n): void {\n  document$\n    .pipe(\n      switchMap(() => getElements<HTMLInputElement>(\n        \".md-toggle--indeterminate\"\n      )),\n      tap(el => {\n        el.indeterminate = true\n        el.checked = false\n      }),\n      mergeMap(el => fromEvent(el, \"change\")\n        .pipe(\n          takeWhile(() => el.classList.contains(\"md-toggle--indeterminate\")),\n          map(() => el)\n        )\n      ),\n      withLatestFrom(tablet$)\n    )\n      .subscribe(([el, tablet]) => {\n        el.classList.remove(\"md-toggle--indeterminate\")\n        if (tablet)\n          el.checked = false\n      })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  filter,\n  fromEvent,\n  map,\n  mergeMap,\n  switchMap,\n  tap\n} from \"rxjs\"\n\nimport { getElements } from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Patch options\n */\ninterface PatchOptions {\n  document$: Observable<Document>      /* Document observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Check whether the given device is an Apple device\n *\n * @returns Test result\n */\nfunction isAppleDevice(): boolean {\n  return /(iPad|iPhone|iPod)/.test(navigator.userAgent)\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Patch all elements with `data-md-scrollfix` attributes\n *\n * This is a year-old patch which ensures that overflow scrolling works at the\n * top and bottom of containers on iOS by ensuring a `1px` scroll offset upon\n * the start of a touch event.\n *\n * @see https://bit.ly/2SCtAOO - Original source\n *\n * @param options - Options\n */\nexport function patchScrollfix(\n  { document$ }: PatchOptions\n): void {\n  document$\n    .pipe(\n      switchMap(() => getElements(\"[data-md-scrollfix]\")),\n      tap(el => el.removeAttribute(\"data-md-scrollfix\")),\n      filter(isAppleDevice),\n      mergeMap(el => fromEvent(el, \"touchstart\")\n        .pipe(\n          map(() => el)\n        )\n      )\n    )\n      .subscribe(el => {\n        const top = el.scrollTop\n\n        /* We're at the top of the container */\n        if (top === 0) {\n          el.scrollTop = 1\n\n        /* We're at the bottom of the container */\n        } else if (top + el.offsetHeight === el.scrollHeight) {\n          el.scrollTop = top - 1\n        }\n      })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  combineLatest,\n  delay,\n  map,\n  of,\n  switchMap,\n  withLatestFrom\n} from \"rxjs\"\n\nimport {\n  Viewport,\n  watchToggle\n} from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Patch options\n */\ninterface PatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  tablet$: Observable<boolean>         /* Media tablet observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Patch the document body to lock when search is open\n *\n * For mobile and tablet viewports, the search is rendered full screen, which\n * leads to scroll leaking when at the top or bottom of the search result. This\n * function locks the body when the search is in full screen mode, and restores\n * the scroll position when leaving.\n *\n * @param options - Options\n */\nexport function patchScrolllock(\n  { viewport$, tablet$ }: PatchOptions\n): void {\n  combineLatest([watchToggle(\"search\"), tablet$])\n    .pipe(\n      map(([active, tablet]) => active && !tablet),\n      switchMap(active => of(active)\n        .pipe(\n          delay(active ? 400 : 100)\n        )\n      ),\n      withLatestFrom(viewport$)\n    )\n      .subscribe(([active, { offset: { y }}]) => {\n        if (active) {\n          document.body.setAttribute(\"data-md-scrolllock\", \"\")\n          document.body.style.top = `-${y}px`\n        } else {\n          const value = -1 * parseInt(document.body.style.top, 10)\n          document.body.removeAttribute(\"data-md-scrolllock\")\n          document.body.style.top = \"\"\n          if (value)\n            window.scrollTo(0, value)\n        }\n      })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Polyfills\n * ------------------------------------------------------------------------- */\n\n/* Polyfill `Object.entries` */\nif (!Object.entries)\n  Object.entries = function (obj: object) {\n    const data: [string, string][] = []\n    for (const key of Object.keys(obj))\n      // @ts-expect-error - ignore property access warning\n      data.push([key, obj[key]])\n\n    /* Return entries */\n    return data\n  }\n\n/* Polyfill `Object.values` */\nif (!Object.values)\n  Object.values = function (obj: object) {\n    const data: string[] = []\n    for (const key of Object.keys(obj))\n      // @ts-expect-error - ignore property access warning\n      data.push(obj[key])\n\n    /* Return values */\n    return data\n  }\n\n/* ------------------------------------------------------------------------- */\n\n/* Polyfills for `Element` */\nif (typeof Element !== \"undefined\") {\n\n  /* Polyfill `Element.scrollTo` */\n  if (!Element.prototype.scrollTo)\n    Element.prototype.scrollTo = function (\n      x?: ScrollToOptions | number, y?: number\n    ): void {\n      if (typeof x === \"object\") {\n        this.scrollLeft = x.left!\n        this.scrollTop = x.top!\n      } else {\n        this.scrollLeft = x!\n        this.scrollTop = y!\n      }\n    }\n\n  /* Polyfill `Element.replaceWith` */\n  if (!Element.prototype.replaceWith)\n    Element.prototype.replaceWith = function (\n      ...nodes: Array<string | Node>\n    ): void {\n      const parent = this.parentNode\n      if (parent) {\n        if (nodes.length === 0)\n          parent.removeChild(this)\n\n        /* Replace children and create text nodes */\n        for (let i = nodes.length - 1; i >= 0; i--) {\n          let node = nodes[i]\n          if (typeof node === \"string\")\n            node = document.createTextNode(node)\n          else if (node.parentNode)\n            node.parentNode.removeChild(node)\n\n          /* Replace child or insert before previous sibling */\n          if (!i)\n            parent.replaceChild(node, this)\n          else\n            parent.insertBefore(this.previousSibling!, node)\n        }\n      }\n    }\n}\n"],
+  "mappings": "2rCAAA,IAAAA,GAAAC,GAAA,CAAAC,GAAAC,KAAA,EAAC,SAAUC,EAAQC,EAAS,CAC1B,OAAOH,IAAY,UAAY,OAAOC,IAAW,YAAcE,EAAQ,EACvE,OAAO,QAAW,YAAc,OAAO,IAAM,OAAOA,CAAO,EAC1DA,EAAQ,CACX,GAAEH,GAAO,UAAY,CAAE,aASrB,SAASI,EAA0BC,EAAO,CACxC,IAAIC,EAAmB,GACnBC,EAA0B,GAC1BC,EAAiC,KAEjCC,EAAsB,CACxB,KAAM,GACN,OAAQ,GACR,IAAK,GACL,IAAK,GACL,MAAO,GACP,SAAU,GACV,OAAQ,GACR,KAAM,GACN,MAAO,GACP,KAAM,GACN,KAAM,GACN,SAAU,GACV,iBAAkB,EACpB,EAOA,SAASC,EAAmBC,EAAI,CAC9B,MACE,GAAAA,GACAA,IAAO,UACPA,EAAG,WAAa,QAChBA,EAAG,WAAa,QAChB,cAAeA,GACf,aAAcA,EAAG,UAKrB,CASA,SAASC,EAA8BD,EAAI,CACzC,IAAIE,GAAOF,EAAG,KACVG,GAAUH,EAAG,QAUjB,MARI,GAAAG,KAAY,SAAWL,EAAoBI,EAAI,GAAK,CAACF,EAAG,UAIxDG,KAAY,YAAc,CAACH,EAAG,UAI9BA,EAAG,kBAKT,CAOA,SAASI,EAAqBJ,EAAI,CAC5BA,EAAG,UAAU,SAAS,eAAe,IAGzCA,EAAG,UAAU,IAAI,eAAe,EAChCA,EAAG,aAAa,2BAA4B,EAAE,EAChD,CAOA,SAASK,EAAwBL,EAAI,CAC9BA,EAAG,aAAa,0BAA0B,IAG/CA,EAAG,UAAU,OAAO,eAAe,EACnCA,EAAG,gBAAgB,0BAA0B,EAC/C,CAUA,SAASM,EAAUC,EAAG,CAChBA,EAAE,SAAWA,EAAE,QAAUA,EAAE,UAI3BR,EAAmBL,EAAM,aAAa,GACxCU,EAAqBV,EAAM,aAAa,EAG1CC,EAAmB,GACrB,CAUA,SAASa,EAAcD,EAAG,CACxBZ,EAAmB,EACrB,CASA,SAASc,EAAQF,EAAG,CAEbR,EAAmBQ,EAAE,MAAM,IAI5BZ,GAAoBM,EAA8BM,EAAE,MAAM,IAC5DH,EAAqBG,EAAE,MAAM,CAEjC,CAMA,SAASG,EAAOH,EAAG,CACZR,EAAmBQ,EAAE,MAAM,IAK9BA,EAAE,OAAO,UAAU,SAAS,eAAe,GAC3CA,EAAE,OAAO,aAAa,0BAA0B,KAMhDX,EAA0B,GAC1B,OAAO,aAAaC,CAA8B,EAClDA,EAAiC,OAAO,WAAW,UAAW,CAC5DD,EAA0B,EAC5B,EAAG,GAAG,EACNS,EAAwBE,EAAE,MAAM,EAEpC,CAOA,SAASI,EAAmBJ,EAAG,CACzB,SAAS,kBAAoB,WAK3BX,IACFD,EAAmB,IAErBiB,EAA+B,EAEnC,CAQA,SAASA,GAAiC,CACxC,SAAS,iBAAiB,YAAaC,CAAoB,EAC3D,SAAS,iBAAiB,YAAaA,CAAoB,EAC3D,SAAS,iBAAiB,UAAWA,CAAoB,EACzD,SAAS,iBAAiB,cAAeA,CAAoB,EAC7D,SAAS,iBAAiB,cAAeA,CAAoB,EAC7D,SAAS,iBAAiB,YAAaA,CAAoB,EAC3D,SAAS,iBAAiB,YAAaA,CAAoB,EAC3D,SAAS,iBAAiB,aAAcA,CAAoB,EAC5D,SAAS,iBAAiB,WAAYA,CAAoB,CAC5D,CAEA,SAASC,IAAoC,CAC3C,SAAS,oBAAoB,YAAaD,CAAoB,EAC9D,SAAS,oBAAoB,YAAaA,CAAoB,EAC9D,SAAS,oBAAoB,UAAWA,CAAoB,EAC5D,SAAS,oBAAoB,cAAeA,CAAoB,EAChE,SAAS,oBAAoB,cAAeA,CAAoB,EAChE,SAAS,oBAAoB,YAAaA,CAAoB,EAC9D,SAAS,oBAAoB,YAAaA,CAAoB,EAC9D,SAAS,oBAAoB,aAAcA,CAAoB,EAC/D,SAAS,oBAAoB,WAAYA,CAAoB,CAC/D,CASA,SAASA,EAAqBN,EAAG,CAG3BA,EAAE,OAAO,UAAYA,EAAE,OAAO,SAAS,YAAY,IAAM,SAI7DZ,EAAmB,GACnBmB,GAAkC,EACpC,CAKA,SAAS,iBAAiB,UAAWR,EAAW,EAAI,EACpD,SAAS,iBAAiB,YAAaE,EAAe,EAAI,EAC1D,SAAS,iBAAiB,cAAeA,EAAe,EAAI,EAC5D,SAAS,iBAAiB,aAAcA,EAAe,EAAI,EAC3D,SAAS,iBAAiB,mBAAoBG,EAAoB,EAAI,EAEtEC,EAA+B,EAM/BlB,EAAM,iBAAiB,QAASe,EAAS,EAAI,EAC7Cf,EAAM,iBAAiB,OAAQgB,EAAQ,EAAI,EAOvChB,EAAM,WAAa,KAAK,wBAA0BA,EAAM,KAI1DA,EAAM,KAAK,aAAa,wBAAyB,EAAE,EAC1CA,EAAM,WAAa,KAAK,gBACjC,SAAS,gBAAgB,UAAU,IAAI,kBAAkB,EACzD,SAAS,gBAAgB,aAAa,wBAAyB,EAAE,EAErE,CAKA,GAAI,OAAO,QAAW,aAAe,OAAO,UAAa,YAAa,CAIpE,OAAO,0BAA4BD,EAInC,IAAIsB,EAEJ,GAAI,CACFA,EAAQ,IAAI,YAAY,8BAA8B,CACxD,OAASC,EAAO,CAEdD,EAAQ,SAAS,YAAY,aAAa,EAC1CA,EAAM,gBAAgB,+BAAgC,GAAO,GAAO,CAAC,CAAC,CACxE,CAEA,OAAO,cAAcA,CAAK,CAC5B,CAEI,OAAO,UAAa,aAGtBtB,EAA0B,QAAQ,CAGtC,CAAE,ICvTF,IAAAwB,GAAAC,GAAA,CAAAC,GAAAC,KAAA;AAAA;AAAA;AAAA;AAAA;AAAA,IAMC,SAA0CC,EAAMC,EAAS,CACtD,OAAOH,IAAY,UAAY,OAAOC,IAAW,SACnDA,GAAO,QAAUE,EAAQ,EAClB,OAAO,QAAW,YAAc,OAAO,IAC9C,OAAO,CAAC,EAAGA,CAAO,EACX,OAAOH,IAAY,SAC1BA,GAAQ,YAAiBG,EAAQ,EAEjCD,EAAK,YAAiBC,EAAQ,CAChC,GAAGH,GAAM,UAAW,CACpB,OAAiB,UAAW,CAClB,IAAII,EAAuB,CAE/B,IACC,SAASC,EAAyBC,EAAqBC,EAAqB,CAEnF,aAGAA,EAAoB,EAAED,EAAqB,CACzC,QAAW,UAAW,CAAE,OAAqBE,EAAW,CAC1D,CAAC,EAGD,IAAIC,EAAeF,EAAoB,GAAG,EACtCG,EAAoCH,EAAoB,EAAEE,CAAY,EAEtEE,EAASJ,EAAoB,GAAG,EAChCK,EAA8BL,EAAoB,EAAEI,CAAM,EAE1DE,EAAaN,EAAoB,GAAG,EACpCO,EAA8BP,EAAoB,EAAEM,CAAU,EAOlE,SAASE,EAAQC,EAAM,CACrB,GAAI,CACF,OAAO,SAAS,YAAYA,CAAI,CAClC,OAASC,EAAK,CACZ,MAAO,EACT,CACF,CAUA,IAAIC,EAAqB,SAA4BC,EAAQ,CAC3D,IAAIC,EAAeN,EAAe,EAAEK,CAAM,EAC1C,OAAAJ,EAAQ,KAAK,EACNK,CACT,EAEiCC,EAAeH,EAOhD,SAASI,EAAkBC,EAAO,CAChC,IAAIC,EAAQ,SAAS,gBAAgB,aAAa,KAAK,IAAM,MACzDC,EAAc,SAAS,cAAc,UAAU,EAEnDA,EAAY,MAAM,SAAW,OAE7BA,EAAY,MAAM,OAAS,IAC3BA,EAAY,MAAM,QAAU,IAC5BA,EAAY,MAAM,OAAS,IAE3BA,EAAY,MAAM,SAAW,WAC7BA,EAAY,MAAMD,EAAQ,QAAU,MAAM,EAAI,UAE9C,IAAIE,EAAY,OAAO,aAAe,SAAS,gBAAgB,UAC/D,OAAAD,EAAY,MAAM,IAAM,GAAG,OAAOC,EAAW,IAAI,EACjDD,EAAY,aAAa,WAAY,EAAE,EACvCA,EAAY,MAAQF,EACbE,CACT,CAYA,IAAIE,EAAiB,SAAwBJ,EAAOK,EAAS,CAC3D,IAAIH,EAAcH,EAAkBC,CAAK,EACzCK,EAAQ,UAAU,YAAYH,CAAW,EACzC,IAAIL,EAAeN,EAAe,EAAEW,CAAW,EAC/C,OAAAV,EAAQ,MAAM,EACdU,EAAY,OAAO,EACZL,CACT,EASIS,GAAsB,SAA6BV,EAAQ,CAC7D,IAAIS,EAAU,UAAU,OAAS,GAAK,UAAU,CAAC,IAAM,OAAY,UAAU,CAAC,EAAI,CAChF,UAAW,SAAS,IACtB,EACIR,EAAe,GAEnB,OAAI,OAAOD,GAAW,SACpBC,EAAeO,EAAeR,EAAQS,CAAO,EACpCT,aAAkB,kBAAoB,CAAC,CAAC,OAAQ,SAAU,MAAO,MAAO,UAAU,EAAE,SAASA,GAAW,KAA4B,OAASA,EAAO,IAAI,EAEjKC,EAAeO,EAAeR,EAAO,MAAOS,CAAO,GAEnDR,EAAeN,EAAe,EAAEK,CAAM,EACtCJ,EAAQ,MAAM,GAGTK,CACT,EAEiCU,EAAgBD,GAEjD,SAASE,EAAQC,EAAK,CAAE,0BAA2B,OAAI,OAAO,QAAW,YAAc,OAAO,OAAO,UAAa,SAAYD,EAAU,SAAiBC,EAAK,CAAE,OAAO,OAAOA,CAAK,EAAYD,EAAU,SAAiBC,EAAK,CAAE,OAAOA,GAAO,OAAO,QAAW,YAAcA,EAAI,cAAgB,QAAUA,IAAQ,OAAO,UAAY,SAAW,OAAOA,CAAK,EAAYD,EAAQC,CAAG,CAAG,CAUzX,IAAIC,GAAyB,UAAkC,CAC7D,IAAIL,EAAU,UAAU,OAAS,GAAK,UAAU,CAAC,IAAM,OAAY,UAAU,CAAC,EAAI,CAAC,EAE/EM,EAAkBN,EAAQ,OAC1BO,EAASD,IAAoB,OAAS,OAASA,EAC/CE,EAAYR,EAAQ,UACpBT,EAASS,EAAQ,OACjBS,GAAOT,EAAQ,KAEnB,GAAIO,IAAW,QAAUA,IAAW,MAClC,MAAM,IAAI,MAAM,oDAAoD,EAItE,GAAIhB,IAAW,OACb,GAAIA,GAAUY,EAAQZ,CAAM,IAAM,UAAYA,EAAO,WAAa,EAAG,CACnE,GAAIgB,IAAW,QAAUhB,EAAO,aAAa,UAAU,EACrD,MAAM,IAAI,MAAM,mFAAmF,EAGrG,GAAIgB,IAAW,QAAUhB,EAAO,aAAa,UAAU,GAAKA,EAAO,aAAa,UAAU,GACxF,MAAM,IAAI,MAAM,uGAAwG,CAE5H,KACE,OAAM,IAAI,MAAM,6CAA6C,EAKjE,GAAIkB,GACF,OAAOP,EAAaO,GAAM,CACxB,UAAWD,CACb,CAAC,EAIH,GAAIjB,EACF,OAAOgB,IAAW,MAAQd,EAAYF,CAAM,EAAIW,EAAaX,EAAQ,CACnE,UAAWiB,CACb,CAAC,CAEL,EAEiCE,GAAmBL,GAEpD,SAASM,GAAiBP,EAAK,CAAE,0BAA2B,OAAI,OAAO,QAAW,YAAc,OAAO,OAAO,UAAa,SAAYO,GAAmB,SAAiBP,EAAK,CAAE,OAAO,OAAOA,CAAK,EAAYO,GAAmB,SAAiBP,EAAK,CAAE,OAAOA,GAAO,OAAO,QAAW,YAAcA,EAAI,cAAgB,QAAUA,IAAQ,OAAO,UAAY,SAAW,OAAOA,CAAK,EAAYO,GAAiBP,CAAG,CAAG,CAE7Z,SAASQ,GAAgBC,EAAUC,EAAa,CAAE,GAAI,EAAED,aAAoBC,GAAgB,MAAM,IAAI,UAAU,mCAAmC,CAAK,CAExJ,SAASC,GAAkBxB,EAAQyB,EAAO,CAAE,QAASC,EAAI,EAAGA,EAAID,EAAM,OAAQC,IAAK,CAAE,IAAIC,EAAaF,EAAMC,CAAC,EAAGC,EAAW,WAAaA,EAAW,YAAc,GAAOA,EAAW,aAAe,GAAU,UAAWA,IAAYA,EAAW,SAAW,IAAM,OAAO,eAAe3B,EAAQ2B,EAAW,IAAKA,CAAU,CAAG,CAAE,CAE5T,SAASC,GAAaL,EAAaM,EAAYC,EAAa,CAAE,OAAID,GAAYL,GAAkBD,EAAY,UAAWM,CAAU,EAAOC,GAAaN,GAAkBD,EAAaO,CAAW,EAAUP,CAAa,CAEtN,SAASQ,GAAUC,EAAUC,EAAY,CAAE,GAAI,OAAOA,GAAe,YAAcA,IAAe,KAAQ,MAAM,IAAI,UAAU,oDAAoD,EAAKD,EAAS,UAAY,OAAO,OAAOC,GAAcA,EAAW,UAAW,CAAE,YAAa,CAAE,MAAOD,EAAU,SAAU,GAAM,aAAc,EAAK,CAAE,CAAC,EAAOC,GAAYC,GAAgBF,EAAUC,CAAU,CAAG,CAEhY,SAASC,GAAgBC,EAAGC,EAAG,CAAE,OAAAF,GAAkB,OAAO,gBAAkB,SAAyBC,EAAGC,EAAG,CAAE,OAAAD,EAAE,UAAYC,EAAUD,CAAG,EAAUD,GAAgBC,EAAGC,CAAC,CAAG,CAEzK,SAASC,GAAaC,EAAS,CAAE,IAAIC,EAA4BC,GAA0B,EAAG,OAAO,UAAgC,CAAE,IAAIC,EAAQC,GAAgBJ,CAAO,EAAGK,EAAQ,GAAIJ,EAA2B,CAAE,IAAIK,EAAYF,GAAgB,IAAI,EAAE,YAAaC,EAAS,QAAQ,UAAUF,EAAO,UAAWG,CAAS,CAAG,MAASD,EAASF,EAAM,MAAM,KAAM,SAAS,EAAK,OAAOI,GAA2B,KAAMF,CAAM,CAAG,CAAG,CAExa,SAASE,GAA2BC,EAAMC,EAAM,CAAE,OAAIA,IAAS3B,GAAiB2B,CAAI,IAAM,UAAY,OAAOA,GAAS,YAAsBA,EAAeC,GAAuBF,CAAI,CAAG,CAEzL,SAASE,GAAuBF,EAAM,CAAE,GAAIA,IAAS,OAAU,MAAM,IAAI,eAAe,2DAA2D,EAAK,OAAOA,CAAM,CAErK,SAASN,IAA4B,CAA0E,GAApE,OAAO,SAAY,aAAe,CAAC,QAAQ,WAA6B,QAAQ,UAAU,KAAM,MAAO,GAAO,GAAI,OAAO,OAAU,WAAY,MAAO,GAAM,GAAI,CAAE,YAAK,UAAU,SAAS,KAAK,QAAQ,UAAU,KAAM,CAAC,EAAG,UAAY,CAAC,CAAC,CAAC,EAAU,EAAM,OAASS,EAAG,CAAE,MAAO,EAAO,CAAE,CAEnU,SAASP,GAAgBP,EAAG,CAAE,OAAAO,GAAkB,OAAO,eAAiB,OAAO,eAAiB,SAAyBP,EAAG,CAAE,OAAOA,EAAE,WAAa,OAAO,eAAeA,CAAC,CAAG,EAAUO,GAAgBP,CAAC,CAAG,CAa5M,SAASe,GAAkBC,EAAQC,EAAS,CAC1C,IAAIC,EAAY,kBAAkB,OAAOF,CAAM,EAE/C,GAAKC,EAAQ,aAAaC,CAAS,EAInC,OAAOD,EAAQ,aAAaC,CAAS,CACvC,CAOA,IAAIC,GAAyB,SAAUC,EAAU,CAC/CxB,GAAUuB,EAAWC,CAAQ,EAE7B,IAAIC,EAASnB,GAAaiB,CAAS,EAMnC,SAASA,EAAUG,EAAShD,EAAS,CACnC,IAAIiD,EAEJ,OAAArC,GAAgB,KAAMiC,CAAS,EAE/BI,EAAQF,EAAO,KAAK,IAAI,EAExBE,EAAM,eAAejD,CAAO,EAE5BiD,EAAM,YAAYD,CAAO,EAElBC,CACT,CAQA,OAAA9B,GAAa0B,EAAW,CAAC,CACvB,IAAK,iBACL,MAAO,UAA0B,CAC/B,IAAI7C,EAAU,UAAU,OAAS,GAAK,UAAU,CAAC,IAAM,OAAY,UAAU,CAAC,EAAI,CAAC,EACnF,KAAK,OAAS,OAAOA,EAAQ,QAAW,WAAaA,EAAQ,OAAS,KAAK,cAC3E,KAAK,OAAS,OAAOA,EAAQ,QAAW,WAAaA,EAAQ,OAAS,KAAK,cAC3E,KAAK,KAAO,OAAOA,EAAQ,MAAS,WAAaA,EAAQ,KAAO,KAAK,YACrE,KAAK,UAAYW,GAAiBX,EAAQ,SAAS,IAAM,SAAWA,EAAQ,UAAY,SAAS,IACnG,CAMF,EAAG,CACD,IAAK,cACL,MAAO,SAAqBgD,EAAS,CACnC,IAAIE,EAAS,KAEb,KAAK,SAAWlE,EAAe,EAAEgE,EAAS,QAAS,SAAUR,GAAG,CAC9D,OAAOU,EAAO,QAAQV,EAAC,CACzB,CAAC,CACH,CAMF,EAAG,CACD,IAAK,UACL,MAAO,SAAiBA,EAAG,CACzB,IAAIQ,EAAUR,EAAE,gBAAkBA,EAAE,cAChCjC,GAAS,KAAK,OAAOyC,CAAO,GAAK,OACjCvC,GAAOC,GAAgB,CACzB,OAAQH,GACR,UAAW,KAAK,UAChB,OAAQ,KAAK,OAAOyC,CAAO,EAC3B,KAAM,KAAK,KAAKA,CAAO,CACzB,CAAC,EAED,KAAK,KAAKvC,GAAO,UAAY,QAAS,CACpC,OAAQF,GACR,KAAME,GACN,QAASuC,EACT,eAAgB,UAA0B,CACpCA,GACFA,EAAQ,MAAM,EAGhB,OAAO,aAAa,EAAE,gBAAgB,CACxC,CACF,CAAC,CACH,CAMF,EAAG,CACD,IAAK,gBACL,MAAO,SAAuBA,EAAS,CACrC,OAAOP,GAAkB,SAAUO,CAAO,CAC5C,CAMF,EAAG,CACD,IAAK,gBACL,MAAO,SAAuBA,EAAS,CACrC,IAAIG,EAAWV,GAAkB,SAAUO,CAAO,EAElD,GAAIG,EACF,OAAO,SAAS,cAAcA,CAAQ,CAE1C,CAQF,EAAG,CACD,IAAK,cAML,MAAO,SAAqBH,EAAS,CACnC,OAAOP,GAAkB,OAAQO,CAAO,CAC1C,CAKF,EAAG,CACD,IAAK,UACL,MAAO,UAAmB,CACxB,KAAK,SAAS,QAAQ,CACxB,CACF,CAAC,EAAG,CAAC,CACH,IAAK,OACL,MAAO,SAAczD,EAAQ,CAC3B,IAAIS,EAAU,UAAU,OAAS,GAAK,UAAU,CAAC,IAAM,OAAY,UAAU,CAAC,EAAI,CAChF,UAAW,SAAS,IACtB,EACA,OAAOE,EAAaX,EAAQS,CAAO,CACrC,CAOF,EAAG,CACD,IAAK,MACL,MAAO,SAAaT,EAAQ,CAC1B,OAAOE,EAAYF,CAAM,CAC3B,CAOF,EAAG,CACD,IAAK,cACL,MAAO,UAAuB,CAC5B,IAAIgB,EAAS,UAAU,OAAS,GAAK,UAAU,CAAC,IAAM,OAAY,UAAU,CAAC,EAAI,CAAC,OAAQ,KAAK,EAC3F6C,EAAU,OAAO7C,GAAW,SAAW,CAACA,CAAM,EAAIA,EAClD8C,GAAU,CAAC,CAAC,SAAS,sBACzB,OAAAD,EAAQ,QAAQ,SAAU7C,GAAQ,CAChC8C,GAAUA,IAAW,CAAC,CAAC,SAAS,sBAAsB9C,EAAM,CAC9D,CAAC,EACM8C,EACT,CACF,CAAC,CAAC,EAEKR,CACT,EAAG/D,EAAqB,CAAE,EAEOF,GAAaiE,EAExC,EAEA,IACC,SAASxE,EAAQ,CAExB,IAAIiF,EAAqB,EAKzB,GAAI,OAAO,SAAY,aAAe,CAAC,QAAQ,UAAU,QAAS,CAC9D,IAAIC,EAAQ,QAAQ,UAEpBA,EAAM,QAAUA,EAAM,iBACNA,EAAM,oBACNA,EAAM,mBACNA,EAAM,kBACNA,EAAM,qBAC1B,CASA,SAASC,EAASb,EAASQ,EAAU,CACjC,KAAOR,GAAWA,EAAQ,WAAaW,GAAoB,CACvD,GAAI,OAAOX,EAAQ,SAAY,YAC3BA,EAAQ,QAAQQ,CAAQ,EAC1B,OAAOR,EAETA,EAAUA,EAAQ,UACtB,CACJ,CAEAtE,EAAO,QAAUmF,CAGX,EAEA,IACC,SAASnF,EAAQoF,EAA0B9E,EAAqB,CAEvE,IAAI6E,EAAU7E,EAAoB,GAAG,EAYrC,SAAS+E,EAAUf,EAASQ,EAAU/D,EAAMuE,EAAUC,EAAY,CAC9D,IAAIC,EAAaC,EAAS,MAAM,KAAM,SAAS,EAE/C,OAAAnB,EAAQ,iBAAiBvD,EAAMyE,EAAYD,CAAU,EAE9C,CACH,QAAS,UAAW,CAChBjB,EAAQ,oBAAoBvD,EAAMyE,EAAYD,CAAU,CAC5D,CACJ,CACJ,CAYA,SAASG,EAASC,EAAUb,EAAU/D,EAAMuE,EAAUC,EAAY,CAE9D,OAAI,OAAOI,EAAS,kBAAqB,WAC9BN,EAAU,MAAM,KAAM,SAAS,EAItC,OAAOtE,GAAS,WAGTsE,EAAU,KAAK,KAAM,QAAQ,EAAE,MAAM,KAAM,SAAS,GAI3D,OAAOM,GAAa,WACpBA,EAAW,SAAS,iBAAiBA,CAAQ,GAI1C,MAAM,UAAU,IAAI,KAAKA,EAAU,SAAUrB,EAAS,CACzD,OAAOe,EAAUf,EAASQ,EAAU/D,EAAMuE,EAAUC,CAAU,CAClE,CAAC,EACL,CAWA,SAASE,EAASnB,EAASQ,EAAU/D,EAAMuE,EAAU,CACjD,OAAO,SAASnB,EAAG,CACfA,EAAE,eAAiBgB,EAAQhB,EAAE,OAAQW,CAAQ,EAEzCX,EAAE,gBACFmB,EAAS,KAAKhB,EAASH,CAAC,CAEhC,CACJ,CAEAnE,EAAO,QAAU0F,CAGX,EAEA,IACC,SAAStF,EAAyBL,EAAS,CAQlDA,EAAQ,KAAO,SAASuB,EAAO,CAC3B,OAAOA,IAAU,QACVA,aAAiB,aACjBA,EAAM,WAAa,CAC9B,EAQAvB,EAAQ,SAAW,SAASuB,EAAO,CAC/B,IAAIP,EAAO,OAAO,UAAU,SAAS,KAAKO,CAAK,EAE/C,OAAOA,IAAU,SACTP,IAAS,qBAAuBA,IAAS,4BACzC,WAAYO,IACZA,EAAM,SAAW,GAAKvB,EAAQ,KAAKuB,EAAM,CAAC,CAAC,EACvD,EAQAvB,EAAQ,OAAS,SAASuB,EAAO,CAC7B,OAAO,OAAOA,GAAU,UACjBA,aAAiB,MAC5B,EAQAvB,EAAQ,GAAK,SAASuB,EAAO,CACzB,IAAIP,EAAO,OAAO,UAAU,SAAS,KAAKO,CAAK,EAE/C,OAAOP,IAAS,mBACpB,CAGM,EAEA,IACC,SAASf,EAAQoF,EAA0B9E,EAAqB,CAEvE,IAAIsF,EAAKtF,EAAoB,GAAG,EAC5BoF,EAAWpF,EAAoB,GAAG,EAWtC,SAASI,EAAOQ,EAAQH,EAAMuE,EAAU,CACpC,GAAI,CAACpE,GAAU,CAACH,GAAQ,CAACuE,EACrB,MAAM,IAAI,MAAM,4BAA4B,EAGhD,GAAI,CAACM,EAAG,OAAO7E,CAAI,EACf,MAAM,IAAI,UAAU,kCAAkC,EAG1D,GAAI,CAAC6E,EAAG,GAAGN,CAAQ,EACf,MAAM,IAAI,UAAU,mCAAmC,EAG3D,GAAIM,EAAG,KAAK1E,CAAM,EACd,OAAO2E,EAAW3E,EAAQH,EAAMuE,CAAQ,EAEvC,GAAIM,EAAG,SAAS1E,CAAM,EACvB,OAAO4E,EAAe5E,EAAQH,EAAMuE,CAAQ,EAE3C,GAAIM,EAAG,OAAO1E,CAAM,EACrB,OAAO6E,EAAe7E,EAAQH,EAAMuE,CAAQ,EAG5C,MAAM,IAAI,UAAU,2EAA2E,CAEvG,CAWA,SAASO,EAAWG,EAAMjF,EAAMuE,EAAU,CACtC,OAAAU,EAAK,iBAAiBjF,EAAMuE,CAAQ,EAE7B,CACH,QAAS,UAAW,CAChBU,EAAK,oBAAoBjF,EAAMuE,CAAQ,CAC3C,CACJ,CACJ,CAWA,SAASQ,EAAeG,EAAUlF,EAAMuE,EAAU,CAC9C,aAAM,UAAU,QAAQ,KAAKW,EAAU,SAASD,EAAM,CAClDA,EAAK,iBAAiBjF,EAAMuE,CAAQ,CACxC,CAAC,EAEM,CACH,QAAS,UAAW,CAChB,MAAM,UAAU,QAAQ,KAAKW,EAAU,SAASD,EAAM,CAClDA,EAAK,oBAAoBjF,EAAMuE,CAAQ,CAC3C,CAAC,CACL,CACJ,CACJ,CAWA,SAASS,EAAejB,EAAU/D,EAAMuE,EAAU,CAC9C,OAAOI,EAAS,SAAS,KAAMZ,EAAU/D,EAAMuE,CAAQ,CAC3D,CAEAtF,EAAO,QAAUU,CAGX,EAEA,IACC,SAASV,EAAQ,CAExB,SAASkG,EAAO5B,EAAS,CACrB,IAAInD,EAEJ,GAAImD,EAAQ,WAAa,SACrBA,EAAQ,MAAM,EAEdnD,EAAemD,EAAQ,cAElBA,EAAQ,WAAa,SAAWA,EAAQ,WAAa,WAAY,CACtE,IAAI6B,EAAa7B,EAAQ,aAAa,UAAU,EAE3C6B,GACD7B,EAAQ,aAAa,WAAY,EAAE,EAGvCA,EAAQ,OAAO,EACfA,EAAQ,kBAAkB,EAAGA,EAAQ,MAAM,MAAM,EAE5C6B,GACD7B,EAAQ,gBAAgB,UAAU,EAGtCnD,EAAemD,EAAQ,KAC3B,KACK,CACGA,EAAQ,aAAa,iBAAiB,GACtCA,EAAQ,MAAM,EAGlB,IAAI8B,EAAY,OAAO,aAAa,EAChCC,EAAQ,SAAS,YAAY,EAEjCA,EAAM,mBAAmB/B,CAAO,EAChC8B,EAAU,gBAAgB,EAC1BA,EAAU,SAASC,CAAK,EAExBlF,EAAeiF,EAAU,SAAS,CACtC,CAEA,OAAOjF,CACX,CAEAnB,EAAO,QAAUkG,CAGX,EAEA,IACC,SAASlG,EAAQ,CAExB,SAASsG,GAAK,CAGd,CAEAA,EAAE,UAAY,CACZ,GAAI,SAAUC,EAAMjB,EAAUkB,EAAK,CACjC,IAAIrC,EAAI,KAAK,IAAM,KAAK,EAAI,CAAC,GAE7B,OAACA,EAAEoC,CAAI,IAAMpC,EAAEoC,CAAI,EAAI,CAAC,IAAI,KAAK,CAC/B,GAAIjB,EACJ,IAAKkB,CACP,CAAC,EAEM,IACT,EAEA,KAAM,SAAUD,EAAMjB,EAAUkB,EAAK,CACnC,IAAIxC,EAAO,KACX,SAASyB,GAAY,CACnBzB,EAAK,IAAIuC,EAAMd,CAAQ,EACvBH,EAAS,MAAMkB,EAAK,SAAS,CAC/B,CAEA,OAAAf,EAAS,EAAIH,EACN,KAAK,GAAGiB,EAAMd,EAAUe,CAAG,CACpC,EAEA,KAAM,SAAUD,EAAM,CACpB,IAAIE,EAAO,CAAC,EAAE,MAAM,KAAK,UAAW,CAAC,EACjCC,IAAW,KAAK,IAAM,KAAK,EAAI,CAAC,IAAIH,CAAI,GAAK,CAAC,GAAG,MAAM,EACvD3D,EAAI,EACJ+D,EAAMD,EAAO,OAEjB,IAAK9D,EAAGA,EAAI+D,EAAK/D,IACf8D,EAAO9D,CAAC,EAAE,GAAG,MAAM8D,EAAO9D,CAAC,EAAE,IAAK6D,CAAI,EAGxC,OAAO,IACT,EAEA,IAAK,SAAUF,EAAMjB,EAAU,CAC7B,IAAInB,EAAI,KAAK,IAAM,KAAK,EAAI,CAAC,GACzByC,EAAOzC,EAAEoC,CAAI,EACbM,EAAa,CAAC,EAElB,GAAID,GAAQtB,EACV,QAAS1C,EAAI,EAAG+D,EAAMC,EAAK,OAAQhE,EAAI+D,EAAK/D,IACtCgE,EAAKhE,CAAC,EAAE,KAAO0C,GAAYsB,EAAKhE,CAAC,EAAE,GAAG,IAAM0C,GAC9CuB,EAAW,KAAKD,EAAKhE,CAAC,CAAC,EAQ7B,OAACiE,EAAW,OACR1C,EAAEoC,CAAI,EAAIM,EACV,OAAO1C,EAAEoC,CAAI,EAEV,IACT,CACF,EAEAvG,EAAO,QAAUsG,EACjBtG,EAAO,QAAQ,YAAcsG,CAGvB,CAEI,EAGIQ,EAA2B,CAAC,EAGhC,SAASxG,EAAoByG,EAAU,CAEtC,GAAGD,EAAyBC,CAAQ,EACnC,OAAOD,EAAyBC,CAAQ,EAAE,QAG3C,IAAI/G,EAAS8G,EAAyBC,CAAQ,EAAI,CAGjD,QAAS,CAAC,CACX,EAGA,OAAA5G,EAAoB4G,CAAQ,EAAE/G,EAAQA,EAAO,QAASM,CAAmB,EAGlEN,EAAO,OACf,CAIA,OAAC,UAAW,CAEXM,EAAoB,EAAI,SAASN,EAAQ,CACxC,IAAIgH,EAAShH,GAAUA,EAAO,WAC7B,UAAW,CAAE,OAAOA,EAAO,OAAY,EACvC,UAAW,CAAE,OAAOA,CAAQ,EAC7B,OAAAM,EAAoB,EAAE0G,EAAQ,CAAE,EAAGA,CAAO,CAAC,EACpCA,CACR,CACD,EAAE,EAGD,UAAW,CAEX1G,EAAoB,EAAI,SAASP,EAASkH,EAAY,CACrD,QAAQC,KAAOD,EACX3G,EAAoB,EAAE2G,EAAYC,CAAG,GAAK,CAAC5G,EAAoB,EAAEP,EAASmH,CAAG,GAC/E,OAAO,eAAenH,EAASmH,EAAK,CAAE,WAAY,GAAM,IAAKD,EAAWC,CAAG,CAAE,CAAC,CAGjF,CACD,EAAE,EAGD,UAAW,CACX5G,EAAoB,EAAI,SAASyB,EAAKoF,EAAM,CAAE,OAAO,OAAO,UAAU,eAAe,KAAKpF,EAAKoF,CAAI,CAAG,CACvG,EAAE,EAMK7G,EAAoB,GAAG,CAC/B,EAAG,EACX,OACD,CAAC,ICz3BD,IAAA8G,GAAAC,GAAA,CAAAC,GAAAC,KAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,GAeA,IAAIC,GAAkB,UAOtBD,GAAO,QAAUE,GAUjB,SAASA,GAAWC,EAAQ,CAC1B,IAAIC,EAAM,GAAKD,EACXE,EAAQJ,GAAgB,KAAKG,CAAG,EAEpC,GAAI,CAACC,EACH,OAAOD,EAGT,IAAIE,EACAC,EAAO,GACPC,EAAQ,EACRC,EAAY,EAEhB,IAAKD,EAAQH,EAAM,MAAOG,EAAQJ,EAAI,OAAQI,IAAS,CACrD,OAAQJ,EAAI,WAAWI,CAAK,EAAG,CAC7B,IAAK,IACHF,EAAS,SACT,MACF,IAAK,IACHA,EAAS,QACT,MACF,IAAK,IACHA,EAAS,QACT,MACF,IAAK,IACHA,EAAS,OACT,MACF,IAAK,IACHA,EAAS,OACT,MACF,QACE,QACJ,CAEIG,IAAcD,IAChBD,GAAQH,EAAI,UAAUK,EAAWD,CAAK,GAGxCC,EAAYD,EAAQ,EACpBD,GAAQD,CACV,CAEA,OAAOG,IAAcD,EACjBD,EAAOH,EAAI,UAAUK,EAAWD,CAAK,EACrCD,CACN,ICvDA,IAAAG,GAAO,SCtBP;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,gFAgBA,IAAIC,GAAgB,SAASC,EAAGC,EAAG,CAC/B,OAAAF,GAAgB,OAAO,gBAClB,CAAE,UAAW,CAAC,CAAE,YAAa,OAAS,SAAUC,EAAGC,EAAG,CAAED,EAAE,UAAYC,CAAG,GAC1E,SAAUD,EAAGC,EAAG,CAAE,QAASC,KAAKD,EAAO,OAAO,UAAU,eAAe,KAAKA,EAAGC,CAAC,IAAGF,EAAEE,CAAC,EAAID,EAAEC,CAAC,EAAG,EAC7FH,GAAcC,EAAGC,CAAC,CAC7B,EAEO,SAASE,GAAUH,EAAGC,EAAG,CAC5B,GAAI,OAAOA,GAAM,YAAcA,IAAM,KACjC,MAAM,IAAI,UAAU,uBAAyB,OAAOA,CAAC,EAAI,+BAA+B,EAC5FF,GAAcC,EAAGC,CAAC,EAClB,SAASG,GAAK,CAAE,KAAK,YAAcJ,CAAG,CACtCA,EAAE,UAAYC,IAAM,KAAO,OAAO,OAAOA,CAAC,GAAKG,EAAG,UAAYH,EAAE,UAAW,IAAIG,EACnF,CAwCO,SAASC,GAAUC,EAASC,EAAYC,EAAGC,EAAW,CACzD,SAASC,EAAMC,EAAO,CAAE,OAAOA,aAAiBH,EAAIG,EAAQ,IAAIH,EAAE,SAAUI,EAAS,CAAEA,EAAQD,CAAK,CAAG,CAAC,CAAG,CAC3G,OAAO,IAAKH,IAAMA,EAAI,UAAU,SAAUI,EAASC,EAAQ,CACvD,SAASC,EAAUH,EAAO,CAAE,GAAI,CAAEI,EAAKN,EAAU,KAAKE,CAAK,CAAC,CAAG,OAASK,EAAG,CAAEH,EAAOG,CAAC,CAAG,CAAE,CAC1F,SAASC,EAASN,EAAO,CAAE,GAAI,CAAEI,EAAKN,EAAU,MAASE,CAAK,CAAC,CAAG,OAASK,EAAG,CAAEH,EAAOG,CAAC,CAAG,CAAE,CAC7F,SAASD,EAAKG,EAAQ,CAAEA,EAAO,KAAON,EAAQM,EAAO,KAAK,EAAIR,EAAMQ,EAAO,KAAK,EAAE,KAAKJ,EAAWG,CAAQ,CAAG,CAC7GF,GAAMN,EAAYA,EAAU,MAAMH,EAASC,GAAc,CAAC,CAAC,GAAG,KAAK,CAAC,CACxE,CAAC,CACL,CAEO,SAASY,GAAYb,EAASc,EAAM,CACvC,IAAIC,EAAI,CAAE,MAAO,EAAG,KAAM,UAAW,CAAE,GAAIC,EAAE,CAAC,EAAI,EAAG,MAAMA,EAAE,CAAC,EAAG,OAAOA,EAAE,CAAC,CAAG,EAAG,KAAM,CAAC,EAAG,IAAK,CAAC,CAAE,EAAGC,EAAGC,EAAGF,EAAGG,EAC/G,OAAOA,EAAI,CAAE,KAAMC,EAAK,CAAC,EAAG,MAASA,EAAK,CAAC,EAAG,OAAUA,EAAK,CAAC,CAAE,EAAG,OAAO,QAAW,aAAeD,EAAE,OAAO,QAAQ,EAAI,UAAW,CAAE,OAAO,IAAM,GAAIA,EACvJ,SAASC,EAAKC,EAAG,CAAE,OAAO,SAAUC,EAAG,CAAE,OAAOb,EAAK,CAACY,EAAGC,CAAC,CAAC,CAAG,CAAG,CACjE,SAASb,EAAKc,EAAI,CACd,GAAIN,EAAG,MAAM,IAAI,UAAU,iCAAiC,EAC5D,KAAOF,GAAG,GAAI,CACV,GAAIE,EAAI,EAAGC,IAAMF,EAAIO,EAAG,CAAC,EAAI,EAAIL,EAAE,OAAYK,EAAG,CAAC,EAAIL,EAAE,SAAcF,EAAIE,EAAE,SAAcF,EAAE,KAAKE,CAAC,EAAG,GAAKA,EAAE,OAAS,EAAEF,EAAIA,EAAE,KAAKE,EAAGK,EAAG,CAAC,CAAC,GAAG,KAAM,OAAOP,EAE3J,OADIE,EAAI,EAAGF,IAAGO,EAAK,CAACA,EAAG,CAAC,EAAI,EAAGP,EAAE,KAAK,GAC9BO,EAAG,CAAC,EAAG,CACX,IAAK,GAAG,IAAK,GAAGP,EAAIO,EAAI,MACxB,IAAK,GAAG,OAAAR,EAAE,QAAgB,CAAE,MAAOQ,EAAG,CAAC,EAAG,KAAM,EAAM,EACtD,IAAK,GAAGR,EAAE,QAASG,EAAIK,EAAG,CAAC,EAAGA,EAAK,CAAC,CAAC,EAAG,SACxC,IAAK,GAAGA,EAAKR,EAAE,IAAI,IAAI,EAAGA,EAAE,KAAK,IAAI,EAAG,SACxC,QACI,GAAMC,EAAID,EAAE,KAAM,EAAAC,EAAIA,EAAE,OAAS,GAAKA,EAAEA,EAAE,OAAS,CAAC,KAAOO,EAAG,CAAC,IAAM,GAAKA,EAAG,CAAC,IAAM,GAAI,CAAER,EAAI,EAAG,QAAU,CAC3G,GAAIQ,EAAG,CAAC,IAAM,IAAM,CAACP,GAAMO,EAAG,CAAC,EAAIP,EAAE,CAAC,GAAKO,EAAG,CAAC,EAAIP,EAAE,CAAC,GAAK,CAAED,EAAE,MAAQQ,EAAG,CAAC,EAAG,KAAO,CACrF,GAAIA,EAAG,CAAC,IAAM,GAAKR,EAAE,MAAQC,EAAE,CAAC,EAAG,CAAED,EAAE,MAAQC,EAAE,CAAC,EAAGA,EAAIO,EAAI,KAAO,CACpE,GAAIP,GAAKD,EAAE,MAAQC,EAAE,CAAC,EAAG,CAAED,EAAE,MAAQC,EAAE,CAAC,EAAGD,EAAE,IAAI,KAAKQ,CAAE,EAAG,KAAO,CAC9DP,EAAE,CAAC,GAAGD,EAAE,IAAI,IAAI,EACpBA,EAAE,KAAK,IAAI,EAAG,QACtB,CACAQ,EAAKT,EAAK,KAAKd,EAASe,CAAC,CAC7B,OAASL,EAAG,CAAEa,EAAK,CAAC,EAAGb,CAAC,EAAGQ,EAAI,CAAG,QAAE,CAAUD,EAAID,EAAI,CAAG,CACzD,GAAIO,EAAG,CAAC,EAAI,EAAG,MAAMA,EAAG,CAAC,EAAG,MAAO,CAAE,MAAOA,EAAG,CAAC,EAAIA,EAAG,CAAC,EAAI,OAAQ,KAAM,EAAK,CACnF,CACJ,CAcO,SAASC,GAASC,EAAG,CACxB,IAAIC,EAAI,OAAO,QAAW,YAAc,OAAO,SAAUC,EAAID,GAAKD,EAAEC,CAAC,EAAGE,EAAI,EAC5E,GAAID,EAAG,OAAOA,EAAE,KAAKF,CAAC,EACtB,GAAIA,GAAK,OAAOA,EAAE,QAAW,SAAU,MAAO,CAC1C,KAAM,UAAY,CACd,OAAIA,GAAKG,GAAKH,EAAE,SAAQA,EAAI,QACrB,CAAE,MAAOA,GAAKA,EAAEG,GAAG,EAAG,KAAM,CAACH,CAAE,CAC1C,CACJ,EACA,MAAM,IAAI,UAAUC,EAAI,0BAA4B,iCAAiC,CACzF,CAEO,SAASG,EAAOJ,EAAGK,EAAG,CACzB,IAAIH,EAAI,OAAO,QAAW,YAAcF,EAAE,OAAO,QAAQ,EACzD,GAAI,CAACE,EAAG,OAAOF,EACf,IAAIG,EAAID,EAAE,KAAKF,CAAC,EAAGM,EAAGC,EAAK,CAAC,EAAGC,EAC/B,GAAI,CACA,MAAQH,IAAM,QAAUA,KAAM,IAAM,EAAEC,EAAIH,EAAE,KAAK,GAAG,MAAMI,EAAG,KAAKD,EAAE,KAAK,CAC7E,OACOG,EAAO,CAAED,EAAI,CAAE,MAAOC,CAAM,CAAG,QACtC,CACI,GAAI,CACIH,GAAK,CAACA,EAAE,OAASJ,EAAIC,EAAE,SAAYD,EAAE,KAAKC,CAAC,CACnD,QACA,CAAU,GAAIK,EAAG,MAAMA,EAAE,KAAO,CACpC,CACA,OAAOD,CACX,CAkBO,SAASG,EAAcC,EAAIC,EAAMC,EAAM,CAC1C,GAAIA,GAAQ,UAAU,SAAW,EAAG,QAASC,EAAI,EAAGC,EAAIH,EAAK,OAAQI,EAAIF,EAAIC,EAAGD,KACxEE,GAAM,EAAEF,KAAKF,MACRI,IAAIA,EAAK,MAAM,UAAU,MAAM,KAAKJ,EAAM,EAAGE,CAAC,GACnDE,EAAGF,CAAC,EAAIF,EAAKE,CAAC,GAGtB,OAAOH,EAAG,OAAOK,GAAM,MAAM,UAAU,MAAM,KAAKJ,CAAI,CAAC,CAC3D,CAEO,SAASK,GAAQC,EAAG,CACvB,OAAO,gBAAgBD,IAAW,KAAK,EAAIC,EAAG,MAAQ,IAAID,GAAQC,CAAC,CACvE,CAEO,SAASC,GAAiBC,EAASC,EAAYC,EAAW,CAC7D,GAAI,CAAC,OAAO,cAAe,MAAM,IAAI,UAAU,sCAAsC,EACrF,IAAIC,EAAID,EAAU,MAAMF,EAASC,GAAc,CAAC,CAAC,EAAGP,EAAGU,EAAI,CAAC,EAC5D,OAAOV,EAAI,CAAC,EAAGW,EAAK,MAAM,EAAGA,EAAK,OAAO,EAAGA,EAAK,QAAQ,EAAGX,EAAE,OAAO,aAAa,EAAI,UAAY,CAAE,OAAO,IAAM,EAAGA,EACpH,SAASW,EAAKC,EAAG,CAAMH,EAAEG,CAAC,IAAGZ,EAAEY,CAAC,EAAI,SAAUR,EAAG,CAAE,OAAO,IAAI,QAAQ,SAAUS,EAAGC,EAAG,CAAEJ,EAAE,KAAK,CAACE,EAAGR,EAAGS,EAAGC,CAAC,CAAC,EAAI,GAAKC,EAAOH,EAAGR,CAAC,CAAG,CAAC,CAAG,EAAG,CACzI,SAASW,EAAOH,EAAGR,EAAG,CAAE,GAAI,CAAEY,EAAKP,EAAEG,CAAC,EAAER,CAAC,CAAC,CAAG,OAASa,EAAG,CAAEC,EAAOR,EAAE,CAAC,EAAE,CAAC,EAAGO,CAAC,CAAG,CAAE,CACjF,SAASD,EAAKG,EAAG,CAAEA,EAAE,iBAAiBhB,GAAU,QAAQ,QAAQgB,EAAE,MAAM,CAAC,EAAE,KAAKC,EAASC,CAAM,EAAIH,EAAOR,EAAE,CAAC,EAAE,CAAC,EAAGS,CAAC,CAAG,CACvH,SAASC,EAAQE,EAAO,CAAEP,EAAO,OAAQO,CAAK,CAAG,CACjD,SAASD,EAAOC,EAAO,CAAEP,EAAO,QAASO,CAAK,CAAG,CACjD,SAASJ,EAAOK,EAAGnB,EAAG,CAAMmB,EAAEnB,CAAC,EAAGM,EAAE,MAAM,EAAGA,EAAE,QAAQK,EAAOL,EAAE,CAAC,EAAE,CAAC,EAAGA,EAAE,CAAC,EAAE,CAAC,CAAC,CAAG,CACrF,CAQO,SAASc,GAAcC,EAAG,CAC7B,GAAI,CAAC,OAAO,cAAe,MAAM,IAAI,UAAU,sCAAsC,EACrF,IAAIC,EAAID,EAAE,OAAO,aAAa,EAAGE,EACjC,OAAOD,EAAIA,EAAE,KAAKD,CAAC,GAAKA,EAAI,OAAOG,IAAa,WAAaA,GAASH,CAAC,EAAIA,EAAE,OAAO,QAAQ,EAAE,EAAGE,EAAI,CAAC,EAAGE,EAAK,MAAM,EAAGA,EAAK,OAAO,EAAGA,EAAK,QAAQ,EAAGF,EAAE,OAAO,aAAa,EAAI,UAAY,CAAE,OAAO,IAAM,EAAGA,GAC9M,SAASE,EAAKC,EAAG,CAAEH,EAAEG,CAAC,EAAIL,EAAEK,CAAC,GAAK,SAAUC,EAAG,CAAE,OAAO,IAAI,QAAQ,SAAUC,EAASC,EAAQ,CAAEF,EAAIN,EAAEK,CAAC,EAAEC,CAAC,EAAGG,EAAOF,EAASC,EAAQF,EAAE,KAAMA,EAAE,KAAK,CAAG,CAAC,CAAG,CAAG,CAC/J,SAASG,EAAOF,EAASC,EAAQE,EAAGJ,EAAG,CAAE,QAAQ,QAAQA,CAAC,EAAE,KAAK,SAASA,EAAG,CAAEC,EAAQ,CAAE,MAAOD,EAAG,KAAMI,CAAE,CAAC,CAAG,EAAGF,CAAM,CAAG,CAC/H,CCtMM,SAAUG,EAAWC,EAAU,CACnC,OAAO,OAAOA,GAAU,UAC1B,CCGM,SAAUC,GAAoBC,EAAgC,CAClE,IAAMC,EAAS,SAACC,EAAa,CAC3B,MAAM,KAAKA,CAAQ,EACnBA,EAAS,MAAQ,IAAI,MAAK,EAAG,KAC/B,EAEMC,EAAWH,EAAWC,CAAM,EAClC,OAAAE,EAAS,UAAY,OAAO,OAAO,MAAM,SAAS,EAClDA,EAAS,UAAU,YAAcA,EAC1BA,CACT,CCDO,IAAMC,GAA+CC,GAC1D,SAACC,EAAM,CACL,OAAA,SAA4CC,EAA0B,CACpED,EAAO,IAAI,EACX,KAAK,QAAUC,EACRA,EAAO,OAAM;EACxBA,EAAO,IAAI,SAACC,EAAKC,EAAC,CAAK,OAAGA,EAAI,EAAC,KAAKD,EAAI,SAAQ,CAAzB,CAA6B,EAAE,KAAK;GAAM,EACzD,GACJ,KAAK,KAAO,sBACZ,KAAK,OAASD,CAChB,CARA,CAQC,ECvBC,SAAUG,GAAaC,EAA6BC,EAAO,CAC/D,GAAID,EAAK,CACP,IAAME,EAAQF,EAAI,QAAQC,CAAI,EAC9B,GAAKC,GAASF,EAAI,OAAOE,EAAO,CAAC,EAErC,CCOA,IAAAC,GAAA,UAAA,CAyBE,SAAAA,EAAoBC,EAA4B,CAA5B,KAAA,gBAAAA,EAdb,KAAA,OAAS,GAER,KAAA,WAAmD,KAMnD,KAAA,YAAqD,IAMV,CAQnD,OAAAD,EAAA,UAAA,YAAA,UAAA,aACME,EAEJ,GAAI,CAAC,KAAK,OAAQ,CAChB,KAAK,OAAS,GAGN,IAAAC,EAAe,KAAI,WAC3B,GAAIA,EAEF,GADA,KAAK,WAAa,KACd,MAAM,QAAQA,CAAU,MAC1B,QAAqBC,EAAAC,GAAAF,CAAU,EAAAG,EAAAF,EAAA,KAAA,EAAA,CAAAE,EAAA,KAAAA,EAAAF,EAAA,KAAA,EAAE,CAA5B,IAAMG,EAAMD,EAAA,MACfC,EAAO,OAAO,IAAI,yGAGpBJ,EAAW,OAAO,IAAI,EAIlB,IAAiBK,EAAqB,KAAI,gBAClD,GAAIC,EAAWD,CAAgB,EAC7B,GAAI,CACFA,EAAgB,QACTE,EAAG,CACVR,EAASQ,aAAaC,GAAsBD,EAAE,OAAS,CAACA,CAAC,EAIrD,IAAAE,EAAgB,KAAI,YAC5B,GAAIA,EAAa,CACf,KAAK,YAAc,SACnB,QAAwBC,EAAAR,GAAAO,CAAW,EAAAE,EAAAD,EAAA,KAAA,EAAA,CAAAC,EAAA,KAAAA,EAAAD,EAAA,KAAA,EAAE,CAAhC,IAAME,EAASD,EAAA,MAClB,GAAI,CACFE,GAAcD,CAAS,QAChBE,EAAK,CACZf,EAASA,GAAM,KAANA,EAAU,CAAA,EACfe,aAAeN,GACjBT,EAAMgB,EAAAA,EAAA,CAAA,EAAAC,EAAOjB,CAAM,CAAA,EAAAiB,EAAKF,EAAI,MAAM,CAAA,EAElCf,EAAO,KAAKe,CAAG,sGAMvB,GAAIf,EACF,MAAM,IAAIS,GAAoBT,CAAM,EAG1C,EAoBAF,EAAA,UAAA,IAAA,SAAIoB,EAAuB,OAGzB,GAAIA,GAAYA,IAAa,KAC3B,GAAI,KAAK,OAGPJ,GAAcI,CAAQ,MACjB,CACL,GAAIA,aAAoBpB,EAAc,CAGpC,GAAIoB,EAAS,QAAUA,EAAS,WAAW,IAAI,EAC7C,OAEFA,EAAS,WAAW,IAAI,GAEzB,KAAK,aAAcC,EAAA,KAAK,eAAW,MAAAA,IAAA,OAAAA,EAAI,CAAA,GAAI,KAAKD,CAAQ,EAG/D,EAOQpB,EAAA,UAAA,WAAR,SAAmBsB,EAAoB,CAC7B,IAAAnB,EAAe,KAAI,WAC3B,OAAOA,IAAemB,GAAW,MAAM,QAAQnB,CAAU,GAAKA,EAAW,SAASmB,CAAM,CAC1F,EASQtB,EAAA,UAAA,WAAR,SAAmBsB,EAAoB,CAC7B,IAAAnB,EAAe,KAAI,WAC3B,KAAK,WAAa,MAAM,QAAQA,CAAU,GAAKA,EAAW,KAAKmB,CAAM,EAAGnB,GAAcA,EAAa,CAACA,EAAYmB,CAAM,EAAIA,CAC5H,EAMQtB,EAAA,UAAA,cAAR,SAAsBsB,EAAoB,CAChC,IAAAnB,EAAe,KAAI,WACvBA,IAAemB,EACjB,KAAK,WAAa,KACT,MAAM,QAAQnB,CAAU,GACjCoB,GAAUpB,EAAYmB,CAAM,CAEhC,EAgBAtB,EAAA,UAAA,OAAA,SAAOoB,EAAsC,CACnC,IAAAR,EAAgB,KAAI,YAC5BA,GAAeW,GAAUX,EAAaQ,CAAQ,EAE1CA,aAAoBpB,GACtBoB,EAAS,cAAc,IAAI,CAE/B,EAlLcpB,EAAA,MAAS,UAAA,CACrB,IAAMwB,EAAQ,IAAIxB,EAClB,OAAAwB,EAAM,OAAS,GACRA,CACT,EAAE,EA+KJxB,GArLA,EAuLO,IAAMyB,GAAqBC,GAAa,MAEzC,SAAUC,GAAeC,EAAU,CACvC,OACEA,aAAiBF,IAChBE,GAAS,WAAYA,GAASC,EAAWD,EAAM,MAAM,GAAKC,EAAWD,EAAM,GAAG,GAAKC,EAAWD,EAAM,WAAW,CAEpH,CAEA,SAASE,GAAcC,EAAwC,CACzDF,EAAWE,CAAS,EACtBA,EAAS,EAETA,EAAU,YAAW,CAEzB,CChNO,IAAMC,GAAuB,CAClC,iBAAkB,KAClB,sBAAuB,KACvB,QAAS,OACT,sCAAuC,GACvC,yBAA0B,ICGrB,IAAMC,GAAmC,CAG9C,WAAA,SAAWC,EAAqBC,EAAgB,SAAEC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,EAAA,CAAA,EAAA,UAAAA,CAAA,EACxC,IAAAC,EAAaL,GAAe,SACpC,OAAIK,GAAQ,MAARA,EAAU,WACLA,EAAS,WAAU,MAAnBA,EAAQC,EAAA,CAAYL,EAASC,CAAO,EAAAK,EAAKJ,CAAI,CAAA,CAAA,EAE/C,WAAU,MAAA,OAAAG,EAAA,CAACL,EAASC,CAAO,EAAAK,EAAKJ,CAAI,CAAA,CAAA,CAC7C,EACA,aAAA,SAAaK,EAAM,CACT,IAAAH,EAAaL,GAAe,SACpC,QAAQK,GAAQ,KAAA,OAARA,EAAU,eAAgB,cAAcG,CAAa,CAC/D,EACA,SAAU,QCjBN,SAAUC,GAAqBC,EAAQ,CAC3CC,GAAgB,WAAW,UAAA,CACjB,IAAAC,EAAqBC,GAAM,iBACnC,GAAID,EAEFA,EAAiBF,CAAG,MAGpB,OAAMA,CAEV,CAAC,CACH,CCtBM,SAAUI,IAAI,CAAK,CCMlB,IAAMC,GAAyB,UAAA,CAAM,OAAAC,GAAmB,IAAK,OAAW,MAAS,CAA5C,EAAsE,EAO5G,SAAUC,GAAkBC,EAAU,CAC1C,OAAOF,GAAmB,IAAK,OAAWE,CAAK,CACjD,CAOM,SAAUC,GAAoBC,EAAQ,CAC1C,OAAOJ,GAAmB,IAAKI,EAAO,MAAS,CACjD,CAQM,SAAUJ,GAAmBK,EAAuBD,EAAYF,EAAU,CAC9E,MAAO,CACL,KAAIG,EACJ,MAAKD,EACL,MAAKF,EAET,CCrCA,IAAII,GAAuD,KASrD,SAAUC,GAAaC,EAAc,CACzC,GAAIC,GAAO,sCAAuC,CAChD,IAAMC,EAAS,CAACJ,GAKhB,GAJII,IACFJ,GAAU,CAAE,YAAa,GAAO,MAAO,IAAI,GAE7CE,EAAE,EACEE,EAAQ,CACJ,IAAAC,EAAyBL,GAAvBM,EAAWD,EAAA,YAAEE,EAAKF,EAAA,MAE1B,GADAL,GAAU,KACNM,EACF,MAAMC,QAMVL,EAAE,CAEN,CAMM,SAAUM,GAAaC,EAAQ,CAC/BN,GAAO,uCAAyCH,KAClDA,GAAQ,YAAc,GACtBA,GAAQ,MAAQS,EAEpB,CCrBA,IAAAC,GAAA,SAAAC,EAAA,CAAmCC,GAAAF,EAAAC,CAAA,EA6BjC,SAAAD,EAAYG,EAA6C,CAAzD,IAAAC,EACEH,EAAA,KAAA,IAAA,GAAO,KATC,OAAAG,EAAA,UAAqB,GAUzBD,GACFC,EAAK,YAAcD,EAGfE,GAAeF,CAAW,GAC5BA,EAAY,IAAIC,CAAI,GAGtBA,EAAK,YAAcE,IAEvB,CAzBO,OAAAN,EAAA,OAAP,SAAiBO,EAAwBC,EAA2BC,EAAqB,CACvF,OAAO,IAAIC,GAAeH,EAAMC,EAAOC,CAAQ,CACjD,EAgCAT,EAAA,UAAA,KAAA,SAAKW,EAAS,CACR,KAAK,UACPC,GAA0BC,GAAiBF,CAAK,EAAG,IAAI,EAEvD,KAAK,MAAMA,CAAM,CAErB,EASAX,EAAA,UAAA,MAAA,SAAMc,EAAS,CACT,KAAK,UACPF,GAA0BG,GAAkBD,CAAG,EAAG,IAAI,GAEtD,KAAK,UAAY,GACjB,KAAK,OAAOA,CAAG,EAEnB,EAQAd,EAAA,UAAA,SAAA,UAAA,CACM,KAAK,UACPY,GAA0BI,GAAuB,IAAI,GAErD,KAAK,UAAY,GACjB,KAAK,UAAS,EAElB,EAEAhB,EAAA,UAAA,YAAA,UAAA,CACO,KAAK,SACR,KAAK,UAAY,GACjBC,EAAA,UAAM,YAAW,KAAA,IAAA,EACjB,KAAK,YAAc,KAEvB,EAEUD,EAAA,UAAA,MAAV,SAAgBW,EAAQ,CACtB,KAAK,YAAY,KAAKA,CAAK,CAC7B,EAEUX,EAAA,UAAA,OAAV,SAAiBc,EAAQ,CACvB,GAAI,CACF,KAAK,YAAY,MAAMA,CAAG,UAE1B,KAAK,YAAW,EAEpB,EAEUd,EAAA,UAAA,UAAV,UAAA,CACE,GAAI,CACF,KAAK,YAAY,SAAQ,UAEzB,KAAK,YAAW,EAEpB,EACFA,CAAA,EApHmCiB,EAAY,EA2H/C,IAAMC,GAAQ,SAAS,UAAU,KAEjC,SAASC,GAAyCC,EAAQC,EAAY,CACpE,OAAOH,GAAM,KAAKE,EAAIC,CAAO,CAC/B,CAMA,IAAAC,GAAA,UAAA,CACE,SAAAA,EAAoBC,EAAqC,CAArC,KAAA,gBAAAA,CAAwC,CAE5D,OAAAD,EAAA,UAAA,KAAA,SAAKE,EAAQ,CACH,IAAAD,EAAoB,KAAI,gBAChC,GAAIA,EAAgB,KAClB,GAAI,CACFA,EAAgB,KAAKC,CAAK,QACnBC,EAAO,CACdC,GAAqBD,CAAK,EAGhC,EAEAH,EAAA,UAAA,MAAA,SAAMK,EAAQ,CACJ,IAAAJ,EAAoB,KAAI,gBAChC,GAAIA,EAAgB,MAClB,GAAI,CACFA,EAAgB,MAAMI,CAAG,QAClBF,EAAO,CACdC,GAAqBD,CAAK,OAG5BC,GAAqBC,CAAG,CAE5B,EAEAL,EAAA,UAAA,SAAA,UAAA,CACU,IAAAC,EAAoB,KAAI,gBAChC,GAAIA,EAAgB,SAClB,GAAI,CACFA,EAAgB,SAAQ,QACjBE,EAAO,CACdC,GAAqBD,CAAK,EAGhC,EACFH,CAAA,EArCA,EAuCAM,GAAA,SAAAC,EAAA,CAAuCC,GAAAF,EAAAC,CAAA,EACrC,SAAAD,EACEG,EACAN,EACAO,EAA8B,CAHhC,IAAAC,EAKEJ,EAAA,KAAA,IAAA,GAAO,KAEHN,EACJ,GAAIW,EAAWH,CAAc,GAAK,CAACA,EAGjCR,EAAkB,CAChB,KAAOQ,GAAc,KAAdA,EAAkB,OACzB,MAAON,GAAK,KAALA,EAAS,OAChB,SAAUO,GAAQ,KAARA,EAAY,YAEnB,CAEL,IAAIG,EACAF,GAAQG,GAAO,0BAIjBD,EAAU,OAAO,OAAOJ,CAAc,EACtCI,EAAQ,YAAc,UAAA,CAAM,OAAAF,EAAK,YAAW,CAAhB,EAC5BV,EAAkB,CAChB,KAAMQ,EAAe,MAAQZ,GAAKY,EAAe,KAAMI,CAAO,EAC9D,MAAOJ,EAAe,OAASZ,GAAKY,EAAe,MAAOI,CAAO,EACjE,SAAUJ,EAAe,UAAYZ,GAAKY,EAAe,SAAUI,CAAO,IAI5EZ,EAAkBQ,EAMtB,OAAAE,EAAK,YAAc,IAAIX,GAAiBC,CAAe,GACzD,CACF,OAAAK,CAAA,EAzCuCS,EAAU,EA2CjD,SAASC,GAAqBC,EAAU,CAClCC,GAAO,sCACTC,GAAaF,CAAK,EAIlBG,GAAqBH,CAAK,CAE9B,CAQA,SAASI,GAAoBC,EAAQ,CACnC,MAAMA,CACR,CAOA,SAASC,GAA0BC,EAA2CC,EAA2B,CAC/F,IAAAC,EAA0BR,GAAM,sBACxCQ,GAAyBC,GAAgB,WAAW,UAAA,CAAM,OAAAD,EAAsBF,EAAcC,CAAU,CAA9C,CAA+C,CAC3G,CAOO,IAAMG,GAA6D,CACxE,OAAQ,GACR,KAAMC,GACN,MAAOR,GACP,SAAUQ,IC5QL,IAAMC,GAA+B,UAAA,CAAM,OAAC,OAAO,QAAW,YAAc,OAAO,YAAe,cAAvD,EAAsE,ECoClH,SAAUC,GAAYC,EAAI,CAC9B,OAAOA,CACT,CCiCM,SAAUC,IAAI,SAACC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACnB,OAAOC,GAAcF,CAAG,CAC1B,CAGM,SAAUE,GAAoBF,EAA+B,CACjE,OAAIA,EAAI,SAAW,EACVG,GAGLH,EAAI,SAAW,EACVA,EAAI,CAAC,EAGP,SAAeI,EAAQ,CAC5B,OAAOJ,EAAI,OAAO,SAACK,EAAWC,EAAuB,CAAK,OAAAA,EAAGD,CAAI,CAAP,EAAUD,CAAY,CAClF,CACF,CC9EA,IAAAG,EAAA,UAAA,CAkBE,SAAAA,EAAYC,EAA6E,CACnFA,IACF,KAAK,WAAaA,EAEtB,CA4BA,OAAAD,EAAA,UAAA,KAAA,SAAQE,EAAyB,CAC/B,IAAMC,EAAa,IAAIH,EACvB,OAAAG,EAAW,OAAS,KACpBA,EAAW,SAAWD,EACfC,CACT,EA6IAH,EAAA,UAAA,UAAA,SACEI,EACAC,EACAC,EAA8B,CAHhC,IAAAC,EAAA,KAKQC,EAAaC,GAAaL,CAAc,EAAIA,EAAiB,IAAIM,GAAeN,EAAgBC,EAAOC,CAAQ,EAErH,OAAAK,GAAa,UAAA,CACL,IAAAC,EAAuBL,EAArBL,EAAQU,EAAA,SAAEC,EAAMD,EAAA,OACxBJ,EAAW,IACTN,EAGIA,EAAS,KAAKM,EAAYK,CAAM,EAChCA,EAIAN,EAAK,WAAWC,CAAU,EAG1BD,EAAK,cAAcC,CAAU,CAAC,CAEtC,CAAC,EAEMA,CACT,EAGUR,EAAA,UAAA,cAAV,SAAwBc,EAAmB,CACzC,GAAI,CACF,OAAO,KAAK,WAAWA,CAAI,QACpBC,EAAK,CAIZD,EAAK,MAAMC,CAAG,EAElB,EA6DAf,EAAA,UAAA,QAAA,SAAQgB,EAA0BC,EAAoC,CAAtE,IAAAV,EAAA,KACE,OAAAU,EAAcC,GAAeD,CAAW,EAEjC,IAAIA,EAAkB,SAACE,EAASC,EAAM,CAC3C,IAAMZ,EAAa,IAAIE,GAAkB,CACvC,KAAM,SAACW,EAAK,CACV,GAAI,CACFL,EAAKK,CAAK,QACHN,EAAK,CACZK,EAAOL,CAAG,EACVP,EAAW,YAAW,EAE1B,EACA,MAAOY,EACP,SAAUD,EACX,EACDZ,EAAK,UAAUC,CAAU,CAC3B,CAAC,CACH,EAGUR,EAAA,UAAA,WAAV,SAAqBQ,EAA2B,OAC9C,OAAOI,EAAA,KAAK,UAAM,MAAAA,IAAA,OAAA,OAAAA,EAAE,UAAUJ,CAAU,CAC1C,EAOAR,EAAA,UAACG,EAAiB,EAAlB,UAAA,CACE,OAAO,IACT,EA4FAH,EAAA,UAAA,KAAA,UAAA,SAAKsB,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACH,OAAOC,GAAcF,CAAU,EAAE,IAAI,CACvC,EA6BAtB,EAAA,UAAA,UAAA,SAAUiB,EAAoC,CAA9C,IAAAV,EAAA,KACE,OAAAU,EAAcC,GAAeD,CAAW,EAEjC,IAAIA,EAAY,SAACE,EAASC,EAAM,CACrC,IAAIC,EACJd,EAAK,UACH,SAACkB,EAAI,CAAK,OAACJ,EAAQI,CAAT,EACV,SAACV,EAAQ,CAAK,OAAAK,EAAOL,CAAG,CAAV,EACd,UAAA,CAAM,OAAAI,EAAQE,CAAK,CAAb,CAAc,CAExB,CAAC,CACH,EA1aOrB,EAAA,OAAkC,SAAIC,EAAwD,CACnG,OAAO,IAAID,EAAcC,CAAS,CACpC,EAyaFD,GA9cA,EAudA,SAAS0B,GAAeC,EAA+C,OACrE,OAAOC,EAAAD,GAAW,KAAXA,EAAeE,GAAO,WAAO,MAAAD,IAAA,OAAAA,EAAI,OAC1C,CAEA,SAASE,GAAcC,EAAU,CAC/B,OAAOA,GAASC,EAAWD,EAAM,IAAI,GAAKC,EAAWD,EAAM,KAAK,GAAKC,EAAWD,EAAM,QAAQ,CAChG,CAEA,SAASE,GAAgBF,EAAU,CACjC,OAAQA,GAASA,aAAiBG,IAAgBJ,GAAWC,CAAK,GAAKI,GAAeJ,CAAK,CAC7F,CCzeM,SAAUK,GAAQC,EAAW,CACjC,OAAOC,EAAWD,GAAM,KAAA,OAANA,EAAQ,IAAI,CAChC,CAMM,SAAUE,EACdC,EAAqF,CAErF,OAAO,SAACH,EAAqB,CAC3B,GAAID,GAAQC,CAAM,EAChB,OAAOA,EAAO,KAAK,SAA+BI,EAA2B,CAC3E,GAAI,CACF,OAAOD,EAAKC,EAAc,IAAI,QACvBC,EAAK,CACZ,KAAK,MAAMA,CAAG,EAElB,CAAC,EAEH,MAAM,IAAI,UAAU,wCAAwC,CAC9D,CACF,CCjBM,SAAUC,EACdC,EACAC,EACAC,EACAC,EACAC,EAAuB,CAEvB,OAAO,IAAIC,GAAmBL,EAAaC,EAAQC,EAAYC,EAASC,CAAU,CACpF,CAMA,IAAAC,GAAA,SAAAC,EAAA,CAA2CC,GAAAF,EAAAC,CAAA,EAiBzC,SAAAD,EACEL,EACAC,EACAC,EACAC,EACQC,EACAI,EAAiC,CAN3C,IAAAC,EAoBEH,EAAA,KAAA,KAAMN,CAAW,GAAC,KAfV,OAAAS,EAAA,WAAAL,EACAK,EAAA,kBAAAD,EAeRC,EAAK,MAAQR,EACT,SAAuCS,EAAQ,CAC7C,GAAI,CACFT,EAAOS,CAAK,QACLC,EAAK,CACZX,EAAY,MAAMW,CAAG,EAEzB,EACAL,EAAA,UAAM,MACVG,EAAK,OAASN,EACV,SAAuCQ,EAAQ,CAC7C,GAAI,CACFR,EAAQQ,CAAG,QACJA,EAAK,CAEZX,EAAY,MAAMW,CAAG,UAGrB,KAAK,YAAW,EAEpB,EACAL,EAAA,UAAM,OACVG,EAAK,UAAYP,EACb,UAAA,CACE,GAAI,CACFA,EAAU,QACHS,EAAK,CAEZX,EAAY,MAAMW,CAAG,UAGrB,KAAK,YAAW,EAEpB,EACAL,EAAA,UAAM,WACZ,CAEA,OAAAD,EAAA,UAAA,YAAA,UAAA,OACE,GAAI,CAAC,KAAK,mBAAqB,KAAK,kBAAiB,EAAI,CAC/C,IAAAO,EAAW,KAAI,OACvBN,EAAA,UAAM,YAAW,KAAA,IAAA,EAEjB,CAACM,KAAUC,EAAA,KAAK,cAAU,MAAAA,IAAA,QAAAA,EAAA,KAAf,IAAI,GAEnB,EACFR,CAAA,EAnF2CS,EAAU,ECd9C,IAAMC,GAAiD,CAG5D,SAAA,SAASC,EAAQ,CACf,IAAIC,EAAU,sBACVC,EAAkD,qBAC9CC,EAAaJ,GAAsB,SACvCI,IACFF,EAAUE,EAAS,sBACnBD,EAASC,EAAS,sBAEpB,IAAMC,EAASH,EAAQ,SAACI,EAAS,CAI/BH,EAAS,OACTF,EAASK,CAAS,CACpB,CAAC,EACD,OAAO,IAAIC,GAAa,UAAA,CAAM,OAAAJ,GAAM,KAAA,OAANA,EAASE,CAAM,CAAf,CAAgB,CAChD,EACA,sBAAqB,UAAA,SAACG,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACZ,IAAAL,EAAaJ,GAAsB,SAC3C,QAAQI,GAAQ,KAAA,OAARA,EAAU,wBAAyB,uBAAsB,MAAA,OAAAM,EAAA,CAAA,EAAAC,EAAIH,CAAI,CAAA,CAAA,CAC3E,EACA,qBAAoB,UAAA,SAACA,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACX,IAAAL,EAAaJ,GAAsB,SAC3C,QAAQI,GAAQ,KAAA,OAARA,EAAU,uBAAwB,sBAAqB,MAAA,OAAAM,EAAA,CAAA,EAAAC,EAAIH,CAAI,CAAA,CAAA,CACzE,EACA,SAAU,QCrBL,IAAMI,GAAuDC,GAClE,SAACC,EAAM,CACL,OAAA,UAAoC,CAClCA,EAAO,IAAI,EACX,KAAK,KAAO,0BACZ,KAAK,QAAU,qBACjB,CAJA,CAIC,ECXL,IAAAC,EAAA,SAAAC,EAAA,CAAgCC,GAAAF,EAAAC,CAAA,EAwB9B,SAAAD,GAAA,CAAA,IAAAG,EAEEF,EAAA,KAAA,IAAA,GAAO,KAzBT,OAAAE,EAAA,OAAS,GAEDA,EAAA,iBAAyC,KAGjDA,EAAA,UAA2B,CAAA,EAE3BA,EAAA,UAAY,GAEZA,EAAA,SAAW,GAEXA,EAAA,YAAmB,MAenB,CAGA,OAAAH,EAAA,UAAA,KAAA,SAAQI,EAAwB,CAC9B,IAAMC,EAAU,IAAIC,GAAiB,KAAM,IAAI,EAC/C,OAAAD,EAAQ,SAAWD,EACZC,CACT,EAGUL,EAAA,UAAA,eAAV,UAAA,CACE,GAAI,KAAK,OACP,MAAM,IAAIO,EAEd,EAEAP,EAAA,UAAA,KAAA,SAAKQ,EAAQ,CAAb,IAAAL,EAAA,KACEM,GAAa,UAAA,SAEX,GADAN,EAAK,eAAc,EACf,CAACA,EAAK,UAAW,CACdA,EAAK,mBACRA,EAAK,iBAAmB,MAAM,KAAKA,EAAK,SAAS,OAEnD,QAAuBO,EAAAC,GAAAR,EAAK,gBAAgB,EAAAS,EAAAF,EAAA,KAAA,EAAA,CAAAE,EAAA,KAAAA,EAAAF,EAAA,KAAA,EAAE,CAAzC,IAAMG,EAAQD,EAAA,MACjBC,EAAS,KAAKL,CAAK,qGAGzB,CAAC,CACH,EAEAR,EAAA,UAAA,MAAA,SAAMc,EAAQ,CAAd,IAAAX,EAAA,KACEM,GAAa,UAAA,CAEX,GADAN,EAAK,eAAc,EACf,CAACA,EAAK,UAAW,CACnBA,EAAK,SAAWA,EAAK,UAAY,GACjCA,EAAK,YAAcW,EAEnB,QADQC,EAAcZ,EAAI,UACnBY,EAAU,QACfA,EAAU,MAAK,EAAI,MAAMD,CAAG,EAGlC,CAAC,CACH,EAEAd,EAAA,UAAA,SAAA,UAAA,CAAA,IAAAG,EAAA,KACEM,GAAa,UAAA,CAEX,GADAN,EAAK,eAAc,EACf,CAACA,EAAK,UAAW,CACnBA,EAAK,UAAY,GAEjB,QADQY,EAAcZ,EAAI,UACnBY,EAAU,QACfA,EAAU,MAAK,EAAI,SAAQ,EAGjC,CAAC,CACH,EAEAf,EAAA,UAAA,YAAA,UAAA,CACE,KAAK,UAAY,KAAK,OAAS,GAC/B,KAAK,UAAY,KAAK,iBAAmB,IAC3C,EAEA,OAAA,eAAIA,EAAA,UAAA,WAAQ,KAAZ,UAAA,OACE,QAAOgB,EAAA,KAAK,aAAS,MAAAA,IAAA,OAAA,OAAAA,EAAE,QAAS,CAClC,kCAGUhB,EAAA,UAAA,cAAV,SAAwBiB,EAAyB,CAC/C,YAAK,eAAc,EACZhB,EAAA,UAAM,cAAa,KAAA,KAACgB,CAAU,CACvC,EAGUjB,EAAA,UAAA,WAAV,SAAqBiB,EAAyB,CAC5C,YAAK,eAAc,EACnB,KAAK,wBAAwBA,CAAU,EAChC,KAAK,gBAAgBA,CAAU,CACxC,EAGUjB,EAAA,UAAA,gBAAV,SAA0BiB,EAA2B,CAArD,IAAAd,EAAA,KACQa,EAAqC,KAAnCE,EAAQF,EAAA,SAAEG,EAASH,EAAA,UAAED,EAASC,EAAA,UACtC,OAAIE,GAAYC,EACPC,IAET,KAAK,iBAAmB,KACxBL,EAAU,KAAKE,CAAU,EAClB,IAAII,GAAa,UAAA,CACtBlB,EAAK,iBAAmB,KACxBmB,GAAUP,EAAWE,CAAU,CACjC,CAAC,EACH,EAGUjB,EAAA,UAAA,wBAAV,SAAkCiB,EAA2B,CACrD,IAAAD,EAAuC,KAArCE,EAAQF,EAAA,SAAEO,EAAWP,EAAA,YAAEG,EAASH,EAAA,UACpCE,EACFD,EAAW,MAAMM,CAAW,EACnBJ,GACTF,EAAW,SAAQ,CAEvB,EAQAjB,EAAA,UAAA,aAAA,UAAA,CACE,IAAMwB,EAAkB,IAAIC,EAC5B,OAAAD,EAAW,OAAS,KACbA,CACT,EAxHOxB,EAAA,OAAkC,SAAI0B,EAA0BC,EAAqB,CAC1F,OAAO,IAAIrB,GAAoBoB,EAAaC,CAAM,CACpD,EAuHF3B,GA7IgCyB,CAAU,EAkJ1C,IAAAG,GAAA,SAAAC,EAAA,CAAyCC,GAAAF,EAAAC,CAAA,EACvC,SAAAD,EAESG,EACPC,EAAsB,CAHxB,IAAAC,EAKEJ,EAAA,KAAA,IAAA,GAAO,KAHA,OAAAI,EAAA,YAAAF,EAIPE,EAAK,OAASD,GAChB,CAEA,OAAAJ,EAAA,UAAA,KAAA,SAAKM,EAAQ,UACXC,GAAAC,EAAA,KAAK,eAAW,MAAAA,IAAA,OAAA,OAAAA,EAAE,QAAI,MAAAD,IAAA,QAAAA,EAAA,KAAAC,EAAGF,CAAK,CAChC,EAEAN,EAAA,UAAA,MAAA,SAAMS,EAAQ,UACZF,GAAAC,EAAA,KAAK,eAAW,MAAAA,IAAA,OAAA,OAAAA,EAAE,SAAK,MAAAD,IAAA,QAAAA,EAAA,KAAAC,EAAGC,CAAG,CAC/B,EAEAT,EAAA,UAAA,SAAA,UAAA,UACEO,GAAAC,EAAA,KAAK,eAAW,MAAAA,IAAA,OAAA,OAAAA,EAAE,YAAQ,MAAAD,IAAA,QAAAA,EAAA,KAAAC,CAAA,CAC5B,EAGUR,EAAA,UAAA,WAAV,SAAqBU,EAAyB,SAC5C,OAAOH,GAAAC,EAAA,KAAK,UAAM,MAAAA,IAAA,OAAA,OAAAA,EAAE,UAAUE,CAAU,KAAC,MAAAH,IAAA,OAAAA,EAAII,EAC/C,EACFX,CAAA,EA1ByCY,CAAO,EC5JzC,IAAMC,GAA+C,CAC1D,IAAG,UAAA,CAGD,OAAQA,GAAsB,UAAY,MAAM,IAAG,CACrD,EACA,SAAU,QCwBZ,IAAAC,GAAA,SAAAC,EAAA,CAAsCC,GAAAF,EAAAC,CAAA,EAUpC,SAAAD,EACUG,EACAC,EACAC,EAA6D,CAF7DF,IAAA,SAAAA,EAAA,KACAC,IAAA,SAAAA,EAAA,KACAC,IAAA,SAAAA,EAAAC,IAHV,IAAAC,EAKEN,EAAA,KAAA,IAAA,GAAO,KAJC,OAAAM,EAAA,YAAAJ,EACAI,EAAA,YAAAH,EACAG,EAAA,mBAAAF,EAZFE,EAAA,QAA0B,CAAA,EAC1BA,EAAA,oBAAsB,GAc5BA,EAAK,oBAAsBH,IAAgB,IAC3CG,EAAK,YAAc,KAAK,IAAI,EAAGJ,CAAW,EAC1CI,EAAK,YAAc,KAAK,IAAI,EAAGH,CAAW,GAC5C,CAEA,OAAAJ,EAAA,UAAA,KAAA,SAAKQ,EAAQ,CACL,IAAAC,EAA+E,KAA7EC,EAASD,EAAA,UAAEE,EAAOF,EAAA,QAAEG,EAAmBH,EAAA,oBAAEJ,EAAkBI,EAAA,mBAAEL,EAAWK,EAAA,YAC3EC,IACHC,EAAQ,KAAKH,CAAK,EAClB,CAACI,GAAuBD,EAAQ,KAAKN,EAAmB,IAAG,EAAKD,CAAW,GAE7E,KAAK,YAAW,EAChBH,EAAA,UAAM,KAAI,KAAA,KAACO,CAAK,CAClB,EAGUR,EAAA,UAAA,WAAV,SAAqBa,EAAyB,CAC5C,KAAK,eAAc,EACnB,KAAK,YAAW,EAQhB,QANMC,EAAe,KAAK,gBAAgBD,CAAU,EAE9CJ,EAAmC,KAAjCG,EAAmBH,EAAA,oBAAEE,EAAOF,EAAA,QAG9BM,EAAOJ,EAAQ,MAAK,EACjBK,EAAI,EAAGA,EAAID,EAAK,QAAU,CAACF,EAAW,OAAQG,GAAKJ,EAAsB,EAAI,EACpFC,EAAW,KAAKE,EAAKC,CAAC,CAAM,EAG9B,YAAK,wBAAwBH,CAAU,EAEhCC,CACT,EAEQd,EAAA,UAAA,YAAR,UAAA,CACQ,IAAAS,EAAoE,KAAlEN,EAAWM,EAAA,YAAEJ,EAAkBI,EAAA,mBAAEE,EAAOF,EAAA,QAAEG,EAAmBH,EAAA,oBAK/DQ,GAAsBL,EAAsB,EAAI,GAAKT,EAK3D,GAJAA,EAAc,KAAYc,EAAqBN,EAAQ,QAAUA,EAAQ,OAAO,EAAGA,EAAQ,OAASM,CAAkB,EAIlH,CAACL,EAAqB,CAKxB,QAJMM,EAAMb,EAAmB,IAAG,EAC9Bc,EAAO,EAGFH,EAAI,EAAGA,EAAIL,EAAQ,QAAWA,EAAQK,CAAC,GAAgBE,EAAKF,GAAK,EACxEG,EAAOH,EAETG,GAAQR,EAAQ,OAAO,EAAGQ,EAAO,CAAC,EAEtC,EACFnB,CAAA,EAzEsCoB,CAAO,EClB7C,IAAAC,GAAA,SAAAC,EAAA,CAA+BC,GAAAF,EAAAC,CAAA,EAC7B,SAAAD,EAAYG,EAAsBC,EAAmD,QACnFH,EAAA,KAAA,IAAA,GAAO,IACT,CAWO,OAAAD,EAAA,UAAA,SAAP,SAAgBK,EAAWC,EAAiB,CAAjB,OAAAA,IAAA,SAAAA,EAAA,GAClB,IACT,EACFN,CAAA,EAjB+BO,EAAY,ECHpC,IAAMC,GAAqC,CAGhD,YAAA,SAAYC,EAAqBC,EAAgB,SAAEC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,EAAA,CAAA,EAAA,UAAAA,CAAA,EACzC,IAAAC,EAAaL,GAAgB,SACrC,OAAIK,GAAQ,MAARA,EAAU,YACLA,EAAS,YAAW,MAApBA,EAAQC,EAAA,CAAaL,EAASC,CAAO,EAAAK,EAAKJ,CAAI,CAAA,CAAA,EAEhD,YAAW,MAAA,OAAAG,EAAA,CAACL,EAASC,CAAO,EAAAK,EAAKJ,CAAI,CAAA,CAAA,CAC9C,EACA,cAAA,SAAcK,EAAM,CACV,IAAAH,EAAaL,GAAgB,SACrC,QAAQK,GAAQ,KAAA,OAARA,EAAU,gBAAiB,eAAeG,CAAa,CACjE,EACA,SAAU,QCrBZ,IAAAC,GAAA,SAAAC,EAAA,CAAoCC,GAAAF,EAAAC,CAAA,EAOlC,SAAAD,EAAsBG,EAAqCC,EAAmD,CAA9G,IAAAC,EACEJ,EAAA,KAAA,KAAME,EAAWC,CAAI,GAAC,KADF,OAAAC,EAAA,UAAAF,EAAqCE,EAAA,KAAAD,EAFjDC,EAAA,QAAmB,IAI7B,CAEO,OAAAL,EAAA,UAAA,SAAP,SAAgBM,EAAWC,EAAiB,OAC1C,GADyBA,IAAA,SAAAA,EAAA,GACrB,KAAK,OACP,OAAO,KAIT,KAAK,MAAQD,EAEb,IAAME,EAAK,KAAK,GACVL,EAAY,KAAK,UAuBvB,OAAIK,GAAM,OACR,KAAK,GAAK,KAAK,eAAeL,EAAWK,EAAID,CAAK,GAKpD,KAAK,QAAU,GAEf,KAAK,MAAQA,EAEb,KAAK,IAAKE,EAAA,KAAK,MAAE,MAAAA,IAAA,OAAAA,EAAI,KAAK,eAAeN,EAAW,KAAK,GAAII,CAAK,EAE3D,IACT,EAEUP,EAAA,UAAA,eAAV,SAAyBG,EAA2BO,EAAmBH,EAAiB,CAAjB,OAAAA,IAAA,SAAAA,EAAA,GAC9DI,GAAiB,YAAYR,EAAU,MAAM,KAAKA,EAAW,IAAI,EAAGI,CAAK,CAClF,EAEUP,EAAA,UAAA,eAAV,SAAyBY,EAA4BJ,EAAkBD,EAAwB,CAE7F,GAFqEA,IAAA,SAAAA,EAAA,GAEjEA,GAAS,MAAQ,KAAK,QAAUA,GAAS,KAAK,UAAY,GAC5D,OAAOC,EAILA,GAAM,MACRG,GAAiB,cAAcH,CAAE,CAIrC,EAMOR,EAAA,UAAA,QAAP,SAAeM,EAAUC,EAAa,CACpC,GAAI,KAAK,OACP,OAAO,IAAI,MAAM,8BAA8B,EAGjD,KAAK,QAAU,GACf,IAAMM,EAAQ,KAAK,SAASP,EAAOC,CAAK,EACxC,GAAIM,EACF,OAAOA,EACE,KAAK,UAAY,IAAS,KAAK,IAAM,OAc9C,KAAK,GAAK,KAAK,eAAe,KAAK,UAAW,KAAK,GAAI,IAAI,EAE/D,EAEUb,EAAA,UAAA,SAAV,SAAmBM,EAAUQ,EAAc,CACzC,IAAIC,EAAmB,GACnBC,EACJ,GAAI,CACF,KAAK,KAAKV,CAAK,QACRW,EAAG,CACVF,EAAU,GAIVC,EAAaC,GAAQ,IAAI,MAAM,oCAAoC,EAErE,GAAIF,EACF,YAAK,YAAW,EACTC,CAEX,EAEAhB,EAAA,UAAA,YAAA,UAAA,CACE,GAAI,CAAC,KAAK,OAAQ,CACV,IAAAS,EAAoB,KAAlBD,EAAEC,EAAA,GAAEN,EAASM,EAAA,UACbS,EAAYf,EAAS,QAE7B,KAAK,KAAO,KAAK,MAAQ,KAAK,UAAY,KAC1C,KAAK,QAAU,GAEfgB,GAAUD,EAAS,IAAI,EACnBV,GAAM,OACR,KAAK,GAAK,KAAK,eAAeL,EAAWK,EAAI,IAAI,GAGnD,KAAK,MAAQ,KACbP,EAAA,UAAM,YAAW,KAAA,IAAA,EAErB,EACFD,CAAA,EA9IoCoB,EAAM,ECgB1C,IAAAC,GAAA,UAAA,CAGE,SAAAA,EAAoBC,EAAoCC,EAAiC,CAAjCA,IAAA,SAAAA,EAAoBF,EAAU,KAAlE,KAAA,oBAAAC,EAClB,KAAK,IAAMC,CACb,CA6BO,OAAAF,EAAA,UAAA,SAAP,SAAmBG,EAAqDC,EAAmBC,EAAS,CAA5B,OAAAD,IAAA,SAAAA,EAAA,GAC/D,IAAI,KAAK,oBAAuB,KAAMD,CAAI,EAAE,SAASE,EAAOD,CAAK,CAC1E,EAnCcJ,EAAA,IAAoBM,GAAsB,IAoC1DN,GArCA,ECnBA,IAAAO,GAAA,SAAAC,EAAA,CAAoCC,GAAAF,EAAAC,CAAA,EAkBlC,SAAAD,EAAYG,EAAgCC,EAAiC,CAAjCA,IAAA,SAAAA,EAAoBC,GAAU,KAA1E,IAAAC,EACEL,EAAA,KAAA,KAAME,EAAiBC,CAAG,GAAC,KAlBtB,OAAAE,EAAA,QAAmC,CAAA,EAOnCA,EAAA,QAAmB,IAY1B,CAEO,OAAAN,EAAA,UAAA,MAAP,SAAaO,EAAwB,CAC3B,IAAAC,EAAY,KAAI,QAExB,GAAI,KAAK,QAAS,CAChBA,EAAQ,KAAKD,CAAM,EACnB,OAGF,IAAIE,EACJ,KAAK,QAAU,GAEf,EACE,IAAKA,EAAQF,EAAO,QAAQA,EAAO,MAAOA,EAAO,KAAK,EACpD,YAEMA,EAASC,EAAQ,MAAK,GAIhC,GAFA,KAAK,QAAU,GAEXC,EAAO,CACT,KAAQF,EAASC,EAAQ,MAAK,GAC5BD,EAAO,YAAW,EAEpB,MAAME,EAEV,EACFT,CAAA,EAhDoCK,EAAS,EC6CtC,IAAMK,GAAiB,IAAIC,GAAeC,EAAW,EAK/CC,GAAQH,GCjDrB,IAAAI,GAAA,SAAAC,EAAA,CAA6CC,GAAAF,EAAAC,CAAA,EAC3C,SAAAD,EAAsBG,EAA8CC,EAAmD,CAAvH,IAAAC,EACEJ,EAAA,KAAA,KAAME,EAAWC,CAAI,GAAC,KADF,OAAAC,EAAA,UAAAF,EAA8CE,EAAA,KAAAD,GAEpE,CAEU,OAAAJ,EAAA,UAAA,eAAV,SAAyBG,EAAoCG,EAAkBC,EAAiB,CAE9F,OAF6EA,IAAA,SAAAA,EAAA,GAEzEA,IAAU,MAAQA,EAAQ,EACrBN,EAAA,UAAM,eAAc,KAAA,KAACE,EAAWG,EAAIC,CAAK,GAGlDJ,EAAU,QAAQ,KAAK,IAAI,EAIpBA,EAAU,aAAeA,EAAU,WAAaK,GAAuB,sBAAsB,UAAA,CAAM,OAAAL,EAAU,MAAM,MAAS,CAAzB,CAA0B,GACtI,EAEUH,EAAA,UAAA,eAAV,SAAyBG,EAAoCG,EAAkBC,EAAiB,OAI9F,GAJ6EA,IAAA,SAAAA,EAAA,GAIzEA,GAAS,KAAOA,EAAQ,EAAI,KAAK,MAAQ,EAC3C,OAAON,EAAA,UAAM,eAAc,KAAA,KAACE,EAAWG,EAAIC,CAAK,EAK1C,IAAAE,EAAYN,EAAS,QACzBG,GAAM,QAAQI,EAAAD,EAAQA,EAAQ,OAAS,CAAC,KAAC,MAAAC,IAAA,OAAA,OAAAA,EAAE,MAAOJ,IACpDE,GAAuB,qBAAqBF,CAAY,EACxDH,EAAU,WAAa,OAI3B,EACFH,CAAA,EApC6CW,EAAW,ECHxD,IAAAC,GAAA,SAAAC,EAAA,CAA6CC,GAAAF,EAAAC,CAAA,EAA7C,SAAAD,GAAA,+CAkCA,CAjCS,OAAAA,EAAA,UAAA,MAAP,SAAaG,EAAyB,CACpC,KAAK,QAAU,GAUf,IAAMC,EAAU,KAAK,WACrB,KAAK,WAAa,OAEV,IAAAC,EAAY,KAAI,QACpBC,EACJH,EAASA,GAAUE,EAAQ,MAAK,EAEhC,EACE,IAAKC,EAAQH,EAAO,QAAQA,EAAO,MAAOA,EAAO,KAAK,EACpD,aAEMA,EAASE,EAAQ,CAAC,IAAMF,EAAO,KAAOC,GAAWC,EAAQ,MAAK,GAIxE,GAFA,KAAK,QAAU,GAEXC,EAAO,CACT,MAAQH,EAASE,EAAQ,CAAC,IAAMF,EAAO,KAAOC,GAAWC,EAAQ,MAAK,GACpEF,EAAO,YAAW,EAEpB,MAAMG,EAEV,EACFN,CAAA,EAlC6CO,EAAc,ECgCpD,IAAMC,GAA0B,IAAIC,GAAwBC,EAAoB,EC8BhF,IAAMC,EAAQ,IAAIC,EAAkB,SAACC,EAAU,CAAK,OAAAA,EAAW,SAAQ,CAAnB,CAAqB,EC9D1E,SAAUC,GAAYC,EAAU,CACpC,OAAOA,GAASC,EAAWD,EAAM,QAAQ,CAC3C,CCDA,SAASE,GAAQC,EAAQ,CACvB,OAAOA,EAAIA,EAAI,OAAS,CAAC,CAC3B,CAEM,SAAUC,GAAkBC,EAAW,CAC3C,OAAOC,EAAWJ,GAAKG,CAAI,CAAC,EAAIA,EAAK,IAAG,EAAK,MAC/C,CAEM,SAAUE,GAAaF,EAAW,CACtC,OAAOG,GAAYN,GAAKG,CAAI,CAAC,EAAIA,EAAK,IAAG,EAAK,MAChD,CAEM,SAAUI,GAAUJ,EAAaK,EAAoB,CACzD,OAAO,OAAOR,GAAKG,CAAI,GAAM,SAAWA,EAAK,IAAG,EAAMK,CACxD,CClBO,IAAMC,GAAe,SAAIC,EAAM,CAAwB,OAAAA,GAAK,OAAOA,EAAE,QAAW,UAAY,OAAOA,GAAM,UAAlD,ECMxD,SAAUC,GAAUC,EAAU,CAClC,OAAOC,EAAWD,GAAK,KAAA,OAALA,EAAO,IAAI,CAC/B,CCHM,SAAUE,GAAoBC,EAAU,CAC5C,OAAOC,EAAWD,EAAME,EAAiB,CAAC,CAC5C,CCLM,SAAUC,GAAmBC,EAAQ,CACzC,OAAO,OAAO,eAAiBC,EAAWD,GAAG,KAAA,OAAHA,EAAM,OAAO,aAAa,CAAC,CACvE,CCAM,SAAUE,GAAiCC,EAAU,CAEzD,OAAO,IAAI,UACT,iBACEA,IAAU,MAAQ,OAAOA,GAAU,SAAW,oBAAsB,IAAIA,EAAK,KAAG,0HACwC,CAE9H,CCXM,SAAUC,IAAiB,CAC/B,OAAI,OAAO,QAAW,YAAc,CAAC,OAAO,SACnC,aAGF,OAAO,QAChB,CAEO,IAAMC,GAAWD,GAAiB,ECJnC,SAAUE,GAAWC,EAAU,CACnC,OAAOC,EAAWD,GAAK,KAAA,OAALA,EAAQE,EAAe,CAAC,CAC5C,CCHM,SAAiBC,GAAsCC,EAAqC,mGAC1FC,EAASD,EAAe,UAAS,2DAGX,MAAA,CAAA,EAAAE,GAAMD,EAAO,KAAI,CAAE,CAAA,gBAArCE,EAAkBC,EAAA,KAAA,EAAhBC,EAAKF,EAAA,MAAEG,EAAIH,EAAA,KACfG,iBAAA,CAAA,EAAA,CAAA,SACF,MAAA,CAAA,EAAAF,EAAA,KAAA,CAAA,qBAEIC,CAAM,CAAA,SAAZ,MAAA,CAAA,EAAAD,EAAA,KAAA,CAAA,SAAA,OAAAA,EAAA,KAAA,mCAGF,OAAAH,EAAO,YAAW,6BAIhB,SAAUM,GAAwBC,EAAQ,CAG9C,OAAOC,EAAWD,GAAG,KAAA,OAAHA,EAAK,SAAS,CAClC,CCPM,SAAUE,EAAaC,EAAyB,CACpD,GAAIA,aAAiBC,EACnB,OAAOD,EAET,GAAIA,GAAS,KAAM,CACjB,GAAIE,GAAoBF,CAAK,EAC3B,OAAOG,GAAsBH,CAAK,EAEpC,GAAII,GAAYJ,CAAK,EACnB,OAAOK,GAAcL,CAAK,EAE5B,GAAIM,GAAUN,CAAK,EACjB,OAAOO,GAAYP,CAAK,EAE1B,GAAIQ,GAAgBR,CAAK,EACvB,OAAOS,GAAkBT,CAAK,EAEhC,GAAIU,GAAWV,CAAK,EAClB,OAAOW,GAAaX,CAAK,EAE3B,GAAIY,GAAqBZ,CAAK,EAC5B,OAAOa,GAAuBb,CAAK,EAIvC,MAAMc,GAAiCd,CAAK,CAC9C,CAMM,SAAUG,GAAyBY,EAAQ,CAC/C,OAAO,IAAId,EAAW,SAACe,EAAyB,CAC9C,IAAMC,EAAMF,EAAIG,EAAiB,EAAC,EAClC,GAAIC,EAAWF,EAAI,SAAS,EAC1B,OAAOA,EAAI,UAAUD,CAAU,EAGjC,MAAM,IAAI,UAAU,gEAAgE,CACtF,CAAC,CACH,CASM,SAAUX,GAAiBe,EAAmB,CAClD,OAAO,IAAInB,EAAW,SAACe,EAAyB,CAU9C,QAASK,EAAI,EAAGA,EAAID,EAAM,QAAU,CAACJ,EAAW,OAAQK,IACtDL,EAAW,KAAKI,EAAMC,CAAC,CAAC,EAE1BL,EAAW,SAAQ,CACrB,CAAC,CACH,CAEM,SAAUT,GAAee,EAAuB,CACpD,OAAO,IAAIrB,EAAW,SAACe,EAAyB,CAC9CM,EACG,KACC,SAACC,EAAK,CACCP,EAAW,SACdA,EAAW,KAAKO,CAAK,EACrBP,EAAW,SAAQ,EAEvB,EACA,SAACQ,EAAQ,CAAK,OAAAR,EAAW,MAAMQ,CAAG,CAApB,CAAqB,EAEpC,KAAK,KAAMC,EAAoB,CACpC,CAAC,CACH,CAEM,SAAUd,GAAgBe,EAAqB,CACnD,OAAO,IAAIzB,EAAW,SAACe,EAAyB,aAC9C,QAAoBW,EAAAC,GAAAF,CAAQ,EAAAG,EAAAF,EAAA,KAAA,EAAA,CAAAE,EAAA,KAAAA,EAAAF,EAAA,KAAA,EAAE,CAAzB,IAAMJ,EAAKM,EAAA,MAEd,GADAb,EAAW,KAAKO,CAAK,EACjBP,EAAW,OACb,yGAGJA,EAAW,SAAQ,CACrB,CAAC,CACH,CAEM,SAAUP,GAAqBqB,EAA+B,CAClE,OAAO,IAAI7B,EAAW,SAACe,EAAyB,CAC9Ce,GAAQD,EAAed,CAAU,EAAE,MAAM,SAACQ,EAAG,CAAK,OAAAR,EAAW,MAAMQ,CAAG,CAApB,CAAqB,CACzE,CAAC,CACH,CAEM,SAAUX,GAA0BmB,EAAqC,CAC7E,OAAOvB,GAAkBwB,GAAmCD,CAAc,CAAC,CAC7E,CAEA,SAAeD,GAAWD,EAAiCd,EAAyB,uIACxDkB,EAAAC,GAAAL,CAAa,gFAIrC,GAJeP,EAAKa,EAAA,MACpBpB,EAAW,KAAKO,CAAK,EAGjBP,EAAW,OACb,MAAA,CAAA,CAAA,6RAGJ,OAAAA,EAAW,SAAQ,WChHf,SAAUqB,GACdC,EACAC,EACAC,EACAC,EACAC,EAAc,CADdD,IAAA,SAAAA,EAAA,GACAC,IAAA,SAAAA,EAAA,IAEA,IAAMC,EAAuBJ,EAAU,SAAS,UAAA,CAC9CC,EAAI,EACAE,EACFJ,EAAmB,IAAI,KAAK,SAAS,KAAMG,CAAK,CAAC,EAEjD,KAAK,YAAW,CAEpB,EAAGA,CAAK,EAIR,GAFAH,EAAmB,IAAIK,CAAoB,EAEvC,CAACD,EAKH,OAAOC,CAEX,CCeM,SAAUC,GAAaC,EAA0BC,EAAS,CAAT,OAAAA,IAAA,SAAAA,EAAA,GAC9CC,EAAQ,SAACC,EAAQC,EAAU,CAChCD,EAAO,UACLE,EACED,EACA,SAACE,EAAK,CAAK,OAAAC,GAAgBH,EAAYJ,EAAW,UAAA,CAAM,OAAAI,EAAW,KAAKE,CAAK,CAArB,EAAwBL,CAAK,CAA1E,EACX,UAAA,CAAM,OAAAM,GAAgBH,EAAYJ,EAAW,UAAA,CAAM,OAAAI,EAAW,SAAQ,CAAnB,EAAuBH,CAAK,CAAzE,EACN,SAACO,EAAG,CAAK,OAAAD,GAAgBH,EAAYJ,EAAW,UAAA,CAAM,OAAAI,EAAW,MAAMI,CAAG,CAApB,EAAuBP,CAAK,CAAzE,CAA0E,CACpF,CAEL,CAAC,CACH,CCPM,SAAUQ,GAAeC,EAA0BC,EAAiB,CAAjB,OAAAA,IAAA,SAAAA,EAAA,GAChDC,EAAQ,SAACC,EAAQC,EAAU,CAChCA,EAAW,IAAIJ,EAAU,SAAS,UAAA,CAAM,OAAAG,EAAO,UAAUC,CAAU,CAA3B,EAA8BH,CAAK,CAAC,CAC9E,CAAC,CACH,CC7DM,SAAUI,GAAsBC,EAA6BC,EAAwB,CACzF,OAAOC,EAAUF,CAAK,EAAE,KAAKG,GAAYF,CAAS,EAAGG,GAAUH,CAAS,CAAC,CAC3E,CCFM,SAAUI,GAAmBC,EAAuBC,EAAwB,CAChF,OAAOC,EAAUF,CAAK,EAAE,KAAKG,GAAYF,CAAS,EAAGG,GAAUH,CAAS,CAAC,CAC3E,CCJM,SAAUI,GAAiBC,EAAqBC,EAAwB,CAC5E,OAAO,IAAIC,EAAc,SAACC,EAAU,CAElC,IAAIC,EAAI,EAER,OAAOH,EAAU,SAAS,UAAA,CACpBG,IAAMJ,EAAM,OAGdG,EAAW,SAAQ,GAInBA,EAAW,KAAKH,EAAMI,GAAG,CAAC,EAIrBD,EAAW,QACd,KAAK,SAAQ,EAGnB,CAAC,CACH,CAAC,CACH,CCfM,SAAUE,GAAoBC,EAAoBC,EAAwB,CAC9E,OAAO,IAAIC,EAAc,SAACC,EAAU,CAClC,IAAIC,EAKJ,OAAAC,GAAgBF,EAAYF,EAAW,UAAA,CAErCG,EAAYJ,EAAcI,EAAe,EAAC,EAE1CC,GACEF,EACAF,EACA,UAAA,OACMK,EACAC,EACJ,GAAI,CAEDC,EAAkBJ,EAAS,KAAI,EAA7BE,EAAKE,EAAA,MAAED,EAAIC,EAAA,WACPC,EAAK,CAEZN,EAAW,MAAMM,CAAG,EACpB,OAGEF,EAKFJ,EAAW,SAAQ,EAGnBA,EAAW,KAAKG,CAAK,CAEzB,EACA,EACA,EAAI,CAER,CAAC,EAMM,UAAA,CAAM,OAAAI,EAAWN,GAAQ,KAAA,OAARA,EAAU,MAAM,GAAKA,EAAS,OAAM,CAA/C,CACf,CAAC,CACH,CCvDM,SAAUO,GAAyBC,EAAyBC,EAAwB,CACxF,GAAI,CAACD,EACH,MAAM,IAAI,MAAM,yBAAyB,EAE3C,OAAO,IAAIE,EAAc,SAACC,EAAU,CAClCC,GAAgBD,EAAYF,EAAW,UAAA,CACrC,IAAMI,EAAWL,EAAM,OAAO,aAAa,EAAC,EAC5CI,GACED,EACAF,EACA,UAAA,CACEI,EAAS,KAAI,EAAG,KAAK,SAACC,EAAM,CACtBA,EAAO,KAGTH,EAAW,SAAQ,EAEnBA,EAAW,KAAKG,EAAO,KAAK,CAEhC,CAAC,CACH,EACA,EACA,EAAI,CAER,CAAC,CACH,CAAC,CACH,CCzBM,SAAUC,GAA8BC,EAA8BC,EAAwB,CAClG,OAAOC,GAAsBC,GAAmCH,CAAK,EAAGC,CAAS,CACnF,CCoBM,SAAUG,GAAaC,EAA2BC,EAAwB,CAC9E,GAAID,GAAS,KAAM,CACjB,GAAIE,GAAoBF,CAAK,EAC3B,OAAOG,GAAmBH,EAAOC,CAAS,EAE5C,GAAIG,GAAYJ,CAAK,EACnB,OAAOK,GAAcL,EAAOC,CAAS,EAEvC,GAAIK,GAAUN,CAAK,EACjB,OAAOO,GAAgBP,EAAOC,CAAS,EAEzC,GAAIO,GAAgBR,CAAK,EACvB,OAAOS,GAAsBT,EAAOC,CAAS,EAE/C,GAAIS,GAAWV,CAAK,EAClB,OAAOW,GAAiBX,EAAOC,CAAS,EAE1C,GAAIW,GAAqBZ,CAAK,EAC5B,OAAOa,GAA2Bb,EAAOC,CAAS,EAGtD,MAAMa,GAAiCd,CAAK,CAC9C,CCoDM,SAAUe,GAAQC,EAA2BC,EAAyB,CAC1E,OAAOA,EAAYC,GAAUF,EAAOC,CAAS,EAAIE,EAAUH,CAAK,CAClE,CCxBM,SAAUI,GAAE,SAAIC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACpB,IAAMC,EAAYC,GAAaH,CAAI,EACnC,OAAOI,GAAKJ,EAAaE,CAAS,CACpC,CCsCM,SAAUG,GAAWC,EAA0BC,EAAyB,CAC5E,IAAMC,EAAeC,EAAWH,CAAmB,EAAIA,EAAsB,UAAA,CAAM,OAAAA,CAAA,EAC7EI,EAAO,SAACC,EAA6B,CAAK,OAAAA,EAAW,MAAMH,EAAY,CAAE,CAA/B,EAChD,OAAO,IAAII,EAAWL,EAAY,SAACI,EAAU,CAAK,OAAAJ,EAAU,SAASG,EAAa,EAAGC,CAAU,CAA7C,EAAiDD,CAAI,CACzG,CCpGO,IAAMG,GAA6BC,GAAiB,SAACC,EAAM,CAAK,OAAA,UAAuB,CAC5FA,EAAO,IAAI,EACX,KAAK,KAAO,aACZ,KAAK,QAAU,yBACjB,CAJuE,CAItE,ECrBK,SAAUC,GAAYC,EAAU,CACpC,OAAOA,aAAiB,MAAQ,CAAC,MAAMA,CAAY,CACrD,CCsCM,SAAUC,EAAUC,EAAyCC,EAAa,CAC9E,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAEhC,IAAIC,EAAQ,EAGZF,EAAO,UACLG,EAAyBF,EAAY,SAACG,EAAQ,CAG5CH,EAAW,KAAKJ,EAAQ,KAAKC,EAASM,EAAOF,GAAO,CAAC,CACvD,CAAC,CAAC,CAEN,CAAC,CACH,CC1DQ,IAAAG,GAAY,MAAK,QAEzB,SAASC,GAAkBC,EAA6BC,EAAW,CAC/D,OAAOH,GAAQG,CAAI,EAAID,EAAE,MAAA,OAAAE,EAAA,CAAA,EAAAC,EAAIF,CAAI,CAAA,CAAA,EAAID,EAAGC,CAAI,CAChD,CAMM,SAAUG,GAAuBJ,EAA2B,CAC9D,OAAOK,EAAI,SAAAJ,EAAI,CAAI,OAAAF,GAAYC,EAAIC,CAAI,CAApB,CAAqB,CAC5C,CCfQ,IAAAK,GAAY,MAAK,QACjBC,GAA0D,OAAM,eAArCC,GAA+B,OAAM,UAAlBC,GAAY,OAAM,KAQlE,SAAUC,GAAqDC,EAAuB,CAC1F,GAAIA,EAAK,SAAW,EAAG,CACrB,IAAMC,EAAQD,EAAK,CAAC,EACpB,GAAIL,GAAQM,CAAK,EACf,MAAO,CAAE,KAAMA,EAAO,KAAM,IAAI,EAElC,GAAIC,GAAOD,CAAK,EAAG,CACjB,IAAME,EAAOL,GAAQG,CAAK,EAC1B,MAAO,CACL,KAAME,EAAK,IAAI,SAACC,EAAG,CAAK,OAAAH,EAAMG,CAAG,CAAT,CAAU,EAClC,KAAID,IAKV,MAAO,CAAE,KAAMH,EAAa,KAAM,IAAI,CACxC,CAEA,SAASE,GAAOG,EAAQ,CACtB,OAAOA,GAAO,OAAOA,GAAQ,UAAYT,GAAeS,CAAG,IAAMR,EACnE,CC7BM,SAAUS,GAAaC,EAAgBC,EAAa,CACxD,OAAOD,EAAK,OAAO,SAACE,EAAQC,EAAKC,EAAC,CAAK,OAAEF,EAAOC,CAAG,EAAIF,EAAOG,CAAC,EAAIF,CAA5B,EAAqC,CAAA,CAAS,CACvF,CCsMM,SAAUG,GAAa,SAAoCC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAC/D,IAAMC,EAAYC,GAAaH,CAAI,EAC7BI,EAAiBC,GAAkBL,CAAI,EAEvCM,EAA8BC,GAAqBP,CAAI,EAA/CQ,EAAWF,EAAA,KAAEG,EAAIH,EAAA,KAE/B,GAAIE,EAAY,SAAW,EAIzB,OAAOE,GAAK,CAAA,EAAIR,CAAgB,EAGlC,IAAMS,EAAS,IAAIC,EACjBC,GACEL,EACAN,EACAO,EAEI,SAACK,EAAM,CAAK,OAAAC,GAAaN,EAAMK,CAAM,CAAzB,EAEZE,EAAQ,CACb,EAGH,OAAOZ,EAAkBO,EAAO,KAAKM,GAAiBb,CAAc,CAAC,EAAsBO,CAC7F,CAEM,SAAUE,GACdL,EACAN,EACAgB,EAAiD,CAAjD,OAAAA,IAAA,SAAAA,EAAAF,IAEO,SAACG,EAA2B,CAGjCC,GACElB,EACA,UAAA,CAaE,QAZQmB,EAAWb,EAAW,OAExBM,EAAS,IAAI,MAAMO,CAAM,EAG3BC,EAASD,EAITE,EAAuBF,aAGlBG,EAAC,CACRJ,GACElB,EACA,UAAA,CACE,IAAMuB,EAASf,GAAKF,EAAYgB,CAAC,EAAGtB,CAAgB,EAChDwB,EAAgB,GACpBD,EAAO,UACLE,EACER,EACA,SAACS,EAAK,CAEJd,EAAOU,CAAC,EAAII,EACPF,IAEHA,EAAgB,GAChBH,KAEGA,GAGHJ,EAAW,KAAKD,EAAeJ,EAAO,MAAK,CAAE,CAAC,CAElD,EACA,UAAA,CACO,EAAEQ,GAGLH,EAAW,SAAQ,CAEvB,CAAC,CACF,CAEL,EACAA,CAAU,GAjCLK,EAAI,EAAGA,EAAIH,EAAQG,MAAnBA,CAAC,CAoCZ,EACAL,CAAU,CAEd,CACF,CAMA,SAASC,GAAclB,EAAsC2B,EAAqBC,EAA0B,CACtG5B,EACF6B,GAAgBD,EAAc5B,EAAW2B,CAAO,EAEhDA,EAAO,CAEX,CC3RM,SAAUG,GACdC,EACAC,EACAC,EACAC,EACAC,EACAC,EACAC,EACAC,EAAgC,CAGhC,IAAMC,EAAc,CAAA,EAEhBC,EAAS,EAETC,EAAQ,EAERC,EAAa,GAKXC,EAAgB,UAAA,CAIhBD,GAAc,CAACH,EAAO,QAAU,CAACC,GACnCR,EAAW,SAAQ,CAEvB,EAGMY,EAAY,SAACC,EAAQ,CAAK,OAACL,EAASN,EAAaY,EAAWD,CAAK,EAAIN,EAAO,KAAKM,CAAK,CAA5D,EAE1BC,EAAa,SAACD,EAAQ,CAI1BT,GAAUJ,EAAW,KAAKa,CAAY,EAItCL,IAKA,IAAIO,EAAgB,GAGpBC,EAAUf,EAAQY,EAAOJ,GAAO,CAAC,EAAE,UACjCQ,EACEjB,EACA,SAACkB,GAAU,CAGTf,GAAY,MAAZA,EAAee,EAAU,EAErBd,EAGFQ,EAAUM,EAAiB,EAG3BlB,EAAW,KAAKkB,EAAU,CAE9B,EACA,UAAA,CAGEH,EAAgB,EAClB,EAEA,OACA,UAAA,CAIE,GAAIA,EAKF,GAAI,CAIFP,IAKA,sBACE,IAAMW,EAAgBZ,EAAO,MAAK,EAI9BF,EACFe,GAAgBpB,EAAYK,EAAmB,UAAA,CAAM,OAAAS,EAAWK,CAAa,CAAxB,CAAyB,EAE9EL,EAAWK,CAAa,GARrBZ,EAAO,QAAUC,EAASN,QAYjCS,EAAa,QACNU,EAAK,CACZrB,EAAW,MAAMqB,CAAG,EAG1B,CAAC,CACF,CAEL,EAGA,OAAAtB,EAAO,UACLkB,EAAyBjB,EAAYY,EAAW,UAAA,CAE9CF,EAAa,GACbC,EAAa,CACf,CAAC,CAAC,EAKG,UAAA,CACLL,GAAmB,MAAnBA,EAAmB,CACrB,CACF,CClEM,SAAUgB,GACdC,EACAC,EACAC,EAA6B,CAE7B,OAFAA,IAAA,SAAAA,EAAA,KAEIC,EAAWF,CAAc,EAEpBF,GAAS,SAACK,EAAGC,EAAC,CAAK,OAAAC,EAAI,SAACC,EAAQC,EAAU,CAAK,OAAAP,EAAeG,EAAGG,EAAGF,EAAGG,CAAE,CAA1B,CAA2B,EAAEC,EAAUT,EAAQI,EAAGC,CAAC,CAAC,CAAC,CAAjF,EAAoFH,CAAU,GAC/G,OAAOD,GAAmB,WACnCC,EAAaD,GAGRS,EAAQ,SAACC,EAAQC,EAAU,CAAK,OAAAC,GAAeF,EAAQC,EAAYZ,EAASE,CAAU,CAAtD,CAAuD,EAChG,CChCM,SAAUY,GAAyCC,EAA6B,CAA7B,OAAAA,IAAA,SAAAA,EAAA,KAChDC,GAASC,GAAUF,CAAU,CACtC,CCNM,SAAUG,IAAS,CACvB,OAAOC,GAAS,CAAC,CACnB,CCmDM,SAAUC,IAAM,SAACC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACrB,OAAOC,GAAS,EAAGC,GAAKH,EAAMI,GAAaJ,CAAI,CAAC,CAAC,CACnD,CC9DM,SAAUK,EAAsCC,EAA0B,CAC9E,OAAO,IAAIC,EAA+B,SAACC,EAAU,CACnDC,EAAUH,EAAiB,CAAE,EAAE,UAAUE,CAAU,CACrD,CAAC,CACH,CChDA,IAAME,GAA0B,CAAC,cAAe,gBAAgB,EAC1DC,GAAqB,CAAC,mBAAoB,qBAAqB,EAC/DC,GAAgB,CAAC,KAAM,KAAK,EAkO5B,SAAUC,EACdC,EACAC,EACAC,EACAC,EAAsC,CAMtC,GAJIC,EAAWF,CAAO,IACpBC,EAAiBD,EACjBA,EAAU,QAERC,EACF,OAAOJ,EAAaC,EAAQC,EAAWC,CAA+B,EAAE,KAAKG,GAAiBF,CAAc,CAAC,EAUzG,IAAAG,EAAAC,EAEJC,GAAcR,CAAM,EAChBH,GAAmB,IAAI,SAACY,EAAU,CAAK,OAAA,SAACC,EAAY,CAAK,OAAAV,EAAOS,CAAU,EAAER,EAAWS,EAASR,CAA+B,CAAtE,CAAlB,CAAyF,EAElIS,GAAwBX,CAAM,EAC5BJ,GAAwB,IAAIgB,GAAwBZ,EAAQC,CAAS,CAAC,EACtEY,GAA0Bb,CAAM,EAChCF,GAAc,IAAIc,GAAwBZ,EAAQC,CAAS,CAAC,EAC5D,CAAA,EAAE,CAAA,EATDa,EAAGR,EAAA,CAAA,EAAES,EAAMT,EAAA,CAAA,EAgBlB,GAAI,CAACQ,GACCE,GAAYhB,CAAM,EACpB,OAAOiB,GAAS,SAACC,EAAc,CAAK,OAAAnB,EAAUmB,EAAWjB,EAAWC,CAA+B,CAA/D,CAAgE,EAClGiB,EAAUnB,CAAM,CAAC,EAOvB,GAAI,CAACc,EACH,MAAM,IAAI,UAAU,sBAAsB,EAG5C,OAAO,IAAIM,EAAc,SAACC,EAAU,CAIlC,IAAMX,EAAU,UAAA,SAACY,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAAmB,OAAAF,EAAW,KAAK,EAAIC,EAAK,OAASA,EAAOA,EAAK,CAAC,CAAC,CAAhD,EAEpC,OAAAR,EAAIJ,CAAO,EAEJ,UAAA,CAAM,OAAAK,EAAQL,CAAO,CAAf,CACf,CAAC,CACH,CASA,SAASE,GAAwBZ,EAAaC,EAAiB,CAC7D,OAAO,SAACQ,EAAkB,CAAK,OAAA,SAACC,EAAY,CAAK,OAAAV,EAAOS,CAAU,EAAER,EAAWS,CAAO,CAArC,CAAlB,CACjC,CAOA,SAASC,GAAwBX,EAAW,CAC1C,OAAOI,EAAWJ,EAAO,WAAW,GAAKI,EAAWJ,EAAO,cAAc,CAC3E,CAOA,SAASa,GAA0Bb,EAAW,CAC5C,OAAOI,EAAWJ,EAAO,EAAE,GAAKI,EAAWJ,EAAO,GAAG,CACvD,CAOA,SAASQ,GAAcR,EAAW,CAChC,OAAOI,EAAWJ,EAAO,gBAAgB,GAAKI,EAAWJ,EAAO,mBAAmB,CACrF,CCnMM,SAAUwB,GACdC,EACAC,EACAC,EAAsC,CAEtC,OAAIA,EACKH,GAAoBC,EAAYC,CAAa,EAAE,KAAKE,GAAiBD,CAAc,CAAC,EAGtF,IAAIE,EAAoB,SAACC,EAAU,CACxC,IAAMC,EAAU,UAAA,SAACC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAAc,OAAAH,EAAW,KAAKE,EAAE,SAAW,EAAIA,EAAE,CAAC,EAAIA,CAAC,CAAzC,EACzBE,EAAWT,EAAWM,CAAO,EACnC,OAAOI,EAAWT,CAAa,EAAI,UAAA,CAAM,OAAAA,EAAcK,EAASG,CAAQ,CAA/B,EAAmC,MAC9E,CAAC,CACH,CCtBM,SAAUE,GACdC,EACAC,EACAC,EAAyC,CAFzCF,IAAA,SAAAA,EAAA,GAEAE,IAAA,SAAAA,EAAAC,IAIA,IAAIC,EAAmB,GAEvB,OAAIH,GAAuB,OAIrBI,GAAYJ,CAAmB,EACjCC,EAAYD,EAIZG,EAAmBH,GAIhB,IAAIK,EAAW,SAACC,EAAU,CAI/B,IAAIC,EAAMC,GAAYT,CAAO,EAAI,CAACA,EAAUE,EAAW,IAAG,EAAKF,EAE3DQ,EAAM,IAERA,EAAM,GAIR,IAAIE,EAAI,EAGR,OAAOR,EAAU,SAAS,UAAA,CACnBK,EAAW,SAEdA,EAAW,KAAKG,GAAG,EAEf,GAAKN,EAGP,KAAK,SAAS,OAAWA,CAAgB,EAGzCG,EAAW,SAAQ,EAGzB,EAAGC,CAAG,CACR,CAAC,CACH,CChGM,SAAUG,GAAK,SAACC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACpB,IAAMC,EAAYC,GAAaH,CAAI,EAC7BI,EAAaC,GAAUL,EAAM,GAAQ,EACrCM,EAAUN,EAChB,OAAQM,EAAQ,OAGZA,EAAQ,SAAW,EAEnBC,EAAUD,EAAQ,CAAC,CAAC,EAEpBE,GAASJ,CAAU,EAAEK,GAAKH,EAASJ,CAAS,CAAC,EAL7CQ,CAMN,CCjEO,IAAMC,GAAQ,IAAIC,EAAkBC,EAAI,ECpCvC,IAAAC,GAAY,MAAK,QAMnB,SAAUC,GAAkBC,EAAiB,CACjD,OAAOA,EAAK,SAAW,GAAKF,GAAQE,EAAK,CAAC,CAAC,EAAIA,EAAK,CAAC,EAAKA,CAC5D,CCoDM,SAAUC,EAAUC,EAAiDC,EAAa,CACtF,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAEhC,IAAIC,EAAQ,EAIZF,EAAO,UAILG,EAAyBF,EAAY,SAACG,EAAK,CAAK,OAAAP,EAAU,KAAKC,EAASM,EAAOF,GAAO,GAAKD,EAAW,KAAKG,CAAK,CAAhE,CAAiE,CAAC,CAEtH,CAAC,CACH,CCxBM,SAAUC,IAAG,SAACC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAClB,IAAMC,EAAiBC,GAAkBH,CAAI,EAEvCI,EAAUC,GAAeL,CAAI,EAEnC,OAAOI,EAAQ,OACX,IAAIE,EAAsB,SAACC,EAAU,CAGnC,IAAIC,EAAuBJ,EAAQ,IAAI,UAAA,CAAM,MAAA,CAAA,CAAA,CAAE,EAK3CK,EAAYL,EAAQ,IAAI,UAAA,CAAM,MAAA,EAAA,CAAK,EAGvCG,EAAW,IAAI,UAAA,CACbC,EAAUC,EAAY,IACxB,CAAC,EAKD,mBAASC,EAAW,CAClBC,EAAUP,EAAQM,CAAW,CAAC,EAAE,UAC9BE,EACEL,EACA,SAACM,EAAK,CAKJ,GAJAL,EAAQE,CAAW,EAAE,KAAKG,CAAK,EAI3BL,EAAQ,MAAM,SAACM,EAAM,CAAK,OAAAA,EAAO,MAAP,CAAa,EAAG,CAC5C,IAAMC,EAAcP,EAAQ,IAAI,SAACM,EAAM,CAAK,OAAAA,EAAO,MAAK,CAAZ,CAAe,EAE3DP,EAAW,KAAKL,EAAiBA,EAAc,MAAA,OAAAc,EAAA,CAAA,EAAAC,EAAIF,CAAM,CAAA,CAAA,EAAIA,CAAM,EAI/DP,EAAQ,KAAK,SAACM,EAAQI,EAAC,CAAK,MAAA,CAACJ,EAAO,QAAUL,EAAUS,CAAC,CAA7B,CAA8B,GAC5DX,EAAW,SAAQ,EAGzB,EACA,UAAA,CAGEE,EAAUC,CAAW,EAAI,GAIzB,CAACF,EAAQE,CAAW,EAAE,QAAUH,EAAW,SAAQ,CACrD,CAAC,CACF,GA9BIG,EAAc,EAAG,CAACH,EAAW,QAAUG,EAAcN,EAAQ,OAAQM,MAArEA,CAAW,EAmCpB,OAAO,UAAA,CACLF,EAAUC,EAAY,IACxB,CACF,CAAC,EACDU,CACN,CC9DM,SAAUC,GAASC,EAAoD,CAC3E,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAW,GACXC,EAAsB,KACtBC,EAA6C,KAC7CC,EAAa,GAEXC,EAAc,UAAA,CAGlB,GAFAF,GAAkB,MAAlBA,EAAoB,YAAW,EAC/BA,EAAqB,KACjBF,EAAU,CACZA,EAAW,GACX,IAAMK,EAAQJ,EACdA,EAAY,KACZF,EAAW,KAAKM,CAAK,EAEvBF,GAAcJ,EAAW,SAAQ,CACnC,EAEMO,EAAkB,UAAA,CACtBJ,EAAqB,KACrBC,GAAcJ,EAAW,SAAQ,CACnC,EAEAD,EAAO,UACLS,EACER,EACA,SAACM,EAAK,CACJL,EAAW,GACXC,EAAYI,EACPH,GACHM,EAAUZ,EAAiBS,CAAK,CAAC,EAAE,UAChCH,EAAqBK,EAAyBR,EAAYK,EAAaE,CAAe,CAAE,CAG/F,EACA,UAAA,CACEH,EAAa,IACZ,CAACH,GAAY,CAACE,GAAsBA,EAAmB,SAAWH,EAAW,SAAQ,CACxF,CAAC,CACF,CAEL,CAAC,CACH,CC3CM,SAAUU,GAAaC,EAAkBC,EAAyC,CAAzC,OAAAA,IAAA,SAAAA,EAAAC,IACtCC,GAAM,UAAA,CAAM,OAAAC,GAAMJ,EAAUC,CAAS,CAAzB,CAA0B,CAC/C,CCEM,SAAUI,GAAeC,EAAoBC,EAAsC,CAAtC,OAAAA,IAAA,SAAAA,EAAA,MAGjDA,EAAmBA,GAAgB,KAAhBA,EAAoBD,EAEhCE,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAiB,CAAA,EACjBC,EAAQ,EAEZH,EAAO,UACLI,EACEH,EACA,SAACI,EAAK,aACAC,EAAuB,KAKvBH,IAAUL,IAAsB,GAClCI,EAAQ,KAAK,CAAA,CAAE,MAIjB,QAAqBK,EAAAC,GAAAN,CAAO,EAAAO,EAAAF,EAAA,KAAA,EAAA,CAAAE,EAAA,KAAAA,EAAAF,EAAA,KAAA,EAAE,CAAzB,IAAMG,EAAMD,EAAA,MACfC,EAAO,KAAKL,CAAK,EAMbR,GAAca,EAAO,SACvBJ,EAASA,GAAM,KAANA,EAAU,CAAA,EACnBA,EAAO,KAAKI,CAAM,uGAItB,GAAIJ,MAIF,QAAqBK,EAAAH,GAAAF,CAAM,EAAAM,EAAAD,EAAA,KAAA,EAAA,CAAAC,EAAA,KAAAA,EAAAD,EAAA,KAAA,EAAE,CAAxB,IAAMD,EAAME,EAAA,MACfC,GAAUX,EAASQ,CAAM,EACzBT,EAAW,KAAKS,CAAM,sGAG5B,EACA,UAAA,aAGE,QAAqBI,EAAAN,GAAAN,CAAO,EAAAa,EAAAD,EAAA,KAAA,EAAA,CAAAC,EAAA,KAAAA,EAAAD,EAAA,KAAA,EAAE,CAAzB,IAAMJ,EAAMK,EAAA,MACfd,EAAW,KAAKS,CAAM,oGAExBT,EAAW,SAAQ,CACrB,EAEA,OACA,UAAA,CAEEC,EAAU,IACZ,CAAC,CACF,CAEL,CAAC,CACH,CCbM,SAAUc,GACdC,EAAgD,CAEhD,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAgC,KAChCC,EAAY,GACZC,EAEJF,EAAWF,EAAO,UAChBK,EAAyBJ,EAAY,OAAW,OAAW,SAACK,EAAG,CAC7DF,EAAgBG,EAAUT,EAASQ,EAAKT,GAAWC,CAAQ,EAAEE,CAAM,CAAC,CAAC,EACjEE,GACFA,EAAS,YAAW,EACpBA,EAAW,KACXE,EAAc,UAAUH,CAAU,GAIlCE,EAAY,EAEhB,CAAC,CAAC,EAGAA,IAMFD,EAAS,YAAW,EACpBA,EAAW,KACXE,EAAe,UAAUH,CAAU,EAEvC,CAAC,CACH,CC/HM,SAAUO,GACdC,EACAC,EACAC,EACAC,EACAC,EAAqC,CAErC,OAAO,SAACC,EAAuBC,EAA2B,CAIxD,IAAIC,EAAWL,EAIXM,EAAaP,EAEbQ,EAAQ,EAGZJ,EAAO,UACLK,EACEJ,EACA,SAACK,EAAK,CAEJ,IAAMC,EAAIH,IAEVD,EAAQD,EAEJP,EAAYQ,EAAOG,EAAOC,CAAC,GAIzBL,EAAW,GAAOI,GAGxBR,GAAcG,EAAW,KAAKE,CAAK,CACrC,EAGAJ,GACG,UAAA,CACCG,GAAYD,EAAW,KAAKE,CAAK,EACjCF,EAAW,SAAQ,CACrB,CAAE,CACL,CAEL,CACF,CCnCM,SAAUO,IAAa,SAAOC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAClC,IAAMC,EAAiBC,GAAkBH,CAAI,EAC7C,OAAOE,EACHE,GAAKL,GAAa,MAAA,OAAAM,EAAA,CAAA,EAAAC,EAAKN,CAAoC,CAAA,CAAA,EAAGO,GAAiBL,CAAc,CAAC,EAC9FM,EAAQ,SAACC,EAAQC,EAAU,CACzBC,GAAiBN,EAAA,CAAEI,CAAM,EAAAH,EAAKM,GAAeZ,CAAI,CAAC,CAAA,CAAA,EAAGU,CAAU,CACjE,CAAC,CACP,CCUM,SAAUG,IAAiB,SAC/BC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAEA,OAAOC,GAAa,MAAA,OAAAC,EAAA,CAAA,EAAAC,EAAIJ,CAAY,CAAA,CAAA,CACtC,CCgBM,SAAUK,GAAgBC,EAAiBC,EAAyC,CAAzC,OAAAA,IAAA,SAAAA,EAAAC,IACxCC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAkC,KAClCC,EAAsB,KACtBC,EAA0B,KAExBC,EAAO,UAAA,CACX,GAAIH,EAAY,CAEdA,EAAW,YAAW,EACtBA,EAAa,KACb,IAAMI,EAAQH,EACdA,EAAY,KACZF,EAAW,KAAKK,CAAK,EAEzB,EACA,SAASC,GAAY,CAInB,IAAMC,EAAaJ,EAAYR,EACzBa,EAAMZ,EAAU,IAAG,EACzB,GAAIY,EAAMD,EAAY,CAEpBN,EAAa,KAAK,SAAS,OAAWM,EAAaC,CAAG,EACtDR,EAAW,IAAIC,CAAU,EACzB,OAGFG,EAAI,CACN,CAEAL,EAAO,UACLU,EACET,EACA,SAACK,EAAQ,CACPH,EAAYG,EACZF,EAAWP,EAAU,IAAG,EAGnBK,IACHA,EAAaL,EAAU,SAASU,EAAcX,CAAO,EACrDK,EAAW,IAAIC,CAAU,EAE7B,EACA,UAAA,CAGEG,EAAI,EACJJ,EAAW,SAAQ,CACrB,EAEA,OACA,UAAA,CAEEE,EAAYD,EAAa,IAC3B,CAAC,CACF,CAEL,CAAC,CACH,CCpFM,SAAUS,GAAqBC,EAAe,CAClD,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAW,GACfF,EAAO,UACLG,EACEF,EACA,SAACG,EAAK,CACJF,EAAW,GACXD,EAAW,KAAKG,CAAK,CACvB,EACA,UAAA,CACOF,GACHD,EAAW,KAAKH,CAAa,EAE/BG,EAAW,SAAQ,CACrB,CAAC,CACF,CAEL,CAAC,CACH,CCXM,SAAUI,GAAQC,EAAa,CACnC,OAAOA,GAAS,EAEZ,UAAA,CAAM,OAAAC,CAAA,EACNC,EAAQ,SAACC,EAAQC,EAAU,CACzB,IAAIC,EAAO,EACXF,EAAO,UACLG,EAAyBF,EAAY,SAACG,EAAK,CAIrC,EAAEF,GAAQL,IACZI,EAAW,KAAKG,CAAK,EAIjBP,GAASK,GACXD,EAAW,SAAQ,EAGzB,CAAC,CAAC,CAEN,CAAC,CACP,CC9BM,SAAUI,IAAc,CAC5B,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChCD,EAAO,UAAUE,EAAyBD,EAAYE,EAAI,CAAC,CAC7D,CAAC,CACH,CCCM,SAAUC,GAASC,EAAQ,CAC/B,OAAOC,EAAI,UAAA,CAAM,OAAAD,CAAA,CAAK,CACxB,CC4CM,SAAUE,GACdC,EACAC,EAAmC,CAEnC,OAAIA,EAEK,SAACC,EAAqB,CAC3B,OAAAC,GAAOF,EAAkB,KAAKG,GAAK,CAAC,EAAGC,GAAc,CAAE,EAAGH,EAAO,KAAKH,GAAUC,CAAqB,CAAC,CAAC,CAAvG,EAGGM,GAAS,SAACC,EAAOC,EAAK,CAAK,OAAAC,EAAUT,EAAsBO,EAAOC,CAAK,CAAC,EAAE,KAAKJ,GAAK,CAAC,EAAGM,GAAMH,CAAK,CAAC,CAAzE,CAA0E,CAC9G,CCzCM,SAAUI,GAASC,EAAoBC,EAAyC,CAAzCA,IAAA,SAAAA,EAAAC,IAC3C,IAAMC,EAAWC,GAAMJ,EAAKC,CAAS,EACrC,OAAOI,GAAU,UAAA,CAAM,OAAAF,CAAA,CAAQ,CACjC,CC0EM,SAAUG,EACdC,EACAC,EAA0D,CAA1D,OAAAA,IAAA,SAAAA,EAA+BC,IAK/BF,EAAaA,GAAU,KAAVA,EAAcG,GAEpBC,EAAQ,SAACC,EAAQC,EAAU,CAGhC,IAAIC,EAEAC,EAAQ,GAEZH,EAAO,UACLI,EAAyBH,EAAY,SAACI,EAAK,CAEzC,IAAMC,EAAaV,EAAYS,CAAK,GAKhCF,GAAS,CAACR,EAAYO,EAAaI,CAAU,KAM/CH,EAAQ,GACRD,EAAcI,EAGdL,EAAW,KAAKI,CAAK,EAEzB,CAAC,CAAC,CAEN,CAAC,CACH,CAEA,SAASP,GAAeS,EAAQC,EAAM,CACpC,OAAOD,IAAMC,CACf,CCjHM,SAAUC,EAA8CC,EAAQC,EAAuC,CAC3G,OAAOC,EAAqB,SAACC,EAAMC,EAAI,CAAK,OAAAH,EAAUA,EAAQE,EAAEH,CAAG,EAAGI,EAAEJ,CAAG,CAAC,EAAIG,EAAEH,CAAG,IAAMI,EAAEJ,CAAG,CAApD,CAAqD,CACnG,CC7BM,SAAUK,GAAgBC,EAA6C,CAA7C,OAAAA,IAAA,SAAAA,EAAAC,IACvBC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAW,GACfF,EAAO,UACLG,EACEF,EACA,SAACG,EAAK,CACJF,EAAW,GACXD,EAAW,KAAKG,CAAK,CACvB,EACA,UAAA,CAAM,OAACF,EAAWD,EAAW,SAAQ,EAAKA,EAAW,MAAMJ,EAAY,CAAE,CAAnE,CAAqE,CAC5E,CAEL,CAAC,CACH,CAEA,SAASC,IAAmB,CAC1B,OAAO,IAAIO,EACb,CCMM,SAAUC,IAAO,SAAIC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACzB,OAAO,SAACC,EAAqB,CAAK,OAAAC,GAAOD,EAAQE,EAAE,MAAA,OAAAC,EAAA,CAAA,EAAAC,EAAIN,CAAM,CAAA,CAAA,CAAA,CAA3B,CACpC,CCHM,SAAUO,EAAYC,EAAoB,CAC9C,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAGhC,GAAI,CACFD,EAAO,UAAUC,CAAU,UAE3BA,EAAW,IAAIH,CAAQ,EAE3B,CAAC,CACH,CCMM,SAAUI,GACdC,EACAC,EAAgB,CAEhB,IAAMC,EAAkB,UAAU,QAAU,EAC5C,OAAO,SAACC,EAAqB,CAC3B,OAAAA,EAAO,KACLH,EAAYI,EAAO,SAACC,EAAG,EAAC,CAAK,OAAAL,EAAUK,EAAG,EAAGF,CAAM,CAAtB,CAAuB,EAAIG,GACxDC,GAAK,CAAC,EACNL,EAAkBM,GAAeP,CAAa,EAAIQ,GAAa,UAAA,CAAM,OAAA,IAAIC,EAAJ,CAAgB,CAAC,CAHxF,CAKJ,CC/CM,SAAUC,GAAYC,EAAa,CACvC,OAAOA,GAAS,EACZ,UAAA,CAAM,OAAAC,CAAA,EACNC,EAAQ,SAACC,EAAQC,EAAU,CAKzB,IAAIC,EAAc,CAAA,EAClBF,EAAO,UACLG,EACEF,EACA,SAACG,EAAK,CAEJF,EAAO,KAAKE,CAAK,EAGjBP,EAAQK,EAAO,QAAUA,EAAO,MAAK,CACvC,EACA,UAAA,aAGE,QAAoBG,EAAAC,GAAAJ,CAAM,EAAAK,EAAAF,EAAA,KAAA,EAAA,CAAAE,EAAA,KAAAA,EAAAF,EAAA,KAAA,EAAE,CAAvB,IAAMD,EAAKG,EAAA,MACdN,EAAW,KAAKG,CAAK,oGAEvBH,EAAW,SAAQ,CACrB,EAEA,OACA,UAAA,CAEEC,EAAS,IACX,CAAC,CACF,CAEL,CAAC,CACP,CC1DM,SAAUM,IAAK,SAAIC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACvB,IAAMC,EAAYC,GAAaH,CAAI,EAC7BI,EAAaC,GAAUL,EAAM,GAAQ,EAC3C,OAAAA,EAAOM,GAAeN,CAAI,EAEnBO,EAAQ,SAACC,EAAQC,EAAU,CAChCC,GAASN,CAAU,EAAEO,GAAIC,EAAA,CAAEJ,CAAM,EAAAK,EAAMb,CAA6B,CAAA,EAAGE,CAAS,CAAC,EAAE,UAAUO,CAAU,CACzG,CAAC,CACH,CCcM,SAAUK,IAAS,SACvBC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAEA,OAAOC,GAAK,MAAA,OAAAC,EAAA,CAAA,EAAAC,EAAIJ,CAAY,CAAA,CAAA,CAC9B,CCmEM,SAAUK,GAAUC,EAAqC,OACzDC,EAAQ,IACRC,EAEJ,OAAIF,GAAiB,OACf,OAAOA,GAAkB,UACxBG,EAA4BH,EAAa,MAAzCC,EAAKE,IAAA,OAAG,IAAQA,EAAED,EAAUF,EAAa,OAE5CC,EAAQD,GAILC,GAAS,EACZ,UAAA,CAAM,OAAAG,CAAA,EACNC,EAAQ,SAACC,EAAQC,EAAU,CACzB,IAAIC,EAAQ,EACRC,EAEEC,EAAc,UAAA,CAGlB,GAFAD,GAAS,MAATA,EAAW,YAAW,EACtBA,EAAY,KACRP,GAAS,KAAM,CACjB,IAAMS,EAAW,OAAOT,GAAU,SAAWU,GAAMV,CAAK,EAAIW,EAAUX,EAAMM,CAAK,CAAC,EAC5EM,EAAqBC,EAAyBR,EAAY,UAAA,CAC9DO,EAAmB,YAAW,EAC9BE,EAAiB,CACnB,CAAC,EACDL,EAAS,UAAUG,CAAkB,OAErCE,EAAiB,CAErB,EAEMA,EAAoB,UAAA,CACxB,IAAIC,EAAY,GAChBR,EAAYH,EAAO,UACjBS,EAAyBR,EAAY,OAAW,UAAA,CAC1C,EAAEC,EAAQP,EACRQ,EACFC,EAAW,EAEXO,EAAY,GAGdV,EAAW,SAAQ,CAEvB,CAAC,CAAC,EAGAU,GACFP,EAAW,CAEf,EAEAM,EAAiB,CACnB,CAAC,CACP,CCpFM,SAAUE,GAAcC,EAA6DC,EAAQ,CAMjG,OAAOC,EAAQC,GAAcH,EAAaC,EAAW,UAAU,QAAU,EAAG,EAAI,CAAC,CACnF,CC+CM,SAAUG,GAASC,EAA4B,CAA5BA,IAAA,SAAAA,EAAA,CAAA,GACf,IAAAC,EAAgHD,EAAO,UAAvHE,EAASD,IAAA,OAAG,UAAA,CAAM,OAAA,IAAIE,CAAJ,EAAgBF,EAAEG,EAA4EJ,EAAO,aAAnFK,EAAYD,IAAA,OAAG,GAAIA,EAAEE,EAAuDN,EAAO,gBAA9DO,EAAeD,IAAA,OAAG,GAAIA,EAAEE,EAA+BR,EAAO,oBAAtCS,EAAmBD,IAAA,OAAG,GAAIA,EAUnH,OAAO,SAACE,EAAa,CACnB,IAAIC,EACAC,EACAC,EACAC,EAAW,EACXC,EAAe,GACfC,EAAa,GAEXC,EAAc,UAAA,CAClBL,GAAe,MAAfA,EAAiB,YAAW,EAC5BA,EAAkB,MACpB,EAGMM,GAAQ,UAAA,CACZD,EAAW,EACXN,EAAaE,EAAU,OACvBE,EAAeC,EAAa,EAC9B,EACMG,EAAsB,UAAA,CAG1B,IAAMC,EAAOT,EACbO,GAAK,EACLE,GAAI,MAAJA,EAAM,YAAW,CACnB,EAEA,OAAOC,EAAc,SAACC,EAAQC,GAAU,CACtCT,IACI,CAACE,GAAc,CAACD,GAClBE,EAAW,EAOb,IAAMO,GAAQX,EAAUA,GAAO,KAAPA,EAAWX,EAAS,EAO5CqB,GAAW,IAAI,UAAA,CACbT,IAKIA,IAAa,GAAK,CAACE,GAAc,CAACD,IACpCH,EAAkBa,GAAYN,EAAqBV,CAAmB,EAE1E,CAAC,EAIDe,GAAK,UAAUD,EAAU,EAGvB,CAACZ,GAIDG,EAAW,IAOXH,EAAa,IAAIe,GAAe,CAC9B,KAAM,SAACC,GAAK,CAAK,OAAAH,GAAK,KAAKG,EAAK,CAAf,EACjB,MAAO,SAACC,GAAG,CACTZ,EAAa,GACbC,EAAW,EACXL,EAAkBa,GAAYP,GAAOb,EAAcuB,EAAG,EACtDJ,GAAK,MAAMI,EAAG,CAChB,EACA,SAAU,UAAA,CACRb,EAAe,GACfE,EAAW,EACXL,EAAkBa,GAAYP,GAAOX,CAAe,EACpDiB,GAAK,SAAQ,CACf,EACD,EACDK,EAAUP,CAAM,EAAE,UAAUX,CAAU,EAE1C,CAAC,EAAED,CAAa,CAClB,CACF,CAEA,SAASe,GACPP,EACAY,EAAoD,SACpDC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,EAAA,CAAA,EAAA,UAAAA,CAAA,EAEA,GAAIF,IAAO,GAAM,CACfZ,EAAK,EACL,OAGF,GAAIY,IAAO,GAIX,KAAMG,EAAe,IAAIP,GAAe,CACtC,KAAM,UAAA,CACJO,EAAa,YAAW,EACxBf,EAAK,CACP,EACD,EAED,OAAOW,EAAUC,EAAE,MAAA,OAAAI,EAAA,CAAA,EAAAC,EAAIJ,CAAI,CAAA,CAAA,CAAA,EAAG,UAAUE,CAAY,EACtD,CChHM,SAAUG,EACdC,EACAC,EACAC,EAAyB,WAErBC,EACAC,EAAW,GACf,OAAIJ,GAAsB,OAAOA,GAAuB,UACnDK,EAA8EL,EAAkB,WAAhGG,EAAUE,IAAA,OAAG,IAAQA,EAAEC,EAAuDN,EAAkB,WAAzEC,EAAUK,IAAA,OAAG,IAAQA,EAAEC,EAAgCP,EAAkB,SAAlDI,EAAQG,IAAA,OAAG,GAAKA,EAAEL,EAAcF,EAAkB,WAEnGG,EAAcH,GAAkB,KAAlBA,EAAsB,IAE/BQ,GAAS,CACd,UAAW,UAAA,CAAM,OAAA,IAAIC,GAAcN,EAAYF,EAAYC,CAAS,CAAnD,EACjB,aAAc,GACd,gBAAiB,GACjB,oBAAqBE,EACtB,CACH,CCxIM,SAAUM,GAAQC,EAAa,CACnC,OAAOC,EAAO,SAACC,EAAGC,EAAK,CAAK,OAAAH,GAASG,CAAT,CAAc,CAC5C,CCaM,SAAUC,GAAaC,EAA8B,CACzD,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAS,GAEPC,EAAiBC,EACrBH,EACA,UAAA,CACEE,GAAc,MAAdA,EAAgB,YAAW,EAC3BD,EAAS,EACX,EACAG,EAAI,EAGNC,EAAUR,CAAQ,EAAE,UAAUK,CAAc,EAE5CH,EAAO,UAAUI,EAAyBH,EAAY,SAACM,EAAK,CAAK,OAAAL,GAAUD,EAAW,KAAKM,CAAK,CAA/B,CAAgC,CAAC,CACpG,CAAC,CACH,CCVM,SAAUC,GAAS,SAAOC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAC9B,IAAMC,EAAYC,GAAaH,CAAM,EACrC,OAAOI,EAAQ,SAACC,EAAQC,EAAU,EAI/BJ,EAAYK,GAAOP,EAAQK,EAAQH,CAAS,EAAIK,GAAOP,EAAQK,CAAM,GAAG,UAAUC,CAAU,CAC/F,CAAC,CACH,CCmBM,SAAUE,EACdC,EACAC,EAA6G,CAE7G,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAyD,KACzDC,EAAQ,EAERC,EAAa,GAIXC,EAAgB,UAAA,CAAM,OAAAD,GAAc,CAACF,GAAmBD,EAAW,SAAQ,CAArD,EAE5BD,EAAO,UACLM,EACEL,EACA,SAACM,EAAK,CAEJL,GAAe,MAAfA,EAAiB,YAAW,EAC5B,IAAIM,EAAa,EACXC,EAAaN,IAEnBO,EAAUb,EAAQU,EAAOE,CAAU,CAAC,EAAE,UACnCP,EAAkBI,EACjBL,EAIA,SAACU,EAAU,CAAK,OAAAV,EAAW,KAAKH,EAAiBA,EAAeS,EAAOI,EAAYF,EAAYD,GAAY,EAAIG,CAAU,CAAzG,EAChB,UAAA,CAIET,EAAkB,KAClBG,EAAa,CACf,CAAC,CACD,CAEN,EACA,UAAA,CACED,EAAa,GACbC,EAAa,CACf,CAAC,CACF,CAEL,CAAC,CACH,CCvFM,SAAUO,EAAaC,EAA8B,CACzD,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChCC,EAAUJ,CAAQ,EAAE,UAAUK,EAAyBF,EAAY,UAAA,CAAM,OAAAA,EAAW,SAAQ,CAAnB,EAAuBG,EAAI,CAAC,EACrG,CAACH,EAAW,QAAUD,EAAO,UAAUC,CAAU,CACnD,CAAC,CACH,CCIM,SAAUI,GAAaC,EAAiDC,EAAiB,CAAjB,OAAAA,IAAA,SAAAA,EAAA,IACrEC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAQ,EACZF,EAAO,UACLG,EAAyBF,EAAY,SAACG,EAAK,CACzC,IAAMC,EAASR,EAAUO,EAAOF,GAAO,GACtCG,GAAUP,IAAcG,EAAW,KAAKG,CAAK,EAC9C,CAACC,GAAUJ,EAAW,SAAQ,CAChC,CAAC,CAAC,CAEN,CAAC,CACH,CCqGM,SAAUK,EACdC,EACAC,EACAC,EAA8B,CAK9B,IAAMC,EACJC,EAAWJ,CAAc,GAAKC,GAASC,EAElC,CAAE,KAAMF,EAA2E,MAAKC,EAAE,SAAQC,CAAA,EACnGF,EAEN,OAAOG,EACHE,EAAQ,SAACC,EAAQC,EAAU,QACzBC,EAAAL,EAAY,aAAS,MAAAK,IAAA,QAAAA,EAAA,KAArBL,CAAW,EACX,IAAIM,EAAU,GACdH,EAAO,UACLI,EACEH,EACA,SAACI,EAAK,QACJH,EAAAL,EAAY,QAAI,MAAAK,IAAA,QAAAA,EAAA,KAAhBL,EAAmBQ,CAAK,EACxBJ,EAAW,KAAKI,CAAK,CACvB,EACA,UAAA,OACEF,EAAU,IACVD,EAAAL,EAAY,YAAQ,MAAAK,IAAA,QAAAA,EAAA,KAApBL,CAAW,EACXI,EAAW,SAAQ,CACrB,EACA,SAACK,EAAG,OACFH,EAAU,IACVD,EAAAL,EAAY,SAAK,MAAAK,IAAA,QAAAA,EAAA,KAAjBL,EAAoBS,CAAG,EACvBL,EAAW,MAAMK,CAAG,CACtB,EACA,UAAA,SACMH,KACFD,EAAAL,EAAY,eAAW,MAAAK,IAAA,QAAAA,EAAA,KAAvBL,CAAW,IAEbU,EAAAV,EAAY,YAAQ,MAAAU,IAAA,QAAAA,EAAA,KAApBV,CAAW,CACb,CAAC,CACF,CAEL,CAAC,EAIDW,EACN,CCnIM,SAAUC,GAAYC,EAAsDC,EAAuB,CACvG,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAC1B,IAAAC,EAAuCJ,GAAM,KAANA,EAAU,CAAA,EAA/CK,EAAAD,EAAA,QAAAE,EAAOD,IAAA,OAAG,GAAIA,EAAEE,EAAAH,EAAA,SAAAI,EAAQD,IAAA,OAAG,GAAKA,EACpCE,EAAW,GACXC,EAAsB,KACtBC,EAAiC,KACjCC,EAAa,GAEXC,EAAgB,UAAA,CACpBF,GAAS,MAATA,EAAW,YAAW,EACtBA,EAAY,KACRH,IACFM,EAAI,EACJF,GAAcT,EAAW,SAAQ,EAErC,EAEMY,EAAoB,UAAA,CACxBJ,EAAY,KACZC,GAAcT,EAAW,SAAQ,CACnC,EAEMa,EAAgB,SAACC,GAAQ,CAC7B,OAACN,EAAYO,EAAUnB,EAAiBkB,EAAK,CAAC,EAAE,UAAUE,EAAyBhB,EAAYU,EAAeE,CAAiB,CAAC,CAAhI,EAEID,EAAO,UAAA,CACX,GAAIL,EAAU,CAIZA,EAAW,GACX,IAAMQ,GAAQP,EACdA,EAAY,KAEZP,EAAW,KAAKc,EAAK,EACrB,CAACL,GAAcI,EAAcC,EAAK,EAEtC,EAEAf,EAAO,UACLiB,EACEhB,EAMA,SAACc,GAAK,CACJR,EAAW,GACXC,EAAYO,GACZ,EAAEN,GAAa,CAACA,EAAU,UAAYL,EAAUQ,EAAI,EAAKE,EAAcC,EAAK,EAC9E,EACA,UAAA,CACEL,EAAa,GACb,EAAEJ,GAAYC,GAAYE,GAAa,CAACA,EAAU,SAAWR,EAAW,SAAQ,CAClF,CAAC,CACF,CAEL,CAAC,CACH,CCxFM,SAAUiB,GACdC,EACAC,EACAC,EAAuB,CADvBD,IAAA,SAAAA,EAAAE,IAGA,IAAMC,EAAYC,GAAML,EAAUC,CAAS,EAC3C,OAAOK,GAAS,UAAA,CAAM,OAAAF,CAAA,EAAWF,CAAM,CACzC,CCJM,SAAUK,IAAc,SAAOC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACnC,IAAMC,EAAUC,GAAkBH,CAAM,EAExC,OAAOI,EAAQ,SAACC,EAAQC,EAAU,CAehC,QAdMC,EAAMP,EAAO,OACbQ,EAAc,IAAI,MAAMD,CAAG,EAI7BE,EAAWT,EAAO,IAAI,UAAA,CAAM,MAAA,EAAA,CAAK,EAGjCU,EAAQ,cAMHC,EAAC,CACRC,EAAUZ,EAAOW,CAAC,CAAC,EAAE,UACnBE,EACEP,EACA,SAACQ,EAAK,CACJN,EAAYG,CAAC,EAAIG,EACb,CAACJ,GAAS,CAACD,EAASE,CAAC,IAEvBF,EAASE,CAAC,EAAI,IAKbD,EAAQD,EAAS,MAAMM,EAAQ,KAAON,EAAW,MAEtD,EAGAO,EAAI,CACL,GAnBIL,EAAI,EAAGA,EAAIJ,EAAKI,MAAhBA,CAAC,EAwBVN,EAAO,UACLQ,EAAyBP,EAAY,SAACQ,EAAK,CACzC,GAAIJ,EAAO,CAET,IAAMO,EAAMC,EAAA,CAAIJ,CAAK,EAAAK,EAAKX,CAAW,CAAA,EACrCF,EAAW,KAAKJ,EAAUA,EAAO,MAAA,OAAAgB,EAAA,CAAA,EAAAC,EAAIF,CAAM,CAAA,CAAA,EAAIA,CAAM,EAEzD,CAAC,CAAC,CAEN,CAAC,CACH,CCxFM,SAAUG,IAAG,SAAOC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACxB,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChCL,GAAS,MAAA,OAAAM,EAAA,CAACF,CAA8B,EAAAG,EAAMN,CAAuC,CAAA,CAAA,EAAE,UAAUI,CAAU,CAC7G,CAAC,CACH,CCCM,SAAUG,IAAO,SAAkCC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACvD,OAAOC,GAAG,MAAA,OAAAC,EAAA,CAAA,EAAAC,EAAIJ,CAAW,CAAA,CAAA,CAC3B,CCYO,SAASK,IAAmC,CACjD,IAAMC,EAAY,IAAIC,GAAwB,CAAC,EAC/C,OAAAC,EAAU,SAAU,mBAAoB,CAAE,KAAM,EAAK,CAAC,EACnD,UAAU,IAAMF,EAAU,KAAK,QAAQ,CAAC,EAGpCA,CACT,CCHO,SAASG,EACdC,EAAkBC,EAAmB,SAChC,CACL,OAAO,MAAM,KAAKA,EAAK,iBAAoBD,CAAQ,CAAC,CACtD,CAuBO,SAASE,EACdF,EAAkBC,EAAmB,SAClC,CACH,IAAME,EAAKC,GAAsBJ,EAAUC,CAAI,EAC/C,GAAI,OAAOE,GAAO,YAChB,MAAM,IAAI,eACR,8BAA8BH,CAAQ,iBACxC,EAGF,OAAOG,CACT,CAsBO,SAASC,GACdJ,EAAkBC,EAAmB,SACtB,CACf,OAAOA,EAAK,cAAiBD,CAAQ,GAAK,MAC5C,CAOO,SAASK,IAA4C,CAnH5D,IAAAC,EAAAC,EAAAC,EAAAC,EAoHE,OACEA,GAAAD,GAAAD,GAAAD,EAAA,SAAS,gBAAT,YAAAA,EAAwB,aAAxB,YAAAC,EAAoC,gBAApC,KAAAC,EACA,SAAS,gBADT,KAAAC,EAEA,MAEJ,CCvEA,IAAMC,GAAYC,EAChBC,EAAU,SAAS,KAAM,SAAS,EAClCA,EAAU,SAAS,KAAM,UAAU,CACrC,EACG,KACCC,GAAa,CAAC,EACdC,EAAU,MAAS,EACnBC,EAAI,IAAMC,GAAiB,GAAK,SAAS,IAAI,EAC7CC,EAAY,CAAC,CACf,EAaK,SAASC,GACdC,EACqB,CACrB,OAAOT,GACJ,KACCK,EAAIK,GAAUD,EAAG,SAASC,CAAM,CAAC,EACjCC,EAAqB,CACvB,CACJ,CCpCO,SAASC,GACdC,EAAiBC,EACI,CACrB,OAAOC,EACLC,EAAUH,EAAI,YAAY,EAAE,KAAKI,EAAI,IAAM,EAAI,CAAC,EAChDD,EAAUH,EAAI,YAAY,EAAE,KAAKI,EAAI,IAAM,EAAK,CAAC,CACnD,EACG,KACCH,EAAWI,GAAaJ,CAAQ,EAAIK,GACpCC,EAAU,EAAK,CACjB,CACJ,CCAO,SAASC,GACdC,EACe,CACf,MAAO,CACL,EAAGA,EAAG,WACN,EAAGA,EAAG,SACR,CACF,CAWO,SAASC,GACdD,EAC2B,CAC3B,OAAOE,EACLC,EAAU,OAAQ,MAAM,EACxBA,EAAU,OAAQ,QAAQ,CAC5B,EACG,KACCC,GAAU,EAAGC,EAAuB,EACpCC,EAAI,IAAMP,GAAiBC,CAAE,CAAC,EAC9BO,EAAUR,GAAiBC,CAAE,CAAC,CAChC,CACJ,CCxCO,SAASQ,GACdC,EACe,CACf,MAAO,CACL,EAAGA,EAAG,WACN,EAAGA,EAAG,SACR,CACF,CAWO,SAASC,GACdD,EAC2B,CAC3B,OAAOE,EACLC,EAAUH,EAAI,QAAQ,EACtBG,EAAU,OAAQ,QAAQ,CAC5B,EACG,KACCC,GAAU,EAAGC,EAAuB,EACpCC,EAAI,IAAMP,GAAwBC,CAAE,CAAC,EACrCO,EAAUR,GAAwBC,CAAE,CAAC,CACvC,CACJ,CCnBA,SAASQ,GAAYC,EAAiBC,EAA8B,CAGlE,GAAI,OAAOA,GAAU,UAAY,OAAOA,GAAU,SAChDD,EAAG,WAAaC,EAAM,SAAS,UAGtBA,aAAiB,KAC1BD,EAAG,YAAYC,CAAK,UAGX,MAAM,QAAQA,CAAK,EAC5B,QAAWC,KAAQD,EACjBF,GAAYC,EAAIE,CAAI,CAE1B,CAyBO,SAASC,EACdC,EAAaC,KAAmCC,EAC7C,CACH,IAAMN,EAAK,SAAS,cAAcI,CAAG,EAGrC,GAAIC,EACF,QAAWE,KAAQ,OAAO,KAAKF,CAAU,EACnC,OAAOA,EAAWE,CAAI,GAAM,cAI5B,OAAOF,EAAWE,CAAI,GAAM,UAC9BP,EAAG,aAAaO,EAAMF,EAAWE,CAAI,CAAC,EAEtCP,EAAG,aAAaO,EAAM,EAAE,GAI9B,QAAWN,KAASK,EAClBP,GAAYC,EAAIC,CAAK,EAGvB,OAAOD,CACT,CC9EO,SAASQ,GAAMC,EAAuB,CAC3C,GAAIA,EAAQ,IAAK,CACf,IAAMC,EAAS,GAAGD,EAAQ,KAAO,IAAO,IACxC,MAAO,KAAKA,EAAQ,MAAY,KAAM,QAAQC,CAAM,CAAC,GACvD,KACE,QAAOD,EAAM,SAAS,CAE1B,CCCO,SAASE,GAAYC,EAA+B,CACzD,IAAMC,EAASC,EAAE,SAAU,CAAE,IAAAF,CAAI,CAAC,EAClC,OAAOG,EAAM,KACX,SAAS,KAAK,YAAYF,CAAM,EACzBG,EACLC,EAAUJ,EAAQ,MAAM,EACxBI,EAAUJ,EAAQ,OAAO,EACtB,KACCK,EAAU,IACRC,GAAW,IAAM,IAAI,eAAe,mBAAmBP,CAAG,EAAE,CAAC,CAC9D,CACH,CACJ,EACG,KACCQ,EAAI,IAAG,EAAY,EACnBC,EAAS,IAAM,SAAS,KAAK,YAAYR,CAAM,CAAC,EAChDS,GAAK,CAAC,CACR,EACH,CACH,CCVA,IAAMC,GAAS,IAAIC,EAiBbC,GAAYC,EAAM,IACtB,OAAO,gBAAmB,YACtBC,GAAY,4CAA4C,EACxDC,EAAG,MAAS,CACjB,EACE,KACCC,EAAI,IAAM,IAAI,eAAeC,GAAW,CACtC,QAAWC,KAASD,EAClBP,GAAO,KAAKQ,CAAK,CACrB,CAAC,CAAC,EACFC,EAAUC,GAAYC,EAAMC,GAAOP,EAAGK,CAAQ,CAAC,EAC5C,KACCG,EAAS,IAAMH,EAAS,WAAW,CAAC,CACtC,CACF,EACAI,EAAY,CAAC,CACf,EAaK,SAASC,GACdC,EACa,CACb,MAAO,CACL,MAAQA,EAAG,YACX,OAAQA,EAAG,YACb,CACF,CAuBO,SAASC,GACdD,EACyB,CACzB,OAAOd,GACJ,KACCgB,EAAIR,GAAYA,EAAS,QAAQM,CAAE,CAAC,EACpCP,EAAUC,GAAYV,GACnB,KACCmB,EAAO,CAAC,CAAE,OAAAC,CAAO,IAAMA,IAAWJ,CAAE,EACpCH,EAAS,IAAMH,EAAS,UAAUM,CAAE,CAAC,EACrCV,EAAI,IAAMS,GAAeC,CAAE,CAAC,CAC9B,CACF,EACAK,EAAUN,GAAeC,CAAE,CAAC,CAC9B,CACJ,CCnHO,SAASM,GACdC,EACa,CACb,MAAO,CACL,MAAQA,EAAG,YACX,OAAQA,EAAG,YACb,CACF,CASO,SAASC,GACdD,EACyB,CACzB,IAAIE,EAASF,EAAG,cAChB,KAAOE,IAEHF,EAAG,aAAgBE,EAAO,aAC1BF,EAAG,cAAgBE,EAAO,eAE1BA,GAAUF,EAAKE,GAAQ,cAK3B,OAAOA,EAASF,EAAK,MACvB,CCfA,IAAMG,GAAS,IAAIC,EAUbC,GAAYC,EAAM,IAAMC,EAC5B,IAAI,qBAAqBC,GAAW,CAClC,QAAWC,KAASD,EAClBL,GAAO,KAAKM,CAAK,CACrB,EAAG,CACD,UAAW,CACb,CAAC,CACH,CAAC,EACE,KACCC,EAAUC,GAAYC,EAAMC,GAAON,EAAGI,CAAQ,CAAC,EAC5C,KACCG,EAAS,IAAMH,EAAS,WAAW,CAAC,CACtC,CACF,EACAI,EAAY,CAAC,CACf,EAaK,SAASC,GACdC,EACqB,CACrB,OAAOZ,GACJ,KACCa,EAAIP,GAAYA,EAAS,QAAQM,CAAE,CAAC,EACpCP,EAAUC,GAAYR,GACnB,KACCgB,EAAO,CAAC,CAAE,OAAAC,CAAO,IAAMA,IAAWH,CAAE,EACpCH,EAAS,IAAMH,EAAS,UAAUM,CAAE,CAAC,EACrCI,EAAI,CAAC,CAAE,eAAAC,CAAe,IAAMA,CAAc,CAC5C,CACF,CACF,CACJ,CAaO,SAASC,GACdN,EAAiBO,EAAY,GACR,CACrB,OAAOC,GAA0BR,CAAE,EAChC,KACCI,EAAI,CAAC,CAAE,EAAAK,CAAE,IAAM,CACb,IAAMC,EAAUC,GAAeX,CAAE,EAC3BY,EAAUC,GAAsBb,CAAE,EACxC,OAAOS,GACLG,EAAQ,OAASF,EAAQ,OAASH,CAEtC,CAAC,EACDO,EAAqB,CACvB,CACJ,CCjFA,IAAMC,GAA4C,CAChD,OAAQC,EAAW,yBAAyB,EAC5C,OAAQA,EAAW,yBAAyB,CAC9C,EAaO,SAASC,GAAUC,EAAuB,CAC/C,OAAOH,GAAQG,CAAI,EAAE,OACvB,CAaO,SAASC,GAAUD,EAAcE,EAAsB,CACxDL,GAAQG,CAAI,EAAE,UAAYE,GAC5BL,GAAQG,CAAI,EAAE,MAAM,CACxB,CAWO,SAASG,GAAYH,EAAmC,CAC7D,IAAMI,EAAKP,GAAQG,CAAI,EACvB,OAAOK,EAAUD,EAAI,QAAQ,EAC1B,KACCE,EAAI,IAAMF,EAAG,OAAO,EACpBG,EAAUH,EAAG,OAAO,CACtB,CACJ,CC9BA,SAASI,GACPC,EAAiBC,EACR,CACT,OAAQD,EAAG,YAAa,CAGtB,KAAK,iBAEH,OAAIA,EAAG,OAAS,QACP,SAAS,KAAKC,CAAI,EAElB,GAGX,KAAK,kBACL,KAAK,oBACH,MAAO,GAGT,QACE,OAAOD,EAAG,iBACd,CACF,CAWO,SAASE,IAAwC,CACtD,OAAOC,EACLC,EAAU,OAAQ,kBAAkB,EAAE,KAAKC,EAAI,IAAM,EAAI,CAAC,EAC1DD,EAAU,OAAQ,gBAAgB,EAAE,KAAKC,EAAI,IAAM,EAAK,CAAC,CAC3D,EACG,KACCC,EAAU,EAAK,CACjB,CACJ,CAOO,SAASC,IAAsC,CACpD,IAAMC,EAAYJ,EAAyB,OAAQ,SAAS,EACzD,KACCK,EAAOC,GAAM,EAAEA,EAAG,SAAWA,EAAG,QAAQ,EACxCL,EAAIK,IAAO,CACT,KAAMC,GAAU,QAAQ,EAAI,SAAW,SACvC,KAAMD,EAAG,IACT,OAAQ,CACNA,EAAG,eAAe,EAClBA,EAAG,gBAAgB,CACrB,CACF,EAAc,EACdD,EAAO,CAAC,CAAE,KAAAG,EAAM,KAAAX,CAAK,IAAM,CACzB,GAAIW,IAAS,SAAU,CACrB,IAAMC,EAASC,GAAiB,EAChC,GAAI,OAAOD,GAAW,YACpB,MAAO,CAACd,GAAwBc,EAAQZ,CAAI,CAChD,CACA,MAAO,EACT,CAAC,EACDc,GAAM,CACR,EAGF,OAAOb,GAAiB,EACrB,KACCc,EAAUH,GAAWA,EAAqBI,EAAZT,CAAiB,CACjD,CACJ,CC1GO,SAASU,IAAmB,CACjC,OAAO,IAAI,IAAI,SAAS,IAAI,CAC9B,CAgBO,SAASC,GACdC,EAA4BC,EAAW,GACjC,CACN,GAAIC,EAAQ,oBAAoB,GAAK,CAACD,EAAU,CAC9C,IAAME,EAAKC,EAAE,IAAK,CAAE,KAAMJ,EAAI,IAAK,CAAC,EACpC,SAAS,KAAK,YAAYG,CAAE,EAC5BA,EAAG,MAAM,EACTA,EAAG,OAAO,CAIZ,MACE,SAAS,KAAOH,EAAI,IAExB,CASO,SAASK,IAA8B,CAC5C,OAAO,IAAIC,CACb,CCxCO,SAASC,IAA0B,CACxC,OAAO,SAAS,KAAK,MAAM,CAAC,CAC9B,CAYO,SAASC,GAAgBC,EAAoB,CAClD,IAAMC,EAAKC,EAAE,IAAK,CAAE,KAAMF,CAAK,CAAC,EAChCC,EAAG,iBAAiB,QAASE,GAAMA,EAAG,gBAAgB,CAAC,EACvDF,EAAG,MAAM,CACX,CAWO,SAASG,GACdC,EACoB,CACpB,OAAOC,EACLC,EAA2B,OAAQ,YAAY,EAC/CF,CACF,EACG,KACCG,EAAIV,EAAe,EACnBW,EAAUX,GAAgB,CAAC,EAC3BY,EAAOV,GAAQA,EAAK,OAAS,CAAC,EAC9BW,EAAY,CAAC,CACf,CACJ,CASO,SAASC,GACdP,EACyB,CACzB,OAAOD,GAAkBC,CAAS,EAC/B,KACCG,EAAIK,GAAMC,GAAmB,QAAQD,CAAE,IAAI,CAAE,EAC7CH,EAAOT,GAAM,OAAOA,GAAO,WAAW,CACxC,CACJ,CCtDO,SAASc,GAAWC,EAAoC,CAC7D,IAAMC,EAAQ,WAAWD,CAAK,EAC9B,OAAOE,GAA0BC,GAC/BF,EAAM,YAAY,IAAME,EAAKF,EAAM,OAAO,CAAC,CAC5C,EACE,KACCG,EAAUH,EAAM,OAAO,CACzB,CACJ,CAOO,SAASI,IAAkC,CAChD,IAAMJ,EAAQ,WAAW,OAAO,EAChC,OAAOK,EACLC,EAAU,OAAQ,aAAa,EAAE,KAAKC,EAAI,IAAM,EAAI,CAAC,EACrDD,EAAU,OAAQ,YAAY,EAAE,KAAKC,EAAI,IAAM,EAAK,CAAC,CACvD,EACG,KACCJ,EAAUH,EAAM,OAAO,CACzB,CACJ,CAcO,SAASQ,GACdC,EAA6BC,EACd,CACf,OAAOD,EACJ,KACCE,EAAUC,GAAUA,EAASF,EAAQ,EAAIG,CAAK,CAChD,CACJ,CC/BO,SAASC,GACdC,EAAmBC,EACD,CAClB,OAAO,IAAIC,EAAiBC,GAAY,CACtC,IAAMC,EAAM,IAAI,eAChB,OAAAA,EAAI,KAAK,MAAO,GAAGJ,CAAG,EAAE,EACxBI,EAAI,aAAe,OAGnBA,EAAI,iBAAiB,OAAQ,IAAM,CAC7BA,EAAI,QAAU,KAAOA,EAAI,OAAS,KACpCD,EAAS,KAAKC,EAAI,QAAQ,EAC1BD,EAAS,SAAS,GAIlBA,EAAS,MAAM,IAAI,MAAMC,EAAI,UAAU,CAAC,CAE5C,CAAC,EAGDA,EAAI,iBAAiB,QAAS,IAAM,CAClCD,EAAS,MAAM,IAAI,MAAM,eAAe,CAAC,CAC3C,CAAC,EAGDC,EAAI,iBAAiB,QAAS,IAAM,CAClCD,EAAS,SAAS,CACpB,CAAC,EAGG,OAAOF,GAAA,YAAAA,EAAS,YAAc,cAChCG,EAAI,iBAAiB,WAAYC,GAAS,CA/FhD,IAAAC,EAgGQ,GAAID,EAAM,iBACRJ,EAAQ,UAAW,KAAMI,EAAM,OAASA,EAAM,MAAS,GAAG,MAIrD,CACL,IAAME,GAASD,EAAAF,EAAI,kBAAkB,gBAAgB,IAAtC,KAAAE,EAA2C,EAC1DL,EAAQ,UAAW,KAAMI,EAAM,OAAS,CAACE,EAAU,GAAG,CACxD,CACF,CAAC,EAGDN,EAAQ,UAAU,KAAK,CAAC,GAI1BG,EAAI,KAAK,EACF,IAAMA,EAAI,MAAM,CACzB,CAAC,CACH,CAcO,SAASI,GACdR,EAAmBC,EACJ,CACf,OAAOF,GAAQC,EAAKC,CAAO,EACxB,KACCQ,EAAUC,GAAOA,EAAI,KAAK,CAAC,EAC3BC,EAAIC,GAAQ,KAAK,MAAMA,CAAI,CAAM,EACjCC,EAAY,CAAC,CACf,CACJ,CAUO,SAASC,GACdd,EAAmBC,EACG,CACtB,IAAMc,EAAM,IAAI,UAChB,OAAOhB,GAAQC,EAAKC,CAAO,EACxB,KACCQ,EAAUC,GAAOA,EAAI,KAAK,CAAC,EAC3BC,EAAID,GAAOK,EAAI,gBAAgBL,EAAK,WAAW,CAAC,EAChDG,EAAY,CAAC,CACf,CACJ,CAUO,SAASG,GACdhB,EAAmBC,EACG,CACtB,IAAMc,EAAM,IAAI,UAChB,OAAOhB,GAAQC,EAAKC,CAAO,EACxB,KACCQ,EAAUC,GAAOA,EAAI,KAAK,CAAC,EAC3BC,EAAID,GAAOK,EAAI,gBAAgBL,EAAK,UAAU,CAAC,EAC/CG,EAAY,CAAC,CACf,CACJ,CC5HO,SAASI,IAAoC,CAClD,MAAO,CACL,EAAG,KAAK,IAAI,EAAG,OAAO,EACtB,EAAG,KAAK,IAAI,EAAG,OAAO,CACxB,CACF,CASO,SAASC,IAAkD,CAChE,OAAOC,EACLC,EAAU,OAAQ,SAAU,CAAE,QAAS,EAAK,CAAC,EAC7CA,EAAU,OAAQ,SAAU,CAAE,QAAS,EAAK,CAAC,CAC/C,EACG,KACCC,EAAIJ,EAAiB,EACrBK,EAAUL,GAAkB,CAAC,CAC/B,CACJ,CC3BO,SAASM,IAAgC,CAC9C,MAAO,CACL,MAAQ,WACR,OAAQ,WACV,CACF,CASO,SAASC,IAA8C,CAC5D,OAAOC,EAAU,OAAQ,SAAU,CAAE,QAAS,EAAK,CAAC,EACjD,KACCC,EAAIH,EAAe,EACnBI,EAAUJ,GAAgB,CAAC,CAC7B,CACJ,CCXO,SAASK,IAAsC,CACpD,OAAOC,EAAc,CACnBC,GAAoB,EACpBC,GAAkB,CACpB,CAAC,EACE,KACCC,EAAI,CAAC,CAACC,EAAQC,CAAI,KAAO,CAAE,OAAAD,EAAQ,KAAAC,CAAK,EAAE,EAC1CC,EAAY,CAAC,CACf,CACJ,CCVO,SAASC,GACdC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EAChB,CACtB,IAAMC,EAAQF,EACX,KACCG,EAAwB,MAAM,CAChC,EAGIC,EAAUC,EAAc,CAACH,EAAOD,CAAO,CAAC,EAC3C,KACCK,EAAI,IAAMC,GAAiBR,CAAE,CAAC,CAChC,EAGF,OAAOM,EAAc,CAACJ,EAASD,EAAWI,CAAO,CAAC,EAC/C,KACCE,EAAI,CAAC,CAAC,CAAE,OAAAE,CAAO,EAAG,CAAE,OAAAC,EAAQ,KAAAC,CAAK,EAAG,CAAE,EAAAC,EAAG,EAAAC,CAAE,CAAC,KAAO,CACjD,OAAQ,CACN,EAAGH,EAAO,EAAIE,EACd,EAAGF,EAAO,EAAIG,EAAIJ,CACpB,EACA,KAAAE,CACF,EAAE,CACJ,CACJ,CCzBA,SAASG,GAAQC,EAA+B,CAC9C,OAAOC,EAA8BD,EAAQ,UAAWE,GAAMA,EAAG,IAAI,CACvE,CAWA,SAASC,GAAQH,EAA4B,CAC3C,IAAMI,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAUE,GAAQN,EAAO,YAAYM,CAAI,CAAC,EAGzCF,CACT,CAgBO,SAASG,GACdC,EAAaR,EAAS,IAAI,OAAOQ,CAAG,EACxB,CACZ,IAAMC,EAAQV,GAAQC,CAAM,EACtBI,EAAQD,GAAQH,CAAM,EAGtBU,EAAU,IAAIL,EACpBK,EAAQ,UAAUN,CAAK,EAGvB,IAAMO,EAAQP,EAAM,KAAKQ,GAAe,EAAGC,GAAQ,EAAI,CAAC,EACxD,OAAOH,EACJ,KACCE,GAAe,EACfE,GAAUL,EAAM,KAAKM,EAAUJ,CAAK,CAAC,CAAC,EACtCK,GAAM,CACR,CACJ,CCLA,IAAMC,GAASC,EAAW,WAAW,EAC/BC,GAAiB,KAAK,MAAMF,GAAO,WAAY,EACrDE,GAAO,KAAO,GAAG,IAAI,IAAIA,GAAO,KAAMC,GAAY,CAAC,CAAC,GAW7C,SAASC,IAAwB,CACtC,OAAOF,EACT,CASO,SAASG,EAAQC,EAAqB,CAC3C,OAAOJ,GAAO,SAAS,SAASI,CAAI,CACtC,CAUO,SAASC,GACdC,EAAkBC,EACV,CACR,OAAO,OAAOA,GAAU,YACpBP,GAAO,aAAaM,CAAG,EAAE,QAAQ,IAAKC,EAAM,SAAS,CAAC,EACtDP,GAAO,aAAaM,CAAG,CAC7B,CC/BO,SAASE,GACdC,EAASC,EAAmB,SACP,CACrB,OAAOC,EAAW,sBAAsBF,CAAI,IAAKC,CAAI,CACvD,CAYO,SAASE,GACdH,EAASC,EAAmB,SACL,CACvB,OAAOG,EAAY,sBAAsBJ,CAAI,IAAKC,CAAI,CACxD,CC7EO,SAASI,GACdC,EACsB,CACtB,IAAMC,EAASC,EAAW,6BAA8BF,CAAE,EAC1D,OAAOG,EAAUF,EAAQ,QAAS,CAAE,KAAM,EAAK,CAAC,EAC7C,KACCG,EAAI,IAAMF,EAAW,cAAeF,CAAE,CAAC,EACvCI,EAAIC,IAAY,CAAE,KAAM,UAAUA,EAAQ,SAAS,CAAE,EAAE,CACzD,CACJ,CASO,SAASC,GACdN,EACiC,CACjC,GAAI,CAACO,EAAQ,kBAAkB,GAAK,CAACP,EAAG,kBACtC,OAAOQ,EAGT,GAAI,CAACR,EAAG,OAAQ,CACd,IAAMK,EAAUH,EAAW,cAAeF,CAAE,EACxC,UAAUK,EAAQ,SAAS,IAAM,SAAS,YAAY,IACxDL,EAAG,OAAS,GAChB,CAGA,OAAOS,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAU,CAAC,CAAE,KAAAE,CAAK,IAAM,CAC5BZ,EAAG,OAAS,GAGZ,SAAiB,aAAcY,CAAI,CACrC,CAAC,EAGMb,GAAcC,CAAE,EACpB,KACCa,EAAIC,GAASJ,EAAM,KAAKI,CAAK,CAAC,EAC9BC,EAAS,IAAML,EAAM,SAAS,CAAC,EAC/BN,EAAIU,GAAUE,EAAA,CAAE,IAAKhB,GAAOc,EAAQ,CACtC,CACJ,CAAC,CACH,CC5BO,SAASG,GACdC,EAAiB,CAAE,QAAAC,CAAQ,EACN,CACrB,OAAOA,EACJ,KACCC,EAAIC,IAAW,CAAE,OAAQA,IAAWH,CAAG,EAAE,CAC3C,CACJ,CAYO,SAASI,GACdJ,EAAiBK,EACe,CAChC,IAAMC,EAAY,IAAIC,EACtB,OAAAD,EAAU,UAAU,CAAC,CAAE,OAAAE,CAAO,IAAM,CAClCR,EAAG,OAASQ,CACd,CAAC,EAGMT,GAAaC,EAAIK,CAAO,EAC5B,KACCI,EAAIC,GAASJ,EAAU,KAAKI,CAAK,CAAC,EAClCC,EAAS,IAAML,EAAU,SAAS,CAAC,EACnCJ,EAAIQ,GAAUE,EAAA,CAAE,IAAKZ,GAAOU,EAAQ,CACtC,CACJ,CCrEO,SAASG,GACdC,EAAaC,EACA,CACb,OAAIA,IAAU,SAEVC,EAAC,OAAI,MAAM,gCAAgC,GAAIF,EAAI,KAAK,WACtDE,EAAC,OAAI,MAAM,+BAA+B,CAC5C,EAIAA,EAAC,OAAI,MAAM,aAAa,GAAIF,EAAI,KAAK,WACnCE,EAAC,OAAI,MAAM,+BAA+B,CAC5C,CAGN,CCxBO,SAASC,GACdC,EAAqBC,EACR,CAIb,GAHAA,EAASA,EAAS,GAAGA,CAAM,eAAeD,CAAE,GAAK,OAG7CC,EAAQ,CACV,IAAMC,EAASD,EAAS,IAAIA,CAAM,GAAK,OACvC,OACEE,EAAC,SAAM,MAAM,gBAAgB,SAAU,GACpCC,GAAcH,CAAM,EACrBE,EAAC,KAAE,KAAMD,EAAQ,MAAM,uBAAuB,SAAU,IACtDC,EAAC,QAAK,wBAAuBH,EAAI,CACnC,CACF,CAEJ,KACE,QACEG,EAAC,SAAM,MAAM,gBAAgB,SAAU,GACpCC,GAAcH,CAAM,EACrBE,EAAC,QAAK,MAAM,uBAAuB,SAAU,IAC3CA,EAAC,QAAK,wBAAuBH,EAAI,CACnC,CACF,CAGN,CC5BO,SAASK,GAAsBC,EAAyB,CAC7D,OACEC,EAAC,UACC,MAAM,uBACN,MAAOC,GAAY,gBAAgB,EACnC,wBAAuB,IAAIF,CAAE,UAC9B,CAEL,CCQA,SAASG,GACPC,EAAsBC,EACT,CACb,IAAMC,EAASD,EAAO,EAChBE,EAASF,EAAO,EAGhBG,EAAU,OAAO,KAAKJ,EAAS,KAAK,EACvC,OAAOK,GAAO,CAACL,EAAS,MAAMK,CAAG,CAAC,EAClC,OAAyB,CAACC,EAAMD,IAAQ,CACvC,GAAGC,EAAMC,EAAC,WAAKF,CAAI,EAAQ,GAC7B,EAAG,CAAC,CAAC,EACJ,MAAM,EAAG,EAAE,EAGRG,EAASC,GAAc,EACvBC,EAAM,IAAI,IAAIV,EAAS,SAAUQ,EAAO,IAAI,EAC9CG,EAAQ,kBAAkB,GAC5BD,EAAI,aAAa,IAAI,IAAK,OAAO,QAAQV,EAAS,KAAK,EACpD,OAAO,CAAC,CAAC,CAAEY,CAAK,IAAMA,CAAK,EAC3B,OAAO,CAACC,EAAW,CAACC,CAAK,IAAM,GAAGD,CAAS,IAAIC,CAAK,GAAG,KAAK,EAAG,EAAE,CACpE,EAGF,GAAM,CAAE,KAAAC,CAAK,EAAIN,GAAc,EAC/B,OACEF,EAAC,KAAE,KAAM,GAAGG,CAAG,GAAI,MAAM,yBAAyB,SAAU,IAC1DH,EAAC,WACC,MAAM,uCACN,gBAAeP,EAAS,MAAM,QAAQ,CAAC,GAEtCE,EAAS,GAAKK,EAAC,OAAI,MAAM,iCAAiC,EAC1DL,EAAS,GAAKK,EAAC,UAAIP,EAAS,KAAM,EAClCE,GAAU,GAAKK,EAAC,UAAIP,EAAS,KAAM,EACnCG,EAAS,GAAKH,EAAS,KAAK,OAAS,GACpCA,EAAS,KAEVA,EAAS,MAAQA,EAAS,KAAK,IAAIgB,GAAO,CACzC,IAAMC,EAAOF,EACTC,KAAOD,EACL,uBAAuBA,EAAKC,CAAG,CAAC,GAChC,cACF,GACJ,OACET,EAAC,QAAK,MAAO,UAAUU,CAAI,IAAKD,CAAI,CAExC,CAAC,EACAb,EAAS,GAAKC,EAAQ,OAAS,GAC9BG,EAAC,KAAE,MAAM,2BACNW,GAAY,4BAA4B,EAAE,KAAG,GAAGd,CACnD,CAEJ,CACF,CAEJ,CAaO,SAASe,GACdC,EACa,CACb,IAAMC,EAAYD,EAAO,CAAC,EAAE,MACtBE,EAAO,CAAC,GAAGF,CAAM,EAEjBZ,EAASC,GAAc,EAGvBP,EAASoB,EAAK,UAAUC,GAErB,CADG,GAAG,IAAI,IAAIA,EAAI,SAAUf,EAAO,IAAI,CAAC,GACrC,SAAS,GAAG,CACvB,EACK,CAACgB,CAAO,EAAIF,EAAK,OAAOpB,EAAQ,CAAC,EAGnCuB,EAAQH,EAAK,UAAUC,GAAOA,EAAI,MAAQF,CAAS,EACnDI,IAAU,KACZA,EAAQH,EAAK,QAGf,IAAMI,EAAOJ,EAAK,MAAM,EAAGG,CAAK,EAC1BE,EAAOL,EAAK,MAAMG,CAAK,EAGvBG,EAAW,CACf7B,GAAqByB,EAAS,EAAc,EAAE,CAACtB,GAAUuB,IAAU,EAAE,EACrE,GAAGC,EAAK,IAAIG,GAAW9B,GAAqB8B,EAAS,CAAW,CAAC,EACjE,GAAGF,EAAK,OAAS,CACfpB,EAAC,WAAQ,MAAM,0BACbA,EAAC,WAAQ,SAAU,IACjBA,EAAC,WACEoB,EAAK,OAAS,GAAKA,EAAK,SAAW,EAChCT,GAAY,wBAAwB,EACpCA,GAAY,2BAA4BS,EAAK,MAAM,CAEzD,CACF,EACC,GAAGA,EAAK,IAAIE,GAAW9B,GAAqB8B,EAAS,CAAW,CAAC,CACpE,CACF,EAAI,CAAC,CACP,EAGA,OACEtB,EAAC,MAAG,MAAM,0BACPqB,CACH,CAEJ,CCrIO,SAASE,GAAkBC,EAAiC,CACjE,OACEC,EAAC,MAAG,MAAM,oBACP,OAAO,QAAQD,CAAK,EAAE,IAAI,CAAC,CAACE,EAAKC,CAAK,IACrCF,EAAC,MAAG,MAAO,oCAAoCC,CAAG,IAC/C,OAAOC,GAAU,SAAWC,GAAMD,CAAK,EAAIA,CAC9C,CACD,CACH,CAEJ,CCAO,SAASE,GACdC,EACa,CACb,IAAMC,EAAU,kCAAkCD,CAAI,GACtD,OACEE,EAAC,OAAI,MAAOD,EAAS,OAAM,IACzBC,EAAC,UAAO,MAAM,gBAAgB,SAAU,GAAI,cAAY,OAAO,CACjE,CAEJ,CCpBO,SAASC,GAAYC,EAAiC,CAC3D,OACEC,EAAC,OAAI,MAAM,0BACTA,EAAC,OAAI,MAAM,qBACRD,CACH,CACF,CAEJ,CCcA,SAASE,GAAcC,EAA+B,CACpD,IAAMC,EAASC,GAAc,EAGvBC,EAAM,IAAI,IAAI,MAAMH,EAAQ,OAAO,IAAKC,EAAO,IAAI,EACzD,OACEG,EAAC,MAAG,MAAM,oBACRA,EAAC,KAAE,KAAM,GAAGD,CAAG,GAAI,MAAM,oBACtBH,EAAQ,KACX,CACF,CAEJ,CAcO,SAASK,GACdC,EAAqBC,EACR,CACb,OAAAD,EAAWA,EAAS,OAAON,GAAQ,CAtFrC,IAAAQ,EAsFwC,SAACA,EAAAR,EAAQ,aAAR,MAAAQ,EAAoB,QAAM,EAE/DJ,EAAC,OAAI,MAAM,cACTA,EAAC,UACC,MAAM,sBACN,aAAYK,GAAY,gBAAgB,GAEvCF,EAAO,KACV,EACAH,EAAC,MAAG,MAAM,oBACPE,EAAS,IAAIP,EAAa,CAC7B,CACF,CAEJ,CCxBA,IAAIW,GAAW,EAiBR,SAASC,GACdC,EAAiBC,EACI,CACrB,SAAS,KAAK,OAAOD,CAAE,EAGvB,GAAM,CAAE,MAAAE,CAAM,EAAIC,GAAeH,CAAE,EACnCA,EAAG,MAAM,YAAY,qBAAsB,GAAGE,CAAK,IAAI,EACvDF,EAAG,OAAO,EAGV,IAAMI,EAAYC,GAAoBJ,CAAI,EACpCK,EACJ,OAAOF,GAAc,YACjBG,GAA0BH,CAAS,EACnCI,EAAG,CAAE,EAAG,EAAG,EAAG,CAAE,CAAC,EAGjBC,EAAUC,EACdC,GAAkBV,CAAI,EACtBW,GAAkBX,CAAI,CACxB,EACG,KACCY,EAAqB,CACvB,EAGF,OAAOC,EAAc,CAACL,EAASH,CAAO,CAAC,EACpC,KACCS,EAAI,CAAC,CAACC,EAAQC,CAAM,IAAM,CACxB,GAAI,CAAE,EAAAC,EAAG,EAAAC,CAAE,EAAIC,GAAiBnB,CAAI,EAC9BoB,EAAOlB,GAAeF,CAAI,EAU1BqB,EAAQrB,EAAK,QAAQ,OAAO,EAClC,OAAIqB,GAASrB,EAAK,gBAChBiB,GAAKI,EAAM,WAAarB,EAAK,cAAc,WAC3CkB,GAAKG,EAAM,UAAarB,EAAK,cAAc,WAEtC,CACL,OAAAe,EACA,OAAQ,CACN,EAAGE,EAAID,EAAO,EAAII,EAAK,MAAS,EAAInB,EAAQ,EAC5C,EAAGiB,EAAIF,EAAO,EAAII,EAAK,OAAS,CAClC,CACF,CACF,CAAC,CACH,CACJ,CASO,SAASE,GACdvB,EACgC,CAChC,IAAMwB,EAAQxB,EAAG,MACjB,GAAI,CAACwB,EAAM,OACT,OAAOC,EAGT,IAAMC,EAAK,aAAa5B,IAAU,GAC5B6B,EAAUC,GAAcF,EAAI,QAAQ,EACpCG,EAAUC,EAAW,cAAeH,CAAO,EACjD,OAAAE,EAAQ,UAAYL,EAGbO,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAU,CAGd,KAAK,CAAE,OAAAE,CAAO,EAAG,CACfP,EAAQ,MAAM,YAAY,iBAAkB,GAAGO,EAAO,CAAC,IAAI,EAC3DP,EAAQ,MAAM,YAAY,iBAAkB,GAAGO,EAAO,CAAC,IAAI,CAC7D,EAGA,UAAW,CACTP,EAAQ,MAAM,eAAe,gBAAgB,EAC7CA,EAAQ,MAAM,eAAe,gBAAgB,CAC/C,CACF,CAAC,EAGDjB,EACEsB,EAAM,KAAKG,EAAO,CAAC,CAAE,OAAAnB,CAAO,IAAMA,CAAM,CAAC,EACzCgB,EAAM,KAAKI,GAAa,GAAG,EAAGD,EAAO,CAAC,CAAE,OAAAnB,CAAO,IAAM,CAACA,CAAM,CAAC,CAC/D,EACG,UAAU,CAGT,KAAK,CAAE,OAAAA,CAAO,EAAG,CACXA,GACFhB,EAAG,sBAAsB,WAAY2B,CAAO,EAC5C3B,EAAG,aAAa,mBAAoB0B,CAAE,EACtC1B,EAAG,gBAAgB,OAAO,IAE1B2B,EAAQ,OAAO,EACf3B,EAAG,gBAAgB,kBAAkB,EACrCA,EAAG,aAAa,QAASwB,CAAK,EAElC,EAGA,UAAW,CACTG,EAAQ,OAAO,EACf3B,EAAG,gBAAgB,kBAAkB,EACrCA,EAAG,aAAa,QAASwB,CAAK,CAChC,CACF,CAAC,EAGHQ,EACG,KACCK,GAAU,GAAIC,EAAuB,CACvC,EACG,UAAU,CAAC,CAAE,OAAAtB,CAAO,IAAM,CACzBW,EAAQ,UAAU,OAAO,qBAAsBX,CAAM,CACvD,CAAC,EAMLgB,EACG,KACCO,GAAa,IAAKD,EAAuB,EACzCH,EAAO,IAAM,CAAC,CAACnC,EAAG,YAAY,EAC9Be,EAAI,IAAMf,EAAG,aAAc,sBAAsB,CAAC,EAClDe,EAAI,CAAC,CAAE,EAAAG,CAAE,IAAMA,CAAC,CAClB,EACC,UAAU,CAGT,KAAKsB,EAAQ,CACPA,EACFb,EAAQ,MAAM,YAAY,iBAAkB,GAAG,CAACa,CAAM,IAAI,EAE1Db,EAAQ,MAAM,eAAe,gBAAgB,CACjD,EAGA,UAAW,CACTA,EAAQ,MAAM,eAAe,gBAAgB,CAC/C,CACF,CAAC,EAGI5B,GAAa4B,EAAS3B,CAAE,EAC5B,KACCyC,EAAIC,GAASV,EAAM,KAAKU,CAAK,CAAC,EAC9BC,EAAS,IAAMX,EAAM,SAAS,CAAC,EAC/BjB,EAAI2B,GAAUE,EAAA,CAAE,IAAK5C,GAAO0C,EAAQ,CACtC,CACJ,CAAC,EACE,KACCG,GAAYC,EAAc,CAC5B,CACJ,CC1KO,SAASC,GACdC,EAAiBC,EACO,CACxB,IAAMC,EAAUC,EAAM,IAAMC,EAAc,CACxCC,GAAmBL,CAAE,EACrBM,GAA0BL,CAAS,CACrC,CAAC,CAAC,EACC,KACCM,EAAI,CAAC,CAAC,CAAE,EAAAC,EAAG,EAAAC,CAAE,EAAGC,CAAM,IAAqB,CACzC,GAAM,CAAE,MAAAC,EAAO,OAAAC,CAAO,EAAIC,GAAeb,CAAE,EAC3C,MAAQ,CACN,EAAGQ,EAAIE,EAAO,EAAIC,EAAS,EAC3B,EAAGF,EAAIC,EAAO,EAAIE,EAAS,CAC7B,CACF,CAAC,CACH,EAGF,OAAOE,GAAkBd,CAAE,EACxB,KACCe,EAAUC,GAAUd,EACjB,KACCK,EAAIU,IAAW,CAAE,OAAAD,EAAQ,OAAAC,CAAO,EAAE,EAClCC,GAAK,CAAC,CAACF,GAAU,GAAQ,CAC3B,CACF,CACF,CACJ,CAWO,SAASG,GACdnB,EAAiBC,EAAwB,CAAE,QAAAmB,CAAQ,EAChB,CACnC,GAAM,CAACC,EAASC,CAAK,EAAI,MAAM,KAAKtB,EAAG,QAAQ,EAG/C,OAAOG,EAAM,IAAM,CACjB,IAAMoB,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,GAAe,EAAGC,GAAQ,EAAI,CAAC,EACxD,OAAAJ,EAAM,UAAU,CAGd,KAAK,CAAE,OAAAN,CAAO,EAAG,CACfjB,EAAG,MAAM,YAAY,iBAAkB,GAAGiB,EAAO,CAAC,IAAI,EACtDjB,EAAG,MAAM,YAAY,iBAAkB,GAAGiB,EAAO,CAAC,IAAI,CACxD,EAGA,UAAW,CACTjB,EAAG,MAAM,eAAe,gBAAgB,EACxCA,EAAG,MAAM,eAAe,gBAAgB,CAC1C,CACF,CAAC,EAGD4B,GAAuB5B,CAAE,EACtB,KACC6B,EAAUJ,CAAK,CACjB,EACG,UAAUK,GAAW,CACpB9B,EAAG,gBAAgB,kBAAmB8B,CAAO,CAC/C,CAAC,EAGLC,EACER,EAAM,KAAKS,EAAO,CAAC,CAAE,OAAAhB,CAAO,IAAMA,CAAM,CAAC,EACzCO,EAAM,KAAKU,GAAa,GAAG,EAAGD,EAAO,CAAC,CAAE,OAAAhB,CAAO,IAAM,CAACA,CAAM,CAAC,CAC/D,EACG,UAAU,CAGT,KAAK,CAAE,OAAAA,CAAO,EAAG,CACXA,EACFhB,EAAG,QAAQqB,CAAO,EAElBA,EAAQ,OAAO,CACnB,EAGA,UAAW,CACTrB,EAAG,QAAQqB,CAAO,CACpB,CACF,CAAC,EAGHE,EACG,KACCW,GAAU,GAAIC,EAAuB,CACvC,EACG,UAAU,CAAC,CAAE,OAAAnB,CAAO,IAAM,CACzBK,EAAQ,UAAU,OAAO,qBAAsBL,CAAM,CACvD,CAAC,EAGLO,EACG,KACCa,GAAa,IAAKD,EAAuB,EACzCH,EAAO,IAAM,CAAC,CAAChC,EAAG,YAAY,EAC9BO,EAAI,IAAMP,EAAG,aAAc,sBAAsB,CAAC,EAClDO,EAAI,CAAC,CAAE,EAAAC,CAAE,IAAMA,CAAC,CAClB,EACG,UAAU,CAGT,KAAK6B,EAAQ,CACPA,EACFrC,EAAG,MAAM,YAAY,iBAAkB,GAAG,CAACqC,CAAM,IAAI,EAErDrC,EAAG,MAAM,eAAe,gBAAgB,CAC5C,EAGA,UAAW,CACTA,EAAG,MAAM,eAAe,gBAAgB,CAC1C,CACF,CAAC,EAGLsC,EAAsBhB,EAAO,OAAO,EACjC,KACCO,EAAUJ,CAAK,EACfO,EAAOO,GAAM,EAAEA,EAAG,SAAWA,EAAG,QAAQ,CAC1C,EACG,UAAUA,GAAM,CACfA,EAAG,gBAAgB,EACnBA,EAAG,eAAe,CACpB,CAAC,EAGLD,EAAsBhB,EAAO,WAAW,EACrC,KACCO,EAAUJ,CAAK,EACfe,GAAejB,CAAK,CACtB,EACG,UAAU,CAAC,CAACgB,EAAI,CAAE,OAAAvB,CAAO,CAAC,IAAM,CA3OzC,IAAAyB,EA8OU,GAAIF,EAAG,SAAW,GAAKA,EAAG,SAAWA,EAAG,QACtCA,EAAG,eAAe,UAGTvB,EAAQ,CACjBuB,EAAG,eAAe,EAGlB,IAAMG,EAAS1C,EAAG,cAAe,QAAQ,gBAAgB,EACrD0C,aAAkB,YACpBA,EAAO,MAAM,GAEbD,EAAAE,GAAiB,IAAjB,MAAAF,EAAoB,MACxB,CACF,CAAC,EAGLrB,EACG,KACCS,EAAUJ,CAAK,EACfO,EAAOY,GAAUA,IAAWvB,CAAO,EACnCwB,GAAM,GAAG,CACX,EACG,UAAU,IAAM7C,EAAG,MAAM,CAAC,EAGxBD,GAAgBC,EAAIC,CAAS,EACjC,KACC6C,EAAIC,GAASxB,EAAM,KAAKwB,CAAK,CAAC,EAC9BC,EAAS,IAAMzB,EAAM,SAAS,CAAC,EAC/BhB,EAAIwC,GAAUE,EAAA,CAAE,IAAKjD,GAAO+C,EAAQ,CACtC,CACJ,CAAC,CACH,CCxMA,SAASG,GAAUC,EAAuC,CACxD,OAAOA,EAAU,UAAY,OACzBC,EAAY,eAAgBD,CAAS,EACrC,CAACA,CAAS,CAChB,CASA,SAASE,GAAYF,EAAgC,CACnD,IAAMG,EAAkB,CAAC,EACzB,QAAWC,KAAML,GAAUC,CAAS,EAAG,CACrC,IAAMK,EAAgB,CAAC,EAGjBC,EAAK,SAAS,mBAAmBF,EAAI,WAAW,SAAS,EAC/D,QAASG,EAAOD,EAAG,SAAS,EAAGC,EAAMA,EAAOD,EAAG,SAAS,EACtDD,EAAM,KAAKE,CAAY,EAGzB,QAASC,KAAQH,EAAO,CACtB,IAAII,EAGJ,KAAQA,EAAQ,gBAAgB,KAAKD,EAAK,WAAY,GAAI,CACxD,GAAM,CAAC,CAAEE,EAAIC,CAAK,EAAIF,EACtB,GAAI,OAAOE,GAAU,YAAa,CAChC,IAAMC,EAASJ,EAAK,UAAUC,EAAM,KAAK,EACzCD,EAAOI,EAAO,UAAUF,EAAG,MAAM,EACjCP,EAAQ,KAAKS,CAAM,CAGrB,KAAO,CACLJ,EAAK,YAAcE,EACnBP,EAAQ,KAAKK,CAAI,EACjB,KACF,CACF,CACF,CACF,CACA,OAAOL,CACT,CAQA,SAASU,GAAKC,EAAqBC,EAA2B,CAC5DA,EAAO,OAAO,GAAG,MAAM,KAAKD,EAAO,UAAU,CAAC,CAChD,CAoBO,SAASE,GACdZ,EAAiBJ,EAAwB,CAAE,QAAAiB,EAAS,OAAAC,CAAO,EACxB,CAGnC,IAAMC,EAASnB,EAAU,QAAQ,MAAM,EACjCoB,EAASD,GAAA,YAAAA,EAAQ,GAGjBE,EAAc,IAAI,IACxB,QAAWT,KAAUV,GAAYF,CAAS,EAAG,CAC3C,GAAM,CAAC,CAAEU,CAAE,EAAIE,EAAO,YAAa,MAAM,WAAW,EAChDU,GAAmB,yBAAyBZ,CAAE,IAAKN,CAAE,IACvDiB,EAAY,IAAIX,EAAIa,GAAiBb,EAAIU,CAAM,CAAC,EAChDR,EAAO,YAAYS,EAAY,IAAIX,CAAE,CAAE,EAE3C,CAGA,OAAIW,EAAY,OAAS,EAChBG,EAGFC,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,GAAe,EAAGC,GAAQ,EAAI,CAAC,EAGlDC,EAAsC,CAAC,EAC7C,OAAW,CAACrB,EAAIsB,CAAU,IAAKX,EAC7BU,EAAM,KAAK,CACTE,EAAW,cAAeD,CAAU,EACpCC,EAAW,yBAAyBvB,CAAE,IAAKN,CAAE,CAC/C,CAAC,EAGH,OAAAc,EAAO,KAAKgB,EAAUN,CAAK,CAAC,EACzB,UAAUO,GAAU,CACnB/B,EAAG,OAAS,CAAC+B,EAGb/B,EAAG,UAAU,OAAO,qBAAsB+B,CAAM,EAGhD,OAAW,CAACC,EAAOC,CAAK,IAAKN,EACtBI,EAGHtB,GAAKuB,EAAOC,CAAK,EAFjBxB,GAAKwB,EAAOD,CAAK,CAGvB,CAAC,EAGIE,EAAM,GAAG,CAAC,GAAGjB,CAAW,EAC5B,IAAI,CAAC,CAAC,CAAEW,CAAU,IACjBO,GAAgBP,EAAYhC,EAAW,CAAE,QAAAiB,CAAQ,CAAC,CACnD,CACH,EACG,KACCuB,EAAS,IAAMd,EAAM,SAAS,CAAC,EAC/Be,GAAM,CACR,CACJ,CAAC,CACH,CC7JA,SAASC,GAASC,EAA0C,CAC1D,GAAIA,EAAG,mBAAoB,CACzB,IAAMC,EAAUD,EAAG,mBACnB,GAAIC,EAAQ,UAAY,KACtB,OAAOA,EAGJ,GAAIA,EAAQ,UAAY,KAAO,CAACA,EAAQ,SAAS,OACpD,OAAOF,GAASE,CAAO,CAC3B,CAIF,CAcO,SAASC,GACdF,EAAiBG,EACkB,CACnC,OAAOC,EAAM,IAAM,CACjB,IAAMC,EAAON,GAASC,CAAE,EACxB,OAAO,OAAOK,GAAS,YACnBC,GAAoBD,EAAML,EAAIG,CAAO,EACrCI,CACN,CAAC,CACH,CCjEA,IAAAC,GAAwB,SA2ExB,IAAIC,GAAW,EAaf,SAASC,GAAkBC,EAA0C,CACnE,GAAIA,EAAG,mBAAoB,CACzB,IAAMC,EAAUD,EAAG,mBACnB,GAAIC,EAAQ,UAAY,KACtB,OAAOA,EAGJ,GAAIA,EAAQ,UAAY,KAAO,CAACA,EAAQ,SAAS,OACpD,OAAOF,GAAkBE,CAAO,CACpC,CAIF,CAgBO,SAASC,GACdF,EACsB,CACtB,OAAOG,GAAiBH,CAAE,EACvB,KACCI,EAAI,CAAC,CAAE,MAAAC,CAAM,KAEJ,CACL,WAFcC,GAAsBN,CAAE,EAElB,MAAQK,CAC9B,EACD,EACDE,EAAwB,YAAY,CACtC,CACJ,CAoBO,SAASC,GACdR,EAAiBS,EACiB,CAClC,GAAM,CAAE,QAASC,CAAM,EAAI,WAAW,SAAS,EAGzCC,EAAWC,EAAM,IAAM,CAC3B,IAAMC,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,GAAS,CAAC,CAAC,EACpCH,EAAM,UAAU,CAAC,CAAE,WAAAI,CAAW,IAAM,CAC9BA,GAAcP,EAChBV,EAAG,aAAa,WAAY,GAAG,EAE/BA,EAAG,gBAAgB,UAAU,CACjC,CAAC,EAGD,IAAMkB,EAAoD,CAAC,EAC3D,GAAI,GAAAC,QAAY,YAAY,IACtBnB,EAAG,QAAQ,OAAO,GACpBoB,EAAQ,mBAAmB,GAAK,CAACpB,EAAG,QAAQ,UAAU,GACrD,CACD,IAAMqB,EAASrB,EAAG,QAAQ,KAAK,EAC/BqB,EAAO,GAAK,UAAUvB,IAAU,GAGhC,IAAMwB,EAASC,GAAsBF,EAAO,EAAE,EAC9CA,EAAO,aAAaC,EAAQtB,CAAE,EAC1BoB,EAAQ,kBAAkB,GAC5BF,EAAS,KAAKM,GAAaF,CAAM,CAAC,CACtC,CAIF,IAAMG,EAAYzB,EAAG,QAAQ,YAAY,EACzC,GAAIyB,aAAqB,YAAa,CACpC,IAAMC,EAAO3B,GAAkB0B,CAAS,EAGxC,GAAI,OAAOC,GAAS,cAClBD,EAAU,UAAU,SAAS,UAAU,GACvCL,EAAQ,uBAAuB,GAC9B,CACD,IAAMO,EAAeC,GAAoBF,EAAM1B,EAAIS,CAAO,EAC1DS,EAAS,KACPf,GAAiBsB,CAAS,EACvB,KACCI,EAAUd,CAAK,EACfX,EAAI,CAAC,CAAE,MAAAC,EAAO,OAAAyB,CAAO,IAAMzB,GAASyB,CAAM,EAC1CC,EAAqB,EACrBC,EAAUC,GAAUA,EAASN,EAAeO,CAAK,CACnD,CACJ,CACF,CACF,CAGA,OAAOhC,GAAeF,CAAE,EACrB,KACCmC,EAAIC,GAASvB,EAAM,KAAKuB,CAAK,CAAC,EAC9BC,EAAS,IAAMxB,EAAM,SAAS,CAAC,EAC/BT,EAAIgC,GAAUE,EAAA,CAAE,IAAKtC,GAAOoC,EAAQ,EACpCG,GAAU,GAAGrB,CAAQ,CACvB,CACJ,CAAC,EAGD,OAAIE,EAAQ,cAAc,EACjBoB,GAAuBxC,CAAE,EAC7B,KACCyC,EAAOC,GAAWA,CAAO,EACzBC,GAAK,CAAC,EACNX,EAAU,IAAMrB,CAAQ,CAC1B,EAGGA,CACT,CC1KO,SAASiC,GACdC,EAAwB,CAAE,QAAAC,EAAS,OAAAC,CAAO,EACrB,CACrB,IAAIC,EAAO,GACX,OAAOC,EAGLH,EACG,KACCI,EAAIC,GAAUA,EAAO,QAAQ,qBAAqB,CAAE,EACpDC,EAAOC,GAAWR,IAAOQ,CAAO,EAChCH,EAAI,KAAO,CACT,OAAQ,OAAQ,OAAQ,EAC1B,EAAa,CACf,EAGFH,EACG,KACCK,EAAOE,GAAUA,GAAU,CAACN,CAAI,EAChCO,EAAI,IAAMP,EAAOH,EAAG,IAAI,EACxBK,EAAII,IAAW,CACb,OAAQA,EAAS,OAAS,OAC5B,EAAa,CACf,CACJ,CACF,CAaO,SAASE,GACdX,EAAwBY,EACQ,CAChC,OAAOC,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAU,CAAC,CAAE,OAAAE,EAAQ,OAAAC,CAAO,IAAM,CACtCjB,EAAG,gBAAgB,OAAQgB,IAAW,MAAM,EACxCC,GACFjB,EAAG,eAAe,CACtB,CAAC,EAGMD,GAAaC,EAAIY,CAAO,EAC5B,KACCF,EAAIQ,GAASJ,EAAM,KAAKI,CAAK,CAAC,EAC9BC,EAAS,IAAML,EAAM,SAAS,CAAC,EAC/BT,EAAIa,GAAUE,EAAA,CAAE,IAAKpB,GAAOkB,EAAQ,CACtC,CACJ,CAAC,CACH,CCzIA,IAAAG,GAAA,ktLCqDA,IAAIC,GAKAC,GAAW,EAWf,SAASC,IAAiC,CACxC,OAAO,OAAO,SAAY,aAAe,mBAAmB,QACxDC,GAAY,sDAAsD,EAClEC,EAAG,MAAS,CAClB,CAaO,SAASC,GACdC,EACgC,CAChC,OAAAA,EAAG,UAAU,OAAO,SAAS,EAC7BN,QAAaE,GAAa,EACvB,KACCK,EAAI,IAAM,QAAQ,WAAW,CAC3B,YAAa,GACb,SAAAC,GACA,SAAU,CACR,cAAe,OACf,gBAAiB,OACjB,aAAc,MAChB,CACF,CAAC,CAAC,EACFC,EAAI,IAAG,EAAY,EACnBC,EAAY,CAAC,CACf,GAGFV,GAAS,UAAU,IAAYW,GAAA,sBAC7BL,EAAG,UAAU,IAAI,SAAS,EAC1B,IAAMM,EAAK,aAAaX,IAAU,GAG5BY,EAAOC,EAAE,MAAO,CAAE,MAAO,SAAU,CAAC,EACpCC,EAAOT,EAAG,YAGV,CAAE,IAAAU,EAAK,GAAAC,CAAG,EAAI,MAAM,QAAQ,OAAOL,EAAIG,CAAI,EAG3CG,EAASL,EAAK,aAAa,CAAE,KAAM,QAAS,CAAC,EACnDK,EAAO,UAAYF,EAGnBV,EAAG,YAAYO,CAAI,EACnBI,GAAA,MAAAA,EAAKC,EACP,EAAC,EAGMlB,GACJ,KACCS,EAAI,KAAO,CAAE,IAAKH,CAAG,EAAE,CACzB,CACJ,CCtFA,IAAMa,GAAWC,EAAE,OAAO,EAgBnB,SAASC,GACdC,EACkC,CAClC,OAAAA,EAAG,YAAYH,EAAQ,EACvBA,GAAS,YAAYI,GAAYD,CAAE,CAAC,EAG7BE,EAAG,CAAE,IAAKF,CAAG,CAAC,CACvB,CC2BO,SAASG,GACdC,EACyB,CACzB,IAAMC,EAAUD,EAAO,KAAKE,GAASA,EAAM,OAAO,GAAKF,EAAO,CAAC,EAC/D,OAAOG,EAAM,GAAGH,EAAO,IAAIE,GAASE,EAAUF,EAAO,QAAQ,EAC1D,KACCG,EAAI,IAAMC,EAA6B,cAAcJ,EAAM,EAAE,IAAI,CAAC,CACpE,CACF,CAAC,EACE,KACCK,EAAUD,EAA6B,cAAcL,EAAQ,EAAE,IAAI,CAAC,EACpEI,EAAIG,IAAW,CAAE,OAAAA,CAAO,EAAE,CAC5B,CACJ,CAUO,SAASC,GACdC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EACF,CACpC,IAAMC,EAAYP,EAAW,iBAAkBI,CAAE,EAC3CV,EAASc,EAA8B,iBAAkBJ,CAAE,EAG3DK,EAAOC,GAAoB,MAAM,EACvCN,EAAG,OAAOK,CAAI,EAGd,IAAME,EAAOD,GAAoB,MAAM,EACvC,OAAAN,EAAG,OAAOO,CAAI,EAGPC,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,GAAe,EAAGC,GAAQ,EAAI,CAAC,EACxDC,EAAc,CAACL,EAAOM,GAAiBf,CAAE,CAAC,CAAC,EACxC,KACCgB,EAAUL,CAAK,EACfM,GAAU,EAAGC,EAAuB,CACtC,EACG,UAAU,CAGT,KAAK,CAAC,CAAE,OAAApB,CAAO,EAAGqB,CAAI,EAAG,CACvB,IAAMC,EAASC,GAAiBvB,CAAM,EAChC,CAAE,MAAAwB,CAAM,EAAIC,GAAezB,CAAM,EAGvCE,EAAG,MAAM,YAAY,mBAAoB,GAAGoB,EAAO,CAAC,IAAI,EACxDpB,EAAG,MAAM,YAAY,uBAAwB,GAAGsB,CAAK,IAAI,EAGzD,IAAME,EAAUC,GAAwBtB,CAAS,GAE/CiB,EAAO,EAAYI,EAAQ,GAC3BJ,EAAO,EAAIE,EAAQE,EAAQ,EAAIL,EAAK,QAEpChB,EAAU,SAAS,CACjB,KAAM,KAAK,IAAI,EAAGiB,EAAO,EAAI,EAAE,EAC/B,SAAU,QACZ,CAAC,CACL,EAGA,UAAW,CACTpB,EAAG,MAAM,eAAe,kBAAkB,EAC1CA,EAAG,MAAM,eAAe,sBAAsB,CAChD,CACF,CAAC,EAGLc,EAAc,CACZY,GAA0BvB,CAAS,EACnCY,GAAiBZ,CAAS,CAC5B,CAAC,EACE,KACCa,EAAUL,CAAK,CACjB,EACG,UAAU,CAAC,CAACS,EAAQD,CAAI,IAAM,CAC7B,IAAMK,EAAUG,GAAsBxB,CAAS,EAC/CE,EAAK,OAASe,EAAO,EAAI,GACzBb,EAAK,OAASa,EAAO,EAAII,EAAQ,MAAQL,EAAK,MAAQ,EACxD,CAAC,EAGL1B,EACEC,EAAUW,EAAM,OAAO,EAAE,KAAKV,EAAI,IAAM,EAAE,CAAC,EAC3CD,EAAUa,EAAM,OAAO,EAAE,KAAKZ,EAAI,IAAM,CAAE,CAAC,CAC7C,EACG,KACCqB,EAAUL,CAAK,CACjB,EACG,UAAUiB,GAAa,CACtB,GAAM,CAAE,MAAAN,CAAM,EAAIC,GAAepB,CAAS,EAC1CA,EAAU,SAAS,CACjB,KAAMmB,EAAQM,EACd,SAAU,QACZ,CAAC,CACH,CAAC,EAGL1B,EACG,KACCc,EAAUL,CAAK,EACfkB,EAAOrC,GAASF,EAAO,SAASE,CAAyB,CAAC,CAC5D,EACG,UAAUA,GAASA,EAAM,MAAM,CAAC,EAGrCW,EAAU,UAAU,IAAI,uBAAuB,EAC/C,QAAWX,KAASF,EAAQ,CAC1B,IAAMwC,EAAQlC,EAA6B,cAAcJ,EAAM,EAAE,IAAI,EACrEsC,EAAM,gBAAgBC,EAAE,IAAK,CAC3B,KAAM,IAAID,EAAM,OAAO,GACvB,SAAU,EACZ,EAAG,GAAG,MAAM,KAAKA,EAAM,UAAU,CAAC,CAAC,EAGnCpC,EAAsBoC,EAAM,kBAAoB,OAAO,EACpD,KACCd,EAAUL,CAAK,EACfkB,EAAOG,GAAM,EAAEA,EAAG,SAAWA,EAAG,QAAQ,EACxCC,EAAID,GAAM,CACRA,EAAG,eAAe,EAClBA,EAAG,gBAAgB,CACrB,CAAC,CACH,EAEG,UAAU,IAAM,CACf,QAAQ,aAAa,CAAC,EAAG,GAAI,IAAIF,EAAM,OAAO,EAAE,EAChDA,EAAM,MAAM,CACd,CAAC,CACP,CAGA,OAAII,EAAQ,mBAAmB,GAC7BzB,EAAM,KACJ0B,GAAK,CAAC,EACNC,GAAenC,CAAS,CAC1B,EACG,UAAU,CAAC,CAAC,CAAE,OAAAH,CAAO,EAAG,CAAE,OAAAsB,CAAO,CAAC,IAAM,CACvC,IAAMiB,EAAMvC,EAAO,UAAU,KAAK,EAClC,GAAIA,EAAO,aAAa,mBAAmB,EACzCA,EAAO,gBAAgB,mBAAmB,MAGrC,CACL,IAAMwC,EAAItC,EAAG,UAAYoB,EAAO,EAGhC,QAAWmB,KAAOnC,EAAY,aAAa,EACzC,QAAWZ,KAASY,EAClB,iBAAkBmC,CACpB,EAAG,CACD,IAAMT,EAAQlC,EAAW,cAAcJ,EAAM,EAAE,IAAI,EACnD,GACEsC,IAAUhC,GACVgC,EAAM,UAAU,KAAK,IAAMO,EAC3B,CACAP,EAAM,aAAa,oBAAqB,EAAE,EAC1CtC,EAAM,MAAM,EACZ,KACF,CACF,CAGF,OAAO,SAAS,CACd,IAAKQ,EAAG,UAAYsC,CACtB,CAAC,EAGD,IAAME,EAAO,SAAmB,QAAQ,GAAK,CAAC,EAC9C,SAAS,SAAU,CAAC,GAAG,IAAI,IAAI,CAACH,EAAK,GAAGG,CAAI,CAAC,CAAC,CAAC,CACjD,CACF,CAAC,EAGL/B,EAAM,KAAKO,EAAUL,CAAK,CAAC,EACxB,UAAU,IAAM,CACf,QAAW8B,KAASrC,EAA8B,eAAgBJ,CAAE,EAClEyC,EAAM,MAAM,CAChB,CAAC,EAGIpD,GAAiBC,CAAM,EAC3B,KACC2C,EAAIS,GAASjC,EAAM,KAAKiC,CAAK,CAAC,EAC9BC,EAAS,IAAMlC,EAAM,SAAS,CAAC,EAC/Bd,EAAI+C,GAAUE,EAAA,CAAE,IAAK5C,GAAO0C,EAAQ,CACtC,CACJ,CAAC,EACE,KACCG,GAAYC,EAAc,CAC5B,CACJ,CCnMO,SAASC,GACdC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,EAAS,OAAAC,CAAO,EACd,CAChC,OAAOC,EAGL,GAAGC,EAAY,4BAA6BL,CAAE,EAC3C,IAAIM,GAASC,GAAqBD,EAAO,CAAE,QAAAJ,EAAS,OAAAC,CAAO,CAAC,CAAC,EAGhE,GAAGE,EAAY,2BAA4BL,CAAE,EAC1C,IAAIM,GAASE,GAAeF,EAAO,CAAE,QAAAJ,EAAS,OAAAC,CAAO,CAAC,CAAC,EAG1D,GAAGE,EAAY,cAAeL,CAAE,EAC7B,IAAIM,GAASG,GAAaH,CAAK,CAAC,EAGnC,GAAGD,EAAY,qBAAsBL,CAAE,EACpC,IAAIM,GAASI,GAAeJ,CAAK,CAAC,EAGrC,GAAGD,EAAY,UAAWL,CAAE,EACzB,IAAIM,GAASK,GAAaL,EAAO,CAAE,QAAAJ,EAAS,OAAAC,CAAO,CAAC,CAAC,EAGxD,GAAGE,EAAY,cAAeL,CAAE,EAC7B,IAAIM,GAASM,GAAiBN,EAAO,CAAE,UAAAL,EAAW,QAAAC,CAAQ,CAAC,CAAC,EAG/D,GAAGG,EAAY,UAAWL,CAAE,EACzB,OAAO,IAAMa,EAAQ,kBAAkB,CAAC,EACxC,IAAIP,GAASQ,GAAaR,CAAK,CAAC,CACrC,CACF,CCtDO,SAASS,GACdC,EAAkB,CAAE,OAAAC,CAAO,EACP,CACpB,OAAOA,EACJ,KACCC,EAAUC,GAAWC,EACnBC,EAAG,EAAI,EACPA,EAAG,EAAK,EAAE,KAAKC,GAAM,GAAI,CAAC,CAC5B,EACG,KACCC,EAAIC,IAAW,CAAE,QAAAL,EAAS,OAAAK,CAAO,EAAE,CACrC,CACF,CACF,CACJ,CAaO,SAASC,GACdC,EAAiBC,EACc,CAC/B,IAAMC,EAAQC,EAAW,cAAeH,CAAE,EAC1C,OAAOI,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAU,CAAC,CAAE,QAAAZ,EAAS,OAAAK,CAAO,IAAM,CACvCE,EAAG,UAAU,OAAO,oBAAqBF,CAAM,EAC/CI,EAAM,YAAcT,CACtB,CAAC,EAGMJ,GAAYW,EAAIC,CAAO,EAC3B,KACCM,EAAIC,GAASH,EAAM,KAAKG,CAAK,CAAC,EAC9BC,EAAS,IAAMJ,EAAM,SAAS,CAAC,EAC/BR,EAAIW,GAAUE,EAAA,CAAE,IAAKV,GAAOQ,EAAQ,CACtC,CACJ,CAAC,CACH,CCrBA,SAASG,GAAS,CAAE,UAAAC,CAAU,EAAsC,CAClE,GAAI,CAACC,EAAQ,iBAAiB,EAC5B,OAAOC,EAAG,EAAK,EAGjB,IAAMC,EAAaH,EAChB,KACCI,EAAI,CAAC,CAAE,OAAQ,CAAE,EAAAC,CAAE,CAAE,IAAMA,CAAC,EAC5BC,GAAY,EAAG,CAAC,EAChBF,EAAI,CAAC,CAACG,EAAGC,CAAC,IAAM,CAACD,EAAIC,EAAGA,CAAC,CAAU,EACnCC,EAAwB,CAAC,CAC3B,EAGIC,EAAUC,EAAc,CAACX,EAAWG,CAAU,CAAC,EAClD,KACCS,EAAO,CAAC,CAAC,CAAE,OAAAC,CAAO,EAAG,CAAC,CAAER,CAAC,CAAC,IAAM,KAAK,IAAIA,EAAIQ,EAAO,CAAC,EAAI,GAAG,EAC5DT,EAAI,CAAC,CAAC,CAAE,CAACU,CAAS,CAAC,IAAMA,CAAS,EAClCC,EAAqB,CACvB,EAGIC,EAAUC,GAAY,QAAQ,EACpC,OAAON,EAAc,CAACX,EAAWgB,CAAO,CAAC,EACtC,KACCZ,EAAI,CAAC,CAAC,CAAE,OAAAS,CAAO,EAAGK,CAAM,IAAML,EAAO,EAAI,KAAO,CAACK,CAAM,EACvDH,EAAqB,EACrBI,EAAUC,GAAUA,EAASV,EAAUR,EAAG,EAAK,CAAC,EAChDmB,EAAU,EAAK,CACjB,CACJ,CAcO,SAASC,GACdC,EAAiBC,EACG,CACpB,OAAOC,EAAM,IAAMd,EAAc,CAC/Be,GAAiBH,CAAE,EACnBxB,GAASyB,CAAO,CAClB,CAAC,CAAC,EACC,KACCpB,EAAI,CAAC,CAAC,CAAE,OAAAuB,CAAO,EAAGC,CAAM,KAAO,CAC7B,OAAAD,EACA,OAAAC,CACF,EAAE,EACFb,EAAqB,CAACR,EAAGC,IACvBD,EAAE,SAAWC,EAAE,QACfD,EAAE,SAAWC,EAAE,MAChB,EACDqB,EAAY,CAAC,CACf,CACJ,CAaO,SAASC,GACdP,EAAiB,CAAE,QAAAQ,EAAS,MAAAC,CAAM,EACO,CACzC,OAAOP,EAAM,IAAM,CACjB,IAAMQ,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,GAAe,EAAGC,GAAQ,EAAI,CAAC,EACxDJ,EACG,KACCxB,EAAwB,QAAQ,EAChC6B,GAAkBP,CAAO,CAC3B,EACG,UAAU,CAAC,CAAC,CAAE,OAAAX,CAAO,EAAG,CAAE,OAAAQ,CAAO,CAAC,IAAM,CACvCL,EAAG,UAAU,OAAO,oBAAqBH,GAAU,CAACQ,CAAM,EAC1DL,EAAG,OAASK,CACd,CAAC,EAGL,IAAMW,EAAWC,GAAKC,EAAY,UAAWlB,CAAE,CAAC,EAC7C,KACCX,EAAO,IAAMX,EAAQ,kBAAkB,CAAC,EACxCyC,GAASC,GAASC,GAAaD,CAAK,CAAC,CACvC,EAGF,OAAAX,EAAM,UAAUC,CAAK,EAGdF,EACJ,KACCc,EAAUV,CAAK,EACf/B,EAAI0C,GAAUC,EAAA,CAAE,IAAKxB,GAAOuB,EAAQ,EACpCE,GAAUT,EAAS,KAAKM,EAAUV,CAAK,CAAC,CAAC,CAC3C,CACJ,CAAC,CACH,CCjIO,SAASc,GACdC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EACb,CACzB,OAAOC,GAAgBH,EAAI,CAAE,UAAAC,EAAW,QAAAC,CAAQ,CAAC,EAC9C,KACCE,EAAI,CAAC,CAAE,OAAQ,CAAE,EAAAC,CAAE,CAAE,IAAM,CACzB,GAAM,CAAE,OAAAC,CAAO,EAAIC,GAAeP,CAAE,EACpC,MAAO,CACL,OAAQK,GAAKC,CACf,CACF,CAAC,EACDE,EAAwB,QAAQ,CAClC,CACJ,CAaO,SAASC,GACdT,EAAiBU,EACmB,CACpC,OAAOC,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClBD,EAAM,UAAU,CAGd,KAAK,CAAE,OAAAE,CAAO,EAAG,CACfd,EAAG,UAAU,OAAO,2BAA4Bc,CAAM,CACxD,EAGA,UAAW,CACTd,EAAG,UAAU,OAAO,0BAA0B,CAChD,CACF,CAAC,EAGD,IAAMe,EAAUC,GAAmB,gBAAgB,EACnD,OAAI,OAAOD,GAAY,YACdE,EAGFlB,GAAiBgB,EAASL,CAAO,EACrC,KACCQ,EAAIC,GAASP,EAAM,KAAKO,CAAK,CAAC,EAC9BC,EAAS,IAAMR,EAAM,SAAS,CAAC,EAC/BR,EAAIe,GAAUE,EAAA,CAAE,IAAKrB,GAAOmB,EAAQ,CACtC,CACJ,CAAC,CACH,CChEO,SAASG,GACdC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EACpB,CAGlB,IAAMC,EAAUD,EACb,KACCE,EAAI,CAAC,CAAE,OAAAC,CAAO,IAAMA,CAAM,EAC1BC,EAAqB,CACvB,EAGIC,EAAUJ,EACb,KACCK,EAAU,IAAMC,GAAiBT,CAAE,EAChC,KACCI,EAAI,CAAC,CAAE,OAAAC,CAAO,KAAO,CACnB,IAAQL,EAAG,UACX,OAAQA,EAAG,UAAYK,CACzB,EAAE,EACFK,EAAwB,QAAQ,CAClC,CACF,CACF,EAGF,OAAOC,EAAc,CAACR,EAASI,EAASN,CAAS,CAAC,EAC/C,KACCG,EAAI,CAAC,CAACQ,EAAQ,CAAE,IAAAC,EAAK,OAAAC,CAAO,EAAG,CAAE,OAAQ,CAAE,EAAAC,CAAE,EAAG,KAAM,CAAE,OAAAV,CAAO,CAAE,CAAC,KAChEA,EAAS,KAAK,IAAI,EAAGA,EACjB,KAAK,IAAI,EAAGQ,EAASE,EAAIH,CAAM,EAC/B,KAAK,IAAI,EAAGP,EAASU,EAAID,CAAM,CACnC,EACO,CACL,OAAQD,EAAMD,EACd,OAAAP,EACA,OAAQQ,EAAMD,GAAUG,CAC1B,EACD,EACDT,EAAqB,CAACU,EAAGC,IACvBD,EAAE,SAAWC,EAAE,QACfD,EAAE,SAAWC,EAAE,QACfD,EAAE,SAAWC,EAAE,MAChB,CACH,CACJ,CCxCO,SAASC,GACdC,EACqB,CACrB,IAAMC,EAAU,SAAkB,WAAW,GAAK,CAChD,MAAOD,EAAO,UAAUE,GAAS,WAC/BA,EAAM,aAAa,qBAAqB,CAC1C,EAAE,OAAO,CACX,EAGMC,EAAQ,KAAK,IAAI,EAAG,KAAK,IAAIF,EAAQ,MAAOD,EAAO,OAAS,CAAC,CAAC,EACpE,OAAOI,EAAG,GAAGJ,CAAM,EAChB,KACCK,GAASH,GAASI,EAAUJ,EAAO,QAAQ,EAAE,KAAKK,EAAI,IAAML,CAAK,CAAC,CAAC,EACnEM,EAAUR,EAAOG,CAAK,CAAC,EACvBI,EAAIL,IAAU,CACZ,MAAOF,EAAO,QAAQE,CAAK,EAC3B,MAAO,CACL,MAASA,EAAM,aAAa,qBAAqB,EACjD,OAASA,EAAM,aAAa,sBAAsB,EAClD,QAASA,EAAM,aAAa,uBAAuB,EACnD,OAASA,EAAM,aAAa,sBAAsB,CACpD,CACF,EAAa,EACbO,EAAY,CAAC,CACf,CACJ,CASO,SAASC,GACdC,EACgC,CAChC,IAAMX,EAASY,EAA8B,QAASD,CAAE,EAClDE,EAAOC,EAAE,OAAQ,CAAE,KAAM,aAAc,CAAC,EAC9C,SAAS,KAAK,YAAYD,CAAI,EAG9B,IAAME,EAASD,EAAE,OAAQ,CAAE,KAAM,cAAe,CAAC,EACjD,SAAS,KAAK,YAAYC,CAAM,EAGhC,IAAMC,EAASC,GAAW,+BAA+B,EACzD,OAAOC,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAUE,GAAW,CAIzB,GAHA,SAAS,KAAK,aAAa,0BAA2B,EAAE,EAGpDA,EAAQ,MAAM,QAAU,yBAA0B,CACpD,IAAMC,EAAQ,WAAW,+BAA+B,EAClDpB,EAAQ,SAAS,cAAcoB,EAAM,QACvC,wDACA,sDACJ,EAGAD,EAAQ,MAAM,OAAUnB,EAAM,aAAa,sBAAsB,EACjEmB,EAAQ,MAAM,QAAUnB,EAAM,aAAa,uBAAuB,EAClEmB,EAAQ,MAAM,OAAUnB,EAAM,aAAa,sBAAsB,CACnE,CAGA,OAAW,CAACqB,EAAKC,CAAK,IAAK,OAAO,QAAQH,EAAQ,KAAK,EACrD,SAAS,KAAK,aAAa,iBAAiBE,CAAG,GAAIC,CAAK,EAG1D,QAASrB,EAAQ,EAAGA,EAAQH,EAAO,OAAQG,IAAS,CAClD,IAAMsB,EAAQzB,EAAOG,CAAK,EAAE,mBACxBsB,aAAiB,cACnBA,EAAM,OAASJ,EAAQ,QAAUlB,EACrC,CAGA,SAAS,YAAakB,CAAO,CAC/B,CAAC,EAGDf,EAAyBK,EAAI,SAAS,EAAE,KACtCe,EAAOC,GAAMA,EAAG,MAAQ,OAAO,EAC/BC,GAAeT,EAAO,CAACU,EAAGR,IAAYA,CAAO,CAC/C,EACG,UAAU,CAAC,CAAE,MAAAlB,CAAM,IAAM,CACxBA,GAASA,EAAQ,GAAKH,EAAO,OAC7BA,EAAOG,CAAK,EAAE,MAAM,EACpBH,EAAOG,CAAK,EAAE,MAAM,CACtB,CAAC,EAGHgB,EACG,KACCZ,EAAI,IAAM,CACR,IAAMuB,EAASC,GAAoB,QAAQ,EACrCC,EAAS,OAAO,iBAAiBF,CAAM,EAG7C,OAAAf,EAAO,QAAUiB,EAAM,YAGhBA,EAAM,gBAAgB,MAAM,MAAM,EACtC,IAAIR,IAAU,CAACA,GAAO,SAAS,EAAE,EAAE,SAAS,EAAG,GAAG,CAAC,EACnD,KAAK,EAAE,CACZ,CAAC,CACH,EACG,UAAUS,GAASpB,EAAK,QAAU,IAAIoB,CAAK,EAAE,EAGlDd,EAAM,KAAKe,GAAUC,EAAc,CAAC,EACjC,UAAU,IAAM,CACf,SAAS,KAAK,gBAAgB,yBAAyB,CACzD,CAAC,EAGIpC,GAAaC,CAAM,EACvB,KACCoC,EAAUpB,EAAO,KAAKqB,GAAK,CAAC,CAAC,CAAC,EAC9BC,GAAO,EACPC,EAAIC,GAASrB,EAAM,KAAKqB,CAAK,CAAC,EAC9BC,EAAS,IAAMtB,EAAM,SAAS,CAAC,EAC/BZ,EAAIiC,GAAUE,EAAA,CAAE,IAAK/B,GAAO6B,EAAQ,CACtC,CACJ,CAAC,CACH,CChJO,SAASG,GACdC,EAAiB,CAAE,UAAAC,CAAU,EACI,CAGjC,OAAOC,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAU,CAAC,CAAE,MAAAE,CAAM,IAAM,CAC7BL,EAAG,MAAM,YAAY,sBAAuB,GAAGK,CAAK,EAAE,CACxD,CAAC,EAGMJ,EACJ,KACCK,EAAID,GAASF,EAAM,KAAK,CAAE,MAAAE,CAAM,CAAC,CAAC,EAClCE,EAAS,IAAMJ,EAAM,SAAS,CAAC,EAC/BK,EAAIH,IAAU,CAAE,IAAKL,EAAI,MAAAK,CAAM,EAAE,CACnC,CACJ,CAAC,CACH,CChEA,IAAAI,GAAwB,SAiCxB,SAASC,GAAQC,EAAyB,CACxCA,EAAG,aAAa,kBAAmB,EAAE,EACrC,IAAMC,EAAOD,EAAG,QAAQ,aAAa,EAC/BE,EAAOD,EACTA,EAAK,aAAa,WAAW,EAC7BD,EAAG,UACP,OAAAA,EAAG,gBAAgB,iBAAiB,EAC7BE,EAAK,QAAQ,CACtB,CAWO,SAASC,GACd,CAAE,OAAAC,CAAO,EACH,CACF,GAAAC,QAAY,YAAY,GAC1B,IAAIC,EAA8BC,GAAc,CAC9C,IAAI,GAAAF,QAAY,iDAAkD,CAChE,KAAML,GACJA,EAAG,aAAa,qBAAqB,GACrCD,GAAQS,EACNR,EAAG,aAAa,uBAAuB,CACzC,CAAC,CAEL,CAAC,EACE,GAAG,UAAWS,GAAMF,EAAW,KAAKE,CAAE,CAAC,CAC5C,CAAC,EACE,KACCC,EAAID,GAAM,CACQA,EAAG,QACX,MAAM,CAChB,CAAC,EACDE,EAAI,IAAMC,GAAY,kBAAkB,CAAC,CAC3C,EACG,UAAUR,CAAM,CAEzB,CCrCA,SAASS,GAAQC,EAAUC,EAAW,CACpC,OAAAD,EAAI,SAAWC,EAAK,SACpBD,EAAI,SAAWC,EAAK,SACbD,CACT,CA2BA,SAASE,GAAQC,EAAoBF,EAAoB,CACvD,IAAMG,EAAmB,IAAI,IAC7B,QAAWC,KAAMC,EAAY,MAAOH,CAAQ,EAAG,CAC7C,IAAMH,EAAMO,EAAW,MAAOF,CAAE,EAG1BG,EAAQ,CAACT,GAAQ,IAAI,IAAIC,EAAI,WAAY,EAAGC,CAAI,CAAC,EACvDG,EAAQ,IAAI,GAAGI,EAAM,CAAC,CAAC,GAAIA,CAAK,EAGhC,QAAWC,KAAQH,EAAY,kBAAmBD,CAAE,EAAG,CACrD,IAAMK,EAAOD,EAAK,aAAa,MAAM,EACjCC,GAAQ,MACVF,EAAM,KAAKT,GAAQ,IAAI,IAAIW,CAAI,EAAGT,CAAI,CAAC,CAC3C,CACF,CAGA,OAAOG,CACT,CAgBO,SAASO,GAAaV,EAAyC,CACpE,OAAOW,GAAW,IAAI,IAAI,cAAeX,CAAI,CAAC,EAC3C,KACCY,EAAIV,GAAYD,GAAQC,EAAU,IAAI,IAAIF,CAAI,CAAC,CAAC,EAChDa,GAAW,IAAMC,EAAG,IAAI,GAAK,CAAC,CAChC,CACJ,CClDA,SAASC,GACPC,EAAgBC,EACC,CACjB,GAAI,EAAED,EAAG,kBAAkB,SACzB,OAAOE,EAIT,IAAMC,EAAKH,EAAG,OAAO,QAAQ,GAAG,EAChC,GAAIG,IAAO,KACT,OAAOD,EAMT,GAAIC,EAAG,QAAUH,EAAG,SAAWA,EAAG,QAChC,OAAOE,EAQT,IAAME,EAAM,IAAI,IAAID,EAAG,IAAI,EAO3B,OANAC,EAAI,OAASA,EAAI,KAAO,GAMnBH,EAAQ,IAAI,GAAGG,CAAG,EAAE,GASzBJ,EAAG,eAAe,EACXK,EAAG,IAAI,IAAIF,EAAG,IAAI,CAAC,GATjBD,CAUX,CASA,SAASI,GAAKC,EAA8C,CAC1D,IAAMC,EAAO,IAAI,IACjB,QAAWL,KAAMM,EAAY,aAAcF,EAAS,IAAI,EACtDC,EAAK,IAAIL,EAAG,UAAWA,CAAE,EAG3B,OAAOK,CACT,CAYA,SAASE,GAAQH,EAA0C,CACzD,QAAWJ,KAAMM,EAAY,gBAAiBF,CAAQ,EACpD,QAAWI,IAAO,CAAC,OAAQ,KAAK,EAAG,CACjC,IAAMC,EAAQT,EAAG,aAAaQ,CAAG,EACjC,GAAIC,GAAS,CAAC,qBAAqB,KAAKA,CAAK,EAAG,CAE9CT,EAAGQ,CAAG,EAAIR,EAAGQ,CAAG,EAChB,KACF,CACF,CAGF,OAAON,EAAGE,CAAQ,CACpB,CASA,SAASM,GAAOC,EAAsC,CACpD,QAAWC,IAAY,CACrB,+BACA,gCACA,mCACA,+BACA,2BACA,2BACA,GAAGC,EAAQ,wBAAwB,EAC/B,CAAC,0BAA0B,EAC3B,CAAC,CACP,EAAG,CACD,IAAMC,EAASC,GAAmBH,CAAQ,EACpCI,EAASD,GAAmBH,EAAUD,CAAI,EAE9C,OAAOG,GAAW,aAClB,OAAOE,GAAW,aAElBF,EAAO,YAAYE,CAAM,CAE7B,CAGA,IAAMX,EAAOF,GAAK,QAAQ,EAC1B,OAAW,CAACc,EAAMjB,CAAE,IAAKG,GAAKQ,CAAI,EAC5BN,EAAK,IAAIY,CAAI,EACfZ,EAAK,OAAOY,CAAI,EAEhB,SAAS,KAAK,YAAYjB,CAAE,EAGhC,QAAWA,KAAMK,EAAK,OAAO,EAAG,CAC9B,IAAMa,EAAOlB,EAAG,aAAa,MAAM,EAI/BkB,IAAS,eAAiBA,IAAS,gBACrClB,EAAG,OAAO,CACd,CAIA,IAAMmB,EAAYC,GAAoB,WAAW,EACjD,OAAOC,GAAOf,EAAY,SAAUa,CAAS,CAAC,EAC3C,KACCG,EAAUtB,GAAM,CACd,IAAMuB,EAASZ,EAAK,cAAc,QAAQ,EAC1C,GAAIX,EAAG,IAAK,CACV,QAAWkB,KAAQlB,EAAG,kBAAkB,EACtCuB,EAAO,aAAaL,EAAMlB,EAAG,aAAakB,CAAI,CAAE,EAClD,OAAAlB,EAAG,YAAYuB,CAAM,EAGd,IAAIC,EAAWC,GAAY,CAChCF,EAAO,OAAS,IAAME,EAAS,SAAS,CAC1C,CAAC,CAGH,KACE,QAAAF,EAAO,YAAcvB,EAAG,YACxBA,EAAG,YAAYuB,CAAM,EACdxB,CAEX,CAAC,EACD2B,GAAe,EACfC,GAAQ,QAAQ,CAClB,CACJ,CAgBO,SAASC,GACd,CAAE,UAAAC,EAAW,UAAAC,EAAW,UAAAC,CAAU,EACZ,CACtB,IAAMC,EAASC,GAAc,EAC7B,GAAI,SAAS,WAAa,QACxB,OAAOlC,EAIT,IAAMmC,EAAWC,GAAaH,EAAO,IAAI,EAUzC9B,EAAG,QAAQ,EACR,UAAUK,EAAO,EAUpB,IAAM6B,EACJC,EAAsB,SAAS,KAAM,OAAO,EACzC,KACCC,GAAkBJ,CAAQ,EAC1BZ,EAAU,CAAC,CAACzB,EAAIC,CAAO,IAAMF,GAAOC,EAAIC,CAAO,CAAC,EAChDyC,GAAM,CACR,EAIEC,EACJH,EAAyB,OAAQ,UAAU,EACxC,KACCI,EAAIC,EAAW,EACfH,GAAM,CACR,EAMJH,EAAS,KAAKO,GAAeb,CAAS,CAAC,EACpC,UAAU,CAAC,CAAC7B,EAAK,CAAE,OAAA2C,CAAO,CAAC,IAAM,CAChC,QAAQ,aAAaA,EAAQ,EAAE,EAC/B,QAAQ,UAAU,KAAM,GAAI3C,CAAG,CACjC,CAAC,EAMH4C,EAAMT,EAAUI,CAAQ,EACrB,UAAUX,CAAS,EActB,IAAMiB,EACJjB,EAAU,KACRkB,EAAwB,UAAU,EAClCzB,EAAUrB,GAAO+C,GAAY/C,EAAK,CAAE,UAAA8B,CAAU,CAAC,EAC5C,KACCkB,GAAW,KACTC,GAAYjD,EAAK,EAAI,EACdF,EACR,CACH,CACF,EAIAuB,EAAUf,EAAO,EACjBe,EAAUZ,EAAM,EAChB6B,GAAM,CACR,EAUF,OAAAM,EACEC,EAAU,KAAKH,GAAed,EAAW,CAACsB,EAAGlD,IAAQA,CAAG,CAAC,EASzD4B,EAAU,KACRkB,EAAwB,UAAU,EAClCzB,EAAU,IAAMO,CAAS,EACzBkB,EAAwB,MAAM,CAChC,EAQAlB,EAAU,KACRuB,EAAqB,CAACC,EAAGC,IACvBD,EAAE,WAAaC,EAAE,UACjBD,EAAE,OAAaC,EAAE,IAClB,EACDhC,EAAU,IAAMc,CAAQ,EACxBmB,EAAI,IAAM,QAAQ,KAAK,CAAC,CAC1B,CACF,EACG,UAAUtD,GAAO,CAzYtB,IAAAuD,EAAAC,EA+YU,QAAQ,QAAU,MAAQ,CAACxD,EAAI,KACjC,OAAO,SAAS,GAAGwD,GAAAD,EAAA,QAAQ,QAAR,YAAAA,EAAe,IAAf,KAAAC,EAAoB,CAAC,GAExC,QAAQ,kBAAoB,OAC5BC,GAAgBzD,EAAI,IAAI,EACxB,QAAQ,kBAAoB,SAEhC,CAAC,EAMH4B,EAAU,UAAU,IAAM,CACxB,QAAQ,kBAAoB,QAC9B,CAAC,EAMDQ,EAAU,OAAQ,cAAc,EAC7B,UAAU,IAAM,CACf,QAAQ,kBAAoB,MAC9B,CAAC,EAMHP,EAAU,KACRiB,EAAwB,QAAQ,EAChCY,GAAa,GAAG,CAClB,EACG,UAAU,CAAC,CAAE,OAAAf,CAAO,IAAM,CACzB,QAAQ,aAAaA,EAAQ,EAAE,CACjC,CAAC,EAGIE,CACT,CCjaA,IAAAc,GAAuB,SAqChB,SAASC,GACdC,EAC0B,CAE1B,IAAMC,EAAQD,EAAO,UAAU,MAAM,GAAG,EAAE,IAAIE,GAC/BA,EAAK,QAAQ,sBAAuB,EAAE,EACvC,SAAW,EAAI,SAAMA,CAClC,EACE,KAAK,GAAG,EAELC,EAAY,IAAI,OAAOF,EAAO,KAAK,EACnCG,EAAY,CAACC,EAAYC,EAAcJ,IACpC,GAAGI,CAAI,2BAA2BJ,CAAI,UAI/C,OAAQK,GAAkB,CACxBA,EAAQA,EACL,QAAQ,gBAAiB,GAAG,EAC5B,KAAK,EAGR,IAAMC,EAAQ,IAAI,OAAO,MAAMR,EAAO,SAAS,MAC7CO,EACG,QAAQ,uBAAwB,MAAM,EACtC,QAAQJ,EAAW,GAAG,CAC3B,IAAK,KAAK,EAGV,OAAOM,MAAS,GAAAC,SAAWD,CAAK,EAC7B,QAAQD,EAAOJ,CAAS,EACxB,QAAQ,8BAA+B,IAAI,CAChD,CACF,CCEO,SAASO,GACdC,EAC+B,CAC/B,OAAOA,EAAQ,OAAS,CAC1B,CASO,SAASC,GACdD,EACgC,CAChC,OAAOA,EAAQ,OAAS,CAC1B,CC1CO,SAASE,GACdC,EAAaC,EACW,CACxB,IAAMC,EAAUC,GAA2BH,CAAG,EAC9C,OAAAI,EACEC,EAAG,SAAS,WAAa,OAAO,EAChCC,GAAY,QAAQ,CACtB,EACG,KACCC,GAAMC,GAAUA,CAAM,EACtBC,EAAU,IAAMR,CAAM,CACxB,EACG,UAAU,CAAC,CAAE,OAAAS,EAAQ,KAAAC,CAAK,IAAMT,EAAQ,KAAK,CAC5C,OACA,KAAM,CACJ,OAAAQ,EACA,KAAAC,EACA,QAAS,CACP,QAASC,EAAQ,gBAAgB,CACnC,CACF,CACF,CAAC,CAAC,EAGCV,CACT,CCxBO,SAASW,GACd,CAAE,UAAAC,CAAU,EACN,CACN,IAAMC,EAASC,GAAc,EACvBC,EAAYC,GAChB,IAAI,IAAI,mBAAoBH,EAAO,IAAI,CACzC,EACG,KACCI,GAAW,IAAMC,CAAK,CACxB,EAGIC,EAAWJ,EACd,KACCK,EAAIC,GAAY,CACd,GAAM,CAAC,CAAEC,CAAO,EAAIT,EAAO,KAAK,MAAM,aAAa,EACnD,OAAOQ,EAAS,KAAK,CAAC,CAAE,QAAAE,EAAS,QAAAC,CAAQ,IACvCD,IAAYD,GAAWE,EAAQ,SAASF,CAAO,CAChD,GAAKD,EAAS,CAAC,CAClB,CAAC,CACH,EAGFN,EACG,KACCK,EAAIC,GAAY,IAAI,IAAIA,EAAS,IAAIE,GAAW,CAC9C,GAAG,IAAI,IAAI,MAAMA,EAAQ,OAAO,IAAKV,EAAO,IAAI,CAAC,GACjDU,CACF,CAAC,CAAC,CAAC,EACHE,EAAUC,GAAQC,EAAsB,SAAS,KAAM,OAAO,EAC3D,KACCC,EAAOC,GAAM,CAACA,EAAG,SAAW,CAACA,EAAG,OAAO,EACvCC,GAAeX,CAAQ,EACvBM,EAAU,CAAC,CAACI,EAAIP,CAAO,IAAM,CAC3B,GAAIO,EAAG,kBAAkB,QAAS,CAChC,IAAME,EAAKF,EAAG,OAAO,QAAQ,GAAG,EAChC,GAAIE,GAAM,CAACA,EAAG,QAAUL,EAAK,IAAIK,EAAG,IAAI,EAAG,CACzC,IAAMC,EAAMD,EAAG,KAWf,MAAI,CAACF,EAAG,OAAO,QAAQ,aAAa,GAClBH,EAAK,IAAIM,CAAG,IACZV,EACPJ,GAEXW,EAAG,eAAe,EACXI,EAAGD,CAAG,EACf,CACF,CACA,OAAOd,CACT,CAAC,EACDO,EAAUO,GAAO,CACf,GAAM,CAAE,QAAAT,CAAQ,EAAIG,EAAK,IAAIM,CAAG,EAChC,OAAOE,GAAa,IAAI,IAAIF,CAAG,CAAC,EAC7B,KACCZ,EAAIe,GAAW,CAEb,IAAMC,EADWC,GAAY,EACP,KAAK,QAAQxB,EAAO,KAAM,EAAE,EAClD,OAAOsB,EAAQ,IAAIC,EAAK,MAAM,GAAG,EAAE,CAAC,CAAC,EACjC,IAAI,IAAI,MAAMb,CAAO,IAAIa,CAAI,GAAIvB,EAAO,IAAI,EAC5C,IAAI,IAAImB,CAAG,CACjB,CAAC,CACH,CACJ,CAAC,CACH,CACF,CACF,EACG,UAAUA,GAAOM,GAAYN,EAAK,EAAI,CAAC,EAG5CO,EAAc,CAACxB,EAAWI,CAAQ,CAAC,EAChC,UAAU,CAAC,CAACE,EAAUC,CAAO,IAAM,CACpBkB,EAAW,mBAAmB,EACtC,YAAYC,GAAsBpB,EAAUC,CAAO,CAAC,CAC5D,CAAC,EAGHV,EAAU,KAAKa,EAAU,IAAMN,CAAQ,CAAC,EACrC,UAAUG,GAAW,CA5J1B,IAAAoB,EA+JM,IAAIC,EAAW,SAAS,aAAc,cAAc,EACpD,GAAIA,IAAa,KAAM,CACrBA,EAAW,GAGX,IAAIC,IAAUF,EAAA7B,EAAO,UAAP,YAAA6B,EAAgB,UAAW,SACpC,MAAM,QAAQE,CAAO,IACxBA,EAAU,CAACA,CAAO,GAGpBC,EAAM,QAAWC,KAAUF,EACzB,QAAWrB,KAAWD,EAAQ,QAAQ,OAAOA,EAAQ,OAAO,EAC1D,GAAI,IAAI,OAAOwB,EAAQ,GAAG,EAAE,KAAKvB,CAAO,EAAG,CACzCoB,EAAW,GACX,MAAME,CACR,CAGJ,SAAS,aAAcF,EAAU,cAAc,CACjD,CAGA,GAAIA,EACF,QAAWI,KAAWC,GAAqB,UAAU,EACnDD,EAAQ,OAAS,EACvB,CAAC,CACL,CCrFO,SAASE,GACdC,EAAsB,CAAE,QAAAC,CAAQ,EACP,CAGzB,GAAM,CAAE,aAAAC,CAAa,EAAIC,GAAY,EACjCD,EAAa,IAAI,GAAG,IACtBE,GAAU,SAAU,EAAI,EAGxBJ,EAAG,MAAQE,EAAa,IAAI,GAAG,EAC/BF,EAAG,MAAM,EAGTK,GAAY,QAAQ,EACjB,KACCC,GAAMC,GAAU,CAACA,CAAM,CACzB,EACG,UAAU,IAAM,CACf,IAAMC,EAAML,GAAY,EACxBK,EAAI,aAAa,OAAO,GAAG,EAC3B,QAAQ,aAAa,CAAC,EAAG,GAAI,GAAGA,CAAG,EAAE,CACvC,CAAC,GAIP,IAAMC,EAASC,GAAkBV,CAAE,EAC7BW,EAASC,EACbX,EAAQ,KAAKK,GAAMO,EAAoB,CAAC,EACxCC,EAAUd,EAAI,OAAO,EACrBS,CACF,EACG,KACCM,EAAI,IAAMf,EAAG,KAAK,EAClBgB,EAAqB,CACvB,EAGF,OAAOC,EAAc,CAACN,EAAQF,CAAM,CAAC,EAClC,KACCM,EAAI,CAAC,CAACG,EAAOC,CAAK,KAAO,CAAE,MAAAD,EAAO,MAAAC,CAAM,EAAE,EAC1CC,EAAY,CAAC,CACf,CACJ,CAUO,SAASC,GACdrB,EAAsB,CAAE,QAAAC,CAAQ,EACsB,CACtD,IAAMqB,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,GAAe,EAAGC,GAAQ,EAAI,CAAC,EAGxDT,EAAc,CACZhB,EAAQ,KAAKK,GAAMO,EAAoB,CAAC,EACxCS,CACF,EAAG,CAACK,EAAGC,IAAUA,CAAK,EACnB,KACCC,EAAwB,OAAO,CACjC,EACG,UAAU,CAAC,CAAE,MAAAX,CAAM,IAAMjB,EAAQ,KAAK,CACrC,OACA,KAAMiB,CACR,CAAC,CAAC,EAGNI,EACG,KACCO,EAAwB,OAAO,CACjC,EACG,UAAU,CAAC,CAAE,MAAAV,CAAM,IAAM,CACpBA,GACFf,GAAU,SAAUe,CAAK,CAC7B,CAAC,EAGLL,EAAUd,EAAG,KAAO,OAAO,EACxB,KACC8B,EAAUN,CAAK,CACjB,EACG,UAAU,IAAMxB,EAAG,MAAM,CAAC,EAM/B,IAAM+B,EAAQC,EAAW,uBAAuB,EAChD,OAAAlB,EAAUiB,EAAO,OAAO,EACrB,UAAU,IAAM/B,EAAG,MAAM,CAAC,EAGtBD,GAAiBC,EAAI,CAAE,QAAAC,CAAQ,CAAC,EACpC,KACCgC,EAAIC,GAASZ,EAAM,KAAKY,CAAK,CAAC,EAC9BC,EAAS,IAAMb,EAAM,SAAS,CAAC,EAC/BP,EAAImB,GAAUE,EAAA,CAAE,IAAKpC,GAAOkC,EAAQ,EACpCd,EAAY,CAAC,CACf,CACJ,CCnHO,SAASiB,GACdC,EAAiB,CAAE,QAAAC,EAAS,OAAAC,CAAO,EACE,CACrC,IAAMC,EAAQ,IAAIC,EACZC,EAAYC,GAAqBN,EAAG,aAAc,EACrD,KACCO,EAAO,OAAO,CAChB,EAGIC,EAAYR,EAAG,cAGfS,EAAOC,EAAW,wBAAyBV,CAAE,EAC7CW,EAAOD,EAAW,uBAAwBV,CAAE,EAGlDY,GAAY,QAAQ,EACjB,UAAUC,GAAUF,EAAK,aACxB,OAAQE,EAAS,OAAS,cAC5B,CAAC,EAGHV,EACG,KACCW,GAAeZ,CAAM,EACrBa,GAAUd,EAAQ,KAAKe,GAAMC,EAAoB,CAAC,CAAC,CACrD,EACG,UAAU,CAAC,CAAC,CAAE,MAAAC,CAAM,EAAG,CAAE,MAAAC,CAAM,CAAC,IAAM,CACrC,OAAQD,EAAM,OAAQ,CAGpB,IAAK,GACHT,EAAK,YAAcU,EAAM,OACrBC,GAAY,oBAAoB,EAChCA,GAAY,2BAA2B,EAC3C,MAGF,IAAK,GACHX,EAAK,YAAcW,GAAY,mBAAmB,EAClD,MAGF,QACE,IAAMC,EAAQC,GAAMJ,EAAM,MAAM,EAChCT,EAAK,YAAcW,GAAY,sBAAuBC,CAAK,CAC/D,CACF,CAAC,EAGL,IAAME,EAAUpB,EACb,KACCqB,EAAI,IAAMb,EAAK,UAAY,EAAE,EAC7Bc,EAAU,CAAC,CAAE,MAAAP,CAAM,IAAMQ,EACvBC,EAAG,GAAGT,EAAM,MAAM,EAAG,EAAE,CAAC,EACxBS,EAAG,GAAGT,EAAM,MAAM,EAAE,CAAC,EAClB,KACCU,GAAY,CAAC,EACbC,GAAQxB,CAAS,EACjBoB,EAAU,CAAC,CAACK,CAAK,IAAMA,CAAK,CAC9B,CACJ,CAAC,EACDC,EAAIC,EAAsB,EAC1BC,GAAM,CACR,EAGF,OAAAV,EAAQ,UAAUW,GAAQvB,EAAK,YAAYuB,CAAI,CAAC,EAChDX,EACG,KACCY,GAASD,GAAQ,CACf,IAAME,EAAUC,GAAmB,UAAWH,CAAI,EAClD,OAAI,OAAOE,GAAY,YACdE,EAGFC,EAAUH,EAAS,QAAQ,EAC/B,KACCI,EAAUrC,CAAK,EACf4B,EAAI,IAAMK,CAAO,CACnB,CACJ,CAAC,CACH,EACG,UAAUA,GAAW,CAElBA,EAAQ,OAAS,IACjBA,EAAQ,WAAa5B,EAAU,WAE/BA,EAAU,SAAS,CAAE,IAAK4B,EAAQ,SAAU,CAAC,CACjD,CAAC,EAGWnC,EACb,KACCM,EAAOkC,EAAqB,EAC5BV,EAAI,CAAC,CAAE,KAAAW,CAAK,IAAMA,CAAI,CACxB,EAIC,KACClB,EAAImB,GAASxC,EAAM,KAAKwC,CAAK,CAAC,EAC9BC,EAAS,IAAMzC,EAAM,SAAS,CAAC,EAC/B4B,EAAIY,GAAUE,EAAA,CAAE,IAAK7C,GAAO2C,EAAQ,CACtC,CACJ,CCpHO,SAASG,GACdC,EAAkB,CAAE,OAAAC,CAAO,EACF,CACzB,OAAOA,EACJ,KACCC,EAAI,CAAC,CAAE,MAAAC,CAAM,IAAM,CACjB,IAAMC,EAAMC,GAAY,EACxB,OAAAD,EAAI,KAAO,GAGXD,EAAQA,EACL,QAAQ,OAAQ,GAAG,EACnB,QAAQ,KAAM,KAAK,EACnB,QAAQ,KAAM,KAAK,EAGtBC,EAAI,OAAS,KAAKD,CAAK,GAChB,CAAE,IAAAC,CAAI,CACf,CAAC,CACH,CACJ,CAUO,SAASE,GACdC,EAAuBC,EACa,CACpC,IAAMC,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,GAAe,EAAGC,GAAQ,EAAI,CAAC,EACxD,OAAAJ,EAAM,UAAU,CAAC,CAAE,IAAAL,CAAI,IAAM,CAC3BG,EAAG,aAAa,sBAAuBA,EAAG,IAAI,EAC9CA,EAAG,KAAO,GAAGH,CAAG,EAClB,CAAC,EAGDU,EAAUP,EAAI,OAAO,EAClB,KACCQ,EAAUJ,CAAK,CACjB,EACG,UAAUK,GAAMA,EAAG,eAAe,CAAC,EAGjCjB,GAAiBQ,EAAIC,CAAO,EAChC,KACCS,EAAIC,GAAST,EAAM,KAAKS,CAAK,CAAC,EAC9BC,EAAS,IAAMV,EAAM,SAAS,CAAC,EAC/BP,EAAIgB,GAAUE,EAAA,CAAE,IAAKb,GAAOW,EAAQ,CACtC,CACJ,CCpDO,SAASG,GACdC,EAAiB,CAAE,QAAAC,EAAS,UAAAC,CAAU,EACA,CACtC,IAAMC,EAAQ,IAAIC,EAGZC,EAASC,GAAoB,cAAc,EAC3CC,EAASC,EACbC,EAAUJ,EAAO,SAAS,EAC1BI,EAAUJ,EAAO,OAAO,CAC1B,EACG,KACCK,GAAUC,EAAc,EACxBC,EAAI,IAAMP,EAAM,KAAK,EACrBQ,EAAqB,CACvB,EAGF,OAAAV,EACG,KACCW,GAAkBP,CAAM,EACxBK,EAAI,CAAC,CAAC,CAAE,QAAAG,CAAQ,EAAGC,CAAK,IAAM,CAC5B,IAAMC,EAAQD,EAAM,MAAM,UAAU,EACpC,GAAID,GAAA,MAAAA,EAAS,QAAUE,EAAMA,EAAM,OAAS,CAAC,EAAG,CAC9C,IAAMC,EAAOH,EAAQA,EAAQ,OAAS,CAAC,EACnCG,EAAK,WAAWD,EAAMA,EAAM,OAAS,CAAC,CAAC,IACzCA,EAAMA,EAAM,OAAS,CAAC,EAAIC,EAC9B,MACED,EAAM,OAAS,EAEjB,OAAOA,CACT,CAAC,CACH,EACG,UAAUA,GAASjB,EAAG,UAAYiB,EAChC,KAAK,EAAE,EACP,QAAQ,MAAO,QAAQ,CAC1B,EAGJf,EACG,KACCiB,EAAO,CAAC,CAAE,KAAAC,CAAK,IAAMA,IAAS,QAAQ,CACxC,EACG,UAAUC,GAAO,CAChB,OAAQA,EAAI,KAAM,CAGhB,IAAK,aAEDrB,EAAG,UAAU,QACbK,EAAM,iBAAmBA,EAAM,MAAM,SAErCA,EAAM,MAAQL,EAAG,WACnB,KACJ,CACF,CAAC,EAGWC,EACb,KACCkB,EAAOG,EAAqB,EAC5BV,EAAI,CAAC,CAAE,KAAAW,CAAK,IAAMA,CAAI,CACxB,EAIC,KACCC,EAAIC,GAAStB,EAAM,KAAKsB,CAAK,CAAC,EAC9BC,EAAS,IAAMvB,EAAM,SAAS,CAAC,EAC/BS,EAAI,KAAO,CAAE,IAAKZ,CAAG,EAAE,CACzB,CACJ,CCjDO,SAAS2B,GACdC,EAAiB,CAAE,OAAAC,EAAQ,UAAAC,CAAU,EACN,CAC/B,IAAMC,EAASC,GAAc,EAC7B,GAAI,CACF,IAAMC,EAAUC,GAAkBH,EAAO,OAAQF,CAAM,EAGjDM,EAASC,GAAoB,eAAgBR,CAAE,EAC/CS,EAASD,GAAoB,gBAAiBR,CAAE,EAGtDU,EAAwBV,EAAI,OAAO,EAChC,KACCW,EAAO,CAAC,CAAE,OAAAC,CAAO,IACfA,aAAkB,SAAW,CAAC,CAACA,EAAO,QAAQ,GAAG,CAClD,CACH,EACG,UAAU,IAAMC,GAAU,SAAU,EAAK,CAAC,EAG/CX,EACG,KACCS,EAAO,CAAC,CAAE,KAAAG,CAAK,IAAMA,IAAS,QAAQ,CACxC,EACG,UAAUC,GAAO,CAChB,IAAMC,EAASC,GAAiB,EAChC,OAAQF,EAAI,KAAM,CAGhB,IAAK,QACH,GAAIC,IAAWT,EAAO,CACpB,IAAMW,EAAU,IAAI,IACpB,QAAWC,KAAUC,EACnB,sBAAuBX,CACzB,EAAG,CACD,IAAMY,EAAUF,EAAO,kBACvBD,EAAQ,IAAIC,EAAQ,WAClBE,EAAQ,aAAa,eAAe,CACtC,CAAC,CACH,CAGA,GAAIH,EAAQ,KAAM,CAChB,GAAM,CAAC,CAACI,CAAI,CAAC,EAAI,CAAC,GAAGJ,CAAO,EAAE,KAAK,CAAC,CAAC,CAAEK,CAAC,EAAG,CAAC,CAAEC,CAAC,IAAMA,EAAID,CAAC,EAC1DD,EAAK,MAAM,CACb,CAGAP,EAAI,MAAM,CACZ,CACA,MAGF,IAAK,SACL,IAAK,MACHF,GAAU,SAAU,EAAK,EACzBN,EAAM,KAAK,EACX,MAGF,IAAK,UACL,IAAK,YACH,GAAI,OAAOS,GAAW,YACpBT,EAAM,MAAM,MACP,CACL,IAAMkB,EAAM,CAAClB,EAAO,GAAGa,EACrB,wDACAX,CACF,CAAC,EACKiB,EAAI,KAAK,IAAI,GACjB,KAAK,IAAI,EAAGD,EAAI,QAAQT,CAAM,CAAC,EAAIS,EAAI,QACrCV,EAAI,OAAS,UAAY,GAAK,IAE9BU,EAAI,MAAM,EACdA,EAAIC,CAAC,EAAE,MAAM,CACf,CAGAX,EAAI,MAAM,EACV,MAGF,QACMR,IAAUU,GAAiB,GAC7BV,EAAM,MAAM,CAClB,CACF,CAAC,EAGLL,EACG,KACCS,EAAO,CAAC,CAAE,KAAAG,CAAK,IAAMA,IAAS,QAAQ,CACxC,EACG,UAAUC,GAAO,CAChB,OAAQA,EAAI,KAAM,CAGhB,IAAK,IACL,IAAK,IACL,IAAK,IACHR,EAAM,MAAM,EACZA,EAAM,OAAO,EAGbQ,EAAI,MAAM,EACV,KACJ,CACF,CAAC,EAGL,IAAMY,EAASC,GAAiBrB,EAAO,CAAE,QAAAF,CAAQ,CAAC,EAClD,OAAOwB,EACLF,EACAG,GAAkBrB,EAAQ,CAAE,QAAAJ,EAAS,OAAAsB,CAAO,CAAC,CAC/C,EACG,KACCI,GAGE,GAAGC,GAAqB,eAAgBhC,CAAE,EACvC,IAAIiC,GAASC,GAAiBD,EAAO,CAAE,OAAAN,CAAO,CAAC,CAAC,EAGnD,GAAGK,GAAqB,iBAAkBhC,CAAE,EACzC,IAAIiC,GAASE,GAAmBF,EAAO,CAAE,QAAA5B,EAAS,UAAAH,CAAU,CAAC,CAAC,CACnE,CACF,CAGJ,OAASkC,EAAK,CACZ,OAAApC,EAAG,OAAS,GACLqC,EACT,CACF,CCnKO,SAASC,GACdC,EAAiB,CAAE,OAAAC,EAAQ,UAAAC,CAAU,EACG,CACxC,OAAOC,EAAc,CACnBF,EACAC,EACG,KACCE,EAAUC,GAAY,CAAC,EACvBC,EAAOC,GAAO,CAAC,CAACA,EAAI,aAAa,IAAI,GAAG,CAAC,CAC3C,CACJ,CAAC,EACE,KACCC,EAAI,CAAC,CAACC,EAAOF,CAAG,IAAMG,GAAuBD,EAAM,MAAM,EACvDF,EAAI,aAAa,IAAI,GAAG,CAC1B,CAAC,EACDC,EAAIG,GAAM,CA1FhB,IAAAC,EA2FQ,IAAMC,EAAQ,IAAI,IAGZC,EAAK,SAAS,mBAAmBd,EAAI,WAAW,SAAS,EAC/D,QAASe,EAAOD,EAAG,SAAS,EAAGC,EAAMA,EAAOD,EAAG,SAAS,EACtD,IAAIF,EAAAG,EAAK,gBAAL,MAAAH,EAAoB,aAAc,CACpC,IAAMI,EAAWD,EAAK,YAChBE,EAAWN,EAAGK,CAAQ,EACxBC,EAAS,OAASD,EAAS,QAC7BH,EAAM,IAAIE,EAAmBE,CAAQ,CACzC,CAIF,OAAW,CAACF,EAAMG,CAAI,IAAKL,EAAO,CAChC,GAAM,CAAE,WAAAM,CAAW,EAAIC,EAAE,OAAQ,KAAMF,CAAI,EAC3CH,EAAK,YAAY,GAAG,MAAM,KAAKI,CAAU,CAAC,CAC5C,CAGA,MAAO,CAAE,IAAKnB,EAAI,MAAAa,CAAM,CAC1B,CAAC,CACH,CACJ,CCPO,SAASQ,GACdC,EAAiB,CAAE,UAAAC,EAAW,MAAAC,CAAM,EACf,CACrB,IAAMC,EAASH,EAAG,QAAqB,UAAU,EAC3CI,EACJD,EAAO,UACPA,EAAO,cAAe,UAGxB,OAAOE,EAAc,CAACH,EAAOD,CAAS,CAAC,EACpC,KACCK,EAAI,CAAC,CAAC,CAAE,OAAAC,EAAQ,OAAAC,CAAO,EAAG,CAAE,OAAQ,CAAE,EAAAC,CAAE,CAAE,CAAC,KACzCD,EAASA,EACL,KAAK,IAAIJ,EAAQ,KAAK,IAAI,EAAGK,EAAIF,CAAM,CAAC,EACxCH,EACG,CACL,OAAAI,EACA,OAAQC,GAAKF,EAASH,CACxB,EACD,EACDM,EAAqB,CAACC,EAAGC,IACvBD,EAAE,SAAWC,EAAE,QACfD,EAAE,SAAWC,EAAE,MAChB,CACH,CACJ,CAuBO,SAASC,GACdb,EAAiBc,EACe,CADf,IAAAC,EAAAD,EAAE,SAAAE,CA5JrB,EA4JmBD,EAAcE,EAAAC,GAAdH,EAAc,CAAZ,YAEnB,IAAMI,EAAQC,EAAW,0BAA2BpB,CAAE,EAChD,CAAE,EAAAS,CAAE,EAAIY,GAAiBF,CAAK,EACpC,OAAOG,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,GAAe,EAAGC,GAAQ,EAAI,CAAC,EAClDC,EAAQL,EACX,KACCM,GAAU,EAAGC,EAAuB,CACtC,EAGF,OAAAF,EAAM,KAAKG,GAAef,CAAO,CAAC,EAC/B,UAAU,CAGT,KAAK,CAAC,CAAE,OAAAR,CAAO,EAAG,CAAE,OAAQD,CAAO,CAAC,EAAG,CACrCY,EAAM,MAAM,OAAS,GAAGX,EAAS,EAAIC,CAAC,KACtCT,EAAG,MAAM,IAAY,GAAGO,CAAM,IAChC,EAGA,UAAW,CACTY,EAAM,MAAM,OAAS,GACrBnB,EAAG,MAAM,IAAY,EACvB,CACF,CAAC,EAGH4B,EAAM,KAAKI,GAAM,CAAC,EACf,UAAU,IAAM,CACf,QAAWC,KAAQC,EAAY,8BAA+BlC,CAAE,EAAG,CACjE,GAAI,CAACiC,EAAK,aACR,SACF,IAAME,EAAYF,EAAK,QAAqB,yBAAyB,EACrE,GAAI,OAAOE,GAAc,YAAa,CACpC,IAAM5B,EAAS0B,EAAK,UAAYE,EAAU,UACpC,CAAE,OAAA3B,CAAO,EAAI4B,GAAeD,CAAS,EAC3CA,EAAU,SAAS,CACjB,IAAK5B,EAASC,EAAS,CACzB,CAAC,CACH,CACF,CACF,CAAC,EAGH6B,GAAKH,EAA8B,kBAAmBlC,CAAE,CAAC,EACtD,KACCsC,GAASC,GAASC,EAAUD,EAAO,OAAO,EACvC,KACCE,GAAUC,EAAc,EACxBpC,EAAI,IAAMiC,CAAK,EACfI,EAAUlB,CAAK,CACjB,CACF,CACF,EACG,UAAUc,GAAS,CAClB,IAAMK,EAAQxB,EAA6B,QAAQmB,EAAM,OAAO,IAAI,EACxDnB,EAAW,qBAAqBmB,EAAM,EAAE,IAAI,EACpD,aAAa,gBAAiB,GAAGK,EAAM,OAAO,EAAE,CACtD,CAAC,EAGE7C,GAAaC,EAAIiB,CAAO,EAC5B,KACC4B,EAAIC,GAASvB,EAAM,KAAKuB,CAAK,CAAC,EAC9BC,EAAS,IAAMxB,EAAM,SAAS,CAAC,EAC/BjB,EAAIwC,GAAUE,EAAA,CAAE,IAAKhD,GAAO8C,EAAQ,CACtC,CACJ,CAAC,CACH,CCxKO,SAASG,GACdC,EAAcC,EACW,CACzB,GAAI,OAAOA,GAAS,YAAa,CAC/B,IAAMC,EAAM,gCAAgCF,CAAI,IAAIC,CAAI,GACxD,OAAOE,GAGLC,GAAqB,GAAGF,CAAG,kBAAkB,EAC1C,KACCG,GAAW,IAAMC,CAAK,EACtBC,EAAIC,IAAY,CACd,QAASA,EAAQ,QACnB,EAAE,EACFC,GAAe,CAAC,CAAC,CACnB,EAGFL,GAAkBF,CAAG,EAClB,KACCG,GAAW,IAAMC,CAAK,EACtBC,EAAIG,IAAS,CACX,MAAOA,EAAK,iBACZ,MAAOA,EAAK,WACd,EAAE,EACFD,GAAe,CAAC,CAAC,CACnB,CACJ,EACG,KACCF,EAAI,CAAC,CAACC,EAASE,CAAI,IAAOC,IAAA,GAAKH,GAAYE,EAAO,CACpD,CAGJ,KAAO,CACL,IAAMR,EAAM,gCAAgCF,CAAI,GAChD,OAAOI,GAAkBF,CAAG,EACzB,KACCK,EAAIG,IAAS,CACX,aAAcA,EAAK,YACrB,EAAE,EACFD,GAAe,CAAC,CAAC,CACnB,CACJ,CACF,CCvDO,SAASG,GACdC,EAAcC,EACW,CACzB,IAAMC,EAAM,WAAWF,CAAI,oBAAoB,mBAAmBC,CAAO,CAAC,GAC1E,OAAOE,GAA2BD,CAAG,EAClC,KACCE,GAAW,IAAMC,CAAK,EACtBC,EAAI,CAAC,CAAE,WAAAC,EAAY,YAAAC,CAAY,KAAO,CACpC,MAAOD,EACP,MAAOC,CACT,EAAE,EACFC,GAAe,CAAC,CAAC,CACnB,CACJ,CCOO,SAASC,GACdC,EACyB,CAGzB,IAAIC,EAAQD,EAAI,MAAM,qCAAqC,EAC3D,GAAIC,EAAO,CACT,GAAM,CAAC,CAAEC,EAAMC,CAAI,EAAIF,EACvB,OAAOG,GAA2BF,EAAMC,CAAI,CAC9C,CAIA,GADAF,EAAQD,EAAI,MAAM,oCAAoC,EAClDC,EAAO,CACT,GAAM,CAAC,CAAEI,EAAMC,CAAI,EAAIL,EACvB,OAAOM,GAA2BF,EAAMC,CAAI,CAC9C,CAGA,OAAOE,CACT,CCpBA,IAAIC,GAgBG,SAASC,GACdC,EACoB,CACpB,OAAOF,QAAWG,EAAM,IAAM,CAC5B,IAAMC,EAAS,SAAsB,WAAY,cAAc,EAC/D,GAAIA,EACF,OAAOC,EAAGD,CAAM,EAKhB,GADYE,GAAqB,SAAS,EAClC,OAAQ,CACd,IAAMC,EAAU,SAA0B,WAAW,EACrD,GAAI,EAAEA,GAAWA,EAAQ,QACvB,OAAOC,CACX,CAGA,OAAOC,GAAiBP,EAAG,IAAI,EAC5B,KACCQ,EAAIC,GAAS,SAAS,WAAYA,EAAO,cAAc,CAAC,CAC1D,CAEN,CAAC,EACE,KACCC,GAAW,IAAMJ,CAAK,EACtBK,EAAOF,GAAS,OAAO,KAAKA,CAAK,EAAE,OAAS,CAAC,EAC7CG,EAAIH,IAAU,CAAE,MAAAA,CAAM,EAAE,EACxBI,EAAY,CAAC,CACf,EACJ,CASO,SAASC,GACdd,EAC+B,CAC/B,IAAMe,EAAQC,EAAW,uBAAwBhB,CAAE,EACnD,OAAOC,EAAM,IAAM,CACjB,IAAMgB,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAU,CAAC,CAAE,MAAAR,CAAM,IAAM,CAC7BM,EAAM,YAAYI,GAAkBV,CAAK,CAAC,EAC1CM,EAAM,UAAU,IAAI,+BAA+B,CACrD,CAAC,EAGMhB,GAAYC,CAAE,EAClB,KACCQ,EAAIY,GAASH,EAAM,KAAKG,CAAK,CAAC,EAC9BC,EAAS,IAAMJ,EAAM,SAAS,CAAC,EAC/BL,EAAIQ,GAAUE,EAAA,CAAE,IAAKtB,GAAOoB,EAAQ,CACtC,CACJ,CAAC,CACH,CCtDO,SAASG,GACdC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EACpB,CAClB,OAAOC,GAAiB,SAAS,IAAI,EAClC,KACCC,EAAU,IAAMC,GAAgBL,EAAI,CAAE,QAAAE,EAAS,UAAAD,CAAU,CAAC,CAAC,EAC3DK,EAAI,CAAC,CAAE,OAAQ,CAAE,EAAAC,CAAE,CAAE,KACZ,CACL,OAAQA,GAAK,EACf,EACD,EACDC,EAAwB,QAAQ,CAClC,CACJ,CAaO,SAASC,GACdT,EAAiBU,EACY,CAC7B,OAAOC,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAU,CAGd,KAAK,CAAE,OAAAE,CAAO,EAAG,CACfd,EAAG,OAASc,CACd,EAGA,UAAW,CACTd,EAAG,OAAS,EACd,CACF,CAAC,GAICe,EAAQ,wBAAwB,EAC5BC,EAAG,CAAE,OAAQ,EAAM,CAAC,EACpBjB,GAAUC,EAAIU,CAAO,GAExB,KACCO,EAAIC,GAASN,EAAM,KAAKM,CAAK,CAAC,EAC9BC,EAAS,IAAMP,EAAM,SAAS,CAAC,EAC/BN,EAAIY,GAAUE,EAAA,CAAE,IAAKpB,GAAOkB,EAAQ,CACtC,CACJ,CAAC,CACH,CCfO,SAASG,GACdC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EACT,CAC7B,IAAMC,EAAQ,IAAI,IAGZC,EAAUC,EAA+B,gBAAiBL,CAAE,EAClE,QAAWM,KAAUF,EAAS,CAC5B,IAAMG,EAAK,mBAAmBD,EAAO,KAAK,UAAU,CAAC,CAAC,EAChDE,EAASC,GAAmB,QAAQF,CAAE,IAAI,EAC5C,OAAOC,GAAW,aACpBL,EAAM,IAAIG,EAAQE,CAAM,CAC5B,CAGA,IAAME,EAAUR,EACb,KACCS,EAAwB,QAAQ,EAChCC,EAAI,CAAC,CAAE,OAAAC,CAAO,IAAM,CAClB,IAAMC,EAAOC,GAAoB,MAAM,EACjCC,EAAOC,EAAW,wBAAyBH,CAAI,EACrD,OAAOD,EAAS,IACdG,EAAK,UACLF,EAAK,UAET,CAAC,EACDI,GAAM,CACR,EAqFF,OAlFmBC,GAAiB,SAAS,IAAI,EAC9C,KACCR,EAAwB,QAAQ,EAGhCS,EAAUC,GAAQC,EAAM,IAAM,CAC5B,IAAIC,EAA4B,CAAC,EACjC,OAAOC,EAAG,CAAC,GAAGrB,CAAK,EAAE,OAAO,CAACsB,EAAO,CAACnB,EAAQE,CAAM,IAAM,CACvD,KAAOe,EAAK,QACGpB,EAAM,IAAIoB,EAAKA,EAAK,OAAS,CAAC,CAAC,EACnC,SAAWf,EAAO,SACzBe,EAAK,IAAI,EAOb,IAAIG,EAASlB,EAAO,UACpB,KAAO,CAACkB,GAAUlB,EAAO,eACvBA,EAASA,EAAO,cAChBkB,EAASlB,EAAO,UAIlB,IAAImB,EAASnB,EAAO,aACpB,KAAOmB,EAAQA,EAASA,EAAO,aAC7BD,GAAUC,EAAO,UAGnB,OAAOF,EAAM,IACX,CAAC,GAAGF,EAAO,CAAC,GAAGA,EAAMjB,CAAM,CAAC,EAAE,QAAQ,EACtCoB,CACF,CACF,EAAG,IAAI,GAAkC,CAAC,CAC5C,CAAC,EACE,KAGCd,EAAIa,GAAS,IAAI,IAAI,CAAC,GAAGA,CAAK,EAAE,KAAK,CAAC,CAAC,CAAEG,CAAC,EAAG,CAAC,CAAEC,CAAC,IAAMD,EAAIC,CAAC,CAAC,CAAC,EAC9DC,GAAkBpB,CAAO,EAGzBU,EAAU,CAAC,CAACK,EAAOM,CAAM,IAAM9B,EAC5B,KACC+B,GAAK,CAAC,CAACC,EAAMC,CAAI,EAAG,CAAE,OAAQ,CAAE,EAAAC,CAAE,EAAG,KAAAC,CAAK,IAAM,CAC9C,IAAMC,EAAOF,EAAIC,EAAK,QAAU,KAAK,MAAMf,EAAK,MAAM,EAGtD,KAAOa,EAAK,QAAQ,CAClB,GAAM,CAAC,CAAER,CAAM,EAAIQ,EAAK,CAAC,EACzB,GAAIR,EAASK,EAASI,GAAKE,EACzBJ,EAAO,CAAC,GAAGA,EAAMC,EAAK,MAAM,CAAE,MAE9B,MAEJ,CAGA,KAAOD,EAAK,QAAQ,CAClB,GAAM,CAAC,CAAEP,CAAM,EAAIO,EAAKA,EAAK,OAAS,CAAC,EACvC,GAAIP,EAASK,GAAUI,GAAK,CAACE,EAC3BH,EAAO,CAACD,EAAK,IAAI,EAAI,GAAGC,CAAI,MAE5B,MAEJ,CAGA,MAAO,CAACD,EAAMC,CAAI,CACpB,EAAG,CAAC,CAAC,EAAG,CAAC,GAAGT,CAAK,CAAC,CAAC,EACnBa,EAAqB,CAACV,EAAGC,IACvBD,EAAE,CAAC,IAAMC,EAAE,CAAC,GACZD,EAAE,CAAC,IAAMC,EAAE,CAAC,CACb,CACH,CACF,CACF,CACF,CACF,EAIC,KACCjB,EAAI,CAAC,CAACqB,EAAMC,CAAI,KAAO,CACrB,KAAMD,EAAK,IAAI,CAAC,CAACV,CAAI,IAAMA,CAAI,EAC/B,KAAMW,EAAK,IAAI,CAAC,CAACX,CAAI,IAAMA,CAAI,CACjC,EAAE,EAGFgB,EAAU,CAAE,KAAM,CAAC,EAAG,KAAM,CAAC,CAAE,CAAC,EAChCC,GAAY,EAAG,CAAC,EAChB5B,EAAI,CAAC,CAACgB,EAAGC,CAAC,IAGJD,EAAE,KAAK,OAASC,EAAE,KAAK,OAClB,CACL,KAAMA,EAAE,KAAK,MAAM,KAAK,IAAI,EAAGD,EAAE,KAAK,OAAS,CAAC,EAAGC,EAAE,KAAK,MAAM,EAChE,KAAM,CAAC,CACT,EAIO,CACL,KAAMA,EAAE,KAAK,MAAM,EAAE,EACrB,KAAMA,EAAE,KAAK,MAAM,EAAGA,EAAE,KAAK,OAASD,EAAE,KAAK,MAAM,CACrD,CAEH,CACH,CACJ,CAYO,SAASa,GACdzC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,EAAS,MAAAwC,EAAO,QAAAC,CAAQ,EACd,CACxC,OAAOrB,EAAM,IAAM,CACjB,IAAMsB,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,GAAe,EAAGC,GAAQ,EAAI,CAAC,EAoBxD,GAnBAJ,EAAM,UAAU,CAAC,CAAE,KAAAX,EAAM,KAAAC,CAAK,IAAM,CAGlC,OAAW,CAAC5B,CAAM,IAAK4B,EACrB5B,EAAO,UAAU,OAAO,sBAAsB,EAC9CA,EAAO,UAAU,OAAO,sBAAsB,EAIhD,OAAW,CAACmB,EAAO,CAACnB,CAAM,CAAC,IAAK2B,EAAK,QAAQ,EAC3C3B,EAAO,UAAU,IAAI,sBAAsB,EAC3CA,EAAO,UAAU,OACf,uBACAmB,IAAUQ,EAAK,OAAS,CAC1B,CAEJ,CAAC,EAGGgB,EAAQ,YAAY,EAAG,CAGzB,IAAMC,EAAUC,EACdlD,EAAU,KAAKmD,GAAa,CAAC,EAAGxC,EAAI,IAAG,EAAY,CAAC,EACpDX,EAAU,KAAKmD,GAAa,GAAG,EAAGxC,EAAI,IAAM,QAAiB,CAAC,CAChE,EAGAgC,EACG,KACCS,EAAO,CAAC,CAAE,KAAApB,CAAK,IAAMA,EAAK,OAAS,CAAC,EACpCH,GAAkBY,EAAM,KAAKY,GAAUC,EAAc,CAAC,CAAC,EACvDC,GAAeN,CAAO,CACxB,EACG,UAAU,CAAC,CAAC,CAAC,CAAE,KAAAjB,CAAK,CAAC,EAAGwB,CAAQ,IAAM,CACrC,GAAM,CAACnD,CAAM,EAAI2B,EAAKA,EAAK,OAAS,CAAC,EACrC,GAAI3B,EAAO,aAAc,CAGvB,IAAMoD,EAAYC,GAAoBrD,CAAM,EAC5C,GAAI,OAAOoD,GAAc,YAAa,CACpC,IAAMhC,EAASpB,EAAO,UAAYoD,EAAU,UACtC,CAAE,OAAA7C,CAAO,EAAI+C,GAAeF,CAAS,EAC3CA,EAAU,SAAS,CACjB,IAAKhC,EAASb,EAAS,EACvB,SAAA4C,CACF,CAAC,CACH,CACF,CACF,CAAC,CACP,CAGA,OAAIR,EAAQ,qBAAqB,GAC/BhD,EACG,KACC4D,EAAUf,CAAK,EACfnC,EAAwB,QAAQ,EAChCyC,GAAa,GAAG,EAChBU,GAAK,CAAC,EACND,EAAUlB,EAAQ,KAAKmB,GAAK,CAAC,CAAC,CAAC,EAC/BC,GAAO,CAAE,MAAO,GAAI,CAAC,EACrBP,GAAeZ,CAAK,CACtB,EACG,UAAU,CAAC,CAAC,CAAE,CAAE,KAAAX,CAAK,CAAC,IAAM,CAC3B,IAAM+B,EAAMC,GAAY,EAGlB3D,EAAS2B,EAAKA,EAAK,OAAS,CAAC,EACnC,GAAI3B,GAAUA,EAAO,OAAQ,CAC3B,GAAM,CAAC4D,CAAM,EAAI5D,EACX,CAAE,KAAA6D,CAAK,EAAI,IAAI,IAAID,EAAO,IAAI,EAChCF,EAAI,OAASG,IACfH,EAAI,KAAOG,EACX,QAAQ,aAAa,CAAC,EAAG,GAAI,GAAGH,CAAG,EAAE,EAIzC,MACEA,EAAI,KAAO,GACX,QAAQ,aAAa,CAAC,EAAG,GAAI,GAAGA,CAAG,EAAE,CAEzC,CAAC,EAGAjE,GAAqBC,EAAI,CAAE,UAAAC,EAAW,QAAAC,CAAQ,CAAC,EACnD,KACCkE,EAAIC,GAASzB,EAAM,KAAKyB,CAAK,CAAC,EAC9BC,EAAS,IAAM1B,EAAM,SAAS,CAAC,EAC/BhC,EAAIyD,GAAUE,EAAA,CAAE,IAAKvE,GAAOqE,EAAQ,CACtC,CACJ,CAAC,CACH,CC9RO,SAASG,GACdC,EAAkB,CAAE,UAAAC,EAAW,MAAAC,EAAO,QAAAC,CAAQ,EACvB,CAGvB,IAAMC,EAAaH,EAChB,KACCI,EAAI,CAAC,CAAE,OAAQ,CAAE,EAAAC,CAAE,CAAE,IAAMA,CAAC,EAC5BC,GAAY,EAAG,CAAC,EAChBF,EAAI,CAAC,CAAC,EAAGG,CAAC,IAAM,EAAIA,GAAKA,EAAI,CAAC,EAC9BC,EAAqB,CACvB,EAGIC,EAAUR,EACb,KACCG,EAAI,CAAC,CAAE,OAAAM,CAAO,IAAMA,CAAM,CAC5B,EAGF,OAAOC,EAAc,CAACF,EAASN,CAAU,CAAC,EACvC,KACCC,EAAI,CAAC,CAACM,EAAQE,CAAS,IAAM,EAAEF,GAAUE,EAAU,EACnDJ,EAAqB,EACrBK,EAAUX,EAAQ,KAAKY,GAAK,CAAC,CAAC,CAAC,EAC/BC,GAAQ,EAAI,EACZC,GAAO,CAAE,MAAO,GAAI,CAAC,EACrBZ,EAAIa,IAAW,CAAE,OAAAA,CAAO,EAAE,CAC5B,CACJ,CAYO,SAASC,GACdC,EAAiB,CAAE,UAAAnB,EAAW,QAAAoB,EAAS,MAAAnB,EAAO,QAAAC,CAAQ,EACpB,CAClC,IAAMmB,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,GAAe,EAAGT,GAAQ,EAAI,CAAC,EACxD,OAAAM,EAAM,UAAU,CAGd,KAAK,CAAE,OAAAJ,CAAO,EAAG,CACfE,EAAG,OAASF,EACRA,GACFE,EAAG,aAAa,WAAY,IAAI,EAChCA,EAAG,KAAK,GAERA,EAAG,gBAAgB,UAAU,CAEjC,EAGA,UAAW,CACTA,EAAG,MAAM,IAAM,GACfA,EAAG,OAAS,GACZA,EAAG,gBAAgB,UAAU,CAC/B,CACF,CAAC,EAGDC,EACG,KACCP,EAAUU,CAAK,EACfE,EAAwB,QAAQ,CAClC,EACG,UAAU,CAAC,CAAE,OAAAC,CAAO,IAAM,CACzBP,EAAG,MAAM,IAAM,GAAGO,EAAS,EAAE,IAC/B,CAAC,EAGLC,EAAUR,EAAI,OAAO,EAClB,UAAUS,GAAM,CACfA,EAAG,eAAe,EAClB,OAAO,SAAS,CAAE,IAAK,CAAE,CAAC,CAC5B,CAAC,EAGI9B,GAAeqB,EAAI,CAAE,UAAAnB,EAAW,MAAAC,EAAO,QAAAC,CAAQ,CAAC,EACpD,KACC2B,EAAIC,GAAST,EAAM,KAAKS,CAAK,CAAC,EAC9BC,EAAS,IAAMV,EAAM,SAAS,CAAC,EAC/BjB,EAAI0B,GAAUE,EAAA,CAAE,IAAKb,GAAOW,EAAQ,CACtC,CACJ,CCtHO,SAASG,GACd,CAAE,UAAAC,CAAU,EACN,CACNA,EACG,KACCC,EAAU,IAAMC,EAAY,cAAc,CAAC,EAC3CC,GAASC,GAAMC,GAAuBD,CAAE,EACrC,KACCE,EAAUN,EAAU,KAAKO,GAAK,CAAC,CAAC,CAAC,EACjCC,EAAOC,GAAWA,CAAO,EACzBC,EAAI,IAAMN,CAAE,EACZO,GAAK,CAAC,CACR,CACF,EACAH,EAAOJ,GAAMA,EAAG,YAAcA,EAAG,WAAW,EAC5CD,GAASC,GAAM,CACb,IAAMQ,EAAOR,EAAG,UACVS,EAAOT,EAAG,QAAQ,GAAG,GAAKA,EAChC,OAAAS,EAAK,MAAQD,EAGNE,GAAaD,CAAI,EACrB,KACCP,EAAUN,EAAU,KAAKO,GAAK,CAAC,CAAC,CAAC,EACjCQ,EAAS,IAAMF,EAAK,gBAAgB,OAAO,CAAC,CAC9C,CACJ,CAAC,CACH,EACG,UAAU,EAGfb,EACG,KACCC,EAAU,IAAMC,EAAY,YAAY,CAAC,EACzCC,GAASC,GAAMU,GAAaV,CAAE,CAAC,CACjC,EACG,UAAU,CACjB,CC3CO,SAASY,GACd,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EACf,CACND,EACG,KACCE,EAAU,IAAMC,EACd,2BACF,CAAC,EACDC,EAAIC,GAAM,CACRA,EAAG,cAAgB,GACnBA,EAAG,QAAU,EACf,CAAC,EACDC,GAASD,GAAME,EAAUF,EAAI,QAAQ,EAClC,KACCG,GAAU,IAAMH,EAAG,UAAU,SAAS,0BAA0B,CAAC,EACjEI,EAAI,IAAMJ,CAAE,CACd,CACF,EACAK,GAAeT,CAAO,CACxB,EACG,UAAU,CAAC,CAACI,EAAIM,CAAM,IAAM,CAC3BN,EAAG,UAAU,OAAO,0BAA0B,EAC1CM,IACFN,EAAG,QAAU,GACjB,CAAC,CACP,CC9BA,SAASO,IAAyB,CAChC,MAAO,qBAAqB,KAAK,UAAU,SAAS,CACtD,CAiBO,SAASC,GACd,CAAE,UAAAC,CAAU,EACN,CACNA,EACG,KACCC,EAAU,IAAMC,EAAY,qBAAqB,CAAC,EAClDC,EAAIC,GAAMA,EAAG,gBAAgB,mBAAmB,CAAC,EACjDC,EAAOP,EAAa,EACpBQ,GAASF,GAAMG,EAAUH,EAAI,YAAY,EACtC,KACCI,EAAI,IAAMJ,CAAE,CACd,CACF,CACF,EACG,UAAUA,GAAM,CACf,IAAMK,EAAML,EAAG,UAGXK,IAAQ,EACVL,EAAG,UAAY,EAGNK,EAAML,EAAG,eAAiBA,EAAG,eACtCA,EAAG,UAAYK,EAAM,EAEzB,CAAC,CACP,CCpCO,SAASC,GACd,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EACf,CACNC,EAAc,CAACC,GAAY,QAAQ,EAAGF,CAAO,CAAC,EAC3C,KACCG,EAAI,CAAC,CAACC,EAAQC,CAAM,IAAMD,GAAU,CAACC,CAAM,EAC3CC,EAAUF,GAAUG,EAAGH,CAAM,EAC1B,KACCI,GAAMJ,EAAS,IAAM,GAAG,CAC1B,CACF,EACAK,GAAeV,CAAS,CAC1B,EACG,UAAU,CAAC,CAACK,EAAQ,CAAE,OAAQ,CAAE,EAAAM,CAAE,CAAC,CAAC,IAAM,CACzC,GAAIN,EACF,SAAS,KAAK,aAAa,qBAAsB,EAAE,EACnD,SAAS,KAAK,MAAM,IAAM,IAAIM,CAAC,SAC1B,CACL,IAAMC,EAAQ,GAAK,SAAS,SAAS,KAAK,MAAM,IAAK,EAAE,EACvD,SAAS,KAAK,gBAAgB,oBAAoB,EAClD,SAAS,KAAK,MAAM,IAAM,GACtBA,GACF,OAAO,SAAS,EAAGA,CAAK,CAC5B,CACF,CAAC,CACP,CC7DK,OAAO,UACV,OAAO,QAAU,SAAUC,EAAa,CACtC,IAAMC,EAA2B,CAAC,EAClC,QAAWC,KAAO,OAAO,KAAKF,CAAG,EAE/BC,EAAK,KAAK,CAACC,EAAKF,EAAIE,CAAG,CAAC,CAAC,EAG3B,OAAOD,CACT,GAGG,OAAO,SACV,OAAO,OAAS,SAAUD,EAAa,CACrC,IAAMC,EAAiB,CAAC,EACxB,QAAWC,KAAO,OAAO,KAAKF,CAAG,EAE/BC,EAAK,KAAKD,EAAIE,CAAG,CAAC,EAGpB,OAAOD,CACT,GAKE,OAAO,SAAY,cAGhB,QAAQ,UAAU,WACrB,QAAQ,UAAU,SAAW,SAC3BE,EAA8BC,EACxB,CACF,OAAOD,GAAM,UACf,KAAK,WAAaA,EAAE,KACpB,KAAK,UAAYA,EAAE,MAEnB,KAAK,WAAaA,EAClB,KAAK,UAAYC,EAErB,GAGG,QAAQ,UAAU,cACrB,QAAQ,UAAU,YAAc,YAC3BC,EACG,CACN,IAAMC,EAAS,KAAK,WACpB,GAAIA,EAAQ,CACND,EAAM,SAAW,GACnBC,EAAO,YAAY,IAAI,EAGzB,QAASC,EAAIF,EAAM,OAAS,EAAGE,GAAK,EAAGA,IAAK,CAC1C,IAAIC,EAAOH,EAAME,CAAC,EACd,OAAOC,GAAS,SAClBA,EAAO,SAAS,eAAeA,CAAI,EAC5BA,EAAK,YACZA,EAAK,WAAW,YAAYA,CAAI,EAG7BD,EAGHD,EAAO,aAAa,KAAK,gBAAkBE,CAAI,EAF/CF,EAAO,aAAaE,EAAM,IAAI,CAGlC,CACF,CACF,IpMMJ,SAASC,IAA4C,CACnD,OAAI,SAAS,WAAa,QACjBC,GACL,GAAG,IAAI,IAAI,yBAA0BC,GAAO,IAAI,CAAC,EACnD,EACG,KAECC,EAAI,IAAM,OAAO,EACjBC,EAAY,CAAC,CACf,EAEKC,GACL,IAAI,IAAI,2BAA4BH,GAAO,IAAI,CACjD,CAEJ,CAOA,SAAS,gBAAgB,UAAU,OAAO,OAAO,EACjD,SAAS,gBAAgB,UAAU,IAAI,IAAI,EAG3C,IAAMI,GAAYC,GAAc,EAC1BC,GAAYC,GAAc,EAC1BC,GAAYC,GAAoBH,EAAS,EACzCI,GAAYC,GAAc,EAG1BC,GAAYC,GAAc,EAC1BC,GAAYC,GAAW,oBAAoB,EAC3CC,GAAYD,GAAW,qBAAqB,EAC5CE,GAAYC,GAAW,EAGvBlB,GAASmB,GAAc,EACvBC,GAAS,SAAS,MAAM,UAAU,QAAQ,EAC5CtB,GAAiB,EACjBuB,GAGEC,GAAS,IAAIC,EACnBC,GAAiB,CAAE,OAAAF,EAAO,CAAC,EAG3B,IAAMG,GAAY,IAAIF,EAGlBG,EAAQ,oBAAoB,GAC9BC,GAAuB,CAAE,UAAArB,GAAW,UAAAM,GAAW,UAAAa,EAAU,CAAC,EACvD,UAAUrB,EAAS,EAzJxB,IAAAwB,KA4JIA,GAAA5B,GAAO,UAAP,YAAA4B,GAAgB,YAAa,QAC/BC,GAAqB,CAAE,UAAAzB,EAAU,CAAC,EAGpC0B,EAAMxB,GAAWE,EAAO,EACrB,KACCuB,GAAM,GAAG,CACX,EACG,UAAU,IAAM,CACfC,GAAU,SAAU,EAAK,EACzBA,GAAU,SAAU,EAAK,CAC3B,CAAC,EAGLtB,GACG,KACCuB,EAAO,CAAC,CAAE,KAAAC,CAAK,IAAMA,IAAS,QAAQ,CACxC,EACG,UAAUC,GAAO,CAChB,OAAQA,EAAI,KAAM,CAGhB,IAAK,IACL,IAAK,IACH,IAAMC,EAAOC,GAAoC,gBAAgB,EAC7D,OAAOD,GAAS,aAClBE,GAAYF,CAAI,EAClB,MAGF,IAAK,IACL,IAAK,IACH,IAAMG,EAAOF,GAAoC,gBAAgB,EAC7D,OAAOE,GAAS,aAClBD,GAAYC,CAAI,EAClB,MAGF,IAAK,QACH,IAAMC,EAASC,GAAiB,EAC5BD,aAAkB,kBACpBA,EAAO,MAAM,CACnB,CACF,CAAC,EAGLE,GAAc,CAAE,UAAAtC,EAAU,CAAC,EAC3BuC,GAAmB,CAAE,UAAAvC,GAAW,QAAAU,EAAQ,CAAC,EACzC8B,GAAe,CAAE,UAAAxC,EAAU,CAAC,EAC5ByC,GAAgB,CAAE,UAAAjC,GAAW,QAAAE,EAAQ,CAAC,EAGtC,IAAMgC,GAAUC,GAAYC,GAAoB,QAAQ,EAAG,CAAE,UAAApC,EAAU,CAAC,EAClEqC,GAAQ7C,GACX,KACCH,EAAI,IAAM+C,GAAoB,MAAM,CAAC,EACrCE,EAAUC,GAAMC,GAAUD,EAAI,CAAE,UAAAvC,GAAW,QAAAkC,EAAQ,CAAC,CAAC,EACrD5C,EAAY,CAAC,CACf,EAGImD,GAAWvB,EAGf,GAAGwB,GAAqB,SAAS,EAC9B,IAAIH,GAAMI,GAAaJ,EAAI,CAAE,QAAA3C,EAAQ,CAAC,CAAC,EAG1C,GAAG8C,GAAqB,QAAQ,EAC7B,IAAIH,GAAMK,GAAYL,EAAI,CAAE,OAAA7B,EAAO,CAAC,CAAC,EAGxC,GAAGgC,GAAqB,QAAQ,EAC7B,IAAIH,GAAMM,GAAYN,EAAI,CAAE,UAAAvC,GAAW,QAAAkC,GAAS,MAAAG,EAAM,CAAC,CAAC,EAG3D,GAAGK,GAAqB,SAAS,EAC9B,IAAIH,GAAMO,GAAaP,CAAE,CAAC,EAG7B,GAAGG,GAAqB,UAAU,EAC/B,IAAIH,GAAMQ,GAAcR,EAAI,CAAE,UAAA1B,EAAU,CAAC,CAAC,EAG7C,GAAG6B,GAAqB,QAAQ,EAC7B,IAAIH,GAAMS,GAAYT,EAAI,CAAE,OAAA/B,GAAQ,UAAAV,EAAU,CAAC,CAAC,EAGnD,GAAG4C,GAAqB,QAAQ,EAC7B,IAAIH,GAAMU,GAAYV,CAAE,CAAC,CAC9B,EAGMW,GAAWC,EAAM,IAAMjC,EAG3B,GAAGwB,GAAqB,UAAU,EAC/B,IAAIH,GAAMa,GAAcb,CAAE,CAAC,EAG9B,GAAGG,GAAqB,SAAS,EAC9B,IAAIH,GAAMc,GAAad,EAAI,CAAE,UAAAvC,GAAW,QAAAJ,GAAS,OAAAS,EAAO,CAAC,CAAC,EAG7D,GAAGqC,GAAqB,SAAS,EAC9B,IAAIH,GAAMzB,EAAQ,kBAAkB,EACjCwC,GAAoBf,EAAI,CAAE,OAAA/B,GAAQ,UAAAd,EAAU,CAAC,EAC7C6D,CACJ,EAGF,GAAGb,GAAqB,cAAc,EACnC,IAAIH,GAAMiB,GAAiBjB,EAAI,CAAE,UAAAvC,GAAW,QAAAkC,EAAQ,CAAC,CAAC,EAGzD,GAAGQ,GAAqB,SAAS,EAC9B,IAAIH,GAAMA,EAAG,aAAa,cAAc,IAAM,aAC3CkB,GAAGrD,GAAS,IAAMsD,GAAanB,EAAI,CAAE,UAAAvC,GAAW,QAAAkC,GAAS,MAAAG,EAAM,CAAC,CAAC,EACjEoB,GAAGvD,GAAS,IAAMwD,GAAanB,EAAI,CAAE,UAAAvC,GAAW,QAAAkC,GAAS,MAAAG,EAAM,CAAC,CAAC,CACrE,EAGF,GAAGK,GAAqB,MAAM,EAC3B,IAAIH,GAAMoB,GAAUpB,EAAI,CAAE,UAAAvC,GAAW,QAAAkC,EAAQ,CAAC,CAAC,EAGlD,GAAGQ,GAAqB,KAAK,EAC1B,IAAIH,GAAMqB,GAAqBrB,EAAI,CAClC,UAAAvC,GAAW,QAAAkC,GAAS,MAAAG,GAAO,QAAAzC,EAC7B,CAAC,CAAC,EAGJ,GAAG8C,GAAqB,KAAK,EAC1B,IAAIH,GAAMsB,GAAetB,EAAI,CAAE,UAAAvC,GAAW,QAAAkC,GAAS,MAAAG,GAAO,QAAAzC,EAAQ,CAAC,CAAC,CACzE,CAAC,EAGKkE,GAAatE,GAChB,KACC8C,EAAU,IAAMY,EAAQ,EACxBa,GAAUtB,EAAQ,EAClBnD,EAAY,CAAC,CACf,EAGFwE,GAAW,UAAU,EAMrB,OAAO,UAAatE,GACpB,OAAO,UAAaE,GACpB,OAAO,QAAaE,GACpB,OAAO,UAAaE,GACpB,OAAO,UAAaE,GACpB,OAAO,QAAaE,GACpB,OAAO,QAAaE,GACpB,OAAO,OAAaC,GACpB,OAAO,OAAaK,GACpB,OAAO,UAAaG,GACpB,OAAO,WAAaiD",
+  "names": ["require_focus_visible", "__commonJSMin", "exports", "module", "global", "factory", "applyFocusVisiblePolyfill", "scope", "hadKeyboardEvent", "hadFocusVisibleRecently", "hadFocusVisibleRecentlyTimeout", "inputTypesAllowlist", "isValidFocusTarget", "el", "focusTriggersKeyboardModality", "type", "tagName", "addFocusVisibleClass", "removeFocusVisibleClass", "onKeyDown", "e", "onPointerDown", "onFocus", "onBlur", "onVisibilityChange", "addInitialPointerMoveListeners", "onInitialPointerMove", "removeInitialPointerMoveListeners", "event", "error", "require_clipboard", "__commonJSMin", "exports", "module", "root", "factory", "__webpack_modules__", "__unused_webpack_module", "__webpack_exports__", "__webpack_require__", "clipboard", "tiny_emitter", "tiny_emitter_default", "listen", "listen_default", "src_select", "select_default", "command", "type", "err", "ClipboardActionCut", "target", "selectedText", "actions_cut", "createFakeElement", "value", "isRTL", "fakeElement", "yPosition", "fakeCopyAction", "options", "ClipboardActionCopy", "actions_copy", "_typeof", "obj", "ClipboardActionDefault", "_options$action", "action", "container", "text", "actions_default", "clipboard_typeof", "_classCallCheck", "instance", "Constructor", "_defineProperties", "props", "i", "descriptor", "_createClass", "protoProps", "staticProps", "_inherits", "subClass", "superClass", "_setPrototypeOf", "o", "p", "_createSuper", "Derived", "hasNativeReflectConstruct", "_isNativeReflectConstruct", "Super", "_getPrototypeOf", "result", "NewTarget", "_possibleConstructorReturn", "self", "call", "_assertThisInitialized", "e", "getAttributeValue", "suffix", "element", "attribute", "Clipboard", "_Emitter", "_super", "trigger", "_this", "_this2", "selector", "actions", "support", "DOCUMENT_NODE_TYPE", "proto", "closest", "__unused_webpack_exports", "_delegate", "callback", "useCapture", "listenerFn", "listener", "delegate", "elements", "is", "listenNode", "listenNodeList", "listenSelector", "node", "nodeList", "select", "isReadOnly", "selection", "range", "E", "name", "ctx", "data", "evtArr", "len", "evts", "liveEvents", "__webpack_module_cache__", "moduleId", "getter", "definition", "key", "prop", "require_escape_html", "__commonJSMin", "exports", "module", "matchHtmlRegExp", "escapeHtml", "string", "str", "match", "escape", "html", "index", "lastIndex", "import_focus_visible", "extendStatics", "d", "b", "p", "__extends", "__", "__awaiter", "thisArg", "_arguments", "P", "generator", "adopt", "value", "resolve", "reject", "fulfilled", "step", "e", "rejected", "result", "__generator", "body", "_", "t", "f", "y", "g", "verb", "n", "v", "op", "__values", "o", "s", "m", "i", "__read", "n", "r", "ar", "e", "error", "__spreadArray", "to", "from", "pack", "i", "l", "ar", "__await", "v", "__asyncGenerator", "thisArg", "_arguments", "generator", "g", "q", "verb", "n", "a", "b", "resume", "step", "e", "settle", "r", "fulfill", "reject", "value", "f", "__asyncValues", "o", "m", "i", "__values", "verb", "n", "v", "resolve", "reject", "settle", "d", "isFunction", "value", "createErrorClass", "createImpl", "_super", "instance", "ctorFunc", "UnsubscriptionError", "createErrorClass", "_super", "errors", "err", "i", "arrRemove", "arr", "item", "index", "Subscription", "initialTeardown", "errors", "_parentage", "_parentage_1", "__values", "_parentage_1_1", "parent_1", "initialFinalizer", "isFunction", "e", "UnsubscriptionError", "_finalizers", "_finalizers_1", "_finalizers_1_1", "finalizer", "execFinalizer", "err", "__spreadArray", "__read", "teardown", "_a", "parent", "arrRemove", "empty", "EMPTY_SUBSCRIPTION", "Subscription", "isSubscription", "value", "isFunction", "execFinalizer", "finalizer", "config", "timeoutProvider", "handler", "timeout", "args", "_i", "delegate", "__spreadArray", "__read", "handle", "reportUnhandledError", "err", "timeoutProvider", "onUnhandledError", "config", "noop", "COMPLETE_NOTIFICATION", "createNotification", "errorNotification", "error", "nextNotification", "value", "kind", "context", "errorContext", "cb", "config", "isRoot", "_a", "errorThrown", "error", "captureError", "err", "Subscriber", "_super", "__extends", "destination", "_this", "isSubscription", "EMPTY_OBSERVER", "next", "error", "complete", "SafeSubscriber", "value", "handleStoppedNotification", "nextNotification", "err", "errorNotification", "COMPLETE_NOTIFICATION", "Subscription", "_bind", "bind", "fn", "thisArg", "ConsumerObserver", "partialObserver", "value", "error", "handleUnhandledError", "err", "SafeSubscriber", "_super", "__extends", "observerOrNext", "complete", "_this", "isFunction", "context_1", "config", "Subscriber", "handleUnhandledError", "error", "config", "captureError", "reportUnhandledError", "defaultErrorHandler", "err", "handleStoppedNotification", "notification", "subscriber", "onStoppedNotification", "timeoutProvider", "EMPTY_OBSERVER", "noop", "observable", "identity", "x", "pipe", "fns", "_i", "pipeFromArray", "identity", "input", "prev", "fn", "Observable", "subscribe", "operator", "observable", "observerOrNext", "error", "complete", "_this", "subscriber", "isSubscriber", "SafeSubscriber", "errorContext", "_a", "source", "sink", "err", "next", "promiseCtor", "getPromiseCtor", "resolve", "reject", "value", "operations", "_i", "pipeFromArray", "x", "getPromiseCtor", "promiseCtor", "_a", "config", "isObserver", "value", "isFunction", "isSubscriber", "Subscriber", "isSubscription", "hasLift", "source", "isFunction", "operate", "init", "liftedSource", "err", "createOperatorSubscriber", "destination", "onNext", "onComplete", "onError", "onFinalize", "OperatorSubscriber", "_super", "__extends", "shouldUnsubscribe", "_this", "value", "err", "closed_1", "_a", "Subscriber", "animationFrameProvider", "callback", "request", "cancel", "delegate", "handle", "timestamp", "Subscription", "args", "_i", "__spreadArray", "__read", "ObjectUnsubscribedError", "createErrorClass", "_super", "Subject", "_super", "__extends", "_this", "operator", "subject", "AnonymousSubject", "ObjectUnsubscribedError", "value", "errorContext", "_b", "__values", "_c", "observer", "err", "observers", "_a", "subscriber", "hasError", "isStopped", "EMPTY_SUBSCRIPTION", "Subscription", "arrRemove", "thrownError", "observable", "Observable", "destination", "source", "AnonymousSubject", "_super", "__extends", "destination", "source", "_this", "value", "_b", "_a", "err", "subscriber", "EMPTY_SUBSCRIPTION", "Subject", "dateTimestampProvider", "ReplaySubject", "_super", "__extends", "_bufferSize", "_windowTime", "_timestampProvider", "dateTimestampProvider", "_this", "value", "_a", "isStopped", "_buffer", "_infiniteTimeWindow", "subscriber", "subscription", "copy", "i", "adjustedBufferSize", "now", "last", "Subject", "Action", "_super", "__extends", "scheduler", "work", "state", "delay", "Subscription", "intervalProvider", "handler", "timeout", "args", "_i", "delegate", "__spreadArray", "__read", "handle", "AsyncAction", "_super", "__extends", "scheduler", "work", "_this", "state", "delay", "id", "_a", "_id", "intervalProvider", "_scheduler", "error", "_delay", "errored", "errorValue", "e", "actions", "arrRemove", "Action", "Scheduler", "schedulerActionCtor", "now", "work", "delay", "state", "dateTimestampProvider", "AsyncScheduler", "_super", "__extends", "SchedulerAction", "now", "Scheduler", "_this", "action", "actions", "error", "asyncScheduler", "AsyncScheduler", "AsyncAction", "async", "AnimationFrameAction", "_super", "__extends", "scheduler", "work", "_this", "id", "delay", "animationFrameProvider", "actions", "_a", "AsyncAction", "AnimationFrameScheduler", "_super", "__extends", "action", "flushId", "actions", "error", "AsyncScheduler", "animationFrameScheduler", "AnimationFrameScheduler", "AnimationFrameAction", "EMPTY", "Observable", "subscriber", "isScheduler", "value", "isFunction", "last", "arr", "popResultSelector", "args", "isFunction", "popScheduler", "isScheduler", "popNumber", "defaultValue", "isArrayLike", "x", "isPromise", "value", "isFunction", "isInteropObservable", "input", "isFunction", "observable", "isAsyncIterable", "obj", "isFunction", "createInvalidObservableTypeError", "input", "getSymbolIterator", "iterator", "isIterable", "input", "isFunction", "iterator", "readableStreamLikeToAsyncGenerator", "readableStream", "reader", "__await", "_a", "_b", "value", "done", "isReadableStreamLike", "obj", "isFunction", "innerFrom", "input", "Observable", "isInteropObservable", "fromInteropObservable", "isArrayLike", "fromArrayLike", "isPromise", "fromPromise", "isAsyncIterable", "fromAsyncIterable", "isIterable", "fromIterable", "isReadableStreamLike", "fromReadableStreamLike", "createInvalidObservableTypeError", "obj", "subscriber", "obs", "observable", "isFunction", "array", "i", "promise", "value", "err", "reportUnhandledError", "iterable", "iterable_1", "__values", "iterable_1_1", "asyncIterable", "process", "readableStream", "readableStreamLikeToAsyncGenerator", "asyncIterable_1", "__asyncValues", "asyncIterable_1_1", "executeSchedule", "parentSubscription", "scheduler", "work", "delay", "repeat", "scheduleSubscription", "observeOn", "scheduler", "delay", "operate", "source", "subscriber", "createOperatorSubscriber", "value", "executeSchedule", "err", "subscribeOn", "scheduler", "delay", "operate", "source", "subscriber", "scheduleObservable", "input", "scheduler", "innerFrom", "subscribeOn", "observeOn", "schedulePromise", "input", "scheduler", "innerFrom", "subscribeOn", "observeOn", "scheduleArray", "input", "scheduler", "Observable", "subscriber", "i", "scheduleIterable", "input", "scheduler", "Observable", "subscriber", "iterator", "executeSchedule", "value", "done", "_a", "err", "isFunction", "scheduleAsyncIterable", "input", "scheduler", "Observable", "subscriber", "executeSchedule", "iterator", "result", "scheduleReadableStreamLike", "input", "scheduler", "scheduleAsyncIterable", "readableStreamLikeToAsyncGenerator", "scheduled", "input", "scheduler", "isInteropObservable", "scheduleObservable", "isArrayLike", "scheduleArray", "isPromise", "schedulePromise", "isAsyncIterable", "scheduleAsyncIterable", "isIterable", "scheduleIterable", "isReadableStreamLike", "scheduleReadableStreamLike", "createInvalidObservableTypeError", "from", "input", "scheduler", "scheduled", "innerFrom", "of", "args", "_i", "scheduler", "popScheduler", "from", "throwError", "errorOrErrorFactory", "scheduler", "errorFactory", "isFunction", "init", "subscriber", "Observable", "EmptyError", "createErrorClass", "_super", "isValidDate", "value", "map", "project", "thisArg", "operate", "source", "subscriber", "index", "createOperatorSubscriber", "value", "isArray", "callOrApply", "fn", "args", "__spreadArray", "__read", "mapOneOrManyArgs", "map", "isArray", "getPrototypeOf", "objectProto", "getKeys", "argsArgArrayOrObject", "args", "first_1", "isPOJO", "keys", "key", "obj", "createObject", "keys", "values", "result", "key", "i", "combineLatest", "args", "_i", "scheduler", "popScheduler", "resultSelector", "popResultSelector", "_a", "argsArgArrayOrObject", "observables", "keys", "from", "result", "Observable", "combineLatestInit", "values", "createObject", "identity", "mapOneOrManyArgs", "valueTransform", "subscriber", "maybeSchedule", "length", "active", "remainingFirstValues", "i", "source", "hasFirstValue", "createOperatorSubscriber", "value", "execute", "subscription", "executeSchedule", "mergeInternals", "source", "subscriber", "project", "concurrent", "onBeforeNext", "expand", "innerSubScheduler", "additionalFinalizer", "buffer", "active", "index", "isComplete", "checkComplete", "outerNext", "value", "doInnerSub", "innerComplete", "innerFrom", "createOperatorSubscriber", "innerValue", "bufferedValue", "executeSchedule", "err", "mergeMap", "project", "resultSelector", "concurrent", "isFunction", "a", "i", "map", "b", "ii", "innerFrom", "operate", "source", "subscriber", "mergeInternals", "mergeAll", "concurrent", "mergeMap", "identity", "concatAll", "mergeAll", "concat", "args", "_i", "concatAll", "from", "popScheduler", "defer", "observableFactory", "Observable", "subscriber", "innerFrom", "nodeEventEmitterMethods", "eventTargetMethods", "jqueryMethods", "fromEvent", "target", "eventName", "options", "resultSelector", "isFunction", "mapOneOrManyArgs", "_a", "__read", "isEventTarget", "methodName", "handler", "isNodeStyleEventEmitter", "toCommonHandlerRegistry", "isJQueryStyleEventEmitter", "add", "remove", "isArrayLike", "mergeMap", "subTarget", "innerFrom", "Observable", "subscriber", "args", "_i", "fromEventPattern", "addHandler", "removeHandler", "resultSelector", "mapOneOrManyArgs", "Observable", "subscriber", "handler", "e", "_i", "retValue", "isFunction", "timer", "dueTime", "intervalOrScheduler", "scheduler", "async", "intervalDuration", "isScheduler", "Observable", "subscriber", "due", "isValidDate", "n", "merge", "args", "_i", "scheduler", "popScheduler", "concurrent", "popNumber", "sources", "innerFrom", "mergeAll", "from", "EMPTY", "NEVER", "Observable", "noop", "isArray", "argsOrArgArray", "args", "filter", "predicate", "thisArg", "operate", "source", "subscriber", "index", "createOperatorSubscriber", "value", "zip", "args", "_i", "resultSelector", "popResultSelector", "sources", "argsOrArgArray", "Observable", "subscriber", "buffers", "completed", "sourceIndex", "innerFrom", "createOperatorSubscriber", "value", "buffer", "result", "__spreadArray", "__read", "i", "EMPTY", "audit", "durationSelector", "operate", "source", "subscriber", "hasValue", "lastValue", "durationSubscriber", "isComplete", "endDuration", "value", "cleanupDuration", "createOperatorSubscriber", "innerFrom", "auditTime", "duration", "scheduler", "asyncScheduler", "audit", "timer", "bufferCount", "bufferSize", "startBufferEvery", "operate", "source", "subscriber", "buffers", "count", "createOperatorSubscriber", "value", "toEmit", "buffers_1", "__values", "buffers_1_1", "buffer", "toEmit_1", "toEmit_1_1", "arrRemove", "buffers_2", "buffers_2_1", "catchError", "selector", "operate", "source", "subscriber", "innerSub", "syncUnsub", "handledResult", "createOperatorSubscriber", "err", "innerFrom", "scanInternals", "accumulator", "seed", "hasSeed", "emitOnNext", "emitBeforeComplete", "source", "subscriber", "hasState", "state", "index", "createOperatorSubscriber", "value", "i", "combineLatest", "args", "_i", "resultSelector", "popResultSelector", "pipe", "__spreadArray", "__read", "mapOneOrManyArgs", "operate", "source", "subscriber", "combineLatestInit", "argsOrArgArray", "combineLatestWith", "otherSources", "_i", "combineLatest", "__spreadArray", "__read", "debounceTime", "dueTime", "scheduler", "asyncScheduler", "operate", "source", "subscriber", "activeTask", "lastValue", "lastTime", "emit", "value", "emitWhenIdle", "targetTime", "now", "createOperatorSubscriber", "defaultIfEmpty", "defaultValue", "operate", "source", "subscriber", "hasValue", "createOperatorSubscriber", "value", "take", "count", "EMPTY", "operate", "source", "subscriber", "seen", "createOperatorSubscriber", "value", "ignoreElements", "operate", "source", "subscriber", "createOperatorSubscriber", "noop", "mapTo", "value", "map", "delayWhen", "delayDurationSelector", "subscriptionDelay", "source", "concat", "take", "ignoreElements", "mergeMap", "value", "index", "innerFrom", "mapTo", "delay", "due", "scheduler", "asyncScheduler", "duration", "timer", "delayWhen", "distinctUntilChanged", "comparator", "keySelector", "identity", "defaultCompare", "operate", "source", "subscriber", "previousKey", "first", "createOperatorSubscriber", "value", "currentKey", "a", "b", "distinctUntilKeyChanged", "key", "compare", "distinctUntilChanged", "x", "y", "throwIfEmpty", "errorFactory", "defaultErrorFactory", "operate", "source", "subscriber", "hasValue", "createOperatorSubscriber", "value", "EmptyError", "endWith", "values", "_i", "source", "concat", "of", "__spreadArray", "__read", "finalize", "callback", "operate", "source", "subscriber", "first", "predicate", "defaultValue", "hasDefaultValue", "source", "filter", "v", "identity", "take", "defaultIfEmpty", "throwIfEmpty", "EmptyError", "takeLast", "count", "EMPTY", "operate", "source", "subscriber", "buffer", "createOperatorSubscriber", "value", "buffer_1", "__values", "buffer_1_1", "merge", "args", "_i", "scheduler", "popScheduler", "concurrent", "popNumber", "argsOrArgArray", "operate", "source", "subscriber", "mergeAll", "from", "__spreadArray", "__read", "mergeWith", "otherSources", "_i", "merge", "__spreadArray", "__read", "repeat", "countOrConfig", "count", "delay", "_a", "EMPTY", "operate", "source", "subscriber", "soFar", "sourceSub", "resubscribe", "notifier", "timer", "innerFrom", "notifierSubscriber_1", "createOperatorSubscriber", "subscribeToSource", "syncUnsub", "scan", "accumulator", "seed", "operate", "scanInternals", "share", "options", "_a", "connector", "Subject", "_b", "resetOnError", "_c", "resetOnComplete", "_d", "resetOnRefCountZero", "wrapperSource", "connection", "resetConnection", "subject", "refCount", "hasCompleted", "hasErrored", "cancelReset", "reset", "resetAndUnsubscribe", "conn", "operate", "source", "subscriber", "dest", "handleReset", "SafeSubscriber", "value", "err", "innerFrom", "on", "args", "_i", "onSubscriber", "__spreadArray", "__read", "shareReplay", "configOrBufferSize", "windowTime", "scheduler", "bufferSize", "refCount", "_a", "_b", "_c", "share", "ReplaySubject", "skip", "count", "filter", "_", "index", "skipUntil", "notifier", "operate", "source", "subscriber", "taking", "skipSubscriber", "createOperatorSubscriber", "noop", "innerFrom", "value", "startWith", "values", "_i", "scheduler", "popScheduler", "operate", "source", "subscriber", "concat", "switchMap", "project", "resultSelector", "operate", "source", "subscriber", "innerSubscriber", "index", "isComplete", "checkComplete", "createOperatorSubscriber", "value", "innerIndex", "outerIndex", "innerFrom", "innerValue", "takeUntil", "notifier", "operate", "source", "subscriber", "innerFrom", "createOperatorSubscriber", "noop", "takeWhile", "predicate", "inclusive", "operate", "source", "subscriber", "index", "createOperatorSubscriber", "value", "result", "tap", "observerOrNext", "error", "complete", "tapObserver", "isFunction", "operate", "source", "subscriber", "_a", "isUnsub", "createOperatorSubscriber", "value", "err", "_b", "identity", "throttle", "durationSelector", "config", "operate", "source", "subscriber", "_a", "_b", "leading", "_c", "trailing", "hasValue", "sendValue", "throttled", "isComplete", "endThrottling", "send", "cleanupThrottling", "startThrottle", "value", "innerFrom", "createOperatorSubscriber", "throttleTime", "duration", "scheduler", "config", "asyncScheduler", "duration$", "timer", "throttle", "withLatestFrom", "inputs", "_i", "project", "popResultSelector", "operate", "source", "subscriber", "len", "otherValues", "hasValue", "ready", "i", "innerFrom", "createOperatorSubscriber", "value", "identity", "noop", "values", "__spreadArray", "__read", "zip", "sources", "_i", "operate", "source", "subscriber", "__spreadArray", "__read", "zipWith", "otherInputs", "_i", "zip", "__spreadArray", "__read", "watchDocument", "document$", "ReplaySubject", "fromEvent", "getElements", "selector", "node", "getElement", "el", "getOptionalElement", "getActiveElement", "_a", "_b", "_c", "_d", "observer$", "merge", "fromEvent", "debounceTime", "startWith", "map", "getActiveElement", "shareReplay", "watchElementFocus", "el", "active", "distinctUntilChanged", "watchElementHover", "el", "duration", "merge", "fromEvent", "map", "debounceTime", "identity", "startWith", "getElementOffset", "el", "watchElementOffset", "merge", "fromEvent", "auditTime", "animationFrameScheduler", "map", "startWith", "getElementContentOffset", "el", "watchElementContentOffset", "merge", "fromEvent", "auditTime", "animationFrameScheduler", "map", "startWith", "appendChild", "el", "child", "node", "h", "tag", "attributes", "children", "attr", "round", "value", "digits", "watchScript", "src", "script", "h", "defer", "merge", "fromEvent", "switchMap", "throwError", "map", "finalize", "take", "entry$", "Subject", "observer$", "defer", "watchScript", "of", "map", "entries", "entry", "switchMap", "observer", "merge", "NEVER", "finalize", "shareReplay", "getElementSize", "el", "watchElementSize", "tap", "filter", "target", "startWith", "getElementContentSize", "el", "getElementContainer", "parent", "entry$", "Subject", "observer$", "defer", "of", "entries", "entry", "switchMap", "observer", "merge", "NEVER", "finalize", "shareReplay", "watchElementVisibility", "el", "tap", "filter", "target", "map", "isIntersecting", "watchElementBoundary", "threshold", "watchElementContentOffset", "y", "visible", "getElementSize", "content", "getElementContentSize", "distinctUntilChanged", "toggles", "getElement", "getToggle", "name", "setToggle", "value", "watchToggle", "el", "fromEvent", "map", "startWith", "isSusceptibleToKeyboard", "el", "type", "watchComposition", "merge", "fromEvent", "map", "startWith", "watchKeyboard", "keyboard$", "filter", "ev", "getToggle", "mode", "active", "getActiveElement", "share", "switchMap", "EMPTY", "getLocation", "setLocation", "url", "navigate", "feature", "el", "h", "watchLocation", "Subject", "getLocationHash", "setLocationHash", "hash", "el", "h", "ev", "watchLocationHash", "location$", "merge", "fromEvent", "map", "startWith", "filter", "shareReplay", "watchLocationTarget", "id", "getOptionalElement", "watchMedia", "query", "media", "fromEventPattern", "next", "startWith", "watchPrint", "merge", "fromEvent", "map", "at", "query$", "factory", "switchMap", "active", "EMPTY", "request", "url", "options", "Observable", "observer", "req", "event", "_a", "length", "requestJSON", "switchMap", "res", "map", "body", "shareReplay", "requestHTML", "dom", "requestXML", "getViewportOffset", "watchViewportOffset", "merge", "fromEvent", "map", "startWith", "getViewportSize", "watchViewportSize", "fromEvent", "map", "startWith", "watchViewport", "combineLatest", "watchViewportOffset", "watchViewportSize", "map", "offset", "size", "shareReplay", "watchViewportAt", "el", "viewport$", "header$", "size$", "distinctUntilKeyChanged", "offset$", "combineLatest", "map", "getElementOffset", "height", "offset", "size", "x", "y", "recv", "worker", "fromEvent", "ev", "send", "send$", "Subject", "data", "watchWorker", "url", "recv$", "worker$", "done$", "ignoreElements", "endWith", "mergeWith", "takeUntil", "share", "script", "getElement", "config", "getLocation", "configuration", "feature", "flag", "translation", "key", "value", "getComponentElement", "type", "node", "getElement", "getComponentElements", "getElements", "watchAnnounce", "el", "button", "getElement", "fromEvent", "map", "content", "mountAnnounce", "feature", "EMPTY", "defer", "push$", "Subject", "hash", "tap", "state", "finalize", "__spreadValues", "watchConsent", "el", "target$", "map", "target", "mountConsent", "options", "internal$", "Subject", "hidden", "tap", "state", "finalize", "__spreadValues", "renderTooltip", "id", "style", "h", "renderAnnotation", "id", "prefix", "anchor", "h", "renderTooltip", "renderClipboardButton", "id", "h", "translation", "renderSearchDocument", "document", "flag", "parent", "teaser", "missing", "key", "list", "h", "config", "configuration", "url", "feature", "match", "highlight", "value", "tags", "tag", "type", "translation", "renderSearchResultItem", "result", "threshold", "docs", "doc", "article", "index", "best", "more", "children", "section", "renderSourceFacts", "facts", "h", "key", "value", "round", "renderTabbedControl", "type", "classes", "h", "renderTable", "table", "h", "renderVersion", "version", "config", "configuration", "url", "h", "renderVersionSelector", "versions", "active", "_a", "translation", "sequence", "watchTooltip", "el", "host", "width", "getElementSize", "container", "getElementContainer", "scroll$", "watchElementContentOffset", "of", "active$", "merge", "watchElementFocus", "watchElementHover", "distinctUntilChanged", "combineLatest", "map", "active", "scroll", "x", "y", "getElementOffset", "size", "table", "mountTooltip", "title", "EMPTY", "id", "tooltip", "renderTooltip", "typeset", "getElement", "defer", "push$", "Subject", "offset", "filter", "debounceTime", "auditTime", "animationFrameScheduler", "throttleTime", "origin", "tap", "state", "finalize", "__spreadValues", "subscribeOn", "asyncScheduler", "watchAnnotation", "el", "container", "offset$", "defer", "combineLatest", "watchElementOffset", "watchElementContentOffset", "map", "x", "y", "scroll", "width", "height", "getElementSize", "watchElementFocus", "switchMap", "active", "offset", "take", "mountAnnotation", "target$", "tooltip", "index", "push$", "Subject", "done$", "ignoreElements", "endWith", "watchElementVisibility", "takeUntil", "visible", "merge", "filter", "debounceTime", "auditTime", "animationFrameScheduler", "throttleTime", "origin", "fromEvent", "ev", "withLatestFrom", "_a", "parent", "getActiveElement", "target", "delay", "tap", "state", "finalize", "__spreadValues", "findHosts", "container", "getElements", "findMarkers", "markers", "el", "nodes", "it", "node", "text", "match", "id", "force", "marker", "swap", "source", "target", "mountAnnotationList", "target$", "print$", "parent", "prefix", "annotations", "getOptionalElement", "renderAnnotation", "EMPTY", "defer", "push$", "Subject", "done$", "ignoreElements", "endWith", "pairs", "annotation", "getElement", "takeUntil", "active", "inner", "child", "merge", "mountAnnotation", "finalize", "share", "findList", "el", "sibling", "mountAnnotationBlock", "options", "defer", "list", "mountAnnotationList", "EMPTY", "import_clipboard", "sequence", "findCandidateList", "el", "sibling", "watchCodeBlock", "watchElementSize", "map", "width", "getElementContentSize", "distinctUntilKeyChanged", "mountCodeBlock", "options", "hover", "factory$", "defer", "push$", "Subject", "done$", "takeLast", "scrollable", "content$", "ClipboardJS", "feature", "parent", "button", "renderClipboardButton", "mountTooltip", "container", "list", "annotations$", "mountAnnotationList", "takeUntil", "height", "distinctUntilChanged", "switchMap", "active", "EMPTY", "tap", "state", "finalize", "__spreadValues", "mergeWith", "watchElementVisibility", "filter", "visible", "take", "watchDetails", "el", "target$", "print$", "open", "merge", "map", "target", "filter", "details", "active", "tap", "mountDetails", "options", "defer", "push$", "Subject", "action", "reveal", "state", "finalize", "__spreadValues", "mermaid_default", "mermaid$", "sequence", "fetchScripts", "watchScript", "of", "mountMermaid", "el", "tap", "mermaid_default", "map", "shareReplay", "__async", "id", "host", "h", "text", "svg", "fn", "shadow", "sentinel", "h", "mountDataTable", "el", "renderTable", "of", "watchContentTabs", "inputs", "initial", "input", "merge", "fromEvent", "map", "getElement", "startWith", "active", "mountContentTabs", "el", "viewport$", "target$", "container", "getElements", "prev", "renderTabbedControl", "next", "defer", "push$", "Subject", "done$", "ignoreElements", "endWith", "combineLatest", "watchElementSize", "takeUntil", "auditTime", "animationFrameScheduler", "size", "offset", "getElementOffset", "width", "getElementSize", "content", "getElementContentOffset", "watchElementContentOffset", "getElementContentSize", "direction", "filter", "label", "h", "ev", "tap", "feature", "skip", "withLatestFrom", "tab", "y", "set", "tabs", "media", "state", "finalize", "__spreadValues", "subscribeOn", "asyncScheduler", "mountContent", "el", "viewport$", "target$", "print$", "merge", "getElements", "child", "mountAnnotationBlock", "mountCodeBlock", "mountMermaid", "mountDataTable", "mountDetails", "mountContentTabs", "feature", "mountTooltip", "watchDialog", "_el", "alert$", "switchMap", "message", "merge", "of", "delay", "map", "active", "mountDialog", "el", "options", "inner", "getElement", "defer", "push$", "Subject", "tap", "state", "finalize", "__spreadValues", "isHidden", "viewport$", "feature", "of", "direction$", "map", "y", "bufferCount", "a", "b", "distinctUntilKeyChanged", "hidden$", "combineLatest", "filter", "offset", "direction", "distinctUntilChanged", "search$", "watchToggle", "search", "switchMap", "active", "startWith", "watchHeader", "el", "options", "defer", "watchElementSize", "height", "hidden", "shareReplay", "mountHeader", "header$", "main$", "push$", "Subject", "done$", "ignoreElements", "endWith", "combineLatestWith", "tooltips", "from", "getElements", "mergeMap", "child", "mountTooltip", "takeUntil", "state", "__spreadValues", "mergeWith", "watchHeaderTitle", "el", "viewport$", "header$", "watchViewportAt", "map", "y", "height", "getElementSize", "distinctUntilKeyChanged", "mountHeaderTitle", "options", "defer", "push$", "Subject", "active", "heading", "getOptionalElement", "EMPTY", "tap", "state", "finalize", "__spreadValues", "watchMain", "el", "viewport$", "header$", "adjust$", "map", "height", "distinctUntilChanged", "border$", "switchMap", "watchElementSize", "distinctUntilKeyChanged", "combineLatest", "header", "top", "bottom", "y", "a", "b", "watchPalette", "inputs", "current", "input", "index", "of", "mergeMap", "fromEvent", "map", "startWith", "shareReplay", "mountPalette", "el", "getElements", "meta", "h", "scheme", "media$", "watchMedia", "defer", "push$", "Subject", "palette", "media", "key", "value", "label", "filter", "ev", "withLatestFrom", "_", "header", "getComponentElement", "style", "color", "observeOn", "asyncScheduler", "takeUntil", "skip", "repeat", "tap", "state", "finalize", "__spreadValues", "mountProgress", "el", "progress$", "defer", "push$", "Subject", "value", "tap", "finalize", "map", "import_clipboard", "extract", "el", "copy", "text", "setupClipboardJS", "alert$", "ClipboardJS", "Observable", "subscriber", "getElement", "ev", "tap", "map", "translation", "resolve", "url", "base", "extract", "document", "sitemap", "el", "getElements", "getElement", "links", "link", "href", "fetchSitemap", "requestXML", "map", "catchError", "of", "handle", "ev", "sitemap", "EMPTY", "el", "url", "of", "head", "document", "tags", "getElements", "resolve", "key", "value", "inject", "next", "selector", "feature", "source", "getOptionalElement", "target", "html", "name", "container", "getComponentElement", "concat", "switchMap", "script", "Observable", "observer", "ignoreElements", "endWith", "setupInstantNavigation", "location$", "viewport$", "progress$", "config", "configuration", "sitemap$", "fetchSitemap", "instant$", "fromEvent", "combineLatestWith", "share", "history$", "map", "getLocation", "withLatestFrom", "offset", "merge", "document$", "distinctUntilKeyChanged", "requestHTML", "catchError", "setLocation", "_", "distinctUntilChanged", "a", "b", "tap", "_a", "_b", "setLocationHash", "debounceTime", "import_escape_html", "setupSearchHighlighter", "config", "regex", "term", "separator", "highlight", "_", "data", "query", "match", "value", "escapeHTML", "isSearchReadyMessage", "message", "isSearchResultMessage", "setupSearchWorker", "url", "index$", "worker$", "watchWorker", "merge", "of", "watchToggle", "first", "active", "switchMap", "config", "docs", "feature", "setupVersionSelector", "document$", "config", "configuration", "versions$", "requestJSON", "catchError", "EMPTY", "current$", "map", "versions", "current", "version", "aliases", "switchMap", "urls", "fromEvent", "filter", "ev", "withLatestFrom", "el", "url", "of", "fetchSitemap", "sitemap", "path", "getLocation", "setLocation", "combineLatest", "getElement", "renderVersionSelector", "_a", "outdated", "ignored", "main", "ignore", "warning", "getComponentElements", "watchSearchQuery", "el", "worker$", "searchParams", "getLocation", "setToggle", "watchToggle", "first", "active", "url", "focus$", "watchElementFocus", "value$", "merge", "isSearchReadyMessage", "fromEvent", "map", "distinctUntilChanged", "combineLatest", "value", "focus", "shareReplay", "mountSearchQuery", "push$", "Subject", "done$", "ignoreElements", "endWith", "_", "query", "distinctUntilKeyChanged", "takeUntil", "label", "getElement", "tap", "state", "finalize", "__spreadValues", "mountSearchResult", "el", "worker$", "query$", "push$", "Subject", "boundary$", "watchElementBoundary", "filter", "container", "meta", "getElement", "list", "watchToggle", "active", "withLatestFrom", "skipUntil", "first", "isSearchReadyMessage", "items", "value", "translation", "count", "round", "render$", "tap", "switchMap", "merge", "of", "bufferCount", "zipWith", "chunk", "map", "renderSearchResultItem", "share", "item", "mergeMap", "details", "getOptionalElement", "EMPTY", "fromEvent", "takeUntil", "isSearchResultMessage", "data", "state", "finalize", "__spreadValues", "watchSearchShare", "_el", "query$", "map", "value", "url", "getLocation", "mountSearchShare", "el", "options", "push$", "Subject", "done$", "ignoreElements", "endWith", "fromEvent", "takeUntil", "ev", "tap", "state", "finalize", "__spreadValues", "mountSearchSuggest", "el", "worker$", "keyboard$", "push$", "Subject", "query", "getComponentElement", "query$", "merge", "fromEvent", "observeOn", "asyncScheduler", "map", "distinctUntilChanged", "combineLatestWith", "suggest", "value", "words", "last", "filter", "mode", "key", "isSearchResultMessage", "data", "tap", "state", "finalize", "mountSearch", "el", "index$", "keyboard$", "config", "configuration", "worker$", "setupSearchWorker", "query", "getComponentElement", "result", "fromEvent", "filter", "target", "setToggle", "mode", "key", "active", "getActiveElement", "anchors", "anchor", "getElements", "article", "best", "a", "b", "els", "i", "query$", "mountSearchQuery", "merge", "mountSearchResult", "mergeWith", "getComponentElements", "child", "mountSearchShare", "mountSearchSuggest", "err", "NEVER", "mountSearchHiglight", "el", "index$", "location$", "combineLatest", "startWith", "getLocation", "filter", "url", "map", "index", "setupSearchHighlighter", "fn", "_a", "nodes", "it", "node", "original", "replaced", "text", "childNodes", "h", "watchSidebar", "el", "viewport$", "main$", "parent", "adjust", "combineLatest", "map", "offset", "height", "y", "distinctUntilChanged", "a", "b", "mountSidebar", "_a", "_b", "header$", "options", "__objRest", "inner", "getElement", "getElementOffset", "defer", "push$", "Subject", "done$", "ignoreElements", "endWith", "next$", "auditTime", "animationFrameScheduler", "withLatestFrom", "first", "item", "getElements", "container", "getElementSize", "from", "mergeMap", "label", "fromEvent", "observeOn", "asyncScheduler", "takeUntil", "input", "tap", "state", "finalize", "__spreadValues", "fetchSourceFactsFromGitHub", "user", "repo", "url", "zip", "requestJSON", "catchError", "EMPTY", "map", "release", "defaultIfEmpty", "info", "__spreadValues", "fetchSourceFactsFromGitLab", "base", "project", "url", "requestJSON", "catchError", "EMPTY", "map", "star_count", "forks_count", "defaultIfEmpty", "fetchSourceFacts", "url", "match", "user", "repo", "fetchSourceFactsFromGitHub", "base", "slug", "fetchSourceFactsFromGitLab", "EMPTY", "fetch$", "watchSource", "el", "defer", "cached", "of", "getComponentElements", "consent", "EMPTY", "fetchSourceFacts", "tap", "facts", "catchError", "filter", "map", "shareReplay", "mountSource", "inner", "getElement", "push$", "Subject", "renderSourceFacts", "state", "finalize", "__spreadValues", "watchTabs", "el", "viewport$", "header$", "watchElementSize", "switchMap", "watchViewportAt", "map", "y", "distinctUntilKeyChanged", "mountTabs", "options", "defer", "push$", "Subject", "hidden", "feature", "of", "tap", "state", "finalize", "__spreadValues", "watchTableOfContents", "el", "viewport$", "header$", "table", "anchors", "getElements", "anchor", "id", "target", "getOptionalElement", "adjust$", "distinctUntilKeyChanged", "map", "height", "main", "getComponentElement", "grid", "getElement", "share", "watchElementSize", "switchMap", "body", "defer", "path", "of", "index", "offset", "parent", "a", "b", "combineLatestWith", "adjust", "scan", "prev", "next", "y", "size", "last", "distinctUntilChanged", "startWith", "bufferCount", "mountTableOfContents", "main$", "target$", "push$", "Subject", "done$", "ignoreElements", "endWith", "feature", "smooth$", "merge", "debounceTime", "filter", "observeOn", "asyncScheduler", "withLatestFrom", "behavior", "container", "getElementContainer", "getElementSize", "takeUntil", "skip", "repeat", "url", "getLocation", "active", "hash", "tap", "state", "finalize", "__spreadValues", "watchBackToTop", "_el", "viewport$", "main$", "target$", "direction$", "map", "y", "bufferCount", "b", "distinctUntilChanged", "active$", "active", "combineLatest", "direction", "takeUntil", "skip", "endWith", "repeat", "hidden", "mountBackToTop", "el", "header$", "push$", "Subject", "done$", "ignoreElements", "distinctUntilKeyChanged", "height", "fromEvent", "ev", "tap", "state", "finalize", "__spreadValues", "patchEllipsis", "document$", "switchMap", "getElements", "mergeMap", "el", "watchElementVisibility", "takeUntil", "skip", "filter", "visible", "map", "take", "text", "host", "mountTooltip", "finalize", "patchIndeterminate", "document$", "tablet$", "switchMap", "getElements", "tap", "el", "mergeMap", "fromEvent", "takeWhile", "map", "withLatestFrom", "tablet", "isAppleDevice", "patchScrollfix", "document$", "switchMap", "getElements", "tap", "el", "filter", "mergeMap", "fromEvent", "map", "top", "patchScrolllock", "viewport$", "tablet$", "combineLatest", "watchToggle", "map", "active", "tablet", "switchMap", "of", "delay", "withLatestFrom", "y", "value", "obj", "data", "key", "x", "y", "nodes", "parent", "i", "node", "fetchSearchIndex", "watchScript", "config", "map", "shareReplay", "requestJSON", "document$", "watchDocument", "location$", "watchLocation", "target$", "watchLocationTarget", "keyboard$", "watchKeyboard", "viewport$", "watchViewport", "tablet$", "watchMedia", "screen$", "print$", "watchPrint", "configuration", "index$", "NEVER", "alert$", "Subject", "setupClipboardJS", "progress$", "feature", "setupInstantNavigation", "_a", "setupVersionSelector", "merge", "delay", "setToggle", "filter", "mode", "key", "prev", "getOptionalElement", "setLocation", "next", "active", "getActiveElement", "patchEllipsis", "patchIndeterminate", "patchScrollfix", "patchScrolllock", "header$", "watchHeader", "getComponentElement", "main$", "switchMap", "el", "watchMain", "control$", "getComponentElements", "mountConsent", "mountDialog", "mountHeader", "mountPalette", "mountProgress", "mountSearch", "mountSource", "content$", "defer", "mountAnnounce", "mountContent", "mountSearchHiglight", "EMPTY", "mountHeaderTitle", "at", "mountSidebar", "mountTabs", "mountTableOfContents", "mountBackToTop", "component$", "mergeWith"]
+}
diff --git a/assets/javascripts/bundle.6c14ae12.min.js b/assets/javascripts/bundle.6c14ae12.min.js
deleted file mode 100644
index bbff776812..0000000000
--- a/assets/javascripts/bundle.6c14ae12.min.js
+++ /dev/null
@@ -1,29 +0,0 @@
-"use strict";(()=>{var wi=Object.create;var ur=Object.defineProperty;var Si=Object.getOwnPropertyDescriptor;var Ti=Object.getOwnPropertyNames,kt=Object.getOwnPropertySymbols,Oi=Object.getPrototypeOf,dr=Object.prototype.hasOwnProperty,Zr=Object.prototype.propertyIsEnumerable;var Xr=(e,t,r)=>t in e?ur(e,t,{enumerable:!0,configurable:!0,writable:!0,value:r}):e[t]=r,R=(e,t)=>{for(var r in t||(t={}))dr.call(t,r)&&Xr(e,r,t[r]);if(kt)for(var r of kt(t))Zr.call(t,r)&&Xr(e,r,t[r]);return e};var eo=(e,t)=>{var r={};for(var o in e)dr.call(e,o)&&t.indexOf(o)<0&&(r[o]=e[o]);if(e!=null&&kt)for(var o of kt(e))t.indexOf(o)<0&&Zr.call(e,o)&&(r[o]=e[o]);return r};var hr=(e,t)=>()=>(t||e((t={exports:{}}).exports,t),t.exports);var Mi=(e,t,r,o)=>{if(t&&typeof t=="object"||typeof t=="function")for(let n of Ti(t))!dr.call(e,n)&&n!==r&&ur(e,n,{get:()=>t[n],enumerable:!(o=Si(t,n))||o.enumerable});return e};var Ht=(e,t,r)=>(r=e!=null?wi(Oi(e)):{},Mi(t||!e||!e.__esModule?ur(r,"default",{value:e,enumerable:!0}):r,e));var ro=hr((br,to)=>{(function(e,t){typeof br=="object"&&typeof to!="undefined"?t():typeof define=="function"&&define.amd?define(t):t()})(br,function(){"use strict";function e(r){var o=!0,n=!1,i=null,s={text:!0,search:!0,url:!0,tel:!0,email:!0,password:!0,number:!0,date:!0,month:!0,week:!0,time:!0,datetime:!0,"datetime-local":!0};function a(C){return!!(C&&C!==document&&C.nodeName!=="HTML"&&C.nodeName!=="BODY"&&"classList"in C&&"contains"in C.classList)}function c(C){var it=C.type,Ue=C.tagName;return!!(Ue==="INPUT"&&s[it]&&!C.readOnly||Ue==="TEXTAREA"&&!C.readOnly||C.isContentEditable)}function p(C){C.classList.contains("focus-visible")||(C.classList.add("focus-visible"),C.setAttribute("data-focus-visible-added",""))}function l(C){C.hasAttribute("data-focus-visible-added")&&(C.classList.remove("focus-visible"),C.removeAttribute("data-focus-visible-added"))}function f(C){C.metaKey||C.altKey||C.ctrlKey||(a(r.activeElement)&&p(r.activeElement),o=!0)}function u(C){o=!1}function d(C){a(C.target)&&(o||c(C.target))&&p(C.target)}function v(C){a(C.target)&&(C.target.classList.contains("focus-visible")||C.target.hasAttribute("data-focus-visible-added"))&&(n=!0,window.clearTimeout(i),i=window.setTimeout(function(){n=!1},100),l(C.target))}function b(C){document.visibilityState==="hidden"&&(n&&(o=!0),z())}function z(){document.addEventListener("mousemove",G),document.addEventListener("mousedown",G),document.addEventListener("mouseup",G),document.addEventListener("pointermove",G),document.addEventListener("pointerdown",G),document.addEventListener("pointerup",G),document.addEventListener("touchmove",G),document.addEventListener("touchstart",G),document.addEventListener("touchend",G)}function K(){document.removeEventListener("mousemove",G),document.removeEventListener("mousedown",G),document.removeEventListener("mouseup",G),document.removeEventListener("pointermove",G),document.removeEventListener("pointerdown",G),document.removeEventListener("pointerup",G),document.removeEventListener("touchmove",G),document.removeEventListener("touchstart",G),document.removeEventListener("touchend",G)}function G(C){C.target.nodeName&&C.target.nodeName.toLowerCase()==="html"||(o=!1,K())}document.addEventListener("keydown",f,!0),document.addEventListener("mousedown",u,!0),document.addEventListener("pointerdown",u,!0),document.addEventListener("touchstart",u,!0),document.addEventListener("visibilitychange",b,!0),z(),r.addEventListener("focus",d,!0),r.addEventListener("blur",v,!0),r.nodeType===Node.DOCUMENT_FRAGMENT_NODE&&r.host?r.host.setAttribute("data-js-focus-visible",""):r.nodeType===Node.DOCUMENT_NODE&&(document.documentElement.classList.add("js-focus-visible"),document.documentElement.setAttribute("data-js-focus-visible",""))}if(typeof window!="undefined"&&typeof document!="undefined"){window.applyFocusVisiblePolyfill=e;var t;try{t=new CustomEvent("focus-visible-polyfill-ready")}catch(r){t=document.createEvent("CustomEvent"),t.initCustomEvent("focus-visible-polyfill-ready",!1,!1,{})}window.dispatchEvent(t)}typeof document!="undefined"&&e(document)})});var Vr=hr((Ot,Dr)=>{/*!
- * clipboard.js v2.0.11
- * https://clipboardjs.com/
- *
- * Licensed MIT © Zeno Rocha
- */(function(t,r){typeof Ot=="object"&&typeof Dr=="object"?Dr.exports=r():typeof define=="function"&&define.amd?define([],r):typeof Ot=="object"?Ot.ClipboardJS=r():t.ClipboardJS=r()})(Ot,function(){return function(){var e={686:function(o,n,i){"use strict";i.d(n,{default:function(){return Ei}});var s=i(279),a=i.n(s),c=i(370),p=i.n(c),l=i(817),f=i.n(l);function u(W){try{return document.execCommand(W)}catch(O){return!1}}var d=function(O){var S=f()(O);return u("cut"),S},v=d;function b(W){var O=document.documentElement.getAttribute("dir")==="rtl",S=document.createElement("textarea");S.style.fontSize="12pt",S.style.border="0",S.style.padding="0",S.style.margin="0",S.style.position="absolute",S.style[O?"right":"left"]="-9999px";var $=window.pageYOffset||document.documentElement.scrollTop;return S.style.top="".concat($,"px"),S.setAttribute("readonly",""),S.value=W,S}var z=function(O,S){var $=b(O);S.container.appendChild($);var F=f()($);return u("copy"),$.remove(),F},K=function(O){var S=arguments.length>1&&arguments[1]!==void 0?arguments[1]:{container:document.body},$="";return typeof O=="string"?$=z(O,S):O instanceof HTMLInputElement&&!["text","search","url","tel","password"].includes(O==null?void 0:O.type)?$=z(O.value,S):($=f()(O),u("copy")),$},G=K;function C(W){"@babel/helpers - typeof";return typeof Symbol=="function"&&typeof Symbol.iterator=="symbol"?C=function(S){return typeof S}:C=function(S){return S&&typeof Symbol=="function"&&S.constructor===Symbol&&S!==Symbol.prototype?"symbol":typeof S},C(W)}var it=function(){var O=arguments.length>0&&arguments[0]!==void 0?arguments[0]:{},S=O.action,$=S===void 0?"copy":S,F=O.container,Q=O.target,_e=O.text;if($!=="copy"&&$!=="cut")throw new Error('Invalid "action" value, use either "copy" or "cut"');if(Q!==void 0)if(Q&&C(Q)==="object"&&Q.nodeType===1){if($==="copy"&&Q.hasAttribute("disabled"))throw new Error('Invalid "target" attribute. Please use "readonly" instead of "disabled" attribute');if($==="cut"&&(Q.hasAttribute("readonly")||Q.hasAttribute("disabled")))throw new Error(`Invalid "target" attribute. You can't cut text from elements with "readonly" or "disabled" attributes`)}else throw new Error('Invalid "target" value, use a valid Element');if(_e)return G(_e,{container:F});if(Q)return $==="cut"?v(Q):G(Q,{container:F})},Ue=it;function Pe(W){"@babel/helpers - typeof";return typeof Symbol=="function"&&typeof Symbol.iterator=="symbol"?Pe=function(S){return typeof S}:Pe=function(S){return S&&typeof Symbol=="function"&&S.constructor===Symbol&&S!==Symbol.prototype?"symbol":typeof S},Pe(W)}function ui(W,O){if(!(W instanceof O))throw new TypeError("Cannot call a class as a function")}function Jr(W,O){for(var S=0;S<O.length;S++){var $=O[S];$.enumerable=$.enumerable||!1,$.configurable=!0,"value"in $&&($.writable=!0),Object.defineProperty(W,$.key,$)}}function di(W,O,S){return O&&Jr(W.prototype,O),S&&Jr(W,S),W}function hi(W,O){if(typeof O!="function"&&O!==null)throw new TypeError("Super expression must either be null or a function");W.prototype=Object.create(O&&O.prototype,{constructor:{value:W,writable:!0,configurable:!0}}),O&&mr(W,O)}function mr(W,O){return mr=Object.setPrototypeOf||function($,F){return $.__proto__=F,$},mr(W,O)}function bi(W){var O=xi();return function(){var $=At(W),F;if(O){var Q=At(this).constructor;F=Reflect.construct($,arguments,Q)}else F=$.apply(this,arguments);return vi(this,F)}}function vi(W,O){return O&&(Pe(O)==="object"||typeof O=="function")?O:gi(W)}function gi(W){if(W===void 0)throw new ReferenceError("this hasn't been initialised - super() hasn't been called");return W}function xi(){if(typeof Reflect=="undefined"||!Reflect.construct||Reflect.construct.sham)return!1;if(typeof Proxy=="function")return!0;try{return Date.prototype.toString.call(Reflect.construct(Date,[],function(){})),!0}catch(W){return!1}}function At(W){return At=Object.setPrototypeOf?Object.getPrototypeOf:function(S){return S.__proto__||Object.getPrototypeOf(S)},At(W)}function fr(W,O){var S="data-clipboard-".concat(W);if(O.hasAttribute(S))return O.getAttribute(S)}var yi=function(W){hi(S,W);var O=bi(S);function S($,F){var Q;return ui(this,S),Q=O.call(this),Q.resolveOptions(F),Q.listenClick($),Q}return di(S,[{key:"resolveOptions",value:function(){var F=arguments.length>0&&arguments[0]!==void 0?arguments[0]:{};this.action=typeof F.action=="function"?F.action:this.defaultAction,this.target=typeof F.target=="function"?F.target:this.defaultTarget,this.text=typeof F.text=="function"?F.text:this.defaultText,this.container=Pe(F.container)==="object"?F.container:document.body}},{key:"listenClick",value:function(F){var Q=this;this.listener=p()(F,"click",function(_e){return Q.onClick(_e)})}},{key:"onClick",value:function(F){var Q=F.delegateTarget||F.currentTarget,_e=this.action(Q)||"copy",Ct=Ue({action:_e,container:this.container,target:this.target(Q),text:this.text(Q)});this.emit(Ct?"success":"error",{action:_e,text:Ct,trigger:Q,clearSelection:function(){Q&&Q.focus(),window.getSelection().removeAllRanges()}})}},{key:"defaultAction",value:function(F){return fr("action",F)}},{key:"defaultTarget",value:function(F){var Q=fr("target",F);if(Q)return document.querySelector(Q)}},{key:"defaultText",value:function(F){return fr("text",F)}},{key:"destroy",value:function(){this.listener.destroy()}}],[{key:"copy",value:function(F){var Q=arguments.length>1&&arguments[1]!==void 0?arguments[1]:{container:document.body};return G(F,Q)}},{key:"cut",value:function(F){return v(F)}},{key:"isSupported",value:function(){var F=arguments.length>0&&arguments[0]!==void 0?arguments[0]:["copy","cut"],Q=typeof F=="string"?[F]:F,_e=!!document.queryCommandSupported;return Q.forEach(function(Ct){_e=_e&&!!document.queryCommandSupported(Ct)}),_e}}]),S}(a()),Ei=yi},828:function(o){var n=9;if(typeof Element!="undefined"&&!Element.prototype.matches){var i=Element.prototype;i.matches=i.matchesSelector||i.mozMatchesSelector||i.msMatchesSelector||i.oMatchesSelector||i.webkitMatchesSelector}function s(a,c){for(;a&&a.nodeType!==n;){if(typeof a.matches=="function"&&a.matches(c))return a;a=a.parentNode}}o.exports=s},438:function(o,n,i){var s=i(828);function a(l,f,u,d,v){var b=p.apply(this,arguments);return l.addEventListener(u,b,v),{destroy:function(){l.removeEventListener(u,b,v)}}}function c(l,f,u,d,v){return typeof l.addEventListener=="function"?a.apply(null,arguments):typeof u=="function"?a.bind(null,document).apply(null,arguments):(typeof l=="string"&&(l=document.querySelectorAll(l)),Array.prototype.map.call(l,function(b){return a(b,f,u,d,v)}))}function p(l,f,u,d){return function(v){v.delegateTarget=s(v.target,f),v.delegateTarget&&d.call(l,v)}}o.exports=c},879:function(o,n){n.node=function(i){return i!==void 0&&i instanceof HTMLElement&&i.nodeType===1},n.nodeList=function(i){var s=Object.prototype.toString.call(i);return i!==void 0&&(s==="[object NodeList]"||s==="[object HTMLCollection]")&&"length"in i&&(i.length===0||n.node(i[0]))},n.string=function(i){return typeof i=="string"||i instanceof String},n.fn=function(i){var s=Object.prototype.toString.call(i);return s==="[object Function]"}},370:function(o,n,i){var s=i(879),a=i(438);function c(u,d,v){if(!u&&!d&&!v)throw new Error("Missing required arguments");if(!s.string(d))throw new TypeError("Second argument must be a String");if(!s.fn(v))throw new TypeError("Third argument must be a Function");if(s.node(u))return p(u,d,v);if(s.nodeList(u))return l(u,d,v);if(s.string(u))return f(u,d,v);throw new TypeError("First argument must be a String, HTMLElement, HTMLCollection, or NodeList")}function p(u,d,v){return u.addEventListener(d,v),{destroy:function(){u.removeEventListener(d,v)}}}function l(u,d,v){return Array.prototype.forEach.call(u,function(b){b.addEventListener(d,v)}),{destroy:function(){Array.prototype.forEach.call(u,function(b){b.removeEventListener(d,v)})}}}function f(u,d,v){return a(document.body,u,d,v)}o.exports=c},817:function(o){function n(i){var s;if(i.nodeName==="SELECT")i.focus(),s=i.value;else if(i.nodeName==="INPUT"||i.nodeName==="TEXTAREA"){var a=i.hasAttribute("readonly");a||i.setAttribute("readonly",""),i.select(),i.setSelectionRange(0,i.value.length),a||i.removeAttribute("readonly"),s=i.value}else{i.hasAttribute("contenteditable")&&i.focus();var c=window.getSelection(),p=document.createRange();p.selectNodeContents(i),c.removeAllRanges(),c.addRange(p),s=c.toString()}return s}o.exports=n},279:function(o){function n(){}n.prototype={on:function(i,s,a){var c=this.e||(this.e={});return(c[i]||(c[i]=[])).push({fn:s,ctx:a}),this},once:function(i,s,a){var c=this;function p(){c.off(i,p),s.apply(a,arguments)}return p._=s,this.on(i,p,a)},emit:function(i){var s=[].slice.call(arguments,1),a=((this.e||(this.e={}))[i]||[]).slice(),c=0,p=a.length;for(c;c<p;c++)a[c].fn.apply(a[c].ctx,s);return this},off:function(i,s){var a=this.e||(this.e={}),c=a[i],p=[];if(c&&s)for(var l=0,f=c.length;l<f;l++)c[l].fn!==s&&c[l].fn._!==s&&p.push(c[l]);return p.length?a[i]=p:delete a[i],this}},o.exports=n,o.exports.TinyEmitter=n}},t={};function r(o){if(t[o])return t[o].exports;var n=t[o]={exports:{}};return e[o](n,n.exports,r),n.exports}return function(){r.n=function(o){var n=o&&o.__esModule?function(){return o.default}:function(){return o};return r.d(n,{a:n}),n}}(),function(){r.d=function(o,n){for(var i in n)r.o(n,i)&&!r.o(o,i)&&Object.defineProperty(o,i,{enumerable:!0,get:n[i]})}}(),function(){r.o=function(o,n){return Object.prototype.hasOwnProperty.call(o,n)}}(),r(686)}().default})});var Un=hr((GE,Nn)=>{"use strict";/*!
- * escape-html
- * Copyright(c) 2012-2013 TJ Holowaychuk
- * Copyright(c) 2015 Andreas Lubbe
- * Copyright(c) 2015 Tiancheng "Timothy" Gu
- * MIT Licensed
- */var Ha=/["'&<>]/;Nn.exports=$a;function $a(e){var t=""+e,r=Ha.exec(t);if(!r)return t;var o,n="",i=0,s=0;for(i=r.index;i<t.length;i++){switch(t.charCodeAt(i)){case 34:o="&quot;";break;case 38:o="&amp;";break;case 39:o="&#39;";break;case 60:o="&lt;";break;case 62:o="&gt;";break;default:continue}s!==i&&(n+=t.substring(s,i)),s=i+1,n+=o}return s!==i?n+t.substring(s,i):n}});var TO=Ht(ro());/*! *****************************************************************************
-Copyright (c) Microsoft Corporation.
-
-Permission to use, copy, modify, and/or distribute this software for any
-purpose with or without fee is hereby granted.
-
-THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES WITH
-REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY
-AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT,
-INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM
-LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR
-OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
-PERFORMANCE OF THIS SOFTWARE.
-***************************************************************************** */var vr=function(e,t){return vr=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(r,o){r.__proto__=o}||function(r,o){for(var n in o)Object.prototype.hasOwnProperty.call(o,n)&&(r[n]=o[n])},vr(e,t)};function ie(e,t){if(typeof t!="function"&&t!==null)throw new TypeError("Class extends value "+String(t)+" is not a constructor or null");vr(e,t);function r(){this.constructor=e}e.prototype=t===null?Object.create(t):(r.prototype=t.prototype,new r)}function oo(e,t,r,o){function n(i){return i instanceof r?i:new r(function(s){s(i)})}return new(r||(r=Promise))(function(i,s){function a(l){try{p(o.next(l))}catch(f){s(f)}}function c(l){try{p(o.throw(l))}catch(f){s(f)}}function p(l){l.done?i(l.value):n(l.value).then(a,c)}p((o=o.apply(e,t||[])).next())})}function $t(e,t){var r={label:0,sent:function(){if(i[0]&1)throw i[1];return i[1]},trys:[],ops:[]},o,n,i,s;return s={next:a(0),throw:a(1),return:a(2)},typeof Symbol=="function"&&(s[Symbol.iterator]=function(){return this}),s;function a(p){return function(l){return c([p,l])}}function c(p){if(o)throw new TypeError("Generator is already executing.");for(;r;)try{if(o=1,n&&(i=p[0]&2?n.return:p[0]?n.throw||((i=n.return)&&i.call(n),0):n.next)&&!(i=i.call(n,p[1])).done)return i;switch(n=0,i&&(p=[p[0]&2,i.value]),p[0]){case 0:case 1:i=p;break;case 4:return r.label++,{value:p[1],done:!1};case 5:r.label++,n=p[1],p=[0];continue;case 7:p=r.ops.pop(),r.trys.pop();continue;default:if(i=r.trys,!(i=i.length>0&&i[i.length-1])&&(p[0]===6||p[0]===2)){r=0;continue}if(p[0]===3&&(!i||p[1]>i[0]&&p[1]<i[3])){r.label=p[1];break}if(p[0]===6&&r.label<i[1]){r.label=i[1],i=p;break}if(i&&r.label<i[2]){r.label=i[2],r.ops.push(p);break}i[2]&&r.ops.pop(),r.trys.pop();continue}p=t.call(e,r)}catch(l){p=[6,l],n=0}finally{o=i=0}if(p[0]&5)throw p[1];return{value:p[0]?p[1]:void 0,done:!0}}}function we(e){var t=typeof Symbol=="function"&&Symbol.iterator,r=t&&e[t],o=0;if(r)return r.call(e);if(e&&typeof e.length=="number")return{next:function(){return e&&o>=e.length&&(e=void 0),{value:e&&e[o++],done:!e}}};throw new TypeError(t?"Object is not iterable.":"Symbol.iterator is not defined.")}function N(e,t){var r=typeof Symbol=="function"&&e[Symbol.iterator];if(!r)return e;var o=r.call(e),n,i=[],s;try{for(;(t===void 0||t-- >0)&&!(n=o.next()).done;)i.push(n.value)}catch(a){s={error:a}}finally{try{n&&!n.done&&(r=o.return)&&r.call(o)}finally{if(s)throw s.error}}return i}function D(e,t,r){if(r||arguments.length===2)for(var o=0,n=t.length,i;o<n;o++)(i||!(o in t))&&(i||(i=Array.prototype.slice.call(t,0,o)),i[o]=t[o]);return e.concat(i||Array.prototype.slice.call(t))}function Ze(e){return this instanceof Ze?(this.v=e,this):new Ze(e)}function no(e,t,r){if(!Symbol.asyncIterator)throw new TypeError("Symbol.asyncIterator is not defined.");var o=r.apply(e,t||[]),n,i=[];return n={},s("next"),s("throw"),s("return"),n[Symbol.asyncIterator]=function(){return this},n;function s(u){o[u]&&(n[u]=function(d){return new Promise(function(v,b){i.push([u,d,v,b])>1||a(u,d)})})}function a(u,d){try{c(o[u](d))}catch(v){f(i[0][3],v)}}function c(u){u.value instanceof Ze?Promise.resolve(u.value.v).then(p,l):f(i[0][2],u)}function p(u){a("next",u)}function l(u){a("throw",u)}function f(u,d){u(d),i.shift(),i.length&&a(i[0][0],i[0][1])}}function io(e){if(!Symbol.asyncIterator)throw new TypeError("Symbol.asyncIterator is not defined.");var t=e[Symbol.asyncIterator],r;return t?t.call(e):(e=typeof we=="function"?we(e):e[Symbol.iterator](),r={},o("next"),o("throw"),o("return"),r[Symbol.asyncIterator]=function(){return this},r);function o(i){r[i]=e[i]&&function(s){return new Promise(function(a,c){s=e[i](s),n(a,c,s.done,s.value)})}}function n(i,s,a,c){Promise.resolve(c).then(function(p){i({value:p,done:a})},s)}}function k(e){return typeof e=="function"}function at(e){var t=function(o){Error.call(o),o.stack=new Error().stack},r=e(t);return r.prototype=Object.create(Error.prototype),r.prototype.constructor=r,r}var Rt=at(function(e){return function(r){e(this),this.message=r?r.length+` errors occurred during unsubscription:
-`+r.map(function(o,n){return n+1+") "+o.toString()}).join(`
-  `):"",this.name="UnsubscriptionError",this.errors=r}});function De(e,t){if(e){var r=e.indexOf(t);0<=r&&e.splice(r,1)}}var Ie=function(){function e(t){this.initialTeardown=t,this.closed=!1,this._parentage=null,this._finalizers=null}return e.prototype.unsubscribe=function(){var t,r,o,n,i;if(!this.closed){this.closed=!0;var s=this._parentage;if(s)if(this._parentage=null,Array.isArray(s))try{for(var a=we(s),c=a.next();!c.done;c=a.next()){var p=c.value;p.remove(this)}}catch(b){t={error:b}}finally{try{c&&!c.done&&(r=a.return)&&r.call(a)}finally{if(t)throw t.error}}else s.remove(this);var l=this.initialTeardown;if(k(l))try{l()}catch(b){i=b instanceof Rt?b.errors:[b]}var f=this._finalizers;if(f){this._finalizers=null;try{for(var u=we(f),d=u.next();!d.done;d=u.next()){var v=d.value;try{ao(v)}catch(b){i=i!=null?i:[],b instanceof Rt?i=D(D([],N(i)),N(b.errors)):i.push(b)}}}catch(b){o={error:b}}finally{try{d&&!d.done&&(n=u.return)&&n.call(u)}finally{if(o)throw o.error}}}if(i)throw new Rt(i)}},e.prototype.add=function(t){var r;if(t&&t!==this)if(this.closed)ao(t);else{if(t instanceof e){if(t.closed||t._hasParent(this))return;t._addParent(this)}(this._finalizers=(r=this._finalizers)!==null&&r!==void 0?r:[]).push(t)}},e.prototype._hasParent=function(t){var r=this._parentage;return r===t||Array.isArray(r)&&r.includes(t)},e.prototype._addParent=function(t){var r=this._parentage;this._parentage=Array.isArray(r)?(r.push(t),r):r?[r,t]:t},e.prototype._removeParent=function(t){var r=this._parentage;r===t?this._parentage=null:Array.isArray(r)&&De(r,t)},e.prototype.remove=function(t){var r=this._finalizers;r&&De(r,t),t instanceof e&&t._removeParent(this)},e.EMPTY=function(){var t=new e;return t.closed=!0,t}(),e}();var gr=Ie.EMPTY;function Pt(e){return e instanceof Ie||e&&"closed"in e&&k(e.remove)&&k(e.add)&&k(e.unsubscribe)}function ao(e){k(e)?e():e.unsubscribe()}var Ae={onUnhandledError:null,onStoppedNotification:null,Promise:void 0,useDeprecatedSynchronousErrorHandling:!1,useDeprecatedNextContext:!1};var st={setTimeout:function(e,t){for(var r=[],o=2;o<arguments.length;o++)r[o-2]=arguments[o];var n=st.delegate;return n!=null&&n.setTimeout?n.setTimeout.apply(n,D([e,t],N(r))):setTimeout.apply(void 0,D([e,t],N(r)))},clearTimeout:function(e){var t=st.delegate;return((t==null?void 0:t.clearTimeout)||clearTimeout)(e)},delegate:void 0};function It(e){st.setTimeout(function(){var t=Ae.onUnhandledError;if(t)t(e);else throw e})}function fe(){}var so=function(){return xr("C",void 0,void 0)}();function co(e){return xr("E",void 0,e)}function po(e){return xr("N",e,void 0)}function xr(e,t,r){return{kind:e,value:t,error:r}}var et=null;function ct(e){if(Ae.useDeprecatedSynchronousErrorHandling){var t=!et;if(t&&(et={errorThrown:!1,error:null}),e(),t){var r=et,o=r.errorThrown,n=r.error;if(et=null,o)throw n}}else e()}function lo(e){Ae.useDeprecatedSynchronousErrorHandling&&et&&(et.errorThrown=!0,et.error=e)}var xt=function(e){ie(t,e);function t(r){var o=e.call(this)||this;return o.isStopped=!1,r?(o.destination=r,Pt(r)&&r.add(o)):o.destination=Ci,o}return t.create=function(r,o,n){return new tt(r,o,n)},t.prototype.next=function(r){this.isStopped?Er(po(r),this):this._next(r)},t.prototype.error=function(r){this.isStopped?Er(co(r),this):(this.isStopped=!0,this._error(r))},t.prototype.complete=function(){this.isStopped?Er(so,this):(this.isStopped=!0,this._complete())},t.prototype.unsubscribe=function(){this.closed||(this.isStopped=!0,e.prototype.unsubscribe.call(this),this.destination=null)},t.prototype._next=function(r){this.destination.next(r)},t.prototype._error=function(r){try{this.destination.error(r)}finally{this.unsubscribe()}},t.prototype._complete=function(){try{this.destination.complete()}finally{this.unsubscribe()}},t}(Ie);var Li=Function.prototype.bind;function yr(e,t){return Li.call(e,t)}var _i=function(){function e(t){this.partialObserver=t}return e.prototype.next=function(t){var r=this.partialObserver;if(r.next)try{r.next(t)}catch(o){Ft(o)}},e.prototype.error=function(t){var r=this.partialObserver;if(r.error)try{r.error(t)}catch(o){Ft(o)}else Ft(t)},e.prototype.complete=function(){var t=this.partialObserver;if(t.complete)try{t.complete()}catch(r){Ft(r)}},e}(),tt=function(e){ie(t,e);function t(r,o,n){var i=e.call(this)||this,s;if(k(r)||!r)s={next:r!=null?r:void 0,error:o!=null?o:void 0,complete:n!=null?n:void 0};else{var a;i&&Ae.useDeprecatedNextContext?(a=Object.create(r),a.unsubscribe=function(){return i.unsubscribe()},s={next:r.next&&yr(r.next,a),error:r.error&&yr(r.error,a),complete:r.complete&&yr(r.complete,a)}):s=r}return i.destination=new _i(s),i}return t}(xt);function Ft(e){Ae.useDeprecatedSynchronousErrorHandling?lo(e):It(e)}function Ai(e){throw e}function Er(e,t){var r=Ae.onStoppedNotification;r&&st.setTimeout(function(){return r(e,t)})}var Ci={closed:!0,next:fe,error:Ai,complete:fe};var pt=function(){return typeof Symbol=="function"&&Symbol.observable||"@@observable"}();function ue(e){return e}function mo(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];return wr(e)}function wr(e){return e.length===0?ue:e.length===1?e[0]:function(r){return e.reduce(function(o,n){return n(o)},r)}}var P=function(){function e(t){t&&(this._subscribe=t)}return e.prototype.lift=function(t){var r=new e;return r.source=this,r.operator=t,r},e.prototype.subscribe=function(t,r,o){var n=this,i=Hi(t)?t:new tt(t,r,o);return ct(function(){var s=n,a=s.operator,c=s.source;i.add(a?a.call(i,c):c?n._subscribe(i):n._trySubscribe(i))}),i},e.prototype._trySubscribe=function(t){try{return this._subscribe(t)}catch(r){t.error(r)}},e.prototype.forEach=function(t,r){var o=this;return r=fo(r),new r(function(n,i){var s=new tt({next:function(a){try{t(a)}catch(c){i(c),s.unsubscribe()}},error:i,complete:n});o.subscribe(s)})},e.prototype._subscribe=function(t){var r;return(r=this.source)===null||r===void 0?void 0:r.subscribe(t)},e.prototype[pt]=function(){return this},e.prototype.pipe=function(){for(var t=[],r=0;r<arguments.length;r++)t[r]=arguments[r];return wr(t)(this)},e.prototype.toPromise=function(t){var r=this;return t=fo(t),new t(function(o,n){var i;r.subscribe(function(s){return i=s},function(s){return n(s)},function(){return o(i)})})},e.create=function(t){return new e(t)},e}();function fo(e){var t;return(t=e!=null?e:Ae.Promise)!==null&&t!==void 0?t:Promise}function ki(e){return e&&k(e.next)&&k(e.error)&&k(e.complete)}function Hi(e){return e&&e instanceof xt||ki(e)&&Pt(e)}function $i(e){return k(e==null?void 0:e.lift)}function g(e){return function(t){if($i(t))return t.lift(function(r){try{return e(r,this)}catch(o){this.error(o)}});throw new TypeError("Unable to lift unknown Observable type")}}function y(e,t,r,o,n){return new Ri(e,t,r,o,n)}var Ri=function(e){ie(t,e);function t(r,o,n,i,s,a){var c=e.call(this,r)||this;return c.onFinalize=s,c.shouldUnsubscribe=a,c._next=o?function(p){try{o(p)}catch(l){r.error(l)}}:e.prototype._next,c._error=i?function(p){try{i(p)}catch(l){r.error(l)}finally{this.unsubscribe()}}:e.prototype._error,c._complete=n?function(){try{n()}catch(p){r.error(p)}finally{this.unsubscribe()}}:e.prototype._complete,c}return t.prototype.unsubscribe=function(){var r;if(!this.shouldUnsubscribe||this.shouldUnsubscribe()){var o=this.closed;e.prototype.unsubscribe.call(this),!o&&((r=this.onFinalize)===null||r===void 0||r.call(this))}},t}(xt);var lt={schedule:function(e){var t=requestAnimationFrame,r=cancelAnimationFrame,o=lt.delegate;o&&(t=o.requestAnimationFrame,r=o.cancelAnimationFrame);var n=t(function(i){r=void 0,e(i)});return new Ie(function(){return r==null?void 0:r(n)})},requestAnimationFrame:function(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=lt.delegate;return((r==null?void 0:r.requestAnimationFrame)||requestAnimationFrame).apply(void 0,D([],N(e)))},cancelAnimationFrame:function(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=lt.delegate;return((r==null?void 0:r.cancelAnimationFrame)||cancelAnimationFrame).apply(void 0,D([],N(e)))},delegate:void 0};var uo=at(function(e){return function(){e(this),this.name="ObjectUnsubscribedError",this.message="object unsubscribed"}});var x=function(e){ie(t,e);function t(){var r=e.call(this)||this;return r.closed=!1,r.currentObservers=null,r.observers=[],r.isStopped=!1,r.hasError=!1,r.thrownError=null,r}return t.prototype.lift=function(r){var o=new ho(this,this);return o.operator=r,o},t.prototype._throwIfClosed=function(){if(this.closed)throw new uo},t.prototype.next=function(r){var o=this;ct(function(){var n,i;if(o._throwIfClosed(),!o.isStopped){o.currentObservers||(o.currentObservers=Array.from(o.observers));try{for(var s=we(o.currentObservers),a=s.next();!a.done;a=s.next()){var c=a.value;c.next(r)}}catch(p){n={error:p}}finally{try{a&&!a.done&&(i=s.return)&&i.call(s)}finally{if(n)throw n.error}}}})},t.prototype.error=function(r){var o=this;ct(function(){if(o._throwIfClosed(),!o.isStopped){o.hasError=o.isStopped=!0,o.thrownError=r;for(var n=o.observers;n.length;)n.shift().error(r)}})},t.prototype.complete=function(){var r=this;ct(function(){if(r._throwIfClosed(),!r.isStopped){r.isStopped=!0;for(var o=r.observers;o.length;)o.shift().complete()}})},t.prototype.unsubscribe=function(){this.isStopped=this.closed=!0,this.observers=this.currentObservers=null},Object.defineProperty(t.prototype,"observed",{get:function(){var r;return((r=this.observers)===null||r===void 0?void 0:r.length)>0},enumerable:!1,configurable:!0}),t.prototype._trySubscribe=function(r){return this._throwIfClosed(),e.prototype._trySubscribe.call(this,r)},t.prototype._subscribe=function(r){return this._throwIfClosed(),this._checkFinalizedStatuses(r),this._innerSubscribe(r)},t.prototype._innerSubscribe=function(r){var o=this,n=this,i=n.hasError,s=n.isStopped,a=n.observers;return i||s?gr:(this.currentObservers=null,a.push(r),new Ie(function(){o.currentObservers=null,De(a,r)}))},t.prototype._checkFinalizedStatuses=function(r){var o=this,n=o.hasError,i=o.thrownError,s=o.isStopped;n?r.error(i):s&&r.complete()},t.prototype.asObservable=function(){var r=new P;return r.source=this,r},t.create=function(r,o){return new ho(r,o)},t}(P);var ho=function(e){ie(t,e);function t(r,o){var n=e.call(this)||this;return n.destination=r,n.source=o,n}return t.prototype.next=function(r){var o,n;(n=(o=this.destination)===null||o===void 0?void 0:o.next)===null||n===void 0||n.call(o,r)},t.prototype.error=function(r){var o,n;(n=(o=this.destination)===null||o===void 0?void 0:o.error)===null||n===void 0||n.call(o,r)},t.prototype.complete=function(){var r,o;(o=(r=this.destination)===null||r===void 0?void 0:r.complete)===null||o===void 0||o.call(r)},t.prototype._subscribe=function(r){var o,n;return(n=(o=this.source)===null||o===void 0?void 0:o.subscribe(r))!==null&&n!==void 0?n:gr},t}(x);var yt={now:function(){return(yt.delegate||Date).now()},delegate:void 0};var Et=function(e){ie(t,e);function t(r,o,n){r===void 0&&(r=1/0),o===void 0&&(o=1/0),n===void 0&&(n=yt);var i=e.call(this)||this;return i._bufferSize=r,i._windowTime=o,i._timestampProvider=n,i._buffer=[],i._infiniteTimeWindow=!0,i._infiniteTimeWindow=o===1/0,i._bufferSize=Math.max(1,r),i._windowTime=Math.max(1,o),i}return t.prototype.next=function(r){var o=this,n=o.isStopped,i=o._buffer,s=o._infiniteTimeWindow,a=o._timestampProvider,c=o._windowTime;n||(i.push(r),!s&&i.push(a.now()+c)),this._trimBuffer(),e.prototype.next.call(this,r)},t.prototype._subscribe=function(r){this._throwIfClosed(),this._trimBuffer();for(var o=this._innerSubscribe(r),n=this,i=n._infiniteTimeWindow,s=n._buffer,a=s.slice(),c=0;c<a.length&&!r.closed;c+=i?1:2)r.next(a[c]);return this._checkFinalizedStatuses(r),o},t.prototype._trimBuffer=function(){var r=this,o=r._bufferSize,n=r._timestampProvider,i=r._buffer,s=r._infiniteTimeWindow,a=(s?1:2)*o;if(o<1/0&&a<i.length&&i.splice(0,i.length-a),!s){for(var c=n.now(),p=0,l=1;l<i.length&&i[l]<=c;l+=2)p=l;p&&i.splice(0,p+1)}},t}(x);var bo=function(e){ie(t,e);function t(r,o){return e.call(this)||this}return t.prototype.schedule=function(r,o){return o===void 0&&(o=0),this},t}(Ie);var wt={setInterval:function(e,t){for(var r=[],o=2;o<arguments.length;o++)r[o-2]=arguments[o];var n=wt.delegate;return n!=null&&n.setInterval?n.setInterval.apply(n,D([e,t],N(r))):setInterval.apply(void 0,D([e,t],N(r)))},clearInterval:function(e){var t=wt.delegate;return((t==null?void 0:t.clearInterval)||clearInterval)(e)},delegate:void 0};var jt=function(e){ie(t,e);function t(r,o){var n=e.call(this,r,o)||this;return n.scheduler=r,n.work=o,n.pending=!1,n}return t.prototype.schedule=function(r,o){var n;if(o===void 0&&(o=0),this.closed)return this;this.state=r;var i=this.id,s=this.scheduler;return i!=null&&(this.id=this.recycleAsyncId(s,i,o)),this.pending=!0,this.delay=o,this.id=(n=this.id)!==null&&n!==void 0?n:this.requestAsyncId(s,this.id,o),this},t.prototype.requestAsyncId=function(r,o,n){return n===void 0&&(n=0),wt.setInterval(r.flush.bind(r,this),n)},t.prototype.recycleAsyncId=function(r,o,n){if(n===void 0&&(n=0),n!=null&&this.delay===n&&this.pending===!1)return o;o!=null&&wt.clearInterval(o)},t.prototype.execute=function(r,o){if(this.closed)return new Error("executing a cancelled action");this.pending=!1;var n=this._execute(r,o);if(n)return n;this.pending===!1&&this.id!=null&&(this.id=this.recycleAsyncId(this.scheduler,this.id,null))},t.prototype._execute=function(r,o){var n=!1,i;try{this.work(r)}catch(s){n=!0,i=s||new Error("Scheduled action threw falsy error")}if(n)return this.unsubscribe(),i},t.prototype.unsubscribe=function(){if(!this.closed){var r=this,o=r.id,n=r.scheduler,i=n.actions;this.work=this.state=this.scheduler=null,this.pending=!1,De(i,this),o!=null&&(this.id=this.recycleAsyncId(n,o,null)),this.delay=null,e.prototype.unsubscribe.call(this)}},t}(bo);var Sr=function(){function e(t,r){r===void 0&&(r=e.now),this.schedulerActionCtor=t,this.now=r}return e.prototype.schedule=function(t,r,o){return r===void 0&&(r=0),new this.schedulerActionCtor(this,t).schedule(o,r)},e.now=yt.now,e}();var Wt=function(e){ie(t,e);function t(r,o){o===void 0&&(o=Sr.now);var n=e.call(this,r,o)||this;return n.actions=[],n._active=!1,n}return t.prototype.flush=function(r){var o=this.actions;if(this._active){o.push(r);return}var n;this._active=!0;do if(n=r.execute(r.state,r.delay))break;while(r=o.shift());if(this._active=!1,n){for(;r=o.shift();)r.unsubscribe();throw n}},t}(Sr);var ae=new Wt(jt),Tr=ae;var vo=function(e){ie(t,e);function t(r,o){var n=e.call(this,r,o)||this;return n.scheduler=r,n.work=o,n}return t.prototype.requestAsyncId=function(r,o,n){return n===void 0&&(n=0),n!==null&&n>0?e.prototype.requestAsyncId.call(this,r,o,n):(r.actions.push(this),r._scheduled||(r._scheduled=lt.requestAnimationFrame(function(){return r.flush(void 0)})))},t.prototype.recycleAsyncId=function(r,o,n){var i;if(n===void 0&&(n=0),n!=null?n>0:this.delay>0)return e.prototype.recycleAsyncId.call(this,r,o,n);var s=r.actions;o!=null&&((i=s[s.length-1])===null||i===void 0?void 0:i.id)!==o&&(lt.cancelAnimationFrame(o),r._scheduled=void 0)},t}(jt);var go=function(e){ie(t,e);function t(){return e!==null&&e.apply(this,arguments)||this}return t.prototype.flush=function(r){this._active=!0;var o=this._scheduled;this._scheduled=void 0;var n=this.actions,i;r=r||n.shift();do if(i=r.execute(r.state,r.delay))break;while((r=n[0])&&r.id===o&&n.shift());if(this._active=!1,i){for(;(r=n[0])&&r.id===o&&n.shift();)r.unsubscribe();throw i}},t}(Wt);var Oe=new go(vo);var L=new P(function(e){return e.complete()});function Nt(e){return e&&k(e.schedule)}function Or(e){return e[e.length-1]}function Qe(e){return k(Or(e))?e.pop():void 0}function Me(e){return Nt(Or(e))?e.pop():void 0}function Ut(e,t){return typeof Or(e)=="number"?e.pop():t}var mt=function(e){return e&&typeof e.length=="number"&&typeof e!="function"};function Dt(e){return k(e==null?void 0:e.then)}function Vt(e){return k(e[pt])}function zt(e){return Symbol.asyncIterator&&k(e==null?void 0:e[Symbol.asyncIterator])}function qt(e){return new TypeError("You provided "+(e!==null&&typeof e=="object"?"an invalid object":"'"+e+"'")+" where a stream was expected. You can provide an Observable, Promise, ReadableStream, Array, AsyncIterable, or Iterable.")}function Pi(){return typeof Symbol!="function"||!Symbol.iterator?"@@iterator":Symbol.iterator}var Kt=Pi();function Qt(e){return k(e==null?void 0:e[Kt])}function Yt(e){return no(this,arguments,function(){var r,o,n,i;return $t(this,function(s){switch(s.label){case 0:r=e.getReader(),s.label=1;case 1:s.trys.push([1,,9,10]),s.label=2;case 2:return[4,Ze(r.read())];case 3:return o=s.sent(),n=o.value,i=o.done,i?[4,Ze(void 0)]:[3,5];case 4:return[2,s.sent()];case 5:return[4,Ze(n)];case 6:return[4,s.sent()];case 7:return s.sent(),[3,2];case 8:return[3,10];case 9:return r.releaseLock(),[7];case 10:return[2]}})})}function Bt(e){return k(e==null?void 0:e.getReader)}function I(e){if(e instanceof P)return e;if(e!=null){if(Vt(e))return Ii(e);if(mt(e))return Fi(e);if(Dt(e))return ji(e);if(zt(e))return xo(e);if(Qt(e))return Wi(e);if(Bt(e))return Ni(e)}throw qt(e)}function Ii(e){return new P(function(t){var r=e[pt]();if(k(r.subscribe))return r.subscribe(t);throw new TypeError("Provided object does not correctly implement Symbol.observable")})}function Fi(e){return new P(function(t){for(var r=0;r<e.length&&!t.closed;r++)t.next(e[r]);t.complete()})}function ji(e){return new P(function(t){e.then(function(r){t.closed||(t.next(r),t.complete())},function(r){return t.error(r)}).then(null,It)})}function Wi(e){return new P(function(t){var r,o;try{for(var n=we(e),i=n.next();!i.done;i=n.next()){var s=i.value;if(t.next(s),t.closed)return}}catch(a){r={error:a}}finally{try{i&&!i.done&&(o=n.return)&&o.call(n)}finally{if(r)throw r.error}}t.complete()})}function xo(e){return new P(function(t){Ui(e,t).catch(function(r){return t.error(r)})})}function Ni(e){return xo(Yt(e))}function Ui(e,t){var r,o,n,i;return oo(this,void 0,void 0,function(){var s,a;return $t(this,function(c){switch(c.label){case 0:c.trys.push([0,5,6,11]),r=io(e),c.label=1;case 1:return[4,r.next()];case 2:if(o=c.sent(),!!o.done)return[3,4];if(s=o.value,t.next(s),t.closed)return[2];c.label=3;case 3:return[3,1];case 4:return[3,11];case 5:return a=c.sent(),n={error:a},[3,11];case 6:return c.trys.push([6,,9,10]),o&&!o.done&&(i=r.return)?[4,i.call(r)]:[3,8];case 7:c.sent(),c.label=8;case 8:return[3,10];case 9:if(n)throw n.error;return[7];case 10:return[7];case 11:return t.complete(),[2]}})})}function ve(e,t,r,o,n){o===void 0&&(o=0),n===void 0&&(n=!1);var i=t.schedule(function(){r(),n?e.add(this.schedule(null,o)):this.unsubscribe()},o);if(e.add(i),!n)return i}function Se(e,t){return t===void 0&&(t=0),g(function(r,o){r.subscribe(y(o,function(n){return ve(o,e,function(){return o.next(n)},t)},function(){return ve(o,e,function(){return o.complete()},t)},function(n){return ve(o,e,function(){return o.error(n)},t)}))})}function rt(e,t){return t===void 0&&(t=0),g(function(r,o){o.add(e.schedule(function(){return r.subscribe(o)},t))})}function yo(e,t){return I(e).pipe(rt(t),Se(t))}function Eo(e,t){return I(e).pipe(rt(t),Se(t))}function wo(e,t){return new P(function(r){var o=0;return t.schedule(function(){o===e.length?r.complete():(r.next(e[o++]),r.closed||this.schedule())})})}function So(e,t){return new P(function(r){var o;return ve(r,t,function(){o=e[Kt](),ve(r,t,function(){var n,i,s;try{n=o.next(),i=n.value,s=n.done}catch(a){r.error(a);return}s?r.complete():r.next(i)},0,!0)}),function(){return k(o==null?void 0:o.return)&&o.return()}})}function Gt(e,t){if(!e)throw new Error("Iterable cannot be null");return new P(function(r){ve(r,t,function(){var o=e[Symbol.asyncIterator]();ve(r,t,function(){o.next().then(function(n){n.done?r.complete():r.next(n.value)})},0,!0)})})}function To(e,t){return Gt(Yt(e),t)}function Oo(e,t){if(e!=null){if(Vt(e))return yo(e,t);if(mt(e))return wo(e,t);if(Dt(e))return Eo(e,t);if(zt(e))return Gt(e,t);if(Qt(e))return So(e,t);if(Bt(e))return To(e,t)}throw qt(e)}function ge(e,t){return t?Oo(e,t):I(e)}function j(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Me(e);return ge(e,r)}function Mr(e,t){var r=k(e)?e:function(){return e},o=function(n){return n.error(r())};return new P(t?function(n){return t.schedule(o,0,n)}:o)}var Jt=at(function(e){return function(){e(this),this.name="EmptyError",this.message="no elements in sequence"}});function Mo(e){return e instanceof Date&&!isNaN(e)}function m(e,t){return g(function(r,o){var n=0;r.subscribe(y(o,function(i){o.next(e.call(t,i,n++))}))})}var Di=Array.isArray;function Vi(e,t){return Di(t)?e.apply(void 0,D([],N(t))):e(t)}function Ye(e){return m(function(t){return Vi(e,t)})}var zi=Array.isArray,qi=Object.getPrototypeOf,Ki=Object.prototype,Qi=Object.keys;function Lo(e){if(e.length===1){var t=e[0];if(zi(t))return{args:t,keys:null};if(Yi(t)){var r=Qi(t);return{args:r.map(function(o){return t[o]}),keys:r}}}return{args:e,keys:null}}function Yi(e){return e&&typeof e=="object"&&qi(e)===Ki}function _o(e,t){return e.reduce(function(r,o,n){return r[o]=t[n],r},{})}function B(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Me(e),o=Qe(e),n=Lo(e),i=n.args,s=n.keys;if(i.length===0)return ge([],r);var a=new P(Lr(i,r,s?function(c){return _o(s,c)}:ue));return o?a.pipe(Ye(o)):a}function Lr(e,t,r){return r===void 0&&(r=ue),function(o){Ao(t,function(){for(var n=e.length,i=new Array(n),s=n,a=n,c=function(l){Ao(t,function(){var f=ge(e[l],t),u=!1;f.subscribe(y(o,function(d){i[l]=d,u||(u=!0,a--),a||o.next(r(i.slice()))},function(){--s||o.complete()}))},o)},p=0;p<n;p++)c(p)},o)}}function Ao(e,t,r){e?ve(r,e,t):t()}function Co(e,t,r,o,n,i,s,a){var c=[],p=0,l=0,f=!1,u=function(){f&&!c.length&&!p&&t.complete()},d=function(b){return p<o?v(b):c.push(b)},v=function(b){i&&t.next(b),p++;var z=!1;I(r(b,l++)).subscribe(y(t,function(K){n==null||n(K),i?d(K):t.next(K)},function(){z=!0},void 0,function(){if(z)try{p--;for(var K=function(){var G=c.shift();s?ve(t,s,function(){return v(G)}):v(G)};c.length&&p<o;)K();u()}catch(G){t.error(G)}}))};return e.subscribe(y(t,d,function(){f=!0,u()})),function(){a==null||a()}}function ce(e,t,r){return r===void 0&&(r=1/0),k(t)?ce(function(o,n){return m(function(i,s){return t(o,i,n,s)})(I(e(o,n)))},r):(typeof t=="number"&&(r=t),g(function(o,n){return Co(o,n,e,r)}))}function ft(e){return e===void 0&&(e=1/0),ce(ue,e)}function ko(){return ft(1)}function Fe(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];return ko()(ge(e,Me(e)))}function H(e){return new P(function(t){I(e()).subscribe(t)})}var Bi=["addListener","removeListener"],Gi=["addEventListener","removeEventListener"],Ji=["on","off"];function h(e,t,r,o){if(k(r)&&(o=r,r=void 0),o)return h(e,t,r).pipe(Ye(o));var n=N(ea(e)?Gi.map(function(a){return function(c){return e[a](t,c,r)}}):Xi(e)?Bi.map(Ho(e,t)):Zi(e)?Ji.map(Ho(e,t)):[],2),i=n[0],s=n[1];if(!i&&mt(e))return ce(function(a){return h(a,t,r)})(I(e));if(!i)throw new TypeError("Invalid event target");return new P(function(a){var c=function(){for(var p=[],l=0;l<arguments.length;l++)p[l]=arguments[l];return a.next(1<p.length?p:p[0])};return i(c),function(){return s(c)}})}function Ho(e,t){return function(r){return function(o){return e[r](t,o)}}}function Xi(e){return k(e.addListener)&&k(e.removeListener)}function Zi(e){return k(e.on)&&k(e.off)}function ea(e){return k(e.addEventListener)&&k(e.removeEventListener)}function Xt(e,t,r){return r?Xt(e,t).pipe(Ye(r)):new P(function(o){var n=function(){for(var s=[],a=0;a<arguments.length;a++)s[a]=arguments[a];return o.next(s.length===1?s[0]:s)},i=e(n);return k(t)?function(){return t(n,i)}:void 0})}function Be(e,t,r){e===void 0&&(e=0),r===void 0&&(r=Tr);var o=-1;return t!=null&&(Nt(t)?r=t:o=t),new P(function(n){var i=Mo(e)?+e-r.now():e;i<0&&(i=0);var s=0;return r.schedule(function(){n.closed||(n.next(s++),0<=o?this.schedule(void 0,o):n.complete())},i)})}function _(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Me(e),o=Ut(e,1/0),n=e;return n.length?n.length===1?I(n[0]):ft(o)(ge(n,r)):L}var Ve=new P(fe);var ta=Array.isArray;function ut(e){return e.length===1&&ta(e[0])?e[0]:e}function M(e,t){return g(function(r,o){var n=0;r.subscribe(y(o,function(i){return e.call(t,i,n++)&&o.next(i)}))})}function St(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Qe(e),o=ut(e);return o.length?new P(function(n){var i=o.map(function(){return[]}),s=o.map(function(){return!1});n.add(function(){i=s=null});for(var a=function(p){I(o[p]).subscribe(y(n,function(l){if(i[p].push(l),i.every(function(u){return u.length})){var f=i.map(function(u){return u.shift()});n.next(r?r.apply(void 0,D([],N(f))):f),i.some(function(u,d){return!u.length&&s[d]})&&n.complete()}},function(){s[p]=!0,!i[p].length&&n.complete()}))},c=0;!n.closed&&c<o.length;c++)a(c);return function(){i=s=null}}):L}function $o(e){return g(function(t,r){var o=!1,n=null,i=null,s=!1,a=function(){if(i==null||i.unsubscribe(),i=null,o){o=!1;var p=n;n=null,r.next(p)}s&&r.complete()},c=function(){i=null,s&&r.complete()};t.subscribe(y(r,function(p){o=!0,n=p,i||I(e(p)).subscribe(i=y(r,a,c))},function(){s=!0,(!o||!i||i.closed)&&r.complete()}))})}function Ce(e,t){return t===void 0&&(t=ae),$o(function(){return Be(e,t)})}function Le(e,t){return t===void 0&&(t=null),t=t!=null?t:e,g(function(r,o){var n=[],i=0;r.subscribe(y(o,function(s){var a,c,p,l,f=null;i++%t===0&&n.push([]);try{for(var u=we(n),d=u.next();!d.done;d=u.next()){var v=d.value;v.push(s),e<=v.length&&(f=f!=null?f:[],f.push(v))}}catch(K){a={error:K}}finally{try{d&&!d.done&&(c=u.return)&&c.call(u)}finally{if(a)throw a.error}}if(f)try{for(var b=we(f),z=b.next();!z.done;z=b.next()){var v=z.value;De(n,v),o.next(v)}}catch(K){p={error:K}}finally{try{z&&!z.done&&(l=b.return)&&l.call(b)}finally{if(p)throw p.error}}},function(){var s,a;try{for(var c=we(n),p=c.next();!p.done;p=c.next()){var l=p.value;o.next(l)}}catch(f){s={error:f}}finally{try{p&&!p.done&&(a=c.return)&&a.call(c)}finally{if(s)throw s.error}}o.complete()},void 0,function(){n=null}))})}function de(e){return g(function(t,r){var o=null,n=!1,i;o=t.subscribe(y(r,void 0,void 0,function(s){i=I(e(s,de(e)(t))),o?(o.unsubscribe(),o=null,i.subscribe(r)):n=!0})),n&&(o.unsubscribe(),o=null,i.subscribe(r))})}function Ro(e,t,r,o,n){return function(i,s){var a=r,c=t,p=0;i.subscribe(y(s,function(l){var f=p++;c=a?e(c,l,f):(a=!0,l),o&&s.next(c)},n&&function(){a&&s.next(c),s.complete()}))}}function _r(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Qe(e);return r?mo(_r.apply(void 0,D([],N(e))),Ye(r)):g(function(o,n){Lr(D([o],N(ut(e))))(n)})}function Ge(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];return _r.apply(void 0,D([],N(e)))}function ke(e,t){return t===void 0&&(t=ae),g(function(r,o){var n=null,i=null,s=null,a=function(){if(n){n.unsubscribe(),n=null;var p=i;i=null,o.next(p)}};function c(){var p=s+e,l=t.now();if(l<p){n=this.schedule(void 0,p-l),o.add(n);return}a()}r.subscribe(y(o,function(p){i=p,s=t.now(),n||(n=t.schedule(c,e),o.add(n))},function(){a(),o.complete()},void 0,function(){i=n=null}))})}function He(e){return g(function(t,r){var o=!1;t.subscribe(y(r,function(n){o=!0,r.next(n)},function(){o||r.next(e),r.complete()}))})}function xe(e){return e<=0?function(){return L}:g(function(t,r){var o=0;t.subscribe(y(r,function(n){++o<=e&&(r.next(n),e<=o&&r.complete())}))})}function Z(){return g(function(e,t){e.subscribe(y(t,fe))})}function Po(e){return m(function(){return e})}function Ar(e,t){return t?function(r){return Fe(t.pipe(xe(1),Z()),r.pipe(Ar(e)))}:ce(function(r,o){return I(e(r,o)).pipe(xe(1),Po(r))})}function ze(e,t){t===void 0&&(t=ae);var r=Be(e,t);return Ar(function(){return r})}function X(e,t){return t===void 0&&(t=ue),e=e!=null?e:ra,g(function(r,o){var n,i=!0;r.subscribe(y(o,function(s){var a=t(s);(i||!e(n,a))&&(i=!1,n=a,o.next(s))}))})}function ra(e,t){return e===t}function ee(e,t){return X(function(r,o){return t?t(r[e],o[e]):r[e]===o[e]})}function Io(e){return e===void 0&&(e=oa),g(function(t,r){var o=!1;t.subscribe(y(r,function(n){o=!0,r.next(n)},function(){return o?r.complete():r.error(e())}))})}function oa(){return new Jt}function re(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];return function(r){return Fe(r,j.apply(void 0,D([],N(e))))}}function A(e){return g(function(t,r){try{t.subscribe(r)}finally{r.add(e)}})}function $e(e,t){var r=arguments.length>=2;return function(o){return o.pipe(e?M(function(n,i){return e(n,i,o)}):ue,xe(1),r?He(t):Io(function(){return new Jt}))}}function Fo(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Me(e),o=Ut(e,1/0);return e=ut(e),g(function(n,i){ft(o)(ge(D([n],N(e)),r)).subscribe(i)})}function qe(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];return Fo.apply(void 0,D([],N(e)))}function Tt(e){var t,r=1/0,o;return e!=null&&(typeof e=="object"?(t=e.count,r=t===void 0?1/0:t,o=e.delay):r=e),r<=0?function(){return L}:g(function(n,i){var s=0,a,c=function(){if(a==null||a.unsubscribe(),a=null,o!=null){var l=typeof o=="number"?Be(o):I(o(s)),f=y(i,function(){f.unsubscribe(),p()});l.subscribe(f)}else p()},p=function(){var l=!1;a=n.subscribe(y(i,void 0,function(){++s<r?a?c():l=!0:i.complete()})),l&&c()};p()})}function Cr(e){return g(function(t,r){var o=!1,n=null;t.subscribe(y(r,function(i){o=!0,n=i})),I(e).subscribe(y(r,function(){if(o){o=!1;var i=n;n=null,r.next(i)}},fe))})}function kr(e,t){return g(Ro(e,t,arguments.length>=2,!0))}function le(e){e===void 0&&(e={});var t=e.connector,r=t===void 0?function(){return new x}:t,o=e.resetOnError,n=o===void 0?!0:o,i=e.resetOnComplete,s=i===void 0?!0:i,a=e.resetOnRefCountZero,c=a===void 0?!0:a;return function(p){var l,f,u,d=0,v=!1,b=!1,z=function(){f==null||f.unsubscribe(),f=void 0},K=function(){z(),l=u=void 0,v=b=!1},G=function(){var C=l;K(),C==null||C.unsubscribe()};return g(function(C,it){d++,!b&&!v&&z();var Ue=u=u!=null?u:r();it.add(function(){d--,d===0&&!b&&!v&&(f=Hr(G,c))}),Ue.subscribe(it),!l&&d>0&&(l=new tt({next:function(Pe){return Ue.next(Pe)},error:function(Pe){b=!0,z(),f=Hr(K,n,Pe),Ue.error(Pe)},complete:function(){v=!0,z(),f=Hr(K,s),Ue.complete()}}),I(C).subscribe(l))})(p)}}function Hr(e,t){for(var r=[],o=2;o<arguments.length;o++)r[o-2]=arguments[o];if(t===!0){e();return}if(t!==!1){var n=new tt({next:function(){n.unsubscribe(),e()}});return I(t.apply(void 0,D([],N(r)))).subscribe(n)}}function J(e,t,r){var o,n,i,s,a=!1;return e&&typeof e=="object"?(o=e.bufferSize,s=o===void 0?1/0:o,n=e.windowTime,t=n===void 0?1/0:n,i=e.refCount,a=i===void 0?!1:i,r=e.scheduler):s=e!=null?e:1/0,le({connector:function(){return new Et(s,t,r)},resetOnError:!0,resetOnComplete:!1,resetOnRefCountZero:a})}function je(e){return M(function(t,r){return e<=r})}function $r(e){return g(function(t,r){var o=!1,n=y(r,function(){n==null||n.unsubscribe(),o=!0},fe);I(e).subscribe(n),t.subscribe(y(r,function(i){return o&&r.next(i)}))})}function V(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Me(e);return g(function(o,n){(r?Fe(e,o,r):Fe(e,o)).subscribe(n)})}function E(e,t){return g(function(r,o){var n=null,i=0,s=!1,a=function(){return s&&!n&&o.complete()};r.subscribe(y(o,function(c){n==null||n.unsubscribe();var p=0,l=i++;I(e(c,l)).subscribe(n=y(o,function(f){return o.next(t?t(c,f,l,p++):f)},function(){n=null,a()}))},function(){s=!0,a()}))})}function Y(e){return g(function(t,r){I(e).subscribe(y(r,function(){return r.complete()},fe)),!r.closed&&t.subscribe(r)})}function Rr(e,t){return t===void 0&&(t=!1),g(function(r,o){var n=0;r.subscribe(y(o,function(i){var s=e(i,n++);(s||t)&&o.next(i),!s&&o.complete()}))})}function w(e,t,r){var o=k(e)||t||r?{next:e,error:t,complete:r}:e;return o?g(function(n,i){var s;(s=o.subscribe)===null||s===void 0||s.call(o);var a=!0;n.subscribe(y(i,function(c){var p;(p=o.next)===null||p===void 0||p.call(o,c),i.next(c)},function(){var c;a=!1,(c=o.complete)===null||c===void 0||c.call(o),i.complete()},function(c){var p;a=!1,(p=o.error)===null||p===void 0||p.call(o,c),i.error(c)},function(){var c,p;a&&((c=o.unsubscribe)===null||c===void 0||c.call(o)),(p=o.finalize)===null||p===void 0||p.call(o)}))}):ue}function jo(e,t){return g(function(r,o){var n=t!=null?t:{},i=n.leading,s=i===void 0?!0:i,a=n.trailing,c=a===void 0?!1:a,p=!1,l=null,f=null,u=!1,d=function(){f==null||f.unsubscribe(),f=null,c&&(z(),u&&o.complete())},v=function(){f=null,u&&o.complete()},b=function(K){return f=I(e(K)).subscribe(y(o,d,v))},z=function(){if(p){p=!1;var K=l;l=null,o.next(K),!u&&b(K)}};r.subscribe(y(o,function(K){p=!0,l=K,!(f&&!f.closed)&&(s?z():b(K))},function(){u=!0,!(c&&p&&f&&!f.closed)&&o.complete()}))})}function Pr(e,t,r){t===void 0&&(t=ae);var o=Be(e,t);return jo(function(){return o},r)}function ne(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];var r=Qe(e);return g(function(o,n){for(var i=e.length,s=new Array(i),a=e.map(function(){return!1}),c=!1,p=function(f){I(e[f]).subscribe(y(n,function(u){s[f]=u,!c&&!a[f]&&(a[f]=!0,(c=a.every(ue))&&(a=null))},fe))},l=0;l<i;l++)p(l);o.subscribe(y(n,function(f){if(c){var u=D([f],N(s));n.next(r?r.apply(void 0,D([],N(u))):u)}}))})}function Wo(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];return g(function(r,o){St.apply(void 0,D([r],N(e))).subscribe(o)})}function Ir(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];return Wo.apply(void 0,D([],N(e)))}function No(){let e=new Et(1);return h(document,"DOMContentLoaded",{once:!0}).subscribe(()=>e.next(document)),e}function q(e,t=document){return Array.from(t.querySelectorAll(e))}function U(e,t=document){let r=se(e,t);if(typeof r=="undefined")throw new ReferenceError(`Missing element: expected "${e}" to be present`);return r}function se(e,t=document){return t.querySelector(e)||void 0}function Re(){return document.activeElement instanceof HTMLElement&&document.activeElement||void 0}var na=_(h(document.body,"focusin"),h(document.body,"focusout")).pipe(ke(1),V(void 0),m(()=>Re()||document.body),J(1));function Zt(e){return na.pipe(m(t=>e.contains(t)),X())}function Je(e){return{x:e.offsetLeft,y:e.offsetTop}}function Uo(e){return _(h(window,"load"),h(window,"resize")).pipe(Ce(0,Oe),m(()=>Je(e)),V(Je(e)))}function er(e){return{x:e.scrollLeft,y:e.scrollTop}}function dt(e){return _(h(e,"scroll"),h(window,"resize")).pipe(Ce(0,Oe),m(()=>er(e)),V(er(e)))}function Do(e,t){if(typeof t=="string"||typeof t=="number")e.innerHTML+=t.toString();else if(t instanceof Node)e.appendChild(t);else if(Array.isArray(t))for(let r of t)Do(e,r)}function T(e,t,...r){let o=document.createElement(e);if(t)for(let n of Object.keys(t))typeof t[n]!="undefined"&&(typeof t[n]!="boolean"?o.setAttribute(n,t[n]):o.setAttribute(n,""));for(let n of r)Do(o,n);return o}function tr(e){if(e>999){let t=+((e-950)%1e3>99);return`${((e+1e-6)/1e3).toFixed(t)}k`}else return e.toString()}function ht(e){let t=T("script",{src:e});return H(()=>(document.head.appendChild(t),_(h(t,"load"),h(t,"error").pipe(E(()=>Mr(()=>new ReferenceError(`Invalid script: ${e}`))))).pipe(m(()=>{}),A(()=>document.head.removeChild(t)),xe(1))))}var Vo=new x,ia=H(()=>typeof ResizeObserver=="undefined"?ht("https://unpkg.com/resize-observer-polyfill"):j(void 0)).pipe(m(()=>new ResizeObserver(e=>{for(let t of e)Vo.next(t)})),E(e=>_(Ve,j(e)).pipe(A(()=>e.disconnect()))),J(1));function he(e){return{width:e.offsetWidth,height:e.offsetHeight}}function ye(e){return ia.pipe(w(t=>t.observe(e)),E(t=>Vo.pipe(M(({target:r})=>r===e),A(()=>t.unobserve(e)),m(()=>he(e)))),V(he(e)))}function bt(e){return{width:e.scrollWidth,height:e.scrollHeight}}function zo(e){let t=e.parentElement;for(;t&&(e.scrollWidth<=t.scrollWidth&&e.scrollHeight<=t.scrollHeight);)t=(e=t).parentElement;return t?e:void 0}var qo=new x,aa=H(()=>j(new IntersectionObserver(e=>{for(let t of e)qo.next(t)},{threshold:0}))).pipe(E(e=>_(Ve,j(e)).pipe(A(()=>e.disconnect()))),J(1));function rr(e){return aa.pipe(w(t=>t.observe(e)),E(t=>qo.pipe(M(({target:r})=>r===e),A(()=>t.unobserve(e)),m(({isIntersecting:r})=>r))))}function Ko(e,t=16){return dt(e).pipe(m(({y:r})=>{let o=he(e),n=bt(e);return r>=n.height-o.height-t}),X())}var or={drawer:U("[data-md-toggle=drawer]"),search:U("[data-md-toggle=search]")};function Qo(e){return or[e].checked}function Ke(e,t){or[e].checked!==t&&or[e].click()}function We(e){let t=or[e];return h(t,"change").pipe(m(()=>t.checked),V(t.checked))}function sa(e,t){switch(e.constructor){case HTMLInputElement:return e.type==="radio"?/^Arrow/.test(t):!0;case HTMLSelectElement:case HTMLTextAreaElement:return!0;default:return e.isContentEditable}}function ca(){return _(h(window,"compositionstart").pipe(m(()=>!0)),h(window,"compositionend").pipe(m(()=>!1))).pipe(V(!1))}function Yo(){let e=h(window,"keydown").pipe(M(t=>!(t.metaKey||t.ctrlKey)),m(t=>({mode:Qo("search")?"search":"global",type:t.key,claim(){t.preventDefault(),t.stopPropagation()}})),M(({mode:t,type:r})=>{if(t==="global"){let o=Re();if(typeof o!="undefined")return!sa(o,r)}return!0}),le());return ca().pipe(E(t=>t?L:e))}function pe(){return new URL(location.href)}function ot(e,t=!1){if(te("navigation.instant")&&!t){let r=T("a",{href:e.href});document.body.appendChild(r),r.click(),r.remove()}else location.href=e.href}function Bo(){return new x}function Go(){return location.hash.slice(1)}function nr(e){let t=T("a",{href:e});t.addEventListener("click",r=>r.stopPropagation()),t.click()}function pa(e){return _(h(window,"hashchange"),e).pipe(m(Go),V(Go()),M(t=>t.length>0),J(1))}function Jo(e){return pa(e).pipe(m(t=>se(`[id="${t}"]`)),M(t=>typeof t!="undefined"))}function Fr(e){let t=matchMedia(e);return Xt(r=>t.addListener(()=>r(t.matches))).pipe(V(t.matches))}function Xo(){let e=matchMedia("print");return _(h(window,"beforeprint").pipe(m(()=>!0)),h(window,"afterprint").pipe(m(()=>!1))).pipe(V(e.matches))}function jr(e,t){return e.pipe(E(r=>r?t():L))}function ir(e,t){return new P(r=>{let o=new XMLHttpRequest;o.open("GET",`${e}`),o.responseType="blob",o.addEventListener("load",()=>{o.status>=200&&o.status<300?(r.next(o.response),r.complete()):r.error(new Error(o.statusText))}),o.addEventListener("error",()=>{r.error(new Error("Network Error"))}),o.addEventListener("abort",()=>{r.error(new Error("Request aborted"))}),typeof(t==null?void 0:t.progress$)!="undefined"&&(o.addEventListener("progress",n=>{if(n.lengthComputable)t.progress$.next(n.loaded/n.total*100);else{let i=Number(o.getResponseHeader("Content-Length"))||0;t.progress$.next(n.loaded/i*100)}}),t.progress$.next(5)),o.send()})}function Ne(e,t){return ir(e,t).pipe(E(r=>r.text()),m(r=>JSON.parse(r)),J(1))}function Zo(e,t){let r=new DOMParser;return ir(e,t).pipe(E(o=>o.text()),m(o=>r.parseFromString(o,"text/xml")),J(1))}function en(){return{x:Math.max(0,scrollX),y:Math.max(0,scrollY)}}function tn(){return _(h(window,"scroll",{passive:!0}),h(window,"resize",{passive:!0})).pipe(m(en),V(en()))}function rn(){return{width:innerWidth,height:innerHeight}}function on(){return h(window,"resize",{passive:!0}).pipe(m(rn),V(rn()))}function nn(){return B([tn(),on()]).pipe(m(([e,t])=>({offset:e,size:t})),J(1))}function ar(e,{viewport$:t,header$:r}){let o=t.pipe(ee("size")),n=B([o,r]).pipe(m(()=>Je(e)));return B([r,t,n]).pipe(m(([{height:i},{offset:s,size:a},{x:c,y:p}])=>({offset:{x:s.x-c,y:s.y-p+i},size:a})))}function la(e){return h(e,"message",t=>t.data)}function ma(e){let t=new x;return t.subscribe(r=>e.postMessage(r)),t}function an(e,t=new Worker(e)){let r=la(t),o=ma(t),n=new x;n.subscribe(o);let i=o.pipe(Z(),re(!0));return n.pipe(Z(),qe(r.pipe(Y(i))),le())}var fa=U("#__config"),vt=JSON.parse(fa.textContent);vt.base=`${new URL(vt.base,pe())}`;function me(){return vt}function te(e){return vt.features.includes(e)}function be(e,t){return typeof t!="undefined"?vt.translations[e].replace("#",t.toString()):vt.translations[e]}function Ee(e,t=document){return U(`[data-md-component=${e}]`,t)}function oe(e,t=document){return q(`[data-md-component=${e}]`,t)}function ua(e){let t=U(".md-typeset > :first-child",e);return h(t,"click",{once:!0}).pipe(m(()=>U(".md-typeset",e)),m(r=>({hash:__md_hash(r.innerHTML)})))}function sn(e){if(!te("announce.dismiss")||!e.childElementCount)return L;if(!e.hidden){let t=U(".md-typeset",e);__md_hash(t.innerHTML)===__md_get("__announce")&&(e.hidden=!0)}return H(()=>{let t=new x;return t.subscribe(({hash:r})=>{e.hidden=!0,__md_set("__announce",r)}),ua(e).pipe(w(r=>t.next(r)),A(()=>t.complete()),m(r=>R({ref:e},r)))})}function da(e,{target$:t}){return t.pipe(m(r=>({hidden:r!==e})))}function cn(e,t){let r=new x;return r.subscribe(({hidden:o})=>{e.hidden=o}),da(e,t).pipe(w(o=>r.next(o)),A(()=>r.complete()),m(o=>R({ref:e},o)))}function ha(e,t){let r=H(()=>B([Uo(e),dt(t)])).pipe(m(([{x:o,y:n},i])=>{let{width:s,height:a}=he(e);return{x:o-i.x+s/2,y:n-i.y+a/2}}));return Zt(e).pipe(E(o=>r.pipe(m(n=>({active:o,offset:n})),xe(+!o||1/0))))}function pn(e,t,{target$:r}){let[o,n]=Array.from(e.children);return H(()=>{let i=new x,s=i.pipe(Z(),re(!0));return i.subscribe({next({offset:a}){e.style.setProperty("--md-tooltip-x",`${a.x}px`),e.style.setProperty("--md-tooltip-y",`${a.y}px`)},complete(){e.style.removeProperty("--md-tooltip-x"),e.style.removeProperty("--md-tooltip-y")}}),rr(e).pipe(Y(s)).subscribe(a=>{e.toggleAttribute("data-md-visible",a)}),_(i.pipe(M(({active:a})=>a)),i.pipe(ke(250),M(({active:a})=>!a))).subscribe({next({active:a}){a?e.prepend(o):o.remove()},complete(){e.prepend(o)}}),i.pipe(Ce(16,Oe)).subscribe(({active:a})=>{o.classList.toggle("md-tooltip--active",a)}),i.pipe(Pr(125,Oe),M(()=>!!e.offsetParent),m(()=>e.offsetParent.getBoundingClientRect()),m(({x:a})=>a)).subscribe({next(a){a?e.style.setProperty("--md-tooltip-0",`${-a}px`):e.style.removeProperty("--md-tooltip-0")},complete(){e.style.removeProperty("--md-tooltip-0")}}),h(n,"click").pipe(Y(s),M(a=>!(a.metaKey||a.ctrlKey))).subscribe(a=>{a.stopPropagation(),a.preventDefault()}),h(n,"mousedown").pipe(Y(s),ne(i)).subscribe(([a,{active:c}])=>{var p;if(a.button!==0||a.metaKey||a.ctrlKey)a.preventDefault();else if(c){a.preventDefault();let l=e.parentElement.closest(".md-annotation");l instanceof HTMLElement?l.focus():(p=Re())==null||p.blur()}}),r.pipe(Y(s),M(a=>a===o),ze(125)).subscribe(()=>e.focus()),ha(e,t).pipe(w(a=>i.next(a)),A(()=>i.complete()),m(a=>R({ref:e},a)))})}function Wr(e){return T("div",{class:"md-tooltip",id:e},T("div",{class:"md-tooltip__inner md-typeset"}))}function ln(e,t){if(t=t?`${t}_annotation_${e}`:void 0,t){let r=t?`#${t}`:void 0;return T("aside",{class:"md-annotation",tabIndex:0},Wr(t),T("a",{href:r,class:"md-annotation__index",tabIndex:-1},T("span",{"data-md-annotation-id":e})))}else return T("aside",{class:"md-annotation",tabIndex:0},Wr(t),T("span",{class:"md-annotation__index",tabIndex:-1},T("span",{"data-md-annotation-id":e})))}function mn(e){return T("button",{class:"md-clipboard md-icon",title:be("clipboard.copy"),"data-clipboard-target":`#${e} > code`})}function Nr(e,t){let r=t&2,o=t&1,n=Object.keys(e.terms).filter(c=>!e.terms[c]).reduce((c,p)=>[...c,T("del",null,p)," "],[]).slice(0,-1),i=me(),s=new URL(e.location,i.base);te("search.highlight")&&s.searchParams.set("h",Object.entries(e.terms).filter(([,c])=>c).reduce((c,[p])=>`${c} ${p}`.trim(),""));let{tags:a}=me();return T("a",{href:`${s}`,class:"md-search-result__link",tabIndex:-1},T("article",{class:"md-search-result__article md-typeset","data-md-score":e.score.toFixed(2)},r>0&&T("div",{class:"md-search-result__icon md-icon"}),r>0&&T("h1",null,e.title),r<=0&&T("h2",null,e.title),o>0&&e.text.length>0&&e.text,e.tags&&e.tags.map(c=>{let p=a?c in a?`md-tag-icon md-tag--${a[c]}`:"md-tag-icon":"";return T("span",{class:`md-tag ${p}`},c)}),o>0&&n.length>0&&T("p",{class:"md-search-result__terms"},be("search.result.term.missing"),": ",...n)))}function fn(e){let t=e[0].score,r=[...e],o=me(),n=r.findIndex(l=>!`${new URL(l.location,o.base)}`.includes("#")),[i]=r.splice(n,1),s=r.findIndex(l=>l.score<t);s===-1&&(s=r.length);let a=r.slice(0,s),c=r.slice(s),p=[Nr(i,2|+(!n&&s===0)),...a.map(l=>Nr(l,1)),...c.length?[T("details",{class:"md-search-result__more"},T("summary",{tabIndex:-1},T("div",null,c.length>0&&c.length===1?be("search.result.more.one"):be("search.result.more.other",c.length))),...c.map(l=>Nr(l,1)))]:[]];return T("li",{class:"md-search-result__item"},p)}function un(e){return T("ul",{class:"md-source__facts"},Object.entries(e).map(([t,r])=>T("li",{class:`md-source__fact md-source__fact--${t}`},typeof r=="number"?tr(r):r)))}function Ur(e){let t=`tabbed-control tabbed-control--${e}`;return T("div",{class:t,hidden:!0},T("button",{class:"tabbed-button",tabIndex:-1,"aria-hidden":"true"}))}function dn(e){return T("div",{class:"md-typeset__scrollwrap"},T("div",{class:"md-typeset__table"},e))}function ba(e){let t=me(),r=new URL(`../${e.version}/`,t.base);return T("li",{class:"md-version__item"},T("a",{href:`${r}`,class:"md-version__link"},e.title))}function hn(e,t){return T("div",{class:"md-version"},T("button",{class:"md-version__current","aria-label":be("select.version")},t.title),T("ul",{class:"md-version__list"},e.map(ba)))}function va(e){return e.tagName==="CODE"?q(".c, .c1, .cm",e):[e]}function ga(e){let t=[];for(let r of va(e)){let o=[],n=document.createNodeIterator(r,NodeFilter.SHOW_TEXT);for(let i=n.nextNode();i;i=n.nextNode())o.push(i);for(let i of o){let s;for(;s=/(\(\d+\))(!)?/.exec(i.textContent);){let[,a,c]=s;if(typeof c=="undefined"){let p=i.splitText(s.index);i=p.splitText(a.length),t.push(p)}else{i.textContent=a,t.push(i);break}}}}return t}function bn(e,t){t.append(...Array.from(e.childNodes))}function sr(e,t,{target$:r,print$:o}){let n=t.closest("[id]"),i=n==null?void 0:n.id,s=new Map;for(let a of ga(t)){let[,c]=a.textContent.match(/\((\d+)\)/);se(`:scope > li:nth-child(${c})`,e)&&(s.set(c,ln(c,i)),a.replaceWith(s.get(c)))}return s.size===0?L:H(()=>{let a=new x,c=a.pipe(Z(),re(!0)),p=[];for(let[l,f]of s)p.push([U(".md-typeset",f),U(`:scope > li:nth-child(${l})`,e)]);return o.pipe(Y(c)).subscribe(l=>{e.hidden=!l,e.classList.toggle("md-annotation-list",l);for(let[f,u]of p)l?bn(f,u):bn(u,f)}),_(...[...s].map(([,l])=>pn(l,t,{target$:r}))).pipe(A(()=>a.complete()),le())})}function vn(e){if(e.nextElementSibling){let t=e.nextElementSibling;if(t.tagName==="OL")return t;if(t.tagName==="P"&&!t.children.length)return vn(t)}}function gn(e,t){return H(()=>{let r=vn(e);return typeof r!="undefined"?sr(r,e,t):L})}var yn=Ht(Vr());var xa=0;function En(e){if(e.nextElementSibling){let t=e.nextElementSibling;if(t.tagName==="OL")return t;if(t.tagName==="P"&&!t.children.length)return En(t)}}function xn(e){return ye(e).pipe(m(({width:t})=>({scrollable:bt(e).width>t})),ee("scrollable"))}function wn(e,t){let{matches:r}=matchMedia("(hover)"),o=H(()=>{let n=new x;if(n.subscribe(({scrollable:s})=>{s&&r?e.setAttribute("tabindex","0"):e.removeAttribute("tabindex")}),yn.default.isSupported()&&(e.closest(".copy")||te("content.code.copy")&&!e.closest(".no-copy"))){let s=e.closest("pre");s.id=`__code_${xa++}`,s.insertBefore(mn(s.id),e)}let i=e.closest(".highlight");if(i instanceof HTMLElement){let s=En(i);if(typeof s!="undefined"&&(i.classList.contains("annotate")||te("content.code.annotate"))){let a=sr(s,e,t);return xn(e).pipe(w(c=>n.next(c)),A(()=>n.complete()),m(c=>R({ref:e},c)),qe(ye(i).pipe(m(({width:c,height:p})=>c&&p),X(),E(c=>c?a:L))))}}return xn(e).pipe(w(s=>n.next(s)),A(()=>n.complete()),m(s=>R({ref:e},s)))});return te("content.lazy")?rr(e).pipe(M(n=>n),xe(1),E(()=>o)):o}function ya(e,{target$:t,print$:r}){let o=!0;return _(t.pipe(m(n=>n.closest("details:not([open])")),M(n=>e===n),m(()=>({action:"open",reveal:!0}))),r.pipe(M(n=>n||!o),w(()=>o=e.open),m(n=>({action:n?"open":"close"}))))}function Sn(e,t){return H(()=>{let r=new x;return r.subscribe(({action:o,reveal:n})=>{e.toggleAttribute("open",o==="open"),n&&e.scrollIntoView()}),ya(e,t).pipe(w(o=>r.next(o)),A(()=>r.complete()),m(o=>R({ref:e},o)))})}var Tn=".node circle,.node ellipse,.node path,.node polygon,.node rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}marker{fill:var(--md-mermaid-edge-color)!important}.edgeLabel .label rect{fill:#0000}.label{color:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.label foreignObject{line-height:normal;overflow:visible}.label div .edgeLabel{color:var(--md-mermaid-label-fg-color)}.edgeLabel,.edgeLabel rect,.label div .edgeLabel{background-color:var(--md-mermaid-label-bg-color)}.edgeLabel,.edgeLabel rect{fill:var(--md-mermaid-label-bg-color);color:var(--md-mermaid-edge-color)}.edgePath .path,.flowchart-link{stroke:var(--md-mermaid-edge-color);stroke-width:.05rem}.edgePath .arrowheadPath{fill:var(--md-mermaid-edge-color);stroke:none}.cluster rect{fill:var(--md-default-fg-color--lightest);stroke:var(--md-default-fg-color--lighter)}.cluster span{color:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}g #flowchart-circleEnd,g #flowchart-circleStart,g #flowchart-crossEnd,g #flowchart-crossStart,g #flowchart-pointEnd,g #flowchart-pointStart{stroke:none}g.classGroup line,g.classGroup rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}g.classGroup text{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.classLabel .box{fill:var(--md-mermaid-label-bg-color);background-color:var(--md-mermaid-label-bg-color);opacity:1}.classLabel .label{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.node .divider{stroke:var(--md-mermaid-node-fg-color)}.relation{stroke:var(--md-mermaid-edge-color)}.cardinality{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.cardinality text{fill:inherit!important}defs #classDiagram-compositionEnd,defs #classDiagram-compositionStart,defs #classDiagram-dependencyEnd,defs #classDiagram-dependencyStart,defs #classDiagram-extensionEnd,defs #classDiagram-extensionStart{fill:var(--md-mermaid-edge-color)!important;stroke:var(--md-mermaid-edge-color)!important}defs #classDiagram-aggregationEnd,defs #classDiagram-aggregationStart{fill:var(--md-mermaid-label-bg-color)!important;stroke:var(--md-mermaid-edge-color)!important}g.stateGroup rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}g.stateGroup .state-title{fill:var(--md-mermaid-label-fg-color)!important;font-family:var(--md-mermaid-font-family)}g.stateGroup .composit{fill:var(--md-mermaid-label-bg-color)}.nodeLabel{color:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.node circle.state-end,.node circle.state-start,.start-state{fill:var(--md-mermaid-edge-color);stroke:none}.end-state-inner,.end-state-outer{fill:var(--md-mermaid-edge-color)}.end-state-inner,.node circle.state-end{stroke:var(--md-mermaid-label-bg-color)}.transition{stroke:var(--md-mermaid-edge-color)}[id^=state-fork] rect,[id^=state-join] rect{fill:var(--md-mermaid-edge-color)!important;stroke:none!important}.statediagram-cluster.statediagram-cluster .inner{fill:var(--md-default-bg-color)}.statediagram-cluster rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}.statediagram-state rect.divider{fill:var(--md-default-fg-color--lightest);stroke:var(--md-default-fg-color--lighter)}defs #statediagram-barbEnd{stroke:var(--md-mermaid-edge-color)}.attributeBoxEven,.attributeBoxOdd{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}.entityBox{fill:var(--md-mermaid-label-bg-color);stroke:var(--md-mermaid-node-fg-color)}.entityLabel{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.relationshipLabelBox{fill:var(--md-mermaid-label-bg-color);fill-opacity:1;background-color:var(--md-mermaid-label-bg-color);opacity:1}.relationshipLabel{fill:var(--md-mermaid-label-fg-color)}.relationshipLine{stroke:var(--md-mermaid-edge-color)}defs #ONE_OR_MORE_END *,defs #ONE_OR_MORE_START *,defs #ONLY_ONE_END *,defs #ONLY_ONE_START *,defs #ZERO_OR_MORE_END *,defs #ZERO_OR_MORE_START *,defs #ZERO_OR_ONE_END *,defs #ZERO_OR_ONE_START *{stroke:var(--md-mermaid-edge-color)!important}defs #ZERO_OR_MORE_END circle,defs #ZERO_OR_MORE_START circle{fill:var(--md-mermaid-label-bg-color)}.actor{fill:var(--md-mermaid-sequence-actor-bg-color);stroke:var(--md-mermaid-sequence-actor-border-color)}text.actor>tspan{fill:var(--md-mermaid-sequence-actor-fg-color);font-family:var(--md-mermaid-font-family)}line{stroke:var(--md-mermaid-sequence-actor-line-color)}.actor-man circle,.actor-man line{fill:var(--md-mermaid-sequence-actorman-bg-color);stroke:var(--md-mermaid-sequence-actorman-line-color)}.messageLine0,.messageLine1{stroke:var(--md-mermaid-sequence-message-line-color)}.note{fill:var(--md-mermaid-sequence-note-bg-color);stroke:var(--md-mermaid-sequence-note-border-color)}.loopText,.loopText>tspan,.messageText,.noteText>tspan{stroke:none;font-family:var(--md-mermaid-font-family)!important}.messageText{fill:var(--md-mermaid-sequence-message-fg-color)}.loopText,.loopText>tspan{fill:var(--md-mermaid-sequence-loop-fg-color)}.noteText>tspan{fill:var(--md-mermaid-sequence-note-fg-color)}#arrowhead path{fill:var(--md-mermaid-sequence-message-line-color);stroke:none}.loopLine{fill:var(--md-mermaid-sequence-loop-bg-color);stroke:var(--md-mermaid-sequence-loop-border-color)}.labelBox{fill:var(--md-mermaid-sequence-label-bg-color);stroke:none}.labelText,.labelText>span{fill:var(--md-mermaid-sequence-label-fg-color);font-family:var(--md-mermaid-font-family)}.sequenceNumber{fill:var(--md-mermaid-sequence-number-fg-color)}rect.rect{fill:var(--md-mermaid-sequence-box-bg-color);stroke:none}rect.rect+text.text{fill:var(--md-mermaid-sequence-box-fg-color)}defs #sequencenumber{fill:var(--md-mermaid-sequence-number-bg-color)!important}";var zr,wa=0;function Sa(){return typeof mermaid=="undefined"||mermaid instanceof Element?ht("https://unpkg.com/mermaid@9.4.3/dist/mermaid.min.js"):j(void 0)}function On(e){return e.classList.remove("mermaid"),zr||(zr=Sa().pipe(w(()=>mermaid.initialize({startOnLoad:!1,themeCSS:Tn,sequence:{actorFontSize:"16px",messageFontSize:"16px",noteFontSize:"16px"}})),m(()=>{}),J(1))),zr.subscribe(()=>{e.classList.add("mermaid");let t=`__mermaid_${wa++}`,r=T("div",{class:"mermaid"}),o=e.textContent;mermaid.mermaidAPI.render(t,o,(n,i)=>{let s=r.attachShadow({mode:"closed"});s.innerHTML=n,e.replaceWith(r),i==null||i(s)})}),zr.pipe(m(()=>({ref:e})))}var Mn=T("table");function Ln(e){return e.replaceWith(Mn),Mn.replaceWith(dn(e)),j({ref:e})}function Ta(e){let t=q(":scope > input",e),r=t.find(o=>o.checked)||t[0];return _(...t.map(o=>h(o,"change").pipe(m(()=>U(`label[for="${o.id}"]`))))).pipe(V(U(`label[for="${r.id}"]`)),m(o=>({active:o})))}function _n(e,{viewport$:t}){let r=Ur("prev");e.append(r);let o=Ur("next");e.append(o);let n=U(".tabbed-labels",e);return H(()=>{let i=new x,s=i.pipe(Z(),re(!0));return B([i,ye(e)]).pipe(Ce(1,Oe),Y(s)).subscribe({next([{active:a},c]){let p=Je(a),{width:l}=he(a);e.style.setProperty("--md-indicator-x",`${p.x}px`),e.style.setProperty("--md-indicator-width",`${l}px`);let f=er(n);(p.x<f.x||p.x+l>f.x+c.width)&&n.scrollTo({left:Math.max(0,p.x-16),behavior:"smooth"})},complete(){e.style.removeProperty("--md-indicator-x"),e.style.removeProperty("--md-indicator-width")}}),B([dt(n),ye(n)]).pipe(Y(s)).subscribe(([a,c])=>{let p=bt(n);r.hidden=a.x<16,o.hidden=a.x>p.width-c.width-16}),_(h(r,"click").pipe(m(()=>-1)),h(o,"click").pipe(m(()=>1))).pipe(Y(s)).subscribe(a=>{let{width:c}=he(n);n.scrollBy({left:c*a,behavior:"smooth"})}),te("content.tabs.link")&&i.pipe(je(1),ne(t)).subscribe(([{active:a},{offset:c}])=>{let p=a.innerText.trim();if(a.hasAttribute("data-md-switching"))a.removeAttribute("data-md-switching");else{let l=e.offsetTop-c.y;for(let u of q("[data-tabs]"))for(let d of q(":scope > input",u)){let v=U(`label[for="${d.id}"]`);if(v!==a&&v.innerText.trim()===p){v.setAttribute("data-md-switching",""),d.click();break}}window.scrollTo({top:e.offsetTop-l});let f=__md_get("__tabs")||[];__md_set("__tabs",[...new Set([p,...f])])}}),i.pipe(Y(s)).subscribe(()=>{for(let a of q("audio, video",e))a.pause()}),Ta(e).pipe(w(a=>i.next(a)),A(()=>i.complete()),m(a=>R({ref:e},a)))}).pipe(rt(ae))}function An(e,{viewport$:t,target$:r,print$:o}){return _(...q(".annotate:not(.highlight)",e).map(n=>gn(n,{target$:r,print$:o})),...q("pre:not(.mermaid) > code",e).map(n=>wn(n,{target$:r,print$:o})),...q("pre.mermaid",e).map(n=>On(n)),...q("table:not([class])",e).map(n=>Ln(n)),...q("details",e).map(n=>Sn(n,{target$:r,print$:o})),...q("[data-tabs]",e).map(n=>_n(n,{viewport$:t})))}function Oa(e,{alert$:t}){return t.pipe(E(r=>_(j(!0),j(!1).pipe(ze(2e3))).pipe(m(o=>({message:r,active:o})))))}function Cn(e,t){let r=U(".md-typeset",e);return H(()=>{let o=new x;return o.subscribe(({message:n,active:i})=>{e.classList.toggle("md-dialog--active",i),r.textContent=n}),Oa(e,t).pipe(w(n=>o.next(n)),A(()=>o.complete()),m(n=>R({ref:e},n)))})}function Ma({viewport$:e}){if(!te("header.autohide"))return j(!1);let t=e.pipe(m(({offset:{y:n}})=>n),Le(2,1),m(([n,i])=>[n<i,i]),ee(0)),r=B([e,t]).pipe(M(([{offset:n},[,i]])=>Math.abs(i-n.y)>100),m(([,[n]])=>n),X()),o=We("search");return B([e,o]).pipe(m(([{offset:n},i])=>n.y>400&&!i),X(),E(n=>n?r:j(!1)),V(!1))}function kn(e,t){return H(()=>B([ye(e),Ma(t)])).pipe(m(([{height:r},o])=>({height:r,hidden:o})),X((r,o)=>r.height===o.height&&r.hidden===o.hidden),J(1))}function Hn(e,{header$:t,main$:r}){return H(()=>{let o=new x,n=o.pipe(Z(),re(!0));return o.pipe(ee("active"),Ge(t)).subscribe(([{active:i},{hidden:s}])=>{e.classList.toggle("md-header--shadow",i&&!s),e.hidden=s}),r.subscribe(o),t.pipe(Y(n),m(i=>R({ref:e},i)))})}function La(e,{viewport$:t,header$:r}){return ar(e,{viewport$:t,header$:r}).pipe(m(({offset:{y:o}})=>{let{height:n}=he(e);return{active:o>=n}}),ee("active"))}function $n(e,t){return H(()=>{let r=new x;r.subscribe({next({active:n}){e.classList.toggle("md-header__title--active",n)},complete(){e.classList.remove("md-header__title--active")}});let o=se(".md-content h1");return typeof o=="undefined"?L:La(o,t).pipe(w(n=>r.next(n)),A(()=>r.complete()),m(n=>R({ref:e},n)))})}function Rn(e,{viewport$:t,header$:r}){let o=r.pipe(m(({height:i})=>i),X()),n=o.pipe(E(()=>ye(e).pipe(m(({height:i})=>({top:e.offsetTop,bottom:e.offsetTop+i})),ee("bottom"))));return B([o,n,t]).pipe(m(([i,{top:s,bottom:a},{offset:{y:c},size:{height:p}}])=>(p=Math.max(0,p-Math.max(0,s-c,i)-Math.max(0,p+c-a)),{offset:s-i,height:p,active:s-i<=c})),X((i,s)=>i.offset===s.offset&&i.height===s.height&&i.active===s.active))}function _a(e){let t=__md_get("__palette")||{index:e.findIndex(r=>matchMedia(r.getAttribute("data-md-color-media")).matches)};return j(...e).pipe(ce(r=>h(r,"change").pipe(m(()=>r))),V(e[Math.max(0,t.index)]),m(r=>({index:e.indexOf(r),color:{scheme:r.getAttribute("data-md-color-scheme"),primary:r.getAttribute("data-md-color-primary"),accent:r.getAttribute("data-md-color-accent")}})),J(1))}function Pn(e){let t=T("meta",{name:"theme-color"});document.head.appendChild(t);let r=T("meta",{name:"color-scheme"});return document.head.appendChild(r),H(()=>{let o=new x;o.subscribe(i=>{document.body.setAttribute("data-md-color-switching","");for(let[s,a]of Object.entries(i.color))document.body.setAttribute(`data-md-color-${s}`,a);for(let s=0;s<n.length;s++){let a=n[s].nextElementSibling;a instanceof HTMLElement&&(a.hidden=i.index!==s)}__md_set("__palette",i)}),o.pipe(m(()=>{let i=Ee("header"),s=window.getComputedStyle(i);return r.content=s.colorScheme,s.backgroundColor.match(/\d+/g).map(a=>(+a).toString(16).padStart(2,"0")).join("")})).subscribe(i=>t.content=`#${i}`),o.pipe(Se(ae)).subscribe(()=>{document.body.removeAttribute("data-md-color-switching")});let n=q("input",e);return _a(n).pipe(w(i=>o.next(i)),A(()=>o.complete()),m(i=>R({ref:e},i)))})}function In(e,{progress$:t}){return H(()=>{let r=new x;return r.subscribe(({value:o})=>{e.style.setProperty("--md-progress-value",`${o}`)}),t.pipe(w(o=>r.next({value:o})),A(()=>r.complete()),m(o=>({ref:e,value:o})))})}var qr=Ht(Vr());function Aa(e){e.setAttribute("data-md-copying","");let t=e.closest("[data-copy]"),r=t?t.getAttribute("data-copy"):e.innerText;return e.removeAttribute("data-md-copying"),r.trimEnd()}function Fn({alert$:e}){qr.default.isSupported()&&new P(t=>{new qr.default("[data-clipboard-target], [data-clipboard-text]",{text:r=>r.getAttribute("data-clipboard-text")||Aa(U(r.getAttribute("data-clipboard-target")))}).on("success",r=>t.next(r))}).pipe(w(t=>{t.trigger.focus()}),m(()=>be("clipboard.copied"))).subscribe(e)}function Ca(e){if(e.length<2)return[""];let[t,r]=[...e].sort((n,i)=>n.length-i.length).map(n=>n.replace(/[^/]+$/,"")),o=0;if(t===r)o=t.length;else for(;t.charCodeAt(o)===r.charCodeAt(o);)o++;return e.map(n=>n.replace(t.slice(0,o),""))}function cr(e){let t=__md_get("__sitemap",sessionStorage,e);if(t)return j(t);{let r=me();return Zo(new URL("sitemap.xml",e||r.base)).pipe(m(o=>Ca(q("loc",o).map(n=>n.textContent))),de(()=>L),He([]),w(o=>__md_set("__sitemap",o,sessionStorage,e)))}}function jn(e){let t=se("[rel=canonical]",e);typeof t!="undefined"&&(t.href=t.href.replace("//localhost:","//127.0.0.1:"));let r=new Map;for(let o of q(":scope > *",e)){let n=o.outerHTML;for(let i of["href","src"]){let s=o.getAttribute(i);if(s===null)continue;let a=new URL(s,t==null?void 0:t.href),c=o.cloneNode();c.setAttribute(i,`${a}`),n=c.outerHTML;break}r.set(n,o)}return r}function Wn({location$:e,viewport$:t,progress$:r}){let o=me();if(location.protocol==="file:")return L;let n=cr().pipe(m(l=>l.map(f=>`${new URL(f,o.base)}`))),i=h(document.body,"click").pipe(ne(n),E(([l,f])=>{if(!(l.target instanceof Element))return L;let u=l.target.closest("a");if(u===null)return L;if(u.target||l.metaKey||l.ctrlKey)return L;let d=new URL(u.href);return d.search=d.hash="",f.includes(`${d}`)?(l.preventDefault(),j(new URL(u.href))):L}),le());i.pipe(xe(1)).subscribe(()=>{let l=se("link[rel=icon]");typeof l!="undefined"&&(l.href=l.href)}),h(window,"beforeunload").subscribe(()=>{history.scrollRestoration="auto"}),i.pipe(ne(t)).subscribe(([l,{offset:f}])=>{history.scrollRestoration="manual",history.replaceState(f,""),history.pushState(null,"",l)}),i.subscribe(e);let s=e.pipe(V(pe()),ee("pathname"),je(1),E(l=>ir(l,{progress$:r}).pipe(de(()=>(ot(l,!0),L))))),a=new DOMParser,c=s.pipe(E(l=>l.text()),E(l=>{let f=a.parseFromString(l,"text/html");for(let b of["[data-md-component=announce]","[data-md-component=container]","[data-md-component=header-topic]","[data-md-component=outdated]","[data-md-component=logo]","[data-md-component=skip]",...te("navigation.tabs.sticky")?["[data-md-component=tabs]"]:[]]){let z=se(b),K=se(b,f);typeof z!="undefined"&&typeof K!="undefined"&&z.replaceWith(K)}let u=jn(document.head),d=jn(f.head);for(let[b,z]of d)z.getAttribute("rel")==="stylesheet"||z.hasAttribute("src")||(u.has(b)?u.delete(b):document.head.appendChild(z));for(let b of u.values())b.getAttribute("rel")==="stylesheet"||b.hasAttribute("src")||b.remove();let v=Ee("container");return Fe(q("script",v)).pipe(E(b=>{let z=f.createElement("script");if(b.src){for(let K of b.getAttributeNames())z.setAttribute(K,b.getAttribute(K));return b.replaceWith(z),new P(K=>{z.onload=()=>K.complete()})}else return z.textContent=b.textContent,b.replaceWith(z),L}),Z(),re(f))}),le());return h(window,"popstate").pipe(m(pe)).subscribe(e),e.pipe(V(pe()),Le(2,1),M(([l,f])=>l.pathname===f.pathname&&l.hash!==f.hash),m(([,l])=>l)).subscribe(l=>{var f,u;history.state!==null||!l.hash?window.scrollTo(0,(u=(f=history.state)==null?void 0:f.y)!=null?u:0):(history.scrollRestoration="auto",nr(l.hash),history.scrollRestoration="manual")}),e.pipe(Cr(i),V(pe()),Le(2,1),M(([l,f])=>l.pathname===f.pathname&&l.hash===f.hash),m(([,l])=>l)).subscribe(l=>{history.scrollRestoration="auto",nr(l.hash),history.scrollRestoration="manual",history.back()}),c.pipe(ne(e)).subscribe(([,l])=>{var f,u;history.state!==null||!l.hash?window.scrollTo(0,(u=(f=history.state)==null?void 0:f.y)!=null?u:0):nr(l.hash)}),t.pipe(ee("offset"),ke(100)).subscribe(({offset:l})=>{history.replaceState(l,"")}),c}var Dn=Ht(Un());function Vn(e){let t=e.separator.split("|").map(n=>n.replace(/(\(\?[!=<][^)]+\))/g,"").length===0?"\uFFFD":n).join("|"),r=new RegExp(t,"img"),o=(n,i,s)=>`${i}<mark data-md-highlight>${s}</mark>`;return n=>{n=n.replace(/[\s*+\-:~^]+/g," ").trim();let i=new RegExp(`(^|${e.separator}|)(${n.replace(/[|\\{}()[\]^$+*?.-]/g,"\\$&").replace(r,"|")})`,"img");return s=>(0,Dn.default)(s).replace(i,o).replace(/<\/mark>(\s+)<mark[^>]*>/img,"$1")}}function Mt(e){return e.type===1}function pr(e){return e.type===3}function zn(e,t){let r=an(e);return _(j(location.protocol!=="file:"),We("search")).pipe($e(o=>o),E(()=>t)).subscribe(({config:o,docs:n})=>r.next({type:0,data:{config:o,docs:n,options:{suggest:te("search.suggest")}}})),r}function qn({document$:e}){let t=me(),r=Ne(new URL("../versions.json",t.base)).pipe(de(()=>L)),o=r.pipe(m(n=>{let[,i]=t.base.match(/([^/]+)\/?$/);return n.find(({version:s,aliases:a})=>s===i||a.includes(i))||n[0]}));r.pipe(m(n=>new Map(n.map(i=>[`${new URL(`../${i.version}/`,t.base)}`,i]))),E(n=>h(document.body,"click").pipe(M(i=>!i.metaKey&&!i.ctrlKey),ne(o),E(([i,s])=>{if(i.target instanceof Element){let a=i.target.closest("a");if(a&&!a.target&&n.has(a.href)){let c=a.href;return!i.target.closest(".md-version")&&n.get(c)===s?L:(i.preventDefault(),j(c))}}return L}),E(i=>{let{version:s}=n.get(i);return cr(new URL(i)).pipe(m(a=>{let p=pe().href.replace(t.base,"");return a.includes(p.split("#")[0])?new URL(`../${s}/${p}`,t.base):new URL(i)}))})))).subscribe(n=>ot(n,!0)),B([r,o]).subscribe(([n,i])=>{U(".md-header__topic").appendChild(hn(n,i))}),e.pipe(E(()=>o)).subscribe(n=>{var s;let i=__md_get("__outdated",sessionStorage);if(i===null){i=!0;let a=((s=t.version)==null?void 0:s.default)||"latest";Array.isArray(a)||(a=[a]);e:for(let c of a)for(let p of n.aliases.concat(n.version))if(new RegExp(c,"i").test(p)){i=!1;break e}__md_set("__outdated",i,sessionStorage)}if(i)for(let a of oe("outdated"))a.hidden=!1})}function Pa(e,{worker$:t}){let{searchParams:r}=pe();r.has("q")&&(Ke("search",!0),e.value=r.get("q"),e.focus(),We("search").pipe($e(i=>!i)).subscribe(()=>{let i=pe();i.searchParams.delete("q"),history.replaceState({},"",`${i}`)}));let o=Zt(e),n=_(t.pipe($e(Mt)),h(e,"keyup"),o).pipe(m(()=>e.value),X());return B([n,o]).pipe(m(([i,s])=>({value:i,focus:s})),J(1))}function Kn(e,{worker$:t}){let r=new x,o=r.pipe(Z(),re(!0));B([t.pipe($e(Mt)),r],(i,s)=>s).pipe(ee("value")).subscribe(({value:i})=>t.next({type:2,data:i})),r.pipe(ee("focus")).subscribe(({focus:i})=>{i&&Ke("search",i)}),h(e.form,"reset").pipe(Y(o)).subscribe(()=>e.focus());let n=U("header [for=__search]");return h(n,"click").subscribe(()=>e.focus()),Pa(e,{worker$:t}).pipe(w(i=>r.next(i)),A(()=>r.complete()),m(i=>R({ref:e},i)),J(1))}function Qn(e,{worker$:t,query$:r}){let o=new x,n=Ko(e.parentElement).pipe(M(Boolean)),i=e.parentElement,s=U(":scope > :first-child",e),a=U(":scope > :last-child",e);We("search").subscribe(l=>a.setAttribute("role",l?"list":"presentation")),o.pipe(ne(r),$r(t.pipe($e(Mt)))).subscribe(([{items:l},{value:f}])=>{switch(l.length){case 0:s.textContent=f.length?be("search.result.none"):be("search.result.placeholder");break;case 1:s.textContent=be("search.result.one");break;default:let u=tr(l.length);s.textContent=be("search.result.other",u)}});let c=o.pipe(w(()=>a.innerHTML=""),E(({items:l})=>_(j(...l.slice(0,10)),j(...l.slice(10)).pipe(Le(4),Ir(n),E(([f])=>f)))),m(fn),le());return c.subscribe(l=>a.appendChild(l)),c.pipe(ce(l=>{let f=se("details",l);return typeof f=="undefined"?L:h(f,"toggle").pipe(Y(o),m(()=>f))})).subscribe(l=>{l.open===!1&&l.offsetTop<=i.scrollTop&&i.scrollTo({top:l.offsetTop})}),t.pipe(M(pr),m(({data:l})=>l)).pipe(w(l=>o.next(l)),A(()=>o.complete()),m(l=>R({ref:e},l)))}function Ia(e,{query$:t}){return t.pipe(m(({value:r})=>{let o=pe();return o.hash="",r=r.replace(/\s+/g,"+").replace(/&/g,"%26").replace(/=/g,"%3D"),o.search=`q=${r}`,{url:o}}))}function Yn(e,t){let r=new x,o=r.pipe(Z(),re(!0));return r.subscribe(({url:n})=>{e.setAttribute("data-clipboard-text",e.href),e.href=`${n}`}),h(e,"click").pipe(Y(o)).subscribe(n=>n.preventDefault()),Ia(e,t).pipe(w(n=>r.next(n)),A(()=>r.complete()),m(n=>R({ref:e},n)))}function Bn(e,{worker$:t,keyboard$:r}){let o=new x,n=Ee("search-query"),i=_(h(n,"keydown"),h(n,"focus")).pipe(Se(ae),m(()=>n.value),X());return o.pipe(Ge(i),m(([{suggest:a},c])=>{let p=c.split(/([\s-]+)/);if(a!=null&&a.length&&p[p.length-1]){let l=a[a.length-1];l.startsWith(p[p.length-1])&&(p[p.length-1]=l)}else p.length=0;return p})).subscribe(a=>e.innerHTML=a.join("").replace(/\s/g,"&nbsp;")),r.pipe(M(({mode:a})=>a==="search")).subscribe(a=>{switch(a.type){case"ArrowRight":e.innerText.length&&n.selectionStart===n.value.length&&(n.value=e.innerText);break}}),t.pipe(M(pr),m(({data:a})=>a)).pipe(w(a=>o.next(a)),A(()=>o.complete()),m(()=>({ref:e})))}function Gn(e,{index$:t,keyboard$:r}){let o=me();try{let n=zn(o.search,t),i=Ee("search-query",e),s=Ee("search-result",e);h(e,"click").pipe(M(({target:c})=>c instanceof Element&&!!c.closest("a"))).subscribe(()=>Ke("search",!1)),r.pipe(M(({mode:c})=>c==="search")).subscribe(c=>{let p=Re();switch(c.type){case"Enter":if(p===i){let l=new Map;for(let f of q(":first-child [href]",s)){let u=f.firstElementChild;l.set(f,parseFloat(u.getAttribute("data-md-score")))}if(l.size){let[[f]]=[...l].sort(([,u],[,d])=>d-u);f.click()}c.claim()}break;case"Escape":case"Tab":Ke("search",!1),i.blur();break;case"ArrowUp":case"ArrowDown":if(typeof p=="undefined")i.focus();else{let l=[i,...q(":not(details) > [href], summary, details[open] [href]",s)],f=Math.max(0,(Math.max(0,l.indexOf(p))+l.length+(c.type==="ArrowUp"?-1:1))%l.length);l[f].focus()}c.claim();break;default:i!==Re()&&i.focus()}}),r.pipe(M(({mode:c})=>c==="global")).subscribe(c=>{switch(c.type){case"f":case"s":case"/":i.focus(),i.select(),c.claim();break}});let a=Kn(i,{worker$:n});return _(a,Qn(s,{worker$:n,query$:a})).pipe(qe(...oe("search-share",e).map(c=>Yn(c,{query$:a})),...oe("search-suggest",e).map(c=>Bn(c,{worker$:n,keyboard$:r}))))}catch(n){return e.hidden=!0,Ve}}function Jn(e,{index$:t,location$:r}){return B([t,r.pipe(V(pe()),M(o=>!!o.searchParams.get("h")))]).pipe(m(([o,n])=>Vn(o.config)(n.searchParams.get("h"))),m(o=>{var s;let n=new Map,i=document.createNodeIterator(e,NodeFilter.SHOW_TEXT);for(let a=i.nextNode();a;a=i.nextNode())if((s=a.parentElement)!=null&&s.offsetHeight){let c=a.textContent,p=o(c);p.length>c.length&&n.set(a,p)}for(let[a,c]of n){let{childNodes:p}=T("span",null,c);a.replaceWith(...Array.from(p))}return{ref:e,nodes:n}}))}function Fa(e,{viewport$:t,main$:r}){let o=e.closest(".md-grid"),n=o.offsetTop-o.parentElement.offsetTop;return B([r,t]).pipe(m(([{offset:i,height:s},{offset:{y:a}}])=>(s=s+Math.min(n,Math.max(0,a-i))-n,{height:s,locked:a>=i+n})),X((i,s)=>i.height===s.height&&i.locked===s.locked))}function Kr(e,o){var n=o,{header$:t}=n,r=eo(n,["header$"]);let i=U(".md-sidebar__scrollwrap",e),{y:s}=Je(i);return H(()=>{let a=new x,c=a.pipe(Z(),re(!0)),p=a.pipe(Ce(0,Oe));return p.pipe(ne(t)).subscribe({next([{height:l},{height:f}]){i.style.height=`${l-2*s}px`,e.style.top=`${f}px`},complete(){i.style.height="",e.style.top=""}}),p.pipe($e()).subscribe(()=>{for(let l of q(".md-nav__link--active[href]",e)){if(!l.clientHeight)continue;let f=l.closest(".md-sidebar__scrollwrap");if(typeof f!="undefined"){let u=l.offsetTop-f.offsetTop,{height:d}=he(f);f.scrollTo({top:u-d/2})}}}),ge(q("label[tabindex]",e)).pipe(ce(l=>h(l,"click").pipe(Se(ae),m(()=>l),Y(c)))).subscribe(l=>{let f=U(`[id="${l.htmlFor}"]`);U(`[aria-labelledby="${l.id}"]`).setAttribute("aria-expanded",`${f.checked}`)}),Fa(e,r).pipe(w(l=>a.next(l)),A(()=>a.complete()),m(l=>R({ref:e},l)))})}function Xn(e,t){if(typeof t!="undefined"){let r=`https://api.github.com/repos/${e}/${t}`;return St(Ne(`${r}/releases/latest`).pipe(de(()=>L),m(o=>({version:o.tag_name})),He({})),Ne(r).pipe(de(()=>L),m(o=>({stars:o.stargazers_count,forks:o.forks_count})),He({}))).pipe(m(([o,n])=>R(R({},o),n)))}else{let r=`https://api.github.com/users/${e}`;return Ne(r).pipe(m(o=>({repositories:o.public_repos})),He({}))}}function Zn(e,t){let r=`https://${e}/api/v4/projects/${encodeURIComponent(t)}`;return Ne(r).pipe(de(()=>L),m(({star_count:o,forks_count:n})=>({stars:o,forks:n})),He({}))}function ei(e){let t=e.match(/^.+github\.com\/([^/]+)\/?([^/]+)?/i);if(t){let[,r,o]=t;return Xn(r,o)}if(t=e.match(/^.+?([^/]*gitlab[^/]+)\/(.+?)\/?$/i),t){let[,r,o]=t;return Zn(r,o)}return L}var ja;function Wa(e){return ja||(ja=H(()=>{let t=__md_get("__source",sessionStorage);if(t)return j(t);if(oe("consent").length){let o=__md_get("__consent");if(!(o&&o.github))return L}return ei(e.href).pipe(w(o=>__md_set("__source",o,sessionStorage)))}).pipe(de(()=>L),M(t=>Object.keys(t).length>0),m(t=>({facts:t})),J(1)))}function ti(e){let t=U(":scope > :last-child",e);return H(()=>{let r=new x;return r.subscribe(({facts:o})=>{t.appendChild(un(o)),t.classList.add("md-source__repository--active")}),Wa(e).pipe(w(o=>r.next(o)),A(()=>r.complete()),m(o=>R({ref:e},o)))})}function Na(e,{viewport$:t,header$:r}){return ye(document.body).pipe(E(()=>ar(e,{header$:r,viewport$:t})),m(({offset:{y:o}})=>({hidden:o>=10})),ee("hidden"))}function ri(e,t){return H(()=>{let r=new x;return r.subscribe({next({hidden:o}){e.hidden=o},complete(){e.hidden=!1}}),(te("navigation.tabs.sticky")?j({hidden:!1}):Na(e,t)).pipe(w(o=>r.next(o)),A(()=>r.complete()),m(o=>R({ref:e},o)))})}function Ua(e,{viewport$:t,header$:r}){let o=new Map,n=q("[href^=\\#]",e);for(let a of n){let c=decodeURIComponent(a.hash.substring(1)),p=se(`[id="${c}"]`);typeof p!="undefined"&&o.set(a,p)}let i=r.pipe(ee("height"),m(({height:a})=>{let c=Ee("main"),p=U(":scope > :first-child",c);return a+.8*(p.offsetTop-c.offsetTop)}),le());return ye(document.body).pipe(ee("height"),E(a=>H(()=>{let c=[];return j([...o].reduce((p,[l,f])=>{for(;c.length&&o.get(c[c.length-1]).tagName>=f.tagName;)c.pop();let u=f.offsetTop;for(;!u&&f.parentElement;)f=f.parentElement,u=f.offsetTop;let d=f.offsetParent;for(;d;d=d.offsetParent)u+=d.offsetTop;return p.set([...c=[...c,l]].reverse(),u)},new Map))}).pipe(m(c=>new Map([...c].sort(([,p],[,l])=>p-l))),Ge(i),E(([c,p])=>t.pipe(kr(([l,f],{offset:{y:u},size:d})=>{let v=u+d.height>=Math.floor(a.height);for(;f.length;){let[,b]=f[0];if(b-p<u||v)l=[...l,f.shift()];else break}for(;l.length;){let[,b]=l[l.length-1];if(b-p>=u&&!v)f=[l.pop(),...f];else break}return[l,f]},[[],[...c]]),X((l,f)=>l[0]===f[0]&&l[1]===f[1])))))).pipe(m(([a,c])=>({prev:a.map(([p])=>p),next:c.map(([p])=>p)})),V({prev:[],next:[]}),Le(2,1),m(([a,c])=>a.prev.length<c.prev.length?{prev:c.prev.slice(Math.max(0,a.prev.length-1),c.prev.length),next:[]}:{prev:c.prev.slice(-1),next:c.next.slice(0,c.next.length-a.next.length)}))}function oi(e,{viewport$:t,header$:r,main$:o,target$:n}){return H(()=>{let i=new x,s=i.pipe(Z(),re(!0));if(i.subscribe(({prev:a,next:c})=>{for(let[p]of c)p.classList.remove("md-nav__link--passed"),p.classList.remove("md-nav__link--active");for(let[p,[l]]of a.entries())l.classList.add("md-nav__link--passed"),l.classList.toggle("md-nav__link--active",p===a.length-1)}),te("toc.follow")){let a=_(t.pipe(ke(1),m(()=>{})),t.pipe(ke(250),m(()=>"smooth")));i.pipe(M(({prev:c})=>c.length>0),Ge(o.pipe(Se(ae))),ne(a)).subscribe(([[{prev:c}],p])=>{let[l]=c[c.length-1];if(l.offsetHeight){let f=zo(l);if(typeof f!="undefined"){let u=l.offsetTop-f.offsetTop,{height:d}=he(f);f.scrollTo({top:u-d/2,behavior:p})}}})}return te("navigation.tracking")&&t.pipe(Y(s),ee("offset"),ke(250),je(1),Y(n.pipe(je(1))),Tt({delay:250}),ne(i)).subscribe(([,{prev:a}])=>{let c=pe(),p=a[a.length-1];if(p&&p.length){let[l]=p,{hash:f}=new URL(l.href);c.hash!==f&&(c.hash=f,history.replaceState({},"",`${c}`))}else c.hash="",history.replaceState({},"",`${c}`)}),Ua(e,{viewport$:t,header$:r}).pipe(w(a=>i.next(a)),A(()=>i.complete()),m(a=>R({ref:e},a)))})}function Da(e,{viewport$:t,main$:r,target$:o}){let n=t.pipe(m(({offset:{y:s}})=>s),Le(2,1),m(([s,a])=>s>a&&a>0),X()),i=r.pipe(m(({active:s})=>s));return B([i,n]).pipe(m(([s,a])=>!(s&&a)),X(),Y(o.pipe(je(1))),re(!0),Tt({delay:250}),m(s=>({hidden:s})))}function ni(e,{viewport$:t,header$:r,main$:o,target$:n}){let i=new x,s=i.pipe(Z(),re(!0));return i.subscribe({next({hidden:a}){e.hidden=a,a?(e.setAttribute("tabindex","-1"),e.blur()):e.removeAttribute("tabindex")},complete(){e.style.top="",e.hidden=!0,e.removeAttribute("tabindex")}}),r.pipe(Y(s),ee("height")).subscribe(({height:a})=>{e.style.top=`${a+16}px`}),h(e,"click").subscribe(a=>{a.preventDefault(),window.scrollTo({top:0})}),Da(e,{viewport$:t,main$:o,target$:n}).pipe(w(a=>i.next(a)),A(()=>i.complete()),m(a=>R({ref:e},a)))}function ii({document$:e,tablet$:t}){e.pipe(E(()=>q(".md-toggle--indeterminate")),w(r=>{r.indeterminate=!0,r.checked=!1}),ce(r=>h(r,"change").pipe(Rr(()=>r.classList.contains("md-toggle--indeterminate")),m(()=>r))),ne(t)).subscribe(([r,o])=>{r.classList.remove("md-toggle--indeterminate"),o&&(r.checked=!1)})}function Va(){return/(iPad|iPhone|iPod)/.test(navigator.userAgent)}function ai({document$:e}){e.pipe(E(()=>q("[data-md-scrollfix]")),w(t=>t.removeAttribute("data-md-scrollfix")),M(Va),ce(t=>h(t,"touchstart").pipe(m(()=>t)))).subscribe(t=>{let r=t.scrollTop;r===0?t.scrollTop=1:r+t.offsetHeight===t.scrollHeight&&(t.scrollTop=r-1)})}function si({viewport$:e,tablet$:t}){B([We("search"),t]).pipe(m(([r,o])=>r&&!o),E(r=>j(r).pipe(ze(r?400:100))),ne(e)).subscribe(([r,{offset:{y:o}}])=>{if(r)document.body.setAttribute("data-md-scrolllock",""),document.body.style.top=`-${o}px`;else{let n=-1*parseInt(document.body.style.top,10);document.body.removeAttribute("data-md-scrolllock"),document.body.style.top="",n&&window.scrollTo(0,n)}})}Object.entries||(Object.entries=function(e){let t=[];for(let r of Object.keys(e))t.push([r,e[r]]);return t});Object.values||(Object.values=function(e){let t=[];for(let r of Object.keys(e))t.push(e[r]);return t});typeof Element!="undefined"&&(Element.prototype.scrollTo||(Element.prototype.scrollTo=function(e,t){typeof e=="object"?(this.scrollLeft=e.left,this.scrollTop=e.top):(this.scrollLeft=e,this.scrollTop=t)}),Element.prototype.replaceWith||(Element.prototype.replaceWith=function(...e){let t=this.parentNode;if(t){e.length===0&&t.removeChild(this);for(let r=e.length-1;r>=0;r--){let o=e[r];typeof o=="string"?o=document.createTextNode(o):o.parentNode&&o.parentNode.removeChild(o),r?t.insertBefore(this.previousSibling,o):t.replaceChild(o,this)}}}));function za(){return location.protocol==="file:"?ht(`${new URL("search/search_index.js",Qr.base)}`).pipe(m(()=>__index),J(1)):Ne(new URL("search/search_index.json",Qr.base))}document.documentElement.classList.remove("no-js");document.documentElement.classList.add("js");var nt=No(),_t=Bo(),gt=Jo(_t),Yr=Yo(),Te=nn(),lr=Fr("(min-width: 960px)"),pi=Fr("(min-width: 1220px)"),li=Xo(),Qr=me(),mi=document.forms.namedItem("search")?za():Ve,Br=new x;Fn({alert$:Br});var Gr=new x;te("navigation.instant")&&Wn({location$:_t,viewport$:Te,progress$:Gr}).subscribe(nt);var ci;((ci=Qr.version)==null?void 0:ci.provider)==="mike"&&qn({document$:nt});_(_t,gt).pipe(ze(125)).subscribe(()=>{Ke("drawer",!1),Ke("search",!1)});Yr.pipe(M(({mode:e})=>e==="global")).subscribe(e=>{switch(e.type){case"p":case",":let t=se("link[rel=prev]");typeof t!="undefined"&&ot(t);break;case"n":case".":let r=se("link[rel=next]");typeof r!="undefined"&&ot(r);break;case"Enter":let o=Re();o instanceof HTMLLabelElement&&o.click()}});ii({document$:nt,tablet$:lr});ai({document$:nt});si({viewport$:Te,tablet$:lr});var Xe=kn(Ee("header"),{viewport$:Te}),Lt=nt.pipe(m(()=>Ee("main")),E(e=>Rn(e,{viewport$:Te,header$:Xe})),J(1)),qa=_(...oe("consent").map(e=>cn(e,{target$:gt})),...oe("dialog").map(e=>Cn(e,{alert$:Br})),...oe("header").map(e=>Hn(e,{viewport$:Te,header$:Xe,main$:Lt})),...oe("palette").map(e=>Pn(e)),...oe("progress").map(e=>In(e,{progress$:Gr})),...oe("search").map(e=>Gn(e,{index$:mi,keyboard$:Yr})),...oe("source").map(e=>ti(e))),Ka=H(()=>_(...oe("announce").map(e=>sn(e)),...oe("content").map(e=>An(e,{viewport$:Te,target$:gt,print$:li})),...oe("content").map(e=>te("search.highlight")?Jn(e,{index$:mi,location$:_t}):L),...oe("header-title").map(e=>$n(e,{viewport$:Te,header$:Xe})),...oe("sidebar").map(e=>e.getAttribute("data-md-type")==="navigation"?jr(pi,()=>Kr(e,{viewport$:Te,header$:Xe,main$:Lt})):jr(lr,()=>Kr(e,{viewport$:Te,header$:Xe,main$:Lt}))),...oe("tabs").map(e=>ri(e,{viewport$:Te,header$:Xe})),...oe("toc").map(e=>oi(e,{viewport$:Te,header$:Xe,main$:Lt,target$:gt})),...oe("top").map(e=>ni(e,{viewport$:Te,header$:Xe,main$:Lt,target$:gt})))),fi=nt.pipe(E(()=>Ka),qe(qa),J(1));fi.subscribe();window.document$=nt;window.location$=_t;window.target$=gt;window.keyboard$=Yr;window.viewport$=Te;window.tablet$=lr;window.screen$=pi;window.print$=li;window.alert$=Br;window.progress$=Gr;window.component$=fi;})();
-//# sourceMappingURL=bundle.6c14ae12.min.js.map
-
diff --git a/assets/javascripts/bundle.6c14ae12.min.js.map b/assets/javascripts/bundle.6c14ae12.min.js.map
deleted file mode 100644
index 49396ad0b4..0000000000
--- a/assets/javascripts/bundle.6c14ae12.min.js.map
+++ /dev/null
@@ -1,7 +0,0 @@
-{
-  "version": 3,
-  "sources": ["node_modules/focus-visible/dist/focus-visible.js", "node_modules/clipboard/dist/clipboard.js", "node_modules/escape-html/index.js", "src/templates/assets/javascripts/bundle.ts", "node_modules/rxjs/node_modules/tslib/tslib.es6.js", "node_modules/rxjs/src/internal/util/isFunction.ts", "node_modules/rxjs/src/internal/util/createErrorClass.ts", "node_modules/rxjs/src/internal/util/UnsubscriptionError.ts", "node_modules/rxjs/src/internal/util/arrRemove.ts", "node_modules/rxjs/src/internal/Subscription.ts", "node_modules/rxjs/src/internal/config.ts", "node_modules/rxjs/src/internal/scheduler/timeoutProvider.ts", "node_modules/rxjs/src/internal/util/reportUnhandledError.ts", "node_modules/rxjs/src/internal/util/noop.ts", "node_modules/rxjs/src/internal/NotificationFactories.ts", "node_modules/rxjs/src/internal/util/errorContext.ts", "node_modules/rxjs/src/internal/Subscriber.ts", "node_modules/rxjs/src/internal/symbol/observable.ts", "node_modules/rxjs/src/internal/util/identity.ts", "node_modules/rxjs/src/internal/util/pipe.ts", "node_modules/rxjs/src/internal/Observable.ts", "node_modules/rxjs/src/internal/util/lift.ts", "node_modules/rxjs/src/internal/operators/OperatorSubscriber.ts", "node_modules/rxjs/src/internal/scheduler/animationFrameProvider.ts", "node_modules/rxjs/src/internal/util/ObjectUnsubscribedError.ts", "node_modules/rxjs/src/internal/Subject.ts", "node_modules/rxjs/src/internal/scheduler/dateTimestampProvider.ts", "node_modules/rxjs/src/internal/ReplaySubject.ts", "node_modules/rxjs/src/internal/scheduler/Action.ts", "node_modules/rxjs/src/internal/scheduler/intervalProvider.ts", "node_modules/rxjs/src/internal/scheduler/AsyncAction.ts", "node_modules/rxjs/src/internal/Scheduler.ts", "node_modules/rxjs/src/internal/scheduler/AsyncScheduler.ts", "node_modules/rxjs/src/internal/scheduler/async.ts", "node_modules/rxjs/src/internal/scheduler/AnimationFrameAction.ts", "node_modules/rxjs/src/internal/scheduler/AnimationFrameScheduler.ts", "node_modules/rxjs/src/internal/scheduler/animationFrame.ts", "node_modules/rxjs/src/internal/observable/empty.ts", "node_modules/rxjs/src/internal/util/isScheduler.ts", "node_modules/rxjs/src/internal/util/args.ts", "node_modules/rxjs/src/internal/util/isArrayLike.ts", "node_modules/rxjs/src/internal/util/isPromise.ts", "node_modules/rxjs/src/internal/util/isInteropObservable.ts", "node_modules/rxjs/src/internal/util/isAsyncIterable.ts", "node_modules/rxjs/src/internal/util/throwUnobservableError.ts", "node_modules/rxjs/src/internal/symbol/iterator.ts", "node_modules/rxjs/src/internal/util/isIterable.ts", "node_modules/rxjs/src/internal/util/isReadableStreamLike.ts", "node_modules/rxjs/src/internal/observable/innerFrom.ts", "node_modules/rxjs/src/internal/util/executeSchedule.ts", "node_modules/rxjs/src/internal/operators/observeOn.ts", "node_modules/rxjs/src/internal/operators/subscribeOn.ts", "node_modules/rxjs/src/internal/scheduled/scheduleObservable.ts", "node_modules/rxjs/src/internal/scheduled/schedulePromise.ts", "node_modules/rxjs/src/internal/scheduled/scheduleArray.ts", "node_modules/rxjs/src/internal/scheduled/scheduleIterable.ts", "node_modules/rxjs/src/internal/scheduled/scheduleAsyncIterable.ts", "node_modules/rxjs/src/internal/scheduled/scheduleReadableStreamLike.ts", "node_modules/rxjs/src/internal/scheduled/scheduled.ts", "node_modules/rxjs/src/internal/observable/from.ts", "node_modules/rxjs/src/internal/observable/of.ts", "node_modules/rxjs/src/internal/observable/throwError.ts", "node_modules/rxjs/src/internal/util/EmptyError.ts", "node_modules/rxjs/src/internal/util/isDate.ts", "node_modules/rxjs/src/internal/operators/map.ts", "node_modules/rxjs/src/internal/util/mapOneOrManyArgs.ts", "node_modules/rxjs/src/internal/util/argsArgArrayOrObject.ts", "node_modules/rxjs/src/internal/util/createObject.ts", "node_modules/rxjs/src/internal/observable/combineLatest.ts", "node_modules/rxjs/src/internal/operators/mergeInternals.ts", "node_modules/rxjs/src/internal/operators/mergeMap.ts", "node_modules/rxjs/src/internal/operators/mergeAll.ts", "node_modules/rxjs/src/internal/operators/concatAll.ts", "node_modules/rxjs/src/internal/observable/concat.ts", "node_modules/rxjs/src/internal/observable/defer.ts", "node_modules/rxjs/src/internal/observable/fromEvent.ts", "node_modules/rxjs/src/internal/observable/fromEventPattern.ts", "node_modules/rxjs/src/internal/observable/timer.ts", "node_modules/rxjs/src/internal/observable/merge.ts", "node_modules/rxjs/src/internal/observable/never.ts", "node_modules/rxjs/src/internal/util/argsOrArgArray.ts", "node_modules/rxjs/src/internal/operators/filter.ts", "node_modules/rxjs/src/internal/observable/zip.ts", "node_modules/rxjs/src/internal/operators/audit.ts", "node_modules/rxjs/src/internal/operators/auditTime.ts", "node_modules/rxjs/src/internal/operators/bufferCount.ts", "node_modules/rxjs/src/internal/operators/catchError.ts", "node_modules/rxjs/src/internal/operators/scanInternals.ts", "node_modules/rxjs/src/internal/operators/combineLatest.ts", "node_modules/rxjs/src/internal/operators/combineLatestWith.ts", "node_modules/rxjs/src/internal/operators/debounceTime.ts", "node_modules/rxjs/src/internal/operators/defaultIfEmpty.ts", "node_modules/rxjs/src/internal/operators/take.ts", "node_modules/rxjs/src/internal/operators/ignoreElements.ts", "node_modules/rxjs/src/internal/operators/mapTo.ts", "node_modules/rxjs/src/internal/operators/delayWhen.ts", "node_modules/rxjs/src/internal/operators/delay.ts", "node_modules/rxjs/src/internal/operators/distinctUntilChanged.ts", "node_modules/rxjs/src/internal/operators/distinctUntilKeyChanged.ts", "node_modules/rxjs/src/internal/operators/throwIfEmpty.ts", "node_modules/rxjs/src/internal/operators/endWith.ts", "node_modules/rxjs/src/internal/operators/finalize.ts", "node_modules/rxjs/src/internal/operators/first.ts", "node_modules/rxjs/src/internal/operators/merge.ts", "node_modules/rxjs/src/internal/operators/mergeWith.ts", "node_modules/rxjs/src/internal/operators/repeat.ts", "node_modules/rxjs/src/internal/operators/sample.ts", "node_modules/rxjs/src/internal/operators/scan.ts", "node_modules/rxjs/src/internal/operators/share.ts", "node_modules/rxjs/src/internal/operators/shareReplay.ts", "node_modules/rxjs/src/internal/operators/skip.ts", "node_modules/rxjs/src/internal/operators/skipUntil.ts", "node_modules/rxjs/src/internal/operators/startWith.ts", "node_modules/rxjs/src/internal/operators/switchMap.ts", "node_modules/rxjs/src/internal/operators/takeUntil.ts", "node_modules/rxjs/src/internal/operators/takeWhile.ts", "node_modules/rxjs/src/internal/operators/tap.ts", "node_modules/rxjs/src/internal/operators/throttle.ts", "node_modules/rxjs/src/internal/operators/throttleTime.ts", "node_modules/rxjs/src/internal/operators/withLatestFrom.ts", "node_modules/rxjs/src/internal/operators/zip.ts", "node_modules/rxjs/src/internal/operators/zipWith.ts", "src/templates/assets/javascripts/browser/document/index.ts", "src/templates/assets/javascripts/browser/element/_/index.ts", "src/templates/assets/javascripts/browser/element/focus/index.ts", "src/templates/assets/javascripts/browser/element/offset/_/index.ts", "src/templates/assets/javascripts/browser/element/offset/content/index.ts", "src/templates/assets/javascripts/utilities/h/index.ts", "src/templates/assets/javascripts/utilities/round/index.ts", "src/templates/assets/javascripts/browser/script/index.ts", "src/templates/assets/javascripts/browser/element/size/_/index.ts", "src/templates/assets/javascripts/browser/element/size/content/index.ts", "src/templates/assets/javascripts/browser/element/visibility/index.ts", "src/templates/assets/javascripts/browser/toggle/index.ts", "src/templates/assets/javascripts/browser/keyboard/index.ts", "src/templates/assets/javascripts/browser/location/_/index.ts", "src/templates/assets/javascripts/browser/location/hash/index.ts", "src/templates/assets/javascripts/browser/media/index.ts", "src/templates/assets/javascripts/browser/request/index.ts", "src/templates/assets/javascripts/browser/viewport/offset/index.ts", "src/templates/assets/javascripts/browser/viewport/size/index.ts", "src/templates/assets/javascripts/browser/viewport/_/index.ts", "src/templates/assets/javascripts/browser/viewport/at/index.ts", "src/templates/assets/javascripts/browser/worker/index.ts", "src/templates/assets/javascripts/_/index.ts", "src/templates/assets/javascripts/components/_/index.ts", "src/templates/assets/javascripts/components/announce/index.ts", "src/templates/assets/javascripts/components/consent/index.ts", "src/templates/assets/javascripts/components/content/annotation/_/index.ts", "src/templates/assets/javascripts/templates/tooltip/index.tsx", "src/templates/assets/javascripts/templates/annotation/index.tsx", "src/templates/assets/javascripts/templates/clipboard/index.tsx", "src/templates/assets/javascripts/templates/search/index.tsx", "src/templates/assets/javascripts/templates/source/index.tsx", "src/templates/assets/javascripts/templates/tabbed/index.tsx", "src/templates/assets/javascripts/templates/table/index.tsx", "src/templates/assets/javascripts/templates/version/index.tsx", "src/templates/assets/javascripts/components/content/annotation/list/index.ts", "src/templates/assets/javascripts/components/content/annotation/block/index.ts", "src/templates/assets/javascripts/components/content/code/_/index.ts", "src/templates/assets/javascripts/components/content/details/index.ts", "src/templates/assets/javascripts/components/content/mermaid/index.css", "src/templates/assets/javascripts/components/content/mermaid/index.ts", "src/templates/assets/javascripts/components/content/table/index.ts", "src/templates/assets/javascripts/components/content/tabs/index.ts", "src/templates/assets/javascripts/components/content/_/index.ts", "src/templates/assets/javascripts/components/dialog/index.ts", "src/templates/assets/javascripts/components/header/_/index.ts", "src/templates/assets/javascripts/components/header/title/index.ts", "src/templates/assets/javascripts/components/main/index.ts", "src/templates/assets/javascripts/components/palette/index.ts", "src/templates/assets/javascripts/components/progress/index.ts", "src/templates/assets/javascripts/integrations/clipboard/index.ts", "src/templates/assets/javascripts/integrations/sitemap/index.ts", "src/templates/assets/javascripts/integrations/instant/index.ts", "src/templates/assets/javascripts/integrations/search/highlighter/index.ts", "src/templates/assets/javascripts/integrations/search/worker/message/index.ts", "src/templates/assets/javascripts/integrations/search/worker/_/index.ts", "src/templates/assets/javascripts/integrations/version/index.ts", "src/templates/assets/javascripts/components/search/query/index.ts", "src/templates/assets/javascripts/components/search/result/index.ts", "src/templates/assets/javascripts/components/search/share/index.ts", "src/templates/assets/javascripts/components/search/suggest/index.ts", "src/templates/assets/javascripts/components/search/_/index.ts", "src/templates/assets/javascripts/components/search/highlight/index.ts", "src/templates/assets/javascripts/components/sidebar/index.ts", "src/templates/assets/javascripts/components/source/facts/github/index.ts", "src/templates/assets/javascripts/components/source/facts/gitlab/index.ts", "src/templates/assets/javascripts/components/source/facts/_/index.ts", "src/templates/assets/javascripts/components/source/_/index.ts", "src/templates/assets/javascripts/components/tabs/index.ts", "src/templates/assets/javascripts/components/toc/index.ts", "src/templates/assets/javascripts/components/top/index.ts", "src/templates/assets/javascripts/patches/indeterminate/index.ts", "src/templates/assets/javascripts/patches/scrollfix/index.ts", "src/templates/assets/javascripts/patches/scrolllock/index.ts", "src/templates/assets/javascripts/polyfills/index.ts"],
-  "sourcesContent": ["(function (global, factory) {\n  typeof exports === 'object' && typeof module !== 'undefined' ? factory() :\n  typeof define === 'function' && define.amd ? define(factory) :\n  (factory());\n}(this, (function () { 'use strict';\n\n  /**\n   * Applies the :focus-visible polyfill at the given scope.\n   * A scope in this case is either the top-level Document or a Shadow Root.\n   *\n   * @param {(Document|ShadowRoot)} scope\n   * @see https://github.com/WICG/focus-visible\n   */\n  function applyFocusVisiblePolyfill(scope) {\n    var hadKeyboardEvent = true;\n    var hadFocusVisibleRecently = false;\n    var hadFocusVisibleRecentlyTimeout = null;\n\n    var inputTypesAllowlist = {\n      text: true,\n      search: true,\n      url: true,\n      tel: true,\n      email: true,\n      password: true,\n      number: true,\n      date: true,\n      month: true,\n      week: true,\n      time: true,\n      datetime: true,\n      'datetime-local': true\n    };\n\n    /**\n     * Helper function for legacy browsers and iframes which sometimes focus\n     * elements like document, body, and non-interactive SVG.\n     * @param {Element} el\n     */\n    function isValidFocusTarget(el) {\n      if (\n        el &&\n        el !== document &&\n        el.nodeName !== 'HTML' &&\n        el.nodeName !== 'BODY' &&\n        'classList' in el &&\n        'contains' in el.classList\n      ) {\n        return true;\n      }\n      return false;\n    }\n\n    /**\n     * Computes whether the given element should automatically trigger the\n     * `focus-visible` class being added, i.e. whether it should always match\n     * `:focus-visible` when focused.\n     * @param {Element} el\n     * @return {boolean}\n     */\n    function focusTriggersKeyboardModality(el) {\n      var type = el.type;\n      var tagName = el.tagName;\n\n      if (tagName === 'INPUT' && inputTypesAllowlist[type] && !el.readOnly) {\n        return true;\n      }\n\n      if (tagName === 'TEXTAREA' && !el.readOnly) {\n        return true;\n      }\n\n      if (el.isContentEditable) {\n        return true;\n      }\n\n      return false;\n    }\n\n    /**\n     * Add the `focus-visible` class to the given element if it was not added by\n     * the author.\n     * @param {Element} el\n     */\n    function addFocusVisibleClass(el) {\n      if (el.classList.contains('focus-visible')) {\n        return;\n      }\n      el.classList.add('focus-visible');\n      el.setAttribute('data-focus-visible-added', '');\n    }\n\n    /**\n     * Remove the `focus-visible` class from the given element if it was not\n     * originally added by the author.\n     * @param {Element} el\n     */\n    function removeFocusVisibleClass(el) {\n      if (!el.hasAttribute('data-focus-visible-added')) {\n        return;\n      }\n      el.classList.remove('focus-visible');\n      el.removeAttribute('data-focus-visible-added');\n    }\n\n    /**\n     * If the most recent user interaction was via the keyboard;\n     * and the key press did not include a meta, alt/option, or control key;\n     * then the modality is keyboard. Otherwise, the modality is not keyboard.\n     * Apply `focus-visible` to any current active element and keep track\n     * of our keyboard modality state with `hadKeyboardEvent`.\n     * @param {KeyboardEvent} e\n     */\n    function onKeyDown(e) {\n      if (e.metaKey || e.altKey || e.ctrlKey) {\n        return;\n      }\n\n      if (isValidFocusTarget(scope.activeElement)) {\n        addFocusVisibleClass(scope.activeElement);\n      }\n\n      hadKeyboardEvent = true;\n    }\n\n    /**\n     * If at any point a user clicks with a pointing device, ensure that we change\n     * the modality away from keyboard.\n     * This avoids the situation where a user presses a key on an already focused\n     * element, and then clicks on a different element, focusing it with a\n     * pointing device, while we still think we're in keyboard modality.\n     * @param {Event} e\n     */\n    function onPointerDown(e) {\n      hadKeyboardEvent = false;\n    }\n\n    /**\n     * On `focus`, add the `focus-visible` class to the target if:\n     * - the target received focus as a result of keyboard navigation, or\n     * - the event target is an element that will likely require interaction\n     *   via the keyboard (e.g. a text box)\n     * @param {Event} e\n     */\n    function onFocus(e) {\n      // Prevent IE from focusing the document or HTML element.\n      if (!isValidFocusTarget(e.target)) {\n        return;\n      }\n\n      if (hadKeyboardEvent || focusTriggersKeyboardModality(e.target)) {\n        addFocusVisibleClass(e.target);\n      }\n    }\n\n    /**\n     * On `blur`, remove the `focus-visible` class from the target.\n     * @param {Event} e\n     */\n    function onBlur(e) {\n      if (!isValidFocusTarget(e.target)) {\n        return;\n      }\n\n      if (\n        e.target.classList.contains('focus-visible') ||\n        e.target.hasAttribute('data-focus-visible-added')\n      ) {\n        // To detect a tab/window switch, we look for a blur event followed\n        // rapidly by a visibility change.\n        // If we don't see a visibility change within 100ms, it's probably a\n        // regular focus change.\n        hadFocusVisibleRecently = true;\n        window.clearTimeout(hadFocusVisibleRecentlyTimeout);\n        hadFocusVisibleRecentlyTimeout = window.setTimeout(function() {\n          hadFocusVisibleRecently = false;\n        }, 100);\n        removeFocusVisibleClass(e.target);\n      }\n    }\n\n    /**\n     * If the user changes tabs, keep track of whether or not the previously\n     * focused element had .focus-visible.\n     * @param {Event} e\n     */\n    function onVisibilityChange(e) {\n      if (document.visibilityState === 'hidden') {\n        // If the tab becomes active again, the browser will handle calling focus\n        // on the element (Safari actually calls it twice).\n        // If this tab change caused a blur on an element with focus-visible,\n        // re-apply the class when the user switches back to the tab.\n        if (hadFocusVisibleRecently) {\n          hadKeyboardEvent = true;\n        }\n        addInitialPointerMoveListeners();\n      }\n    }\n\n    /**\n     * Add a group of listeners to detect usage of any pointing devices.\n     * These listeners will be added when the polyfill first loads, and anytime\n     * the window is blurred, so that they are active when the window regains\n     * focus.\n     */\n    function addInitialPointerMoveListeners() {\n      document.addEventListener('mousemove', onInitialPointerMove);\n      document.addEventListener('mousedown', onInitialPointerMove);\n      document.addEventListener('mouseup', onInitialPointerMove);\n      document.addEventListener('pointermove', onInitialPointerMove);\n      document.addEventListener('pointerdown', onInitialPointerMove);\n      document.addEventListener('pointerup', onInitialPointerMove);\n      document.addEventListener('touchmove', onInitialPointerMove);\n      document.addEventListener('touchstart', onInitialPointerMove);\n      document.addEventListener('touchend', onInitialPointerMove);\n    }\n\n    function removeInitialPointerMoveListeners() {\n      document.removeEventListener('mousemove', onInitialPointerMove);\n      document.removeEventListener('mousedown', onInitialPointerMove);\n      document.removeEventListener('mouseup', onInitialPointerMove);\n      document.removeEventListener('pointermove', onInitialPointerMove);\n      document.removeEventListener('pointerdown', onInitialPointerMove);\n      document.removeEventListener('pointerup', onInitialPointerMove);\n      document.removeEventListener('touchmove', onInitialPointerMove);\n      document.removeEventListener('touchstart', onInitialPointerMove);\n      document.removeEventListener('touchend', onInitialPointerMove);\n    }\n\n    /**\n     * When the polfyill first loads, assume the user is in keyboard modality.\n     * If any event is received from a pointing device (e.g. mouse, pointer,\n     * touch), turn off keyboard modality.\n     * This accounts for situations where focus enters the page from the URL bar.\n     * @param {Event} e\n     */\n    function onInitialPointerMove(e) {\n      // Work around a Safari quirk that fires a mousemove on <html> whenever the\n      // window blurs, even if you're tabbing out of the page. \u00AF\\_(\u30C4)_/\u00AF\n      if (e.target.nodeName && e.target.nodeName.toLowerCase() === 'html') {\n        return;\n      }\n\n      hadKeyboardEvent = false;\n      removeInitialPointerMoveListeners();\n    }\n\n    // For some kinds of state, we are interested in changes at the global scope\n    // only. For example, global pointer input, global key presses and global\n    // visibility change should affect the state at every scope:\n    document.addEventListener('keydown', onKeyDown, true);\n    document.addEventListener('mousedown', onPointerDown, true);\n    document.addEventListener('pointerdown', onPointerDown, true);\n    document.addEventListener('touchstart', onPointerDown, true);\n    document.addEventListener('visibilitychange', onVisibilityChange, true);\n\n    addInitialPointerMoveListeners();\n\n    // For focus and blur, we specifically care about state changes in the local\n    // scope. This is because focus / blur events that originate from within a\n    // shadow root are not re-dispatched from the host element if it was already\n    // the active element in its own scope:\n    scope.addEventListener('focus', onFocus, true);\n    scope.addEventListener('blur', onBlur, true);\n\n    // We detect that a node is a ShadowRoot by ensuring that it is a\n    // DocumentFragment and also has a host property. This check covers native\n    // implementation and polyfill implementation transparently. If we only cared\n    // about the native implementation, we could just check if the scope was\n    // an instance of a ShadowRoot.\n    if (scope.nodeType === Node.DOCUMENT_FRAGMENT_NODE && scope.host) {\n      // Since a ShadowRoot is a special kind of DocumentFragment, it does not\n      // have a root element to add a class to. So, we add this attribute to the\n      // host element instead:\n      scope.host.setAttribute('data-js-focus-visible', '');\n    } else if (scope.nodeType === Node.DOCUMENT_NODE) {\n      document.documentElement.classList.add('js-focus-visible');\n      document.documentElement.setAttribute('data-js-focus-visible', '');\n    }\n  }\n\n  // It is important to wrap all references to global window and document in\n  // these checks to support server-side rendering use cases\n  // @see https://github.com/WICG/focus-visible/issues/199\n  if (typeof window !== 'undefined' && typeof document !== 'undefined') {\n    // Make the polyfill helper globally available. This can be used as a signal\n    // to interested libraries that wish to coordinate with the polyfill for e.g.,\n    // applying the polyfill to a shadow root:\n    window.applyFocusVisiblePolyfill = applyFocusVisiblePolyfill;\n\n    // Notify interested libraries of the polyfill's presence, in case the\n    // polyfill was loaded lazily:\n    var event;\n\n    try {\n      event = new CustomEvent('focus-visible-polyfill-ready');\n    } catch (error) {\n      // IE11 does not support using CustomEvent as a constructor directly:\n      event = document.createEvent('CustomEvent');\n      event.initCustomEvent('focus-visible-polyfill-ready', false, false, {});\n    }\n\n    window.dispatchEvent(event);\n  }\n\n  if (typeof document !== 'undefined') {\n    // Apply the polyfill to the global document, so that no JavaScript\n    // coordination is required to use the polyfill in the top-level document:\n    applyFocusVisiblePolyfill(document);\n  }\n\n})));\n", "/*!\n * clipboard.js v2.0.11\n * https://clipboardjs.com/\n *\n * Licensed MIT \u00A9 Zeno Rocha\n */\n(function webpackUniversalModuleDefinition(root, factory) {\n\tif(typeof exports === 'object' && typeof module === 'object')\n\t\tmodule.exports = factory();\n\telse if(typeof define === 'function' && define.amd)\n\t\tdefine([], factory);\n\telse if(typeof exports === 'object')\n\t\texports[\"ClipboardJS\"] = factory();\n\telse\n\t\troot[\"ClipboardJS\"] = factory();\n})(this, function() {\nreturn /******/ (function() { // webpackBootstrap\n/******/ \tvar __webpack_modules__ = ({\n\n/***/ 686:\n/***/ (function(__unused_webpack_module, __webpack_exports__, __webpack_require__) {\n\n\"use strict\";\n\n// EXPORTS\n__webpack_require__.d(__webpack_exports__, {\n  \"default\": function() { return /* binding */ clipboard; }\n});\n\n// EXTERNAL MODULE: ./node_modules/tiny-emitter/index.js\nvar tiny_emitter = __webpack_require__(279);\nvar tiny_emitter_default = /*#__PURE__*/__webpack_require__.n(tiny_emitter);\n// EXTERNAL MODULE: ./node_modules/good-listener/src/listen.js\nvar listen = __webpack_require__(370);\nvar listen_default = /*#__PURE__*/__webpack_require__.n(listen);\n// EXTERNAL MODULE: ./node_modules/select/src/select.js\nvar src_select = __webpack_require__(817);\nvar select_default = /*#__PURE__*/__webpack_require__.n(src_select);\n;// CONCATENATED MODULE: ./src/common/command.js\n/**\n * Executes a given operation type.\n * @param {String} type\n * @return {Boolean}\n */\nfunction command(type) {\n  try {\n    return document.execCommand(type);\n  } catch (err) {\n    return false;\n  }\n}\n;// CONCATENATED MODULE: ./src/actions/cut.js\n\n\n/**\n * Cut action wrapper.\n * @param {String|HTMLElement} target\n * @return {String}\n */\n\nvar ClipboardActionCut = function ClipboardActionCut(target) {\n  var selectedText = select_default()(target);\n  command('cut');\n  return selectedText;\n};\n\n/* harmony default export */ var actions_cut = (ClipboardActionCut);\n;// CONCATENATED MODULE: ./src/common/create-fake-element.js\n/**\n * Creates a fake textarea element with a value.\n * @param {String} value\n * @return {HTMLElement}\n */\nfunction createFakeElement(value) {\n  var isRTL = document.documentElement.getAttribute('dir') === 'rtl';\n  var fakeElement = document.createElement('textarea'); // Prevent zooming on iOS\n\n  fakeElement.style.fontSize = '12pt'; // Reset box model\n\n  fakeElement.style.border = '0';\n  fakeElement.style.padding = '0';\n  fakeElement.style.margin = '0'; // Move element out of screen horizontally\n\n  fakeElement.style.position = 'absolute';\n  fakeElement.style[isRTL ? 'right' : 'left'] = '-9999px'; // Move element to the same position vertically\n\n  var yPosition = window.pageYOffset || document.documentElement.scrollTop;\n  fakeElement.style.top = \"\".concat(yPosition, \"px\");\n  fakeElement.setAttribute('readonly', '');\n  fakeElement.value = value;\n  return fakeElement;\n}\n;// CONCATENATED MODULE: ./src/actions/copy.js\n\n\n\n/**\n * Create fake copy action wrapper using a fake element.\n * @param {String} target\n * @param {Object} options\n * @return {String}\n */\n\nvar fakeCopyAction = function fakeCopyAction(value, options) {\n  var fakeElement = createFakeElement(value);\n  options.container.appendChild(fakeElement);\n  var selectedText = select_default()(fakeElement);\n  command('copy');\n  fakeElement.remove();\n  return selectedText;\n};\n/**\n * Copy action wrapper.\n * @param {String|HTMLElement} target\n * @param {Object} options\n * @return {String}\n */\n\n\nvar ClipboardActionCopy = function ClipboardActionCopy(target) {\n  var options = arguments.length > 1 && arguments[1] !== undefined ? arguments[1] : {\n    container: document.body\n  };\n  var selectedText = '';\n\n  if (typeof target === 'string') {\n    selectedText = fakeCopyAction(target, options);\n  } else if (target instanceof HTMLInputElement && !['text', 'search', 'url', 'tel', 'password'].includes(target === null || target === void 0 ? void 0 : target.type)) {\n    // If input type doesn't support `setSelectionRange`. Simulate it. https://developer.mozilla.org/en-US/docs/Web/API/HTMLInputElement/setSelectionRange\n    selectedText = fakeCopyAction(target.value, options);\n  } else {\n    selectedText = select_default()(target);\n    command('copy');\n  }\n\n  return selectedText;\n};\n\n/* harmony default export */ var actions_copy = (ClipboardActionCopy);\n;// CONCATENATED MODULE: ./src/actions/default.js\nfunction _typeof(obj) { \"@babel/helpers - typeof\"; if (typeof Symbol === \"function\" && typeof Symbol.iterator === \"symbol\") { _typeof = function _typeof(obj) { return typeof obj; }; } else { _typeof = function _typeof(obj) { return obj && typeof Symbol === \"function\" && obj.constructor === Symbol && obj !== Symbol.prototype ? \"symbol\" : typeof obj; }; } return _typeof(obj); }\n\n\n\n/**\n * Inner function which performs selection from either `text` or `target`\n * properties and then executes copy or cut operations.\n * @param {Object} options\n */\n\nvar ClipboardActionDefault = function ClipboardActionDefault() {\n  var options = arguments.length > 0 && arguments[0] !== undefined ? arguments[0] : {};\n  // Defines base properties passed from constructor.\n  var _options$action = options.action,\n      action = _options$action === void 0 ? 'copy' : _options$action,\n      container = options.container,\n      target = options.target,\n      text = options.text; // Sets the `action` to be performed which can be either 'copy' or 'cut'.\n\n  if (action !== 'copy' && action !== 'cut') {\n    throw new Error('Invalid \"action\" value, use either \"copy\" or \"cut\"');\n  } // Sets the `target` property using an element that will be have its content copied.\n\n\n  if (target !== undefined) {\n    if (target && _typeof(target) === 'object' && target.nodeType === 1) {\n      if (action === 'copy' && target.hasAttribute('disabled')) {\n        throw new Error('Invalid \"target\" attribute. Please use \"readonly\" instead of \"disabled\" attribute');\n      }\n\n      if (action === 'cut' && (target.hasAttribute('readonly') || target.hasAttribute('disabled'))) {\n        throw new Error('Invalid \"target\" attribute. You can\\'t cut text from elements with \"readonly\" or \"disabled\" attributes');\n      }\n    } else {\n      throw new Error('Invalid \"target\" value, use a valid Element');\n    }\n  } // Define selection strategy based on `text` property.\n\n\n  if (text) {\n    return actions_copy(text, {\n      container: container\n    });\n  } // Defines which selection strategy based on `target` property.\n\n\n  if (target) {\n    return action === 'cut' ? actions_cut(target) : actions_copy(target, {\n      container: container\n    });\n  }\n};\n\n/* harmony default export */ var actions_default = (ClipboardActionDefault);\n;// CONCATENATED MODULE: ./src/clipboard.js\nfunction clipboard_typeof(obj) { \"@babel/helpers - typeof\"; if (typeof Symbol === \"function\" && typeof Symbol.iterator === \"symbol\") { clipboard_typeof = function _typeof(obj) { return typeof obj; }; } else { clipboard_typeof = function _typeof(obj) { return obj && typeof Symbol === \"function\" && obj.constructor === Symbol && obj !== Symbol.prototype ? \"symbol\" : typeof obj; }; } return clipboard_typeof(obj); }\n\nfunction _classCallCheck(instance, Constructor) { if (!(instance instanceof Constructor)) { throw new TypeError(\"Cannot call a class as a function\"); } }\n\nfunction _defineProperties(target, props) { for (var i = 0; i < props.length; i++) { var descriptor = props[i]; descriptor.enumerable = descriptor.enumerable || false; descriptor.configurable = true; if (\"value\" in descriptor) descriptor.writable = true; Object.defineProperty(target, descriptor.key, descriptor); } }\n\nfunction _createClass(Constructor, protoProps, staticProps) { if (protoProps) _defineProperties(Constructor.prototype, protoProps); if (staticProps) _defineProperties(Constructor, staticProps); return Constructor; }\n\nfunction _inherits(subClass, superClass) { if (typeof superClass !== \"function\" && superClass !== null) { throw new TypeError(\"Super expression must either be null or a function\"); } subClass.prototype = Object.create(superClass && superClass.prototype, { constructor: { value: subClass, writable: true, configurable: true } }); if (superClass) _setPrototypeOf(subClass, superClass); }\n\nfunction _setPrototypeOf(o, p) { _setPrototypeOf = Object.setPrototypeOf || function _setPrototypeOf(o, p) { o.__proto__ = p; return o; }; return _setPrototypeOf(o, p); }\n\nfunction _createSuper(Derived) { var hasNativeReflectConstruct = _isNativeReflectConstruct(); return function _createSuperInternal() { var Super = _getPrototypeOf(Derived), result; if (hasNativeReflectConstruct) { var NewTarget = _getPrototypeOf(this).constructor; result = Reflect.construct(Super, arguments, NewTarget); } else { result = Super.apply(this, arguments); } return _possibleConstructorReturn(this, result); }; }\n\nfunction _possibleConstructorReturn(self, call) { if (call && (clipboard_typeof(call) === \"object\" || typeof call === \"function\")) { return call; } return _assertThisInitialized(self); }\n\nfunction _assertThisInitialized(self) { if (self === void 0) { throw new ReferenceError(\"this hasn't been initialised - super() hasn't been called\"); } return self; }\n\nfunction _isNativeReflectConstruct() { if (typeof Reflect === \"undefined\" || !Reflect.construct) return false; if (Reflect.construct.sham) return false; if (typeof Proxy === \"function\") return true; try { Date.prototype.toString.call(Reflect.construct(Date, [], function () {})); return true; } catch (e) { return false; } }\n\nfunction _getPrototypeOf(o) { _getPrototypeOf = Object.setPrototypeOf ? Object.getPrototypeOf : function _getPrototypeOf(o) { return o.__proto__ || Object.getPrototypeOf(o); }; return _getPrototypeOf(o); }\n\n\n\n\n\n\n/**\n * Helper function to retrieve attribute value.\n * @param {String} suffix\n * @param {Element} element\n */\n\nfunction getAttributeValue(suffix, element) {\n  var attribute = \"data-clipboard-\".concat(suffix);\n\n  if (!element.hasAttribute(attribute)) {\n    return;\n  }\n\n  return element.getAttribute(attribute);\n}\n/**\n * Base class which takes one or more elements, adds event listeners to them,\n * and instantiates a new `ClipboardAction` on each click.\n */\n\n\nvar Clipboard = /*#__PURE__*/function (_Emitter) {\n  _inherits(Clipboard, _Emitter);\n\n  var _super = _createSuper(Clipboard);\n\n  /**\n   * @param {String|HTMLElement|HTMLCollection|NodeList} trigger\n   * @param {Object} options\n   */\n  function Clipboard(trigger, options) {\n    var _this;\n\n    _classCallCheck(this, Clipboard);\n\n    _this = _super.call(this);\n\n    _this.resolveOptions(options);\n\n    _this.listenClick(trigger);\n\n    return _this;\n  }\n  /**\n   * Defines if attributes would be resolved using internal setter functions\n   * or custom functions that were passed in the constructor.\n   * @param {Object} options\n   */\n\n\n  _createClass(Clipboard, [{\n    key: \"resolveOptions\",\n    value: function resolveOptions() {\n      var options = arguments.length > 0 && arguments[0] !== undefined ? arguments[0] : {};\n      this.action = typeof options.action === 'function' ? options.action : this.defaultAction;\n      this.target = typeof options.target === 'function' ? options.target : this.defaultTarget;\n      this.text = typeof options.text === 'function' ? options.text : this.defaultText;\n      this.container = clipboard_typeof(options.container) === 'object' ? options.container : document.body;\n    }\n    /**\n     * Adds a click event listener to the passed trigger.\n     * @param {String|HTMLElement|HTMLCollection|NodeList} trigger\n     */\n\n  }, {\n    key: \"listenClick\",\n    value: function listenClick(trigger) {\n      var _this2 = this;\n\n      this.listener = listen_default()(trigger, 'click', function (e) {\n        return _this2.onClick(e);\n      });\n    }\n    /**\n     * Defines a new `ClipboardAction` on each click event.\n     * @param {Event} e\n     */\n\n  }, {\n    key: \"onClick\",\n    value: function onClick(e) {\n      var trigger = e.delegateTarget || e.currentTarget;\n      var action = this.action(trigger) || 'copy';\n      var text = actions_default({\n        action: action,\n        container: this.container,\n        target: this.target(trigger),\n        text: this.text(trigger)\n      }); // Fires an event based on the copy operation result.\n\n      this.emit(text ? 'success' : 'error', {\n        action: action,\n        text: text,\n        trigger: trigger,\n        clearSelection: function clearSelection() {\n          if (trigger) {\n            trigger.focus();\n          }\n\n          window.getSelection().removeAllRanges();\n        }\n      });\n    }\n    /**\n     * Default `action` lookup function.\n     * @param {Element} trigger\n     */\n\n  }, {\n    key: \"defaultAction\",\n    value: function defaultAction(trigger) {\n      return getAttributeValue('action', trigger);\n    }\n    /**\n     * Default `target` lookup function.\n     * @param {Element} trigger\n     */\n\n  }, {\n    key: \"defaultTarget\",\n    value: function defaultTarget(trigger) {\n      var selector = getAttributeValue('target', trigger);\n\n      if (selector) {\n        return document.querySelector(selector);\n      }\n    }\n    /**\n     * Allow fire programmatically a copy action\n     * @param {String|HTMLElement} target\n     * @param {Object} options\n     * @returns Text copied.\n     */\n\n  }, {\n    key: \"defaultText\",\n\n    /**\n     * Default `text` lookup function.\n     * @param {Element} trigger\n     */\n    value: function defaultText(trigger) {\n      return getAttributeValue('text', trigger);\n    }\n    /**\n     * Destroy lifecycle.\n     */\n\n  }, {\n    key: \"destroy\",\n    value: function destroy() {\n      this.listener.destroy();\n    }\n  }], [{\n    key: \"copy\",\n    value: function copy(target) {\n      var options = arguments.length > 1 && arguments[1] !== undefined ? arguments[1] : {\n        container: document.body\n      };\n      return actions_copy(target, options);\n    }\n    /**\n     * Allow fire programmatically a cut action\n     * @param {String|HTMLElement} target\n     * @returns Text cutted.\n     */\n\n  }, {\n    key: \"cut\",\n    value: function cut(target) {\n      return actions_cut(target);\n    }\n    /**\n     * Returns the support of the given action, or all actions if no action is\n     * given.\n     * @param {String} [action]\n     */\n\n  }, {\n    key: \"isSupported\",\n    value: function isSupported() {\n      var action = arguments.length > 0 && arguments[0] !== undefined ? arguments[0] : ['copy', 'cut'];\n      var actions = typeof action === 'string' ? [action] : action;\n      var support = !!document.queryCommandSupported;\n      actions.forEach(function (action) {\n        support = support && !!document.queryCommandSupported(action);\n      });\n      return support;\n    }\n  }]);\n\n  return Clipboard;\n}((tiny_emitter_default()));\n\n/* harmony default export */ var clipboard = (Clipboard);\n\n/***/ }),\n\n/***/ 828:\n/***/ (function(module) {\n\nvar DOCUMENT_NODE_TYPE = 9;\n\n/**\n * A polyfill for Element.matches()\n */\nif (typeof Element !== 'undefined' && !Element.prototype.matches) {\n    var proto = Element.prototype;\n\n    proto.matches = proto.matchesSelector ||\n                    proto.mozMatchesSelector ||\n                    proto.msMatchesSelector ||\n                    proto.oMatchesSelector ||\n                    proto.webkitMatchesSelector;\n}\n\n/**\n * Finds the closest parent that matches a selector.\n *\n * @param {Element} element\n * @param {String} selector\n * @return {Function}\n */\nfunction closest (element, selector) {\n    while (element && element.nodeType !== DOCUMENT_NODE_TYPE) {\n        if (typeof element.matches === 'function' &&\n            element.matches(selector)) {\n          return element;\n        }\n        element = element.parentNode;\n    }\n}\n\nmodule.exports = closest;\n\n\n/***/ }),\n\n/***/ 438:\n/***/ (function(module, __unused_webpack_exports, __webpack_require__) {\n\nvar closest = __webpack_require__(828);\n\n/**\n * Delegates event to a selector.\n *\n * @param {Element} element\n * @param {String} selector\n * @param {String} type\n * @param {Function} callback\n * @param {Boolean} useCapture\n * @return {Object}\n */\nfunction _delegate(element, selector, type, callback, useCapture) {\n    var listenerFn = listener.apply(this, arguments);\n\n    element.addEventListener(type, listenerFn, useCapture);\n\n    return {\n        destroy: function() {\n            element.removeEventListener(type, listenerFn, useCapture);\n        }\n    }\n}\n\n/**\n * Delegates event to a selector.\n *\n * @param {Element|String|Array} [elements]\n * @param {String} selector\n * @param {String} type\n * @param {Function} callback\n * @param {Boolean} useCapture\n * @return {Object}\n */\nfunction delegate(elements, selector, type, callback, useCapture) {\n    // Handle the regular Element usage\n    if (typeof elements.addEventListener === 'function') {\n        return _delegate.apply(null, arguments);\n    }\n\n    // Handle Element-less usage, it defaults to global delegation\n    if (typeof type === 'function') {\n        // Use `document` as the first parameter, then apply arguments\n        // This is a short way to .unshift `arguments` without running into deoptimizations\n        return _delegate.bind(null, document).apply(null, arguments);\n    }\n\n    // Handle Selector-based usage\n    if (typeof elements === 'string') {\n        elements = document.querySelectorAll(elements);\n    }\n\n    // Handle Array-like based usage\n    return Array.prototype.map.call(elements, function (element) {\n        return _delegate(element, selector, type, callback, useCapture);\n    });\n}\n\n/**\n * Finds closest match and invokes callback.\n *\n * @param {Element} element\n * @param {String} selector\n * @param {String} type\n * @param {Function} callback\n * @return {Function}\n */\nfunction listener(element, selector, type, callback) {\n    return function(e) {\n        e.delegateTarget = closest(e.target, selector);\n\n        if (e.delegateTarget) {\n            callback.call(element, e);\n        }\n    }\n}\n\nmodule.exports = delegate;\n\n\n/***/ }),\n\n/***/ 879:\n/***/ (function(__unused_webpack_module, exports) {\n\n/**\n * Check if argument is a HTML element.\n *\n * @param {Object} value\n * @return {Boolean}\n */\nexports.node = function(value) {\n    return value !== undefined\n        && value instanceof HTMLElement\n        && value.nodeType === 1;\n};\n\n/**\n * Check if argument is a list of HTML elements.\n *\n * @param {Object} value\n * @return {Boolean}\n */\nexports.nodeList = function(value) {\n    var type = Object.prototype.toString.call(value);\n\n    return value !== undefined\n        && (type === '[object NodeList]' || type === '[object HTMLCollection]')\n        && ('length' in value)\n        && (value.length === 0 || exports.node(value[0]));\n};\n\n/**\n * Check if argument is a string.\n *\n * @param {Object} value\n * @return {Boolean}\n */\nexports.string = function(value) {\n    return typeof value === 'string'\n        || value instanceof String;\n};\n\n/**\n * Check if argument is a function.\n *\n * @param {Object} value\n * @return {Boolean}\n */\nexports.fn = function(value) {\n    var type = Object.prototype.toString.call(value);\n\n    return type === '[object Function]';\n};\n\n\n/***/ }),\n\n/***/ 370:\n/***/ (function(module, __unused_webpack_exports, __webpack_require__) {\n\nvar is = __webpack_require__(879);\nvar delegate = __webpack_require__(438);\n\n/**\n * Validates all params and calls the right\n * listener function based on its target type.\n *\n * @param {String|HTMLElement|HTMLCollection|NodeList} target\n * @param {String} type\n * @param {Function} callback\n * @return {Object}\n */\nfunction listen(target, type, callback) {\n    if (!target && !type && !callback) {\n        throw new Error('Missing required arguments');\n    }\n\n    if (!is.string(type)) {\n        throw new TypeError('Second argument must be a String');\n    }\n\n    if (!is.fn(callback)) {\n        throw new TypeError('Third argument must be a Function');\n    }\n\n    if (is.node(target)) {\n        return listenNode(target, type, callback);\n    }\n    else if (is.nodeList(target)) {\n        return listenNodeList(target, type, callback);\n    }\n    else if (is.string(target)) {\n        return listenSelector(target, type, callback);\n    }\n    else {\n        throw new TypeError('First argument must be a String, HTMLElement, HTMLCollection, or NodeList');\n    }\n}\n\n/**\n * Adds an event listener to a HTML element\n * and returns a remove listener function.\n *\n * @param {HTMLElement} node\n * @param {String} type\n * @param {Function} callback\n * @return {Object}\n */\nfunction listenNode(node, type, callback) {\n    node.addEventListener(type, callback);\n\n    return {\n        destroy: function() {\n            node.removeEventListener(type, callback);\n        }\n    }\n}\n\n/**\n * Add an event listener to a list of HTML elements\n * and returns a remove listener function.\n *\n * @param {NodeList|HTMLCollection} nodeList\n * @param {String} type\n * @param {Function} callback\n * @return {Object}\n */\nfunction listenNodeList(nodeList, type, callback) {\n    Array.prototype.forEach.call(nodeList, function(node) {\n        node.addEventListener(type, callback);\n    });\n\n    return {\n        destroy: function() {\n            Array.prototype.forEach.call(nodeList, function(node) {\n                node.removeEventListener(type, callback);\n            });\n        }\n    }\n}\n\n/**\n * Add an event listener to a selector\n * and returns a remove listener function.\n *\n * @param {String} selector\n * @param {String} type\n * @param {Function} callback\n * @return {Object}\n */\nfunction listenSelector(selector, type, callback) {\n    return delegate(document.body, selector, type, callback);\n}\n\nmodule.exports = listen;\n\n\n/***/ }),\n\n/***/ 817:\n/***/ (function(module) {\n\nfunction select(element) {\n    var selectedText;\n\n    if (element.nodeName === 'SELECT') {\n        element.focus();\n\n        selectedText = element.value;\n    }\n    else if (element.nodeName === 'INPUT' || element.nodeName === 'TEXTAREA') {\n        var isReadOnly = element.hasAttribute('readonly');\n\n        if (!isReadOnly) {\n            element.setAttribute('readonly', '');\n        }\n\n        element.select();\n        element.setSelectionRange(0, element.value.length);\n\n        if (!isReadOnly) {\n            element.removeAttribute('readonly');\n        }\n\n        selectedText = element.value;\n    }\n    else {\n        if (element.hasAttribute('contenteditable')) {\n            element.focus();\n        }\n\n        var selection = window.getSelection();\n        var range = document.createRange();\n\n        range.selectNodeContents(element);\n        selection.removeAllRanges();\n        selection.addRange(range);\n\n        selectedText = selection.toString();\n    }\n\n    return selectedText;\n}\n\nmodule.exports = select;\n\n\n/***/ }),\n\n/***/ 279:\n/***/ (function(module) {\n\nfunction E () {\n  // Keep this empty so it's easier to inherit from\n  // (via https://github.com/lipsmack from https://github.com/scottcorgan/tiny-emitter/issues/3)\n}\n\nE.prototype = {\n  on: function (name, callback, ctx) {\n    var e = this.e || (this.e = {});\n\n    (e[name] || (e[name] = [])).push({\n      fn: callback,\n      ctx: ctx\n    });\n\n    return this;\n  },\n\n  once: function (name, callback, ctx) {\n    var self = this;\n    function listener () {\n      self.off(name, listener);\n      callback.apply(ctx, arguments);\n    };\n\n    listener._ = callback\n    return this.on(name, listener, ctx);\n  },\n\n  emit: function (name) {\n    var data = [].slice.call(arguments, 1);\n    var evtArr = ((this.e || (this.e = {}))[name] || []).slice();\n    var i = 0;\n    var len = evtArr.length;\n\n    for (i; i < len; i++) {\n      evtArr[i].fn.apply(evtArr[i].ctx, data);\n    }\n\n    return this;\n  },\n\n  off: function (name, callback) {\n    var e = this.e || (this.e = {});\n    var evts = e[name];\n    var liveEvents = [];\n\n    if (evts && callback) {\n      for (var i = 0, len = evts.length; i < len; i++) {\n        if (evts[i].fn !== callback && evts[i].fn._ !== callback)\n          liveEvents.push(evts[i]);\n      }\n    }\n\n    // Remove event from queue to prevent memory leak\n    // Suggested by https://github.com/lazd\n    // Ref: https://github.com/scottcorgan/tiny-emitter/commit/c6ebfaa9bc973b33d110a84a307742b7cf94c953#commitcomment-5024910\n\n    (liveEvents.length)\n      ? e[name] = liveEvents\n      : delete e[name];\n\n    return this;\n  }\n};\n\nmodule.exports = E;\nmodule.exports.TinyEmitter = E;\n\n\n/***/ })\n\n/******/ \t});\n/************************************************************************/\n/******/ \t// The module cache\n/******/ \tvar __webpack_module_cache__ = {};\n/******/ \t\n/******/ \t// The require function\n/******/ \tfunction __webpack_require__(moduleId) {\n/******/ \t\t// Check if module is in cache\n/******/ \t\tif(__webpack_module_cache__[moduleId]) {\n/******/ \t\t\treturn __webpack_module_cache__[moduleId].exports;\n/******/ \t\t}\n/******/ \t\t// Create a new module (and put it into the cache)\n/******/ \t\tvar module = __webpack_module_cache__[moduleId] = {\n/******/ \t\t\t// no module.id needed\n/******/ \t\t\t// no module.loaded needed\n/******/ \t\t\texports: {}\n/******/ \t\t};\n/******/ \t\n/******/ \t\t// Execute the module function\n/******/ \t\t__webpack_modules__[moduleId](module, module.exports, __webpack_require__);\n/******/ \t\n/******/ \t\t// Return the exports of the module\n/******/ \t\treturn module.exports;\n/******/ \t}\n/******/ \t\n/************************************************************************/\n/******/ \t/* webpack/runtime/compat get default export */\n/******/ \t!function() {\n/******/ \t\t// getDefaultExport function for compatibility with non-harmony modules\n/******/ \t\t__webpack_require__.n = function(module) {\n/******/ \t\t\tvar getter = module && module.__esModule ?\n/******/ \t\t\t\tfunction() { return module['default']; } :\n/******/ \t\t\t\tfunction() { return module; };\n/******/ \t\t\t__webpack_require__.d(getter, { a: getter });\n/******/ \t\t\treturn getter;\n/******/ \t\t};\n/******/ \t}();\n/******/ \t\n/******/ \t/* webpack/runtime/define property getters */\n/******/ \t!function() {\n/******/ \t\t// define getter functions for harmony exports\n/******/ \t\t__webpack_require__.d = function(exports, definition) {\n/******/ \t\t\tfor(var key in definition) {\n/******/ \t\t\t\tif(__webpack_require__.o(definition, key) && !__webpack_require__.o(exports, key)) {\n/******/ \t\t\t\t\tObject.defineProperty(exports, key, { enumerable: true, get: definition[key] });\n/******/ \t\t\t\t}\n/******/ \t\t\t}\n/******/ \t\t};\n/******/ \t}();\n/******/ \t\n/******/ \t/* webpack/runtime/hasOwnProperty shorthand */\n/******/ \t!function() {\n/******/ \t\t__webpack_require__.o = function(obj, prop) { return Object.prototype.hasOwnProperty.call(obj, prop); }\n/******/ \t}();\n/******/ \t\n/************************************************************************/\n/******/ \t// module exports must be returned from runtime so entry inlining is disabled\n/******/ \t// startup\n/******/ \t// Load entry module and return exports\n/******/ \treturn __webpack_require__(686);\n/******/ })()\n.default;\n});", "/*!\n * escape-html\n * Copyright(c) 2012-2013 TJ Holowaychuk\n * Copyright(c) 2015 Andreas Lubbe\n * Copyright(c) 2015 Tiancheng \"Timothy\" Gu\n * MIT Licensed\n */\n\n'use strict';\n\n/**\n * Module variables.\n * @private\n */\n\nvar matchHtmlRegExp = /[\"'&<>]/;\n\n/**\n * Module exports.\n * @public\n */\n\nmodule.exports = escapeHtml;\n\n/**\n * Escape special characters in the given string of html.\n *\n * @param  {string} string The string to escape for inserting into HTML\n * @return {string}\n * @public\n */\n\nfunction escapeHtml(string) {\n  var str = '' + string;\n  var match = matchHtmlRegExp.exec(str);\n\n  if (!match) {\n    return str;\n  }\n\n  var escape;\n  var html = '';\n  var index = 0;\n  var lastIndex = 0;\n\n  for (index = match.index; index < str.length; index++) {\n    switch (str.charCodeAt(index)) {\n      case 34: // \"\n        escape = '&quot;';\n        break;\n      case 38: // &\n        escape = '&amp;';\n        break;\n      case 39: // '\n        escape = '&#39;';\n        break;\n      case 60: // <\n        escape = '&lt;';\n        break;\n      case 62: // >\n        escape = '&gt;';\n        break;\n      default:\n        continue;\n    }\n\n    if (lastIndex !== index) {\n      html += str.substring(lastIndex, index);\n    }\n\n    lastIndex = index + 1;\n    html += escape;\n  }\n\n  return lastIndex !== index\n    ? html + str.substring(lastIndex, index)\n    : html;\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport \"focus-visible\"\n\nimport {\n  EMPTY,\n  NEVER,\n  Observable,\n  Subject,\n  defer,\n  delay,\n  filter,\n  map,\n  merge,\n  mergeWith,\n  shareReplay,\n  switchMap\n} from \"rxjs\"\n\nimport { configuration, feature } from \"./_\"\nimport {\n  at,\n  getActiveElement,\n  getOptionalElement,\n  requestJSON,\n  setLocation,\n  setToggle,\n  watchDocument,\n  watchKeyboard,\n  watchLocation,\n  watchLocationTarget,\n  watchMedia,\n  watchPrint,\n  watchScript,\n  watchViewport\n} from \"./browser\"\nimport {\n  getComponentElement,\n  getComponentElements,\n  mountAnnounce,\n  mountBackToTop,\n  mountConsent,\n  mountContent,\n  mountDialog,\n  mountHeader,\n  mountHeaderTitle,\n  mountPalette,\n  mountProgress,\n  mountSearch,\n  mountSearchHiglight,\n  mountSidebar,\n  mountSource,\n  mountTableOfContents,\n  mountTabs,\n  watchHeader,\n  watchMain\n} from \"./components\"\nimport {\n  SearchIndex,\n  setupClipboardJS,\n  setupInstantNavigation,\n  setupVersionSelector\n} from \"./integrations\"\nimport {\n  patchIndeterminate,\n  patchScrollfix,\n  patchScrolllock\n} from \"./patches\"\nimport \"./polyfills\"\n\n/* ----------------------------------------------------------------------------\n * Functions - @todo refactor\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch search index\n *\n * @returns Search index observable\n */\nfunction fetchSearchIndex(): Observable<SearchIndex> {\n  if (location.protocol === \"file:\") {\n    return watchScript(\n      `${new URL(\"search/search_index.js\", config.base)}`\n    )\n      .pipe(\n        // @ts-ignore - @todo fix typings\n        map(() => __index),\n        shareReplay(1)\n      )\n  } else {\n    return requestJSON<SearchIndex>(\n      new URL(\"search/search_index.json\", config.base)\n    )\n  }\n}\n\n/* ----------------------------------------------------------------------------\n * Application\n * ------------------------------------------------------------------------- */\n\n/* Yay, JavaScript is available */\ndocument.documentElement.classList.remove(\"no-js\")\ndocument.documentElement.classList.add(\"js\")\n\n/* Set up navigation observables and subjects */\nconst document$ = watchDocument()\nconst location$ = watchLocation()\nconst target$   = watchLocationTarget(location$)\nconst keyboard$ = watchKeyboard()\n\n/* Set up media observables */\nconst viewport$ = watchViewport()\nconst tablet$   = watchMedia(\"(min-width: 960px)\")\nconst screen$   = watchMedia(\"(min-width: 1220px)\")\nconst print$    = watchPrint()\n\n/* Retrieve search index, if search is enabled */\nconst config = configuration()\nconst index$ = document.forms.namedItem(\"search\")\n  ? fetchSearchIndex()\n  : NEVER\n\n/* Set up Clipboard.js integration */\nconst alert$ = new Subject<string>()\nsetupClipboardJS({ alert$ })\n\n/* Set up progress indicator */\nconst progress$ = new Subject<number>()\n\n/* Set up instant navigation, if enabled */\nif (feature(\"navigation.instant\"))\n  setupInstantNavigation({ location$, viewport$, progress$ })\n    .subscribe(document$)\n\n/* Set up version selector */\nif (config.version?.provider === \"mike\")\n  setupVersionSelector({ document$ })\n\n/* Always close drawer and search on navigation */\nmerge(location$, target$)\n  .pipe(\n    delay(125)\n  )\n    .subscribe(() => {\n      setToggle(\"drawer\", false)\n      setToggle(\"search\", false)\n    })\n\n/* Set up global keyboard handlers */\nkeyboard$\n  .pipe(\n    filter(({ mode }) => mode === \"global\")\n  )\n    .subscribe(key => {\n      switch (key.type) {\n\n        /* Go to previous page */\n        case \"p\":\n        case \",\":\n          const prev = getOptionalElement<HTMLLinkElement>(\"link[rel=prev]\")\n          if (typeof prev !== \"undefined\")\n            setLocation(prev)\n          break\n\n        /* Go to next page */\n        case \"n\":\n        case \".\":\n          const next = getOptionalElement<HTMLLinkElement>(\"link[rel=next]\")\n          if (typeof next !== \"undefined\")\n            setLocation(next)\n          break\n\n        /* Expand navigation, see https://bit.ly/3ZjG5io */\n        case \"Enter\":\n          const active = getActiveElement()\n          if (active instanceof HTMLLabelElement)\n            active.click()\n      }\n    })\n\n/* Set up patches */\npatchIndeterminate({ document$, tablet$ })\npatchScrollfix({ document$ })\npatchScrolllock({ viewport$, tablet$ })\n\n/* Set up header and main area observable */\nconst header$ = watchHeader(getComponentElement(\"header\"), { viewport$ })\nconst main$ = document$\n  .pipe(\n    map(() => getComponentElement(\"main\")),\n    switchMap(el => watchMain(el, { viewport$, header$ })),\n    shareReplay(1)\n  )\n\n/* Set up control component observables */\nconst control$ = merge(\n\n  /* Consent */\n  ...getComponentElements(\"consent\")\n    .map(el => mountConsent(el, { target$ })),\n\n  /* Dialog */\n  ...getComponentElements(\"dialog\")\n    .map(el => mountDialog(el, { alert$ })),\n\n  /* Header */\n  ...getComponentElements(\"header\")\n    .map(el => mountHeader(el, { viewport$, header$, main$ })),\n\n  /* Color palette */\n  ...getComponentElements(\"palette\")\n    .map(el => mountPalette(el)),\n\n  /* Progress bar */\n  ...getComponentElements(\"progress\")\n    .map(el => mountProgress(el, { progress$ })),\n\n  /* Search */\n  ...getComponentElements(\"search\")\n    .map(el => mountSearch(el, { index$, keyboard$ })),\n\n  /* Repository information */\n  ...getComponentElements(\"source\")\n    .map(el => mountSource(el))\n)\n\n/* Set up content component observables */\nconst content$ = defer(() => merge(\n\n  /* Announcement bar */\n  ...getComponentElements(\"announce\")\n    .map(el => mountAnnounce(el)),\n\n  /* Content */\n  ...getComponentElements(\"content\")\n    .map(el => mountContent(el, { viewport$, target$, print$ })),\n\n  /* Search highlighting */\n  ...getComponentElements(\"content\")\n    .map(el => feature(\"search.highlight\")\n      ? mountSearchHiglight(el, { index$, location$ })\n      : EMPTY\n    ),\n\n  /* Header title */\n  ...getComponentElements(\"header-title\")\n    .map(el => mountHeaderTitle(el, { viewport$, header$ })),\n\n  /* Sidebar */\n  ...getComponentElements(\"sidebar\")\n    .map(el => el.getAttribute(\"data-md-type\") === \"navigation\"\n      ? at(screen$, () => mountSidebar(el, { viewport$, header$, main$ }))\n      : at(tablet$, () => mountSidebar(el, { viewport$, header$, main$ }))\n    ),\n\n  /* Navigation tabs */\n  ...getComponentElements(\"tabs\")\n    .map(el => mountTabs(el, { viewport$, header$ })),\n\n  /* Table of contents */\n  ...getComponentElements(\"toc\")\n    .map(el => mountTableOfContents(el, {\n      viewport$, header$, main$, target$\n    })),\n\n  /* Back-to-top button */\n  ...getComponentElements(\"top\")\n    .map(el => mountBackToTop(el, { viewport$, header$, main$, target$ }))\n))\n\n/* Set up component observables */\nconst component$ = document$\n  .pipe(\n    switchMap(() => content$),\n    mergeWith(control$),\n    shareReplay(1)\n  )\n\n/* Subscribe to all components */\ncomponent$.subscribe()\n\n/* ----------------------------------------------------------------------------\n * Exports\n * ------------------------------------------------------------------------- */\n\nwindow.document$  = document$          /* Document observable */\nwindow.location$  = location$          /* Location subject */\nwindow.target$    = target$            /* Location target observable */\nwindow.keyboard$  = keyboard$          /* Keyboard observable */\nwindow.viewport$  = viewport$          /* Viewport observable */\nwindow.tablet$    = tablet$            /* Media tablet observable */\nwindow.screen$    = screen$            /* Media screen observable */\nwindow.print$     = print$             /* Media print observable */\nwindow.alert$     = alert$             /* Alert subject */\nwindow.progress$  = progress$          /* Progress indicator subject */\nwindow.component$ = component$         /* Component observable */\n", "/*! *****************************************************************************\r\nCopyright (c) Microsoft Corporation.\r\n\r\nPermission to use, copy, modify, and/or distribute this software for any\r\npurpose with or without fee is hereby granted.\r\n\r\nTHE SOFTWARE IS PROVIDED \"AS IS\" AND THE AUTHOR DISCLAIMS ALL WARRANTIES WITH\r\nREGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY\r\nAND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT,\r\nINDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM\r\nLOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR\r\nOTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR\r\nPERFORMANCE OF THIS SOFTWARE.\r\n***************************************************************************** */\r\n/* global Reflect, Promise */\r\n\r\nvar extendStatics = function(d, b) {\r\n    extendStatics = Object.setPrototypeOf ||\r\n        ({ __proto__: [] } instanceof Array && function (d, b) { d.__proto__ = b; }) ||\r\n        function (d, b) { for (var p in b) if (Object.prototype.hasOwnProperty.call(b, p)) d[p] = b[p]; };\r\n    return extendStatics(d, b);\r\n};\r\n\r\nexport function __extends(d, b) {\r\n    if (typeof b !== \"function\" && b !== null)\r\n        throw new TypeError(\"Class extends value \" + String(b) + \" is not a constructor or null\");\r\n    extendStatics(d, b);\r\n    function __() { this.constructor = d; }\r\n    d.prototype = b === null ? Object.create(b) : (__.prototype = b.prototype, new __());\r\n}\r\n\r\nexport var __assign = function() {\r\n    __assign = Object.assign || function __assign(t) {\r\n        for (var s, i = 1, n = arguments.length; i < n; i++) {\r\n            s = arguments[i];\r\n            for (var p in s) if (Object.prototype.hasOwnProperty.call(s, p)) t[p] = s[p];\r\n        }\r\n        return t;\r\n    }\r\n    return __assign.apply(this, arguments);\r\n}\r\n\r\nexport function __rest(s, e) {\r\n    var t = {};\r\n    for (var p in s) if (Object.prototype.hasOwnProperty.call(s, p) && e.indexOf(p) < 0)\r\n        t[p] = s[p];\r\n    if (s != null && typeof Object.getOwnPropertySymbols === \"function\")\r\n        for (var i = 0, p = Object.getOwnPropertySymbols(s); i < p.length; i++) {\r\n            if (e.indexOf(p[i]) < 0 && Object.prototype.propertyIsEnumerable.call(s, p[i]))\r\n                t[p[i]] = s[p[i]];\r\n        }\r\n    return t;\r\n}\r\n\r\nexport function __decorate(decorators, target, key, desc) {\r\n    var c = arguments.length, r = c < 3 ? target : desc === null ? desc = Object.getOwnPropertyDescriptor(target, key) : desc, d;\r\n    if (typeof Reflect === \"object\" && typeof Reflect.decorate === \"function\") r = Reflect.decorate(decorators, target, key, desc);\r\n    else for (var i = decorators.length - 1; i >= 0; i--) if (d = decorators[i]) r = (c < 3 ? d(r) : c > 3 ? d(target, key, r) : d(target, key)) || r;\r\n    return c > 3 && r && Object.defineProperty(target, key, r), r;\r\n}\r\n\r\nexport function __param(paramIndex, decorator) {\r\n    return function (target, key) { decorator(target, key, paramIndex); }\r\n}\r\n\r\nexport function __metadata(metadataKey, metadataValue) {\r\n    if (typeof Reflect === \"object\" && typeof Reflect.metadata === \"function\") return Reflect.metadata(metadataKey, metadataValue);\r\n}\r\n\r\nexport function __awaiter(thisArg, _arguments, P, generator) {\r\n    function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }\r\n    return new (P || (P = Promise))(function (resolve, reject) {\r\n        function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }\r\n        function rejected(value) { try { step(generator[\"throw\"](value)); } catch (e) { reject(e); } }\r\n        function step(result) { result.done ? resolve(result.value) : adopt(result.value).then(fulfilled, rejected); }\r\n        step((generator = generator.apply(thisArg, _arguments || [])).next());\r\n    });\r\n}\r\n\r\nexport function __generator(thisArg, body) {\r\n    var _ = { label: 0, sent: function() { if (t[0] & 1) throw t[1]; return t[1]; }, trys: [], ops: [] }, f, y, t, g;\r\n    return g = { next: verb(0), \"throw\": verb(1), \"return\": verb(2) }, typeof Symbol === \"function\" && (g[Symbol.iterator] = function() { return this; }), g;\r\n    function verb(n) { return function (v) { return step([n, v]); }; }\r\n    function step(op) {\r\n        if (f) throw new TypeError(\"Generator is already executing.\");\r\n        while (_) try {\r\n            if (f = 1, y && (t = op[0] & 2 ? y[\"return\"] : op[0] ? y[\"throw\"] || ((t = y[\"return\"]) && t.call(y), 0) : y.next) && !(t = t.call(y, op[1])).done) return t;\r\n            if (y = 0, t) op = [op[0] & 2, t.value];\r\n            switch (op[0]) {\r\n                case 0: case 1: t = op; break;\r\n                case 4: _.label++; return { value: op[1], done: false };\r\n                case 5: _.label++; y = op[1]; op = [0]; continue;\r\n                case 7: op = _.ops.pop(); _.trys.pop(); continue;\r\n                default:\r\n                    if (!(t = _.trys, t = t.length > 0 && t[t.length - 1]) && (op[0] === 6 || op[0] === 2)) { _ = 0; continue; }\r\n                    if (op[0] === 3 && (!t || (op[1] > t[0] && op[1] < t[3]))) { _.label = op[1]; break; }\r\n                    if (op[0] === 6 && _.label < t[1]) { _.label = t[1]; t = op; break; }\r\n                    if (t && _.label < t[2]) { _.label = t[2]; _.ops.push(op); break; }\r\n                    if (t[2]) _.ops.pop();\r\n                    _.trys.pop(); continue;\r\n            }\r\n            op = body.call(thisArg, _);\r\n        } catch (e) { op = [6, e]; y = 0; } finally { f = t = 0; }\r\n        if (op[0] & 5) throw op[1]; return { value: op[0] ? op[1] : void 0, done: true };\r\n    }\r\n}\r\n\r\nexport var __createBinding = Object.create ? (function(o, m, k, k2) {\r\n    if (k2 === undefined) k2 = k;\r\n    Object.defineProperty(o, k2, { enumerable: true, get: function() { return m[k]; } });\r\n}) : (function(o, m, k, k2) {\r\n    if (k2 === undefined) k2 = k;\r\n    o[k2] = m[k];\r\n});\r\n\r\nexport function __exportStar(m, o) {\r\n    for (var p in m) if (p !== \"default\" && !Object.prototype.hasOwnProperty.call(o, p)) __createBinding(o, m, p);\r\n}\r\n\r\nexport function __values(o) {\r\n    var s = typeof Symbol === \"function\" && Symbol.iterator, m = s && o[s], i = 0;\r\n    if (m) return m.call(o);\r\n    if (o && typeof o.length === \"number\") return {\r\n        next: function () {\r\n            if (o && i >= o.length) o = void 0;\r\n            return { value: o && o[i++], done: !o };\r\n        }\r\n    };\r\n    throw new TypeError(s ? \"Object is not iterable.\" : \"Symbol.iterator is not defined.\");\r\n}\r\n\r\nexport function __read(o, n) {\r\n    var m = typeof Symbol === \"function\" && o[Symbol.iterator];\r\n    if (!m) return o;\r\n    var i = m.call(o), r, ar = [], e;\r\n    try {\r\n        while ((n === void 0 || n-- > 0) && !(r = i.next()).done) ar.push(r.value);\r\n    }\r\n    catch (error) { e = { error: error }; }\r\n    finally {\r\n        try {\r\n            if (r && !r.done && (m = i[\"return\"])) m.call(i);\r\n        }\r\n        finally { if (e) throw e.error; }\r\n    }\r\n    return ar;\r\n}\r\n\r\n/** @deprecated */\r\nexport function __spread() {\r\n    for (var ar = [], i = 0; i < arguments.length; i++)\r\n        ar = ar.concat(__read(arguments[i]));\r\n    return ar;\r\n}\r\n\r\n/** @deprecated */\r\nexport function __spreadArrays() {\r\n    for (var s = 0, i = 0, il = arguments.length; i < il; i++) s += arguments[i].length;\r\n    for (var r = Array(s), k = 0, i = 0; i < il; i++)\r\n        for (var a = arguments[i], j = 0, jl = a.length; j < jl; j++, k++)\r\n            r[k] = a[j];\r\n    return r;\r\n}\r\n\r\nexport function __spreadArray(to, from, pack) {\r\n    if (pack || arguments.length === 2) for (var i = 0, l = from.length, ar; i < l; i++) {\r\n        if (ar || !(i in from)) {\r\n            if (!ar) ar = Array.prototype.slice.call(from, 0, i);\r\n            ar[i] = from[i];\r\n        }\r\n    }\r\n    return to.concat(ar || Array.prototype.slice.call(from));\r\n}\r\n\r\nexport function __await(v) {\r\n    return this instanceof __await ? (this.v = v, this) : new __await(v);\r\n}\r\n\r\nexport function __asyncGenerator(thisArg, _arguments, generator) {\r\n    if (!Symbol.asyncIterator) throw new TypeError(\"Symbol.asyncIterator is not defined.\");\r\n    var g = generator.apply(thisArg, _arguments || []), i, q = [];\r\n    return i = {}, verb(\"next\"), verb(\"throw\"), verb(\"return\"), i[Symbol.asyncIterator] = function () { return this; }, i;\r\n    function verb(n) { if (g[n]) i[n] = function (v) { return new Promise(function (a, b) { q.push([n, v, a, b]) > 1 || resume(n, v); }); }; }\r\n    function resume(n, v) { try { step(g[n](v)); } catch (e) { settle(q[0][3], e); } }\r\n    function step(r) { r.value instanceof __await ? Promise.resolve(r.value.v).then(fulfill, reject) : settle(q[0][2], r); }\r\n    function fulfill(value) { resume(\"next\", value); }\r\n    function reject(value) { resume(\"throw\", value); }\r\n    function settle(f, v) { if (f(v), q.shift(), q.length) resume(q[0][0], q[0][1]); }\r\n}\r\n\r\nexport function __asyncDelegator(o) {\r\n    var i, p;\r\n    return i = {}, verb(\"next\"), verb(\"throw\", function (e) { throw e; }), verb(\"return\"), i[Symbol.iterator] = function () { return this; }, i;\r\n    function verb(n, f) { i[n] = o[n] ? function (v) { return (p = !p) ? { value: __await(o[n](v)), done: n === \"return\" } : f ? f(v) : v; } : f; }\r\n}\r\n\r\nexport function __asyncValues(o) {\r\n    if (!Symbol.asyncIterator) throw new TypeError(\"Symbol.asyncIterator is not defined.\");\r\n    var m = o[Symbol.asyncIterator], i;\r\n    return m ? m.call(o) : (o = typeof __values === \"function\" ? __values(o) : o[Symbol.iterator](), i = {}, verb(\"next\"), verb(\"throw\"), verb(\"return\"), i[Symbol.asyncIterator] = function () { return this; }, i);\r\n    function verb(n) { i[n] = o[n] && function (v) { return new Promise(function (resolve, reject) { v = o[n](v), settle(resolve, reject, v.done, v.value); }); }; }\r\n    function settle(resolve, reject, d, v) { Promise.resolve(v).then(function(v) { resolve({ value: v, done: d }); }, reject); }\r\n}\r\n\r\nexport function __makeTemplateObject(cooked, raw) {\r\n    if (Object.defineProperty) { Object.defineProperty(cooked, \"raw\", { value: raw }); } else { cooked.raw = raw; }\r\n    return cooked;\r\n};\r\n\r\nvar __setModuleDefault = Object.create ? (function(o, v) {\r\n    Object.defineProperty(o, \"default\", { enumerable: true, value: v });\r\n}) : function(o, v) {\r\n    o[\"default\"] = v;\r\n};\r\n\r\nexport function __importStar(mod) {\r\n    if (mod && mod.__esModule) return mod;\r\n    var result = {};\r\n    if (mod != null) for (var k in mod) if (k !== \"default\" && Object.prototype.hasOwnProperty.call(mod, k)) __createBinding(result, mod, k);\r\n    __setModuleDefault(result, mod);\r\n    return result;\r\n}\r\n\r\nexport function __importDefault(mod) {\r\n    return (mod && mod.__esModule) ? mod : { default: mod };\r\n}\r\n\r\nexport function __classPrivateFieldGet(receiver, state, kind, f) {\r\n    if (kind === \"a\" && !f) throw new TypeError(\"Private accessor was defined without a getter\");\r\n    if (typeof state === \"function\" ? receiver !== state || !f : !state.has(receiver)) throw new TypeError(\"Cannot read private member from an object whose class did not declare it\");\r\n    return kind === \"m\" ? f : kind === \"a\" ? f.call(receiver) : f ? f.value : state.get(receiver);\r\n}\r\n\r\nexport function __classPrivateFieldSet(receiver, state, value, kind, f) {\r\n    if (kind === \"m\") throw new TypeError(\"Private method is not writable\");\r\n    if (kind === \"a\" && !f) throw new TypeError(\"Private accessor was defined without a setter\");\r\n    if (typeof state === \"function\" ? receiver !== state || !f : !state.has(receiver)) throw new TypeError(\"Cannot write private member to an object whose class did not declare it\");\r\n    return (kind === \"a\" ? f.call(receiver, value) : f ? f.value = value : state.set(receiver, value)), value;\r\n}\r\n", "/**\n * Returns true if the object is a function.\n * @param value The value to check\n */\nexport function isFunction(value: any): value is (...args: any[]) => any {\n  return typeof value === 'function';\n}\n", "/**\n * Used to create Error subclasses until the community moves away from ES5.\n *\n * This is because compiling from TypeScript down to ES5 has issues with subclassing Errors\n * as well as other built-in types: https://github.com/Microsoft/TypeScript/issues/12123\n *\n * @param createImpl A factory function to create the actual constructor implementation. The returned\n * function should be a named function that calls `_super` internally.\n */\nexport function createErrorClass<T>(createImpl: (_super: any) => any): T {\n  const _super = (instance: any) => {\n    Error.call(instance);\n    instance.stack = new Error().stack;\n  };\n\n  const ctorFunc = createImpl(_super);\n  ctorFunc.prototype = Object.create(Error.prototype);\n  ctorFunc.prototype.constructor = ctorFunc;\n  return ctorFunc;\n}\n", "import { createErrorClass } from './createErrorClass';\n\nexport interface UnsubscriptionError extends Error {\n  readonly errors: any[];\n}\n\nexport interface UnsubscriptionErrorCtor {\n  /**\n   * @deprecated Internal implementation detail. Do not construct error instances.\n   * Cannot be tagged as internal: https://github.com/ReactiveX/rxjs/issues/6269\n   */\n  new (errors: any[]): UnsubscriptionError;\n}\n\n/**\n * An error thrown when one or more errors have occurred during the\n * `unsubscribe` of a {@link Subscription}.\n */\nexport const UnsubscriptionError: UnsubscriptionErrorCtor = createErrorClass(\n  (_super) =>\n    function UnsubscriptionErrorImpl(this: any, errors: (Error | string)[]) {\n      _super(this);\n      this.message = errors\n        ? `${errors.length} errors occurred during unsubscription:\n${errors.map((err, i) => `${i + 1}) ${err.toString()}`).join('\\n  ')}`\n        : '';\n      this.name = 'UnsubscriptionError';\n      this.errors = errors;\n    }\n);\n", "/**\n * Removes an item from an array, mutating it.\n * @param arr The array to remove the item from\n * @param item The item to remove\n */\nexport function arrRemove<T>(arr: T[] | undefined | null, item: T) {\n  if (arr) {\n    const index = arr.indexOf(item);\n    0 <= index && arr.splice(index, 1);\n  }\n}\n", "import { isFunction } from './util/isFunction';\nimport { UnsubscriptionError } from './util/UnsubscriptionError';\nimport { SubscriptionLike, TeardownLogic, Unsubscribable } from './types';\nimport { arrRemove } from './util/arrRemove';\n\n/**\n * Represents a disposable resource, such as the execution of an Observable. A\n * Subscription has one important method, `unsubscribe`, that takes no argument\n * and just disposes the resource held by the subscription.\n *\n * Additionally, subscriptions may be grouped together through the `add()`\n * method, which will attach a child Subscription to the current Subscription.\n * When a Subscription is unsubscribed, all its children (and its grandchildren)\n * will be unsubscribed as well.\n *\n * @class Subscription\n */\nexport class Subscription implements SubscriptionLike {\n  /** @nocollapse */\n  public static EMPTY = (() => {\n    const empty = new Subscription();\n    empty.closed = true;\n    return empty;\n  })();\n\n  /**\n   * A flag to indicate whether this Subscription has already been unsubscribed.\n   */\n  public closed = false;\n\n  private _parentage: Subscription[] | Subscription | null = null;\n\n  /**\n   * The list of registered finalizers to execute upon unsubscription. Adding and removing from this\n   * list occurs in the {@link #add} and {@link #remove} methods.\n   */\n  private _finalizers: Exclude<TeardownLogic, void>[] | null = null;\n\n  /**\n   * @param initialTeardown A function executed first as part of the finalization\n   * process that is kicked off when {@link #unsubscribe} is called.\n   */\n  constructor(private initialTeardown?: () => void) {}\n\n  /**\n   * Disposes the resources held by the subscription. May, for instance, cancel\n   * an ongoing Observable execution or cancel any other type of work that\n   * started when the Subscription was created.\n   * @return {void}\n   */\n  unsubscribe(): void {\n    let errors: any[] | undefined;\n\n    if (!this.closed) {\n      this.closed = true;\n\n      // Remove this from it's parents.\n      const { _parentage } = this;\n      if (_parentage) {\n        this._parentage = null;\n        if (Array.isArray(_parentage)) {\n          for (const parent of _parentage) {\n            parent.remove(this);\n          }\n        } else {\n          _parentage.remove(this);\n        }\n      }\n\n      const { initialTeardown: initialFinalizer } = this;\n      if (isFunction(initialFinalizer)) {\n        try {\n          initialFinalizer();\n        } catch (e) {\n          errors = e instanceof UnsubscriptionError ? e.errors : [e];\n        }\n      }\n\n      const { _finalizers } = this;\n      if (_finalizers) {\n        this._finalizers = null;\n        for (const finalizer of _finalizers) {\n          try {\n            execFinalizer(finalizer);\n          } catch (err) {\n            errors = errors ?? [];\n            if (err instanceof UnsubscriptionError) {\n              errors = [...errors, ...err.errors];\n            } else {\n              errors.push(err);\n            }\n          }\n        }\n      }\n\n      if (errors) {\n        throw new UnsubscriptionError(errors);\n      }\n    }\n  }\n\n  /**\n   * Adds a finalizer to this subscription, so that finalization will be unsubscribed/called\n   * when this subscription is unsubscribed. If this subscription is already {@link #closed},\n   * because it has already been unsubscribed, then whatever finalizer is passed to it\n   * will automatically be executed (unless the finalizer itself is also a closed subscription).\n   *\n   * Closed Subscriptions cannot be added as finalizers to any subscription. Adding a closed\n   * subscription to a any subscription will result in no operation. (A noop).\n   *\n   * Adding a subscription to itself, or adding `null` or `undefined` will not perform any\n   * operation at all. (A noop).\n   *\n   * `Subscription` instances that are added to this instance will automatically remove themselves\n   * if they are unsubscribed. Functions and {@link Unsubscribable} objects that you wish to remove\n   * will need to be removed manually with {@link #remove}\n   *\n   * @param teardown The finalization logic to add to this subscription.\n   */\n  add(teardown: TeardownLogic): void {\n    // Only add the finalizer if it's not undefined\n    // and don't add a subscription to itself.\n    if (teardown && teardown !== this) {\n      if (this.closed) {\n        // If this subscription is already closed,\n        // execute whatever finalizer is handed to it automatically.\n        execFinalizer(teardown);\n      } else {\n        if (teardown instanceof Subscription) {\n          // We don't add closed subscriptions, and we don't add the same subscription\n          // twice. Subscription unsubscribe is idempotent.\n          if (teardown.closed || teardown._hasParent(this)) {\n            return;\n          }\n          teardown._addParent(this);\n        }\n        (this._finalizers = this._finalizers ?? []).push(teardown);\n      }\n    }\n  }\n\n  /**\n   * Checks to see if a this subscription already has a particular parent.\n   * This will signal that this subscription has already been added to the parent in question.\n   * @param parent the parent to check for\n   */\n  private _hasParent(parent: Subscription) {\n    const { _parentage } = this;\n    return _parentage === parent || (Array.isArray(_parentage) && _parentage.includes(parent));\n  }\n\n  /**\n   * Adds a parent to this subscription so it can be removed from the parent if it\n   * unsubscribes on it's own.\n   *\n   * NOTE: THIS ASSUMES THAT {@link _hasParent} HAS ALREADY BEEN CHECKED.\n   * @param parent The parent subscription to add\n   */\n  private _addParent(parent: Subscription) {\n    const { _parentage } = this;\n    this._parentage = Array.isArray(_parentage) ? (_parentage.push(parent), _parentage) : _parentage ? [_parentage, parent] : parent;\n  }\n\n  /**\n   * Called on a child when it is removed via {@link #remove}.\n   * @param parent The parent to remove\n   */\n  private _removeParent(parent: Subscription) {\n    const { _parentage } = this;\n    if (_parentage === parent) {\n      this._parentage = null;\n    } else if (Array.isArray(_parentage)) {\n      arrRemove(_parentage, parent);\n    }\n  }\n\n  /**\n   * Removes a finalizer from this subscription that was previously added with the {@link #add} method.\n   *\n   * Note that `Subscription` instances, when unsubscribed, will automatically remove themselves\n   * from every other `Subscription` they have been added to. This means that using the `remove` method\n   * is not a common thing and should be used thoughtfully.\n   *\n   * If you add the same finalizer instance of a function or an unsubscribable object to a `Subscription` instance\n   * more than once, you will need to call `remove` the same number of times to remove all instances.\n   *\n   * All finalizer instances are removed to free up memory upon unsubscription.\n   *\n   * @param teardown The finalizer to remove from this subscription\n   */\n  remove(teardown: Exclude<TeardownLogic, void>): void {\n    const { _finalizers } = this;\n    _finalizers && arrRemove(_finalizers, teardown);\n\n    if (teardown instanceof Subscription) {\n      teardown._removeParent(this);\n    }\n  }\n}\n\nexport const EMPTY_SUBSCRIPTION = Subscription.EMPTY;\n\nexport function isSubscription(value: any): value is Subscription {\n  return (\n    value instanceof Subscription ||\n    (value && 'closed' in value && isFunction(value.remove) && isFunction(value.add) && isFunction(value.unsubscribe))\n  );\n}\n\nfunction execFinalizer(finalizer: Unsubscribable | (() => void)) {\n  if (isFunction(finalizer)) {\n    finalizer();\n  } else {\n    finalizer.unsubscribe();\n  }\n}\n", "import { Subscriber } from './Subscriber';\nimport { ObservableNotification } from './types';\n\n/**\n * The {@link GlobalConfig} object for RxJS. It is used to configure things\n * like how to react on unhandled errors.\n */\nexport const config: GlobalConfig = {\n  onUnhandledError: null,\n  onStoppedNotification: null,\n  Promise: undefined,\n  useDeprecatedSynchronousErrorHandling: false,\n  useDeprecatedNextContext: false,\n};\n\n/**\n * The global configuration object for RxJS, used to configure things\n * like how to react on unhandled errors. Accessible via {@link config}\n * object.\n */\nexport interface GlobalConfig {\n  /**\n   * A registration point for unhandled errors from RxJS. These are errors that\n   * cannot were not handled by consuming code in the usual subscription path. For\n   * example, if you have this configured, and you subscribe to an observable without\n   * providing an error handler, errors from that subscription will end up here. This\n   * will _always_ be called asynchronously on another job in the runtime. This is because\n   * we do not want errors thrown in this user-configured handler to interfere with the\n   * behavior of the library.\n   */\n  onUnhandledError: ((err: any) => void) | null;\n\n  /**\n   * A registration point for notifications that cannot be sent to subscribers because they\n   * have completed, errored or have been explicitly unsubscribed. By default, next, complete\n   * and error notifications sent to stopped subscribers are noops. However, sometimes callers\n   * might want a different behavior. For example, with sources that attempt to report errors\n   * to stopped subscribers, a caller can configure RxJS to throw an unhandled error instead.\n   * This will _always_ be called asynchronously on another job in the runtime. This is because\n   * we do not want errors thrown in this user-configured handler to interfere with the\n   * behavior of the library.\n   */\n  onStoppedNotification: ((notification: ObservableNotification<any>, subscriber: Subscriber<any>) => void) | null;\n\n  /**\n   * The promise constructor used by default for {@link Observable#toPromise toPromise} and {@link Observable#forEach forEach}\n   * methods.\n   *\n   * @deprecated As of version 8, RxJS will no longer support this sort of injection of a\n   * Promise constructor. If you need a Promise implementation other than native promises,\n   * please polyfill/patch Promise as you see appropriate. Will be removed in v8.\n   */\n  Promise?: PromiseConstructorLike;\n\n  /**\n   * If true, turns on synchronous error rethrowing, which is a deprecated behavior\n   * in v6 and higher. This behavior enables bad patterns like wrapping a subscribe\n   * call in a try/catch block. It also enables producer interference, a nasty bug\n   * where a multicast can be broken for all observers by a downstream consumer with\n   * an unhandled error. DO NOT USE THIS FLAG UNLESS IT'S NEEDED TO BUY TIME\n   * FOR MIGRATION REASONS.\n   *\n   * @deprecated As of version 8, RxJS will no longer support synchronous throwing\n   * of unhandled errors. All errors will be thrown on a separate call stack to prevent bad\n   * behaviors described above. Will be removed in v8.\n   */\n  useDeprecatedSynchronousErrorHandling: boolean;\n\n  /**\n   * If true, enables an as-of-yet undocumented feature from v5: The ability to access\n   * `unsubscribe()` via `this` context in `next` functions created in observers passed\n   * to `subscribe`.\n   *\n   * This is being removed because the performance was severely problematic, and it could also cause\n   * issues when types other than POJOs are passed to subscribe as subscribers, as they will likely have\n   * their `this` context overwritten.\n   *\n   * @deprecated As of version 8, RxJS will no longer support altering the\n   * context of next functions provided as part of an observer to Subscribe. Instead,\n   * you will have access to a subscription or a signal or token that will allow you to do things like\n   * unsubscribe and test closed status. Will be removed in v8.\n   */\n  useDeprecatedNextContext: boolean;\n}\n", "import type { TimerHandle } from './timerHandle';\ntype SetTimeoutFunction = (handler: () => void, timeout?: number, ...args: any[]) => TimerHandle;\ntype ClearTimeoutFunction = (handle: TimerHandle) => void;\n\ninterface TimeoutProvider {\n  setTimeout: SetTimeoutFunction;\n  clearTimeout: ClearTimeoutFunction;\n  delegate:\n    | {\n        setTimeout: SetTimeoutFunction;\n        clearTimeout: ClearTimeoutFunction;\n      }\n    | undefined;\n}\n\nexport const timeoutProvider: TimeoutProvider = {\n  // When accessing the delegate, use the variable rather than `this` so that\n  // the functions can be called without being bound to the provider.\n  setTimeout(handler: () => void, timeout?: number, ...args) {\n    const { delegate } = timeoutProvider;\n    if (delegate?.setTimeout) {\n      return delegate.setTimeout(handler, timeout, ...args);\n    }\n    return setTimeout(handler, timeout, ...args);\n  },\n  clearTimeout(handle) {\n    const { delegate } = timeoutProvider;\n    return (delegate?.clearTimeout || clearTimeout)(handle as any);\n  },\n  delegate: undefined,\n};\n", "import { config } from '../config';\nimport { timeoutProvider } from '../scheduler/timeoutProvider';\n\n/**\n * Handles an error on another job either with the user-configured {@link onUnhandledError},\n * or by throwing it on that new job so it can be picked up by `window.onerror`, `process.on('error')`, etc.\n *\n * This should be called whenever there is an error that is out-of-band with the subscription\n * or when an error hits a terminal boundary of the subscription and no error handler was provided.\n *\n * @param err the error to report\n */\nexport function reportUnhandledError(err: any) {\n  timeoutProvider.setTimeout(() => {\n    const { onUnhandledError } = config;\n    if (onUnhandledError) {\n      // Execute the user-configured error handler.\n      onUnhandledError(err);\n    } else {\n      // Throw so it is picked up by the runtime's uncaught error mechanism.\n      throw err;\n    }\n  });\n}\n", "/* tslint:disable:no-empty */\nexport function noop() { }\n", "import { CompleteNotification, NextNotification, ErrorNotification } from './types';\n\n/**\n * A completion object optimized for memory use and created to be the\n * same \"shape\" as other notifications in v8.\n * @internal\n */\nexport const COMPLETE_NOTIFICATION = (() => createNotification('C', undefined, undefined) as CompleteNotification)();\n\n/**\n * Internal use only. Creates an optimized error notification that is the same \"shape\"\n * as other notifications.\n * @internal\n */\nexport function errorNotification(error: any): ErrorNotification {\n  return createNotification('E', undefined, error) as any;\n}\n\n/**\n * Internal use only. Creates an optimized next notification that is the same \"shape\"\n * as other notifications.\n * @internal\n */\nexport function nextNotification<T>(value: T) {\n  return createNotification('N', value, undefined) as NextNotification<T>;\n}\n\n/**\n * Ensures that all notifications created internally have the same \"shape\" in v8.\n *\n * TODO: This is only exported to support a crazy legacy test in `groupBy`.\n * @internal\n */\nexport function createNotification(kind: 'N' | 'E' | 'C', value: any, error: any) {\n  return {\n    kind,\n    value,\n    error,\n  };\n}\n", "import { config } from '../config';\n\nlet context: { errorThrown: boolean; error: any } | null = null;\n\n/**\n * Handles dealing with errors for super-gross mode. Creates a context, in which\n * any synchronously thrown errors will be passed to {@link captureError}. Which\n * will record the error such that it will be rethrown after the call back is complete.\n * TODO: Remove in v8\n * @param cb An immediately executed function.\n */\nexport function errorContext(cb: () => void) {\n  if (config.useDeprecatedSynchronousErrorHandling) {\n    const isRoot = !context;\n    if (isRoot) {\n      context = { errorThrown: false, error: null };\n    }\n    cb();\n    if (isRoot) {\n      const { errorThrown, error } = context!;\n      context = null;\n      if (errorThrown) {\n        throw error;\n      }\n    }\n  } else {\n    // This is the general non-deprecated path for everyone that\n    // isn't crazy enough to use super-gross mode (useDeprecatedSynchronousErrorHandling)\n    cb();\n  }\n}\n\n/**\n * Captures errors only in super-gross mode.\n * @param err the error to capture\n */\nexport function captureError(err: any) {\n  if (config.useDeprecatedSynchronousErrorHandling && context) {\n    context.errorThrown = true;\n    context.error = err;\n  }\n}\n", "import { isFunction } from './util/isFunction';\nimport { Observer, ObservableNotification } from './types';\nimport { isSubscription, Subscription } from './Subscription';\nimport { config } from './config';\nimport { reportUnhandledError } from './util/reportUnhandledError';\nimport { noop } from './util/noop';\nimport { nextNotification, errorNotification, COMPLETE_NOTIFICATION } from './NotificationFactories';\nimport { timeoutProvider } from './scheduler/timeoutProvider';\nimport { captureError } from './util/errorContext';\n\n/**\n * Implements the {@link Observer} interface and extends the\n * {@link Subscription} class. While the {@link Observer} is the public API for\n * consuming the values of an {@link Observable}, all Observers get converted to\n * a Subscriber, in order to provide Subscription-like capabilities such as\n * `unsubscribe`. Subscriber is a common type in RxJS, and crucial for\n * implementing operators, but it is rarely used as a public API.\n *\n * @class Subscriber<T>\n */\nexport class Subscriber<T> extends Subscription implements Observer<T> {\n  /**\n   * A static factory for a Subscriber, given a (potentially partial) definition\n   * of an Observer.\n   * @param next The `next` callback of an Observer.\n   * @param error The `error` callback of an\n   * Observer.\n   * @param complete The `complete` callback of an\n   * Observer.\n   * @return A Subscriber wrapping the (partially defined)\n   * Observer represented by the given arguments.\n   * @nocollapse\n   * @deprecated Do not use. Will be removed in v8. There is no replacement for this\n   * method, and there is no reason to be creating instances of `Subscriber` directly.\n   * If you have a specific use case, please file an issue.\n   */\n  static create<T>(next?: (x?: T) => void, error?: (e?: any) => void, complete?: () => void): Subscriber<T> {\n    return new SafeSubscriber(next, error, complete);\n  }\n\n  /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n  protected isStopped: boolean = false;\n  /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n  protected destination: Subscriber<any> | Observer<any>; // this `any` is the escape hatch to erase extra type param (e.g. R)\n\n  /**\n   * @deprecated Internal implementation detail, do not use directly. Will be made internal in v8.\n   * There is no reason to directly create an instance of Subscriber. This type is exported for typings reasons.\n   */\n  constructor(destination?: Subscriber<any> | Observer<any>) {\n    super();\n    if (destination) {\n      this.destination = destination;\n      // Automatically chain subscriptions together here.\n      // if destination is a Subscription, then it is a Subscriber.\n      if (isSubscription(destination)) {\n        destination.add(this);\n      }\n    } else {\n      this.destination = EMPTY_OBSERVER;\n    }\n  }\n\n  /**\n   * The {@link Observer} callback to receive notifications of type `next` from\n   * the Observable, with a value. The Observable may call this method 0 or more\n   * times.\n   * @param {T} [value] The `next` value.\n   * @return {void}\n   */\n  next(value?: T): void {\n    if (this.isStopped) {\n      handleStoppedNotification(nextNotification(value), this);\n    } else {\n      this._next(value!);\n    }\n  }\n\n  /**\n   * The {@link Observer} callback to receive notifications of type `error` from\n   * the Observable, with an attached `Error`. Notifies the Observer that\n   * the Observable has experienced an error condition.\n   * @param {any} [err] The `error` exception.\n   * @return {void}\n   */\n  error(err?: any): void {\n    if (this.isStopped) {\n      handleStoppedNotification(errorNotification(err), this);\n    } else {\n      this.isStopped = true;\n      this._error(err);\n    }\n  }\n\n  /**\n   * The {@link Observer} callback to receive a valueless notification of type\n   * `complete` from the Observable. Notifies the Observer that the Observable\n   * has finished sending push-based notifications.\n   * @return {void}\n   */\n  complete(): void {\n    if (this.isStopped) {\n      handleStoppedNotification(COMPLETE_NOTIFICATION, this);\n    } else {\n      this.isStopped = true;\n      this._complete();\n    }\n  }\n\n  unsubscribe(): void {\n    if (!this.closed) {\n      this.isStopped = true;\n      super.unsubscribe();\n      this.destination = null!;\n    }\n  }\n\n  protected _next(value: T): void {\n    this.destination.next(value);\n  }\n\n  protected _error(err: any): void {\n    try {\n      this.destination.error(err);\n    } finally {\n      this.unsubscribe();\n    }\n  }\n\n  protected _complete(): void {\n    try {\n      this.destination.complete();\n    } finally {\n      this.unsubscribe();\n    }\n  }\n}\n\n/**\n * This bind is captured here because we want to be able to have\n * compatibility with monoid libraries that tend to use a method named\n * `bind`. In particular, a library called Monio requires this.\n */\nconst _bind = Function.prototype.bind;\n\nfunction bind<Fn extends (...args: any[]) => any>(fn: Fn, thisArg: any): Fn {\n  return _bind.call(fn, thisArg);\n}\n\n/**\n * Internal optimization only, DO NOT EXPOSE.\n * @internal\n */\nclass ConsumerObserver<T> implements Observer<T> {\n  constructor(private partialObserver: Partial<Observer<T>>) {}\n\n  next(value: T): void {\n    const { partialObserver } = this;\n    if (partialObserver.next) {\n      try {\n        partialObserver.next(value);\n      } catch (error) {\n        handleUnhandledError(error);\n      }\n    }\n  }\n\n  error(err: any): void {\n    const { partialObserver } = this;\n    if (partialObserver.error) {\n      try {\n        partialObserver.error(err);\n      } catch (error) {\n        handleUnhandledError(error);\n      }\n    } else {\n      handleUnhandledError(err);\n    }\n  }\n\n  complete(): void {\n    const { partialObserver } = this;\n    if (partialObserver.complete) {\n      try {\n        partialObserver.complete();\n      } catch (error) {\n        handleUnhandledError(error);\n      }\n    }\n  }\n}\n\nexport class SafeSubscriber<T> extends Subscriber<T> {\n  constructor(\n    observerOrNext?: Partial<Observer<T>> | ((value: T) => void) | null,\n    error?: ((e?: any) => void) | null,\n    complete?: (() => void) | null\n  ) {\n    super();\n\n    let partialObserver: Partial<Observer<T>>;\n    if (isFunction(observerOrNext) || !observerOrNext) {\n      // The first argument is a function, not an observer. The next\n      // two arguments *could* be observers, or they could be empty.\n      partialObserver = {\n        next: (observerOrNext ?? undefined) as (((value: T) => void) | undefined),\n        error: error ?? undefined,\n        complete: complete ?? undefined,\n      };\n    } else {\n      // The first argument is a partial observer.\n      let context: any;\n      if (this && config.useDeprecatedNextContext) {\n        // This is a deprecated path that made `this.unsubscribe()` available in\n        // next handler functions passed to subscribe. This only exists behind a flag\n        // now, as it is *very* slow.\n        context = Object.create(observerOrNext);\n        context.unsubscribe = () => this.unsubscribe();\n        partialObserver = {\n          next: observerOrNext.next && bind(observerOrNext.next, context),\n          error: observerOrNext.error && bind(observerOrNext.error, context),\n          complete: observerOrNext.complete && bind(observerOrNext.complete, context),\n        };\n      } else {\n        // The \"normal\" path. Just use the partial observer directly.\n        partialObserver = observerOrNext;\n      }\n    }\n\n    // Wrap the partial observer to ensure it's a full observer, and\n    // make sure proper error handling is accounted for.\n    this.destination = new ConsumerObserver(partialObserver);\n  }\n}\n\nfunction handleUnhandledError(error: any) {\n  if (config.useDeprecatedSynchronousErrorHandling) {\n    captureError(error);\n  } else {\n    // Ideal path, we report this as an unhandled error,\n    // which is thrown on a new call stack.\n    reportUnhandledError(error);\n  }\n}\n\n/**\n * An error handler used when no error handler was supplied\n * to the SafeSubscriber -- meaning no error handler was supplied\n * do the `subscribe` call on our observable.\n * @param err The error to handle\n */\nfunction defaultErrorHandler(err: any) {\n  throw err;\n}\n\n/**\n * A handler for notifications that cannot be sent to a stopped subscriber.\n * @param notification The notification being sent\n * @param subscriber The stopped subscriber\n */\nfunction handleStoppedNotification(notification: ObservableNotification<any>, subscriber: Subscriber<any>) {\n  const { onStoppedNotification } = config;\n  onStoppedNotification && timeoutProvider.setTimeout(() => onStoppedNotification(notification, subscriber));\n}\n\n/**\n * The observer used as a stub for subscriptions where the user did not\n * pass any arguments to `subscribe`. Comes with the default error handling\n * behavior.\n */\nexport const EMPTY_OBSERVER: Readonly<Observer<any>> & { closed: true } = {\n  closed: true,\n  next: noop,\n  error: defaultErrorHandler,\n  complete: noop,\n};\n", "/**\n * Symbol.observable or a string \"@@observable\". Used for interop\n *\n * @deprecated We will no longer be exporting this symbol in upcoming versions of RxJS.\n * Instead polyfill and use Symbol.observable directly *or* use https://www.npmjs.com/package/symbol-observable\n */\nexport const observable: string | symbol = (() => (typeof Symbol === 'function' && Symbol.observable) || '@@observable')();\n", "/**\n * This function takes one parameter and just returns it. Simply put,\n * this is like `<T>(x: T): T => x`.\n *\n * ## Examples\n *\n * This is useful in some cases when using things like `mergeMap`\n *\n * ```ts\n * import { interval, take, map, range, mergeMap, identity } from 'rxjs';\n *\n * const source$ = interval(1000).pipe(take(5));\n *\n * const result$ = source$.pipe(\n *   map(i => range(i)),\n *   mergeMap(identity) // same as mergeMap(x => x)\n * );\n *\n * result$.subscribe({\n *   next: console.log\n * });\n * ```\n *\n * Or when you want to selectively apply an operator\n *\n * ```ts\n * import { interval, take, identity } from 'rxjs';\n *\n * const shouldLimit = () => Math.random() < 0.5;\n *\n * const source$ = interval(1000);\n *\n * const result$ = source$.pipe(shouldLimit() ? take(5) : identity);\n *\n * result$.subscribe({\n *   next: console.log\n * });\n * ```\n *\n * @param x Any value that is returned by this function\n * @returns The value passed as the first parameter to this function\n */\nexport function identity<T>(x: T): T {\n  return x;\n}\n", "import { identity } from './identity';\nimport { UnaryFunction } from '../types';\n\nexport function pipe(): typeof identity;\nexport function pipe<T, A>(fn1: UnaryFunction<T, A>): UnaryFunction<T, A>;\nexport function pipe<T, A, B>(fn1: UnaryFunction<T, A>, fn2: UnaryFunction<A, B>): UnaryFunction<T, B>;\nexport function pipe<T, A, B, C>(fn1: UnaryFunction<T, A>, fn2: UnaryFunction<A, B>, fn3: UnaryFunction<B, C>): UnaryFunction<T, C>;\nexport function pipe<T, A, B, C, D>(\n  fn1: UnaryFunction<T, A>,\n  fn2: UnaryFunction<A, B>,\n  fn3: UnaryFunction<B, C>,\n  fn4: UnaryFunction<C, D>\n): UnaryFunction<T, D>;\nexport function pipe<T, A, B, C, D, E>(\n  fn1: UnaryFunction<T, A>,\n  fn2: UnaryFunction<A, B>,\n  fn3: UnaryFunction<B, C>,\n  fn4: UnaryFunction<C, D>,\n  fn5: UnaryFunction<D, E>\n): UnaryFunction<T, E>;\nexport function pipe<T, A, B, C, D, E, F>(\n  fn1: UnaryFunction<T, A>,\n  fn2: UnaryFunction<A, B>,\n  fn3: UnaryFunction<B, C>,\n  fn4: UnaryFunction<C, D>,\n  fn5: UnaryFunction<D, E>,\n  fn6: UnaryFunction<E, F>\n): UnaryFunction<T, F>;\nexport function pipe<T, A, B, C, D, E, F, G>(\n  fn1: UnaryFunction<T, A>,\n  fn2: UnaryFunction<A, B>,\n  fn3: UnaryFunction<B, C>,\n  fn4: UnaryFunction<C, D>,\n  fn5: UnaryFunction<D, E>,\n  fn6: UnaryFunction<E, F>,\n  fn7: UnaryFunction<F, G>\n): UnaryFunction<T, G>;\nexport function pipe<T, A, B, C, D, E, F, G, H>(\n  fn1: UnaryFunction<T, A>,\n  fn2: UnaryFunction<A, B>,\n  fn3: UnaryFunction<B, C>,\n  fn4: UnaryFunction<C, D>,\n  fn5: UnaryFunction<D, E>,\n  fn6: UnaryFunction<E, F>,\n  fn7: UnaryFunction<F, G>,\n  fn8: UnaryFunction<G, H>\n): UnaryFunction<T, H>;\nexport function pipe<T, A, B, C, D, E, F, G, H, I>(\n  fn1: UnaryFunction<T, A>,\n  fn2: UnaryFunction<A, B>,\n  fn3: UnaryFunction<B, C>,\n  fn4: UnaryFunction<C, D>,\n  fn5: UnaryFunction<D, E>,\n  fn6: UnaryFunction<E, F>,\n  fn7: UnaryFunction<F, G>,\n  fn8: UnaryFunction<G, H>,\n  fn9: UnaryFunction<H, I>\n): UnaryFunction<T, I>;\nexport function pipe<T, A, B, C, D, E, F, G, H, I>(\n  fn1: UnaryFunction<T, A>,\n  fn2: UnaryFunction<A, B>,\n  fn3: UnaryFunction<B, C>,\n  fn4: UnaryFunction<C, D>,\n  fn5: UnaryFunction<D, E>,\n  fn6: UnaryFunction<E, F>,\n  fn7: UnaryFunction<F, G>,\n  fn8: UnaryFunction<G, H>,\n  fn9: UnaryFunction<H, I>,\n  ...fns: UnaryFunction<any, any>[]\n): UnaryFunction<T, unknown>;\n\n/**\n * pipe() can be called on one or more functions, each of which can take one argument (\"UnaryFunction\")\n * and uses it to return a value.\n * It returns a function that takes one argument, passes it to the first UnaryFunction, and then\n * passes the result to the next one, passes that result to the next one, and so on.  \n */\nexport function pipe(...fns: Array<UnaryFunction<any, any>>): UnaryFunction<any, any> {\n  return pipeFromArray(fns);\n}\n\n/** @internal */\nexport function pipeFromArray<T, R>(fns: Array<UnaryFunction<T, R>>): UnaryFunction<T, R> {\n  if (fns.length === 0) {\n    return identity as UnaryFunction<any, any>;\n  }\n\n  if (fns.length === 1) {\n    return fns[0];\n  }\n\n  return function piped(input: T): R {\n    return fns.reduce((prev: any, fn: UnaryFunction<T, R>) => fn(prev), input as any);\n  };\n}\n", "import { Operator } from './Operator';\nimport { SafeSubscriber, Subscriber } from './Subscriber';\nimport { isSubscription, Subscription } from './Subscription';\nimport { TeardownLogic, OperatorFunction, Subscribable, Observer } from './types';\nimport { observable as Symbol_observable } from './symbol/observable';\nimport { pipeFromArray } from './util/pipe';\nimport { config } from './config';\nimport { isFunction } from './util/isFunction';\nimport { errorContext } from './util/errorContext';\n\n/**\n * A representation of any set of values over any amount of time. This is the most basic building block\n * of RxJS.\n *\n * @class Observable<T>\n */\nexport class Observable<T> implements Subscribable<T> {\n  /**\n   * @deprecated Internal implementation detail, do not use directly. Will be made internal in v8.\n   */\n  source: Observable<any> | undefined;\n\n  /**\n   * @deprecated Internal implementation detail, do not use directly. Will be made internal in v8.\n   */\n  operator: Operator<any, T> | undefined;\n\n  /**\n   * @constructor\n   * @param {Function} subscribe the function that is called when the Observable is\n   * initially subscribed to. This function is given a Subscriber, to which new values\n   * can be `next`ed, or an `error` method can be called to raise an error, or\n   * `complete` can be called to notify of a successful completion.\n   */\n  constructor(subscribe?: (this: Observable<T>, subscriber: Subscriber<T>) => TeardownLogic) {\n    if (subscribe) {\n      this._subscribe = subscribe;\n    }\n  }\n\n  // HACK: Since TypeScript inherits static properties too, we have to\n  // fight against TypeScript here so Subject can have a different static create signature\n  /**\n   * Creates a new Observable by calling the Observable constructor\n   * @owner Observable\n   * @method create\n   * @param {Function} subscribe? the subscriber function to be passed to the Observable constructor\n   * @return {Observable} a new observable\n   * @nocollapse\n   * @deprecated Use `new Observable()` instead. Will be removed in v8.\n   */\n  static create: (...args: any[]) => any = <T>(subscribe?: (subscriber: Subscriber<T>) => TeardownLogic) => {\n    return new Observable<T>(subscribe);\n  };\n\n  /**\n   * Creates a new Observable, with this Observable instance as the source, and the passed\n   * operator defined as the new observable's operator.\n   * @method lift\n   * @param operator the operator defining the operation to take on the observable\n   * @return a new observable with the Operator applied\n   * @deprecated Internal implementation detail, do not use directly. Will be made internal in v8.\n   * If you have implemented an operator using `lift`, it is recommended that you create an\n   * operator by simply returning `new Observable()` directly. See \"Creating new operators from\n   * scratch\" section here: https://rxjs.dev/guide/operators\n   */\n  lift<R>(operator?: Operator<T, R>): Observable<R> {\n    const observable = new Observable<R>();\n    observable.source = this;\n    observable.operator = operator;\n    return observable;\n  }\n\n  subscribe(observerOrNext?: Partial<Observer<T>> | ((value: T) => void)): Subscription;\n  /** @deprecated Instead of passing separate callback arguments, use an observer argument. Signatures taking separate callback arguments will be removed in v8. Details: https://rxjs.dev/deprecations/subscribe-arguments */\n  subscribe(next?: ((value: T) => void) | null, error?: ((error: any) => void) | null, complete?: (() => void) | null): Subscription;\n  /**\n   * Invokes an execution of an Observable and registers Observer handlers for notifications it will emit.\n   *\n   * <span class=\"informal\">Use it when you have all these Observables, but still nothing is happening.</span>\n   *\n   * `subscribe` is not a regular operator, but a method that calls Observable's internal `subscribe` function. It\n   * might be for example a function that you passed to Observable's constructor, but most of the time it is\n   * a library implementation, which defines what will be emitted by an Observable, and when it be will emitted. This means\n   * that calling `subscribe` is actually the moment when Observable starts its work, not when it is created, as it is often\n   * the thought.\n   *\n   * Apart from starting the execution of an Observable, this method allows you to listen for values\n   * that an Observable emits, as well as for when it completes or errors. You can achieve this in two\n   * of the following ways.\n   *\n   * The first way is creating an object that implements {@link Observer} interface. It should have methods\n   * defined by that interface, but note that it should be just a regular JavaScript object, which you can create\n   * yourself in any way you want (ES6 class, classic function constructor, object literal etc.). In particular, do\n   * not attempt to use any RxJS implementation details to create Observers - you don't need them. Remember also\n   * that your object does not have to implement all methods. If you find yourself creating a method that doesn't\n   * do anything, you can simply omit it. Note however, if the `error` method is not provided and an error happens,\n   * it will be thrown asynchronously. Errors thrown asynchronously cannot be caught using `try`/`catch`. Instead,\n   * use the {@link onUnhandledError} configuration option or use a runtime handler (like `window.onerror` or\n   * `process.on('error)`) to be notified of unhandled errors. Because of this, it's recommended that you provide\n   * an `error` method to avoid missing thrown errors.\n   *\n   * The second way is to give up on Observer object altogether and simply provide callback functions in place of its methods.\n   * This means you can provide three functions as arguments to `subscribe`, where the first function is equivalent\n   * of a `next` method, the second of an `error` method and the third of a `complete` method. Just as in case of an Observer,\n   * if you do not need to listen for something, you can omit a function by passing `undefined` or `null`,\n   * since `subscribe` recognizes these functions by where they were placed in function call. When it comes\n   * to the `error` function, as with an Observer, if not provided, errors emitted by an Observable will be thrown asynchronously.\n   *\n   * You can, however, subscribe with no parameters at all. This may be the case where you're not interested in terminal events\n   * and you also handled emissions internally by using operators (e.g. using `tap`).\n   *\n   * Whichever style of calling `subscribe` you use, in both cases it returns a Subscription object.\n   * This object allows you to call `unsubscribe` on it, which in turn will stop the work that an Observable does and will clean\n   * up all resources that an Observable used. Note that cancelling a subscription will not call `complete` callback\n   * provided to `subscribe` function, which is reserved for a regular completion signal that comes from an Observable.\n   *\n   * Remember that callbacks provided to `subscribe` are not guaranteed to be called asynchronously.\n   * It is an Observable itself that decides when these functions will be called. For example {@link of}\n   * by default emits all its values synchronously. Always check documentation for how given Observable\n   * will behave when subscribed and if its default behavior can be modified with a `scheduler`.\n   *\n   * #### Examples\n   *\n   * Subscribe with an {@link guide/observer Observer}\n   *\n   * ```ts\n   * import { of } from 'rxjs';\n   *\n   * const sumObserver = {\n   *   sum: 0,\n   *   next(value) {\n   *     console.log('Adding: ' + value);\n   *     this.sum = this.sum + value;\n   *   },\n   *   error() {\n   *     // We actually could just remove this method,\n   *     // since we do not really care about errors right now.\n   *   },\n   *   complete() {\n   *     console.log('Sum equals: ' + this.sum);\n   *   }\n   * };\n   *\n   * of(1, 2, 3) // Synchronously emits 1, 2, 3 and then completes.\n   *   .subscribe(sumObserver);\n   *\n   * // Logs:\n   * // 'Adding: 1'\n   * // 'Adding: 2'\n   * // 'Adding: 3'\n   * // 'Sum equals: 6'\n   * ```\n   *\n   * Subscribe with functions ({@link deprecations/subscribe-arguments deprecated})\n   *\n   * ```ts\n   * import { of } from 'rxjs'\n   *\n   * let sum = 0;\n   *\n   * of(1, 2, 3).subscribe(\n   *   value => {\n   *     console.log('Adding: ' + value);\n   *     sum = sum + value;\n   *   },\n   *   undefined,\n   *   () => console.log('Sum equals: ' + sum)\n   * );\n   *\n   * // Logs:\n   * // 'Adding: 1'\n   * // 'Adding: 2'\n   * // 'Adding: 3'\n   * // 'Sum equals: 6'\n   * ```\n   *\n   * Cancel a subscription\n   *\n   * ```ts\n   * import { interval } from 'rxjs';\n   *\n   * const subscription = interval(1000).subscribe({\n   *   next(num) {\n   *     console.log(num)\n   *   },\n   *   complete() {\n   *     // Will not be called, even when cancelling subscription.\n   *     console.log('completed!');\n   *   }\n   * });\n   *\n   * setTimeout(() => {\n   *   subscription.unsubscribe();\n   *   console.log('unsubscribed!');\n   * }, 2500);\n   *\n   * // Logs:\n   * // 0 after 1s\n   * // 1 after 2s\n   * // 'unsubscribed!' after 2.5s\n   * ```\n   *\n   * @param {Observer|Function} observerOrNext (optional) Either an observer with methods to be called,\n   * or the first of three possible handlers, which is the handler for each value emitted from the subscribed\n   * Observable.\n   * @param {Function} error (optional) A handler for a terminal event resulting from an error. If no error handler is provided,\n   * the error will be thrown asynchronously as unhandled.\n   * @param {Function} complete (optional) A handler for a terminal event resulting from successful completion.\n   * @return {Subscription} a subscription reference to the registered handlers\n   * @method subscribe\n   */\n  subscribe(\n    observerOrNext?: Partial<Observer<T>> | ((value: T) => void) | null,\n    error?: ((error: any) => void) | null,\n    complete?: (() => void) | null\n  ): Subscription {\n    const subscriber = isSubscriber(observerOrNext) ? observerOrNext : new SafeSubscriber(observerOrNext, error, complete);\n\n    errorContext(() => {\n      const { operator, source } = this;\n      subscriber.add(\n        operator\n          ? // We're dealing with a subscription in the\n            // operator chain to one of our lifted operators.\n            operator.call(subscriber, source)\n          : source\n          ? // If `source` has a value, but `operator` does not, something that\n            // had intimate knowledge of our API, like our `Subject`, must have\n            // set it. We're going to just call `_subscribe` directly.\n            this._subscribe(subscriber)\n          : // In all other cases, we're likely wrapping a user-provided initializer\n            // function, so we need to catch errors and handle them appropriately.\n            this._trySubscribe(subscriber)\n      );\n    });\n\n    return subscriber;\n  }\n\n  /** @internal */\n  protected _trySubscribe(sink: Subscriber<T>): TeardownLogic {\n    try {\n      return this._subscribe(sink);\n    } catch (err) {\n      // We don't need to return anything in this case,\n      // because it's just going to try to `add()` to a subscription\n      // above.\n      sink.error(err);\n    }\n  }\n\n  /**\n   * Used as a NON-CANCELLABLE means of subscribing to an observable, for use with\n   * APIs that expect promises, like `async/await`. You cannot unsubscribe from this.\n   *\n   * **WARNING**: Only use this with observables you *know* will complete. If the source\n   * observable does not complete, you will end up with a promise that is hung up, and\n   * potentially all of the state of an async function hanging out in memory. To avoid\n   * this situation, look into adding something like {@link timeout}, {@link take},\n   * {@link takeWhile}, or {@link takeUntil} amongst others.\n   *\n   * #### Example\n   *\n   * ```ts\n   * import { interval, take } from 'rxjs';\n   *\n   * const source$ = interval(1000).pipe(take(4));\n   *\n   * async function getTotal() {\n   *   let total = 0;\n   *\n   *   await source$.forEach(value => {\n   *     total += value;\n   *     console.log('observable -> ' + value);\n   *   });\n   *\n   *   return total;\n   * }\n   *\n   * getTotal().then(\n   *   total => console.log('Total: ' + total)\n   * );\n   *\n   * // Expected:\n   * // 'observable -> 0'\n   * // 'observable -> 1'\n   * // 'observable -> 2'\n   * // 'observable -> 3'\n   * // 'Total: 6'\n   * ```\n   *\n   * @param next a handler for each value emitted by the observable\n   * @return a promise that either resolves on observable completion or\n   *  rejects with the handled error\n   */\n  forEach(next: (value: T) => void): Promise<void>;\n\n  /**\n   * @param next a handler for each value emitted by the observable\n   * @param promiseCtor a constructor function used to instantiate the Promise\n   * @return a promise that either resolves on observable completion or\n   *  rejects with the handled error\n   * @deprecated Passing a Promise constructor will no longer be available\n   * in upcoming versions of RxJS. This is because it adds weight to the library, for very\n   * little benefit. If you need this functionality, it is recommended that you either\n   * polyfill Promise, or you create an adapter to convert the returned native promise\n   * to whatever promise implementation you wanted. Will be removed in v8.\n   */\n  forEach(next: (value: T) => void, promiseCtor: PromiseConstructorLike): Promise<void>;\n\n  forEach(next: (value: T) => void, promiseCtor?: PromiseConstructorLike): Promise<void> {\n    promiseCtor = getPromiseCtor(promiseCtor);\n\n    return new promiseCtor<void>((resolve, reject) => {\n      const subscriber = new SafeSubscriber<T>({\n        next: (value) => {\n          try {\n            next(value);\n          } catch (err) {\n            reject(err);\n            subscriber.unsubscribe();\n          }\n        },\n        error: reject,\n        complete: resolve,\n      });\n      this.subscribe(subscriber);\n    }) as Promise<void>;\n  }\n\n  /** @internal */\n  protected _subscribe(subscriber: Subscriber<any>): TeardownLogic {\n    return this.source?.subscribe(subscriber);\n  }\n\n  /**\n   * An interop point defined by the es7-observable spec https://github.com/zenparsing/es-observable\n   * @method Symbol.observable\n   * @return {Observable} this instance of the observable\n   */\n  [Symbol_observable]() {\n    return this;\n  }\n\n  /* tslint:disable:max-line-length */\n  pipe(): Observable<T>;\n  pipe<A>(op1: OperatorFunction<T, A>): Observable<A>;\n  pipe<A, B>(op1: OperatorFunction<T, A>, op2: OperatorFunction<A, B>): Observable<B>;\n  pipe<A, B, C>(op1: OperatorFunction<T, A>, op2: OperatorFunction<A, B>, op3: OperatorFunction<B, C>): Observable<C>;\n  pipe<A, B, C, D>(\n    op1: OperatorFunction<T, A>,\n    op2: OperatorFunction<A, B>,\n    op3: OperatorFunction<B, C>,\n    op4: OperatorFunction<C, D>\n  ): Observable<D>;\n  pipe<A, B, C, D, E>(\n    op1: OperatorFunction<T, A>,\n    op2: OperatorFunction<A, B>,\n    op3: OperatorFunction<B, C>,\n    op4: OperatorFunction<C, D>,\n    op5: OperatorFunction<D, E>\n  ): Observable<E>;\n  pipe<A, B, C, D, E, F>(\n    op1: OperatorFunction<T, A>,\n    op2: OperatorFunction<A, B>,\n    op3: OperatorFunction<B, C>,\n    op4: OperatorFunction<C, D>,\n    op5: OperatorFunction<D, E>,\n    op6: OperatorFunction<E, F>\n  ): Observable<F>;\n  pipe<A, B, C, D, E, F, G>(\n    op1: OperatorFunction<T, A>,\n    op2: OperatorFunction<A, B>,\n    op3: OperatorFunction<B, C>,\n    op4: OperatorFunction<C, D>,\n    op5: OperatorFunction<D, E>,\n    op6: OperatorFunction<E, F>,\n    op7: OperatorFunction<F, G>\n  ): Observable<G>;\n  pipe<A, B, C, D, E, F, G, H>(\n    op1: OperatorFunction<T, A>,\n    op2: OperatorFunction<A, B>,\n    op3: OperatorFunction<B, C>,\n    op4: OperatorFunction<C, D>,\n    op5: OperatorFunction<D, E>,\n    op6: OperatorFunction<E, F>,\n    op7: OperatorFunction<F, G>,\n    op8: OperatorFunction<G, H>\n  ): Observable<H>;\n  pipe<A, B, C, D, E, F, G, H, I>(\n    op1: OperatorFunction<T, A>,\n    op2: OperatorFunction<A, B>,\n    op3: OperatorFunction<B, C>,\n    op4: OperatorFunction<C, D>,\n    op5: OperatorFunction<D, E>,\n    op6: OperatorFunction<E, F>,\n    op7: OperatorFunction<F, G>,\n    op8: OperatorFunction<G, H>,\n    op9: OperatorFunction<H, I>\n  ): Observable<I>;\n  pipe<A, B, C, D, E, F, G, H, I>(\n    op1: OperatorFunction<T, A>,\n    op2: OperatorFunction<A, B>,\n    op3: OperatorFunction<B, C>,\n    op4: OperatorFunction<C, D>,\n    op5: OperatorFunction<D, E>,\n    op6: OperatorFunction<E, F>,\n    op7: OperatorFunction<F, G>,\n    op8: OperatorFunction<G, H>,\n    op9: OperatorFunction<H, I>,\n    ...operations: OperatorFunction<any, any>[]\n  ): Observable<unknown>;\n  /* tslint:enable:max-line-length */\n\n  /**\n   * Used to stitch together functional operators into a chain.\n   * @method pipe\n   * @return {Observable} the Observable result of all of the operators having\n   * been called in the order they were passed in.\n   *\n   * ## Example\n   *\n   * ```ts\n   * import { interval, filter, map, scan } from 'rxjs';\n   *\n   * interval(1000)\n   *   .pipe(\n   *     filter(x => x % 2 === 0),\n   *     map(x => x + x),\n   *     scan((acc, x) => acc + x)\n   *   )\n   *   .subscribe(x => console.log(x));\n   * ```\n   */\n  pipe(...operations: OperatorFunction<any, any>[]): Observable<any> {\n    return pipeFromArray(operations)(this);\n  }\n\n  /* tslint:disable:max-line-length */\n  /** @deprecated Replaced with {@link firstValueFrom} and {@link lastValueFrom}. Will be removed in v8. Details: https://rxjs.dev/deprecations/to-promise */\n  toPromise(): Promise<T | undefined>;\n  /** @deprecated Replaced with {@link firstValueFrom} and {@link lastValueFrom}. Will be removed in v8. Details: https://rxjs.dev/deprecations/to-promise */\n  toPromise(PromiseCtor: typeof Promise): Promise<T | undefined>;\n  /** @deprecated Replaced with {@link firstValueFrom} and {@link lastValueFrom}. Will be removed in v8. Details: https://rxjs.dev/deprecations/to-promise */\n  toPromise(PromiseCtor: PromiseConstructorLike): Promise<T | undefined>;\n  /* tslint:enable:max-line-length */\n\n  /**\n   * Subscribe to this Observable and get a Promise resolving on\n   * `complete` with the last emission (if any).\n   *\n   * **WARNING**: Only use this with observables you *know* will complete. If the source\n   * observable does not complete, you will end up with a promise that is hung up, and\n   * potentially all of the state of an async function hanging out in memory. To avoid\n   * this situation, look into adding something like {@link timeout}, {@link take},\n   * {@link takeWhile}, or {@link takeUntil} amongst others.\n   *\n   * @method toPromise\n   * @param [promiseCtor] a constructor function used to instantiate\n   * the Promise\n   * @return A Promise that resolves with the last value emit, or\n   * rejects on an error. If there were no emissions, Promise\n   * resolves with undefined.\n   * @deprecated Replaced with {@link firstValueFrom} and {@link lastValueFrom}. Will be removed in v8. Details: https://rxjs.dev/deprecations/to-promise\n   */\n  toPromise(promiseCtor?: PromiseConstructorLike): Promise<T | undefined> {\n    promiseCtor = getPromiseCtor(promiseCtor);\n\n    return new promiseCtor((resolve, reject) => {\n      let value: T | undefined;\n      this.subscribe(\n        (x: T) => (value = x),\n        (err: any) => reject(err),\n        () => resolve(value)\n      );\n    }) as Promise<T | undefined>;\n  }\n}\n\n/**\n * Decides between a passed promise constructor from consuming code,\n * A default configured promise constructor, and the native promise\n * constructor and returns it. If nothing can be found, it will throw\n * an error.\n * @param promiseCtor The optional promise constructor to passed by consuming code\n */\nfunction getPromiseCtor(promiseCtor: PromiseConstructorLike | undefined) {\n  return promiseCtor ?? config.Promise ?? Promise;\n}\n\nfunction isObserver<T>(value: any): value is Observer<T> {\n  return value && isFunction(value.next) && isFunction(value.error) && isFunction(value.complete);\n}\n\nfunction isSubscriber<T>(value: any): value is Subscriber<T> {\n  return (value && value instanceof Subscriber) || (isObserver(value) && isSubscription(value));\n}\n", "import { Observable } from '../Observable';\nimport { Subscriber } from '../Subscriber';\nimport { OperatorFunction } from '../types';\nimport { isFunction } from './isFunction';\n\n/**\n * Used to determine if an object is an Observable with a lift function.\n */\nexport function hasLift(source: any): source is { lift: InstanceType<typeof Observable>['lift'] } {\n  return isFunction(source?.lift);\n}\n\n/**\n * Creates an `OperatorFunction`. Used to define operators throughout the library in a concise way.\n * @param init The logic to connect the liftedSource to the subscriber at the moment of subscription.\n */\nexport function operate<T, R>(\n  init: (liftedSource: Observable<T>, subscriber: Subscriber<R>) => (() => void) | void\n): OperatorFunction<T, R> {\n  return (source: Observable<T>) => {\n    if (hasLift(source)) {\n      return source.lift(function (this: Subscriber<R>, liftedSource: Observable<T>) {\n        try {\n          return init(liftedSource, this);\n        } catch (err) {\n          this.error(err);\n        }\n      });\n    }\n    throw new TypeError('Unable to lift unknown Observable type');\n  };\n}\n", "import { Subscriber } from '../Subscriber';\n\n/**\n * Creates an instance of an `OperatorSubscriber`.\n * @param destination The downstream subscriber.\n * @param onNext Handles next values, only called if this subscriber is not stopped or closed. Any\n * error that occurs in this function is caught and sent to the `error` method of this subscriber.\n * @param onError Handles errors from the subscription, any errors that occur in this handler are caught\n * and send to the `destination` error handler.\n * @param onComplete Handles completion notification from the subscription. Any errors that occur in\n * this handler are sent to the `destination` error handler.\n * @param onFinalize Additional teardown logic here. This will only be called on teardown if the\n * subscriber itself is not already closed. This is called after all other teardown logic is executed.\n */\nexport function createOperatorSubscriber<T>(\n  destination: Subscriber<any>,\n  onNext?: (value: T) => void,\n  onComplete?: () => void,\n  onError?: (err: any) => void,\n  onFinalize?: () => void\n): Subscriber<T> {\n  return new OperatorSubscriber(destination, onNext, onComplete, onError, onFinalize);\n}\n\n/**\n * A generic helper for allowing operators to be created with a Subscriber and\n * use closures to capture necessary state from the operator function itself.\n */\nexport class OperatorSubscriber<T> extends Subscriber<T> {\n  /**\n   * Creates an instance of an `OperatorSubscriber`.\n   * @param destination The downstream subscriber.\n   * @param onNext Handles next values, only called if this subscriber is not stopped or closed. Any\n   * error that occurs in this function is caught and sent to the `error` method of this subscriber.\n   * @param onError Handles errors from the subscription, any errors that occur in this handler are caught\n   * and send to the `destination` error handler.\n   * @param onComplete Handles completion notification from the subscription. Any errors that occur in\n   * this handler are sent to the `destination` error handler.\n   * @param onFinalize Additional finalization logic here. This will only be called on finalization if the\n   * subscriber itself is not already closed. This is called after all other finalization logic is executed.\n   * @param shouldUnsubscribe An optional check to see if an unsubscribe call should truly unsubscribe.\n   * NOTE: This currently **ONLY** exists to support the strange behavior of {@link groupBy}, where unsubscription\n   * to the resulting observable does not actually disconnect from the source if there are active subscriptions\n   * to any grouped observable. (DO NOT EXPOSE OR USE EXTERNALLY!!!)\n   */\n  constructor(\n    destination: Subscriber<any>,\n    onNext?: (value: T) => void,\n    onComplete?: () => void,\n    onError?: (err: any) => void,\n    private onFinalize?: () => void,\n    private shouldUnsubscribe?: () => boolean\n  ) {\n    // It's important - for performance reasons - that all of this class's\n    // members are initialized and that they are always initialized in the same\n    // order. This will ensure that all OperatorSubscriber instances have the\n    // same hidden class in V8. This, in turn, will help keep the number of\n    // hidden classes involved in property accesses within the base class as\n    // low as possible. If the number of hidden classes involved exceeds four,\n    // the property accesses will become megamorphic and performance penalties\n    // will be incurred - i.e. inline caches won't be used.\n    //\n    // The reasons for ensuring all instances have the same hidden class are\n    // further discussed in this blog post from Benedikt Meurer:\n    // https://benediktmeurer.de/2018/03/23/impact-of-polymorphism-on-component-based-frameworks-like-react/\n    super(destination);\n    this._next = onNext\n      ? function (this: OperatorSubscriber<T>, value: T) {\n          try {\n            onNext(value);\n          } catch (err) {\n            destination.error(err);\n          }\n        }\n      : super._next;\n    this._error = onError\n      ? function (this: OperatorSubscriber<T>, err: any) {\n          try {\n            onError(err);\n          } catch (err) {\n            // Send any errors that occur down stream.\n            destination.error(err);\n          } finally {\n            // Ensure finalization.\n            this.unsubscribe();\n          }\n        }\n      : super._error;\n    this._complete = onComplete\n      ? function (this: OperatorSubscriber<T>) {\n          try {\n            onComplete();\n          } catch (err) {\n            // Send any errors that occur down stream.\n            destination.error(err);\n          } finally {\n            // Ensure finalization.\n            this.unsubscribe();\n          }\n        }\n      : super._complete;\n  }\n\n  unsubscribe() {\n    if (!this.shouldUnsubscribe || this.shouldUnsubscribe()) {\n      const { closed } = this;\n      super.unsubscribe();\n      // Execute additional teardown if we have any and we didn't already do so.\n      !closed && this.onFinalize?.();\n    }\n  }\n}\n", "import { Subscription } from '../Subscription';\n\ninterface AnimationFrameProvider {\n  schedule(callback: FrameRequestCallback): Subscription;\n  requestAnimationFrame: typeof requestAnimationFrame;\n  cancelAnimationFrame: typeof cancelAnimationFrame;\n  delegate:\n    | {\n        requestAnimationFrame: typeof requestAnimationFrame;\n        cancelAnimationFrame: typeof cancelAnimationFrame;\n      }\n    | undefined;\n}\n\nexport const animationFrameProvider: AnimationFrameProvider = {\n  // When accessing the delegate, use the variable rather than `this` so that\n  // the functions can be called without being bound to the provider.\n  schedule(callback) {\n    let request = requestAnimationFrame;\n    let cancel: typeof cancelAnimationFrame | undefined = cancelAnimationFrame;\n    const { delegate } = animationFrameProvider;\n    if (delegate) {\n      request = delegate.requestAnimationFrame;\n      cancel = delegate.cancelAnimationFrame;\n    }\n    const handle = request((timestamp) => {\n      // Clear the cancel function. The request has been fulfilled, so\n      // attempting to cancel the request upon unsubscription would be\n      // pointless.\n      cancel = undefined;\n      callback(timestamp);\n    });\n    return new Subscription(() => cancel?.(handle));\n  },\n  requestAnimationFrame(...args) {\n    const { delegate } = animationFrameProvider;\n    return (delegate?.requestAnimationFrame || requestAnimationFrame)(...args);\n  },\n  cancelAnimationFrame(...args) {\n    const { delegate } = animationFrameProvider;\n    return (delegate?.cancelAnimationFrame || cancelAnimationFrame)(...args);\n  },\n  delegate: undefined,\n};\n", "import { createErrorClass } from './createErrorClass';\n\nexport interface ObjectUnsubscribedError extends Error {}\n\nexport interface ObjectUnsubscribedErrorCtor {\n  /**\n   * @deprecated Internal implementation detail. Do not construct error instances.\n   * Cannot be tagged as internal: https://github.com/ReactiveX/rxjs/issues/6269\n   */\n  new (): ObjectUnsubscribedError;\n}\n\n/**\n * An error thrown when an action is invalid because the object has been\n * unsubscribed.\n *\n * @see {@link Subject}\n * @see {@link BehaviorSubject}\n *\n * @class ObjectUnsubscribedError\n */\nexport const ObjectUnsubscribedError: ObjectUnsubscribedErrorCtor = createErrorClass(\n  (_super) =>\n    function ObjectUnsubscribedErrorImpl(this: any) {\n      _super(this);\n      this.name = 'ObjectUnsubscribedError';\n      this.message = 'object unsubscribed';\n    }\n);\n", "import { Operator } from './Operator';\nimport { Observable } from './Observable';\nimport { Subscriber } from './Subscriber';\nimport { Subscription, EMPTY_SUBSCRIPTION } from './Subscription';\nimport { Observer, SubscriptionLike, TeardownLogic } from './types';\nimport { ObjectUnsubscribedError } from './util/ObjectUnsubscribedError';\nimport { arrRemove } from './util/arrRemove';\nimport { errorContext } from './util/errorContext';\n\n/**\n * A Subject is a special type of Observable that allows values to be\n * multicasted to many Observers. Subjects are like EventEmitters.\n *\n * Every Subject is an Observable and an Observer. You can subscribe to a\n * Subject, and you can call next to feed values as well as error and complete.\n */\nexport class Subject<T> extends Observable<T> implements SubscriptionLike {\n  closed = false;\n\n  private currentObservers: Observer<T>[] | null = null;\n\n  /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n  observers: Observer<T>[] = [];\n  /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n  isStopped = false;\n  /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n  hasError = false;\n  /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n  thrownError: any = null;\n\n  /**\n   * Creates a \"subject\" by basically gluing an observer to an observable.\n   *\n   * @nocollapse\n   * @deprecated Recommended you do not use. Will be removed at some point in the future. Plans for replacement still under discussion.\n   */\n  static create: (...args: any[]) => any = <T>(destination: Observer<T>, source: Observable<T>): AnonymousSubject<T> => {\n    return new AnonymousSubject<T>(destination, source);\n  };\n\n  constructor() {\n    // NOTE: This must be here to obscure Observable's constructor.\n    super();\n  }\n\n  /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n  lift<R>(operator: Operator<T, R>): Observable<R> {\n    const subject = new AnonymousSubject(this, this);\n    subject.operator = operator as any;\n    return subject as any;\n  }\n\n  /** @internal */\n  protected _throwIfClosed() {\n    if (this.closed) {\n      throw new ObjectUnsubscribedError();\n    }\n  }\n\n  next(value: T) {\n    errorContext(() => {\n      this._throwIfClosed();\n      if (!this.isStopped) {\n        if (!this.currentObservers) {\n          this.currentObservers = Array.from(this.observers);\n        }\n        for (const observer of this.currentObservers) {\n          observer.next(value);\n        }\n      }\n    });\n  }\n\n  error(err: any) {\n    errorContext(() => {\n      this._throwIfClosed();\n      if (!this.isStopped) {\n        this.hasError = this.isStopped = true;\n        this.thrownError = err;\n        const { observers } = this;\n        while (observers.length) {\n          observers.shift()!.error(err);\n        }\n      }\n    });\n  }\n\n  complete() {\n    errorContext(() => {\n      this._throwIfClosed();\n      if (!this.isStopped) {\n        this.isStopped = true;\n        const { observers } = this;\n        while (observers.length) {\n          observers.shift()!.complete();\n        }\n      }\n    });\n  }\n\n  unsubscribe() {\n    this.isStopped = this.closed = true;\n    this.observers = this.currentObservers = null!;\n  }\n\n  get observed() {\n    return this.observers?.length > 0;\n  }\n\n  /** @internal */\n  protected _trySubscribe(subscriber: Subscriber<T>): TeardownLogic {\n    this._throwIfClosed();\n    return super._trySubscribe(subscriber);\n  }\n\n  /** @internal */\n  protected _subscribe(subscriber: Subscriber<T>): Subscription {\n    this._throwIfClosed();\n    this._checkFinalizedStatuses(subscriber);\n    return this._innerSubscribe(subscriber);\n  }\n\n  /** @internal */\n  protected _innerSubscribe(subscriber: Subscriber<any>) {\n    const { hasError, isStopped, observers } = this;\n    if (hasError || isStopped) {\n      return EMPTY_SUBSCRIPTION;\n    }\n    this.currentObservers = null;\n    observers.push(subscriber);\n    return new Subscription(() => {\n      this.currentObservers = null;\n      arrRemove(observers, subscriber);\n    });\n  }\n\n  /** @internal */\n  protected _checkFinalizedStatuses(subscriber: Subscriber<any>) {\n    const { hasError, thrownError, isStopped } = this;\n    if (hasError) {\n      subscriber.error(thrownError);\n    } else if (isStopped) {\n      subscriber.complete();\n    }\n  }\n\n  /**\n   * Creates a new Observable with this Subject as the source. You can do this\n   * to create custom Observer-side logic of the Subject and conceal it from\n   * code that uses the Observable.\n   * @return {Observable} Observable that the Subject casts to\n   */\n  asObservable(): Observable<T> {\n    const observable: any = new Observable<T>();\n    observable.source = this;\n    return observable;\n  }\n}\n\n/**\n * @class AnonymousSubject<T>\n */\nexport class AnonymousSubject<T> extends Subject<T> {\n  constructor(\n    /** @deprecated Internal implementation detail, do not use directly. Will be made internal in v8. */\n    public destination?: Observer<T>,\n    source?: Observable<T>\n  ) {\n    super();\n    this.source = source;\n  }\n\n  next(value: T) {\n    this.destination?.next?.(value);\n  }\n\n  error(err: any) {\n    this.destination?.error?.(err);\n  }\n\n  complete() {\n    this.destination?.complete?.();\n  }\n\n  /** @internal */\n  protected _subscribe(subscriber: Subscriber<T>): Subscription {\n    return this.source?.subscribe(subscriber) ?? EMPTY_SUBSCRIPTION;\n  }\n}\n", "import { TimestampProvider } from '../types';\n\ninterface DateTimestampProvider extends TimestampProvider {\n  delegate: TimestampProvider | undefined;\n}\n\nexport const dateTimestampProvider: DateTimestampProvider = {\n  now() {\n    // Use the variable rather than `this` so that the function can be called\n    // without being bound to the provider.\n    return (dateTimestampProvider.delegate || Date).now();\n  },\n  delegate: undefined,\n};\n", "import { Subject } from './Subject';\nimport { TimestampProvider } from './types';\nimport { Subscriber } from './Subscriber';\nimport { Subscription } from './Subscription';\nimport { dateTimestampProvider } from './scheduler/dateTimestampProvider';\n\n/**\n * A variant of {@link Subject} that \"replays\" old values to new subscribers by emitting them when they first subscribe.\n *\n * `ReplaySubject` has an internal buffer that will store a specified number of values that it has observed. Like `Subject`,\n * `ReplaySubject` \"observes\" values by having them passed to its `next` method. When it observes a value, it will store that\n * value for a time determined by the configuration of the `ReplaySubject`, as passed to its constructor.\n *\n * When a new subscriber subscribes to the `ReplaySubject` instance, it will synchronously emit all values in its buffer in\n * a First-In-First-Out (FIFO) manner. The `ReplaySubject` will also complete, if it has observed completion; and it will\n * error if it has observed an error.\n *\n * There are two main configuration items to be concerned with:\n *\n * 1. `bufferSize` - This will determine how many items are stored in the buffer, defaults to infinite.\n * 2. `windowTime` - The amount of time to hold a value in the buffer before removing it from the buffer.\n *\n * Both configurations may exist simultaneously. So if you would like to buffer a maximum of 3 values, as long as the values\n * are less than 2 seconds old, you could do so with a `new ReplaySubject(3, 2000)`.\n *\n * ### Differences with BehaviorSubject\n *\n * `BehaviorSubject` is similar to `new ReplaySubject(1)`, with a couple of exceptions:\n *\n * 1. `BehaviorSubject` comes \"primed\" with a single value upon construction.\n * 2. `ReplaySubject` will replay values, even after observing an error, where `BehaviorSubject` will not.\n *\n * @see {@link Subject}\n * @see {@link BehaviorSubject}\n * @see {@link shareReplay}\n */\nexport class ReplaySubject<T> extends Subject<T> {\n  private _buffer: (T | number)[] = [];\n  private _infiniteTimeWindow = true;\n\n  /**\n   * @param bufferSize The size of the buffer to replay on subscription\n   * @param windowTime The amount of time the buffered items will stay buffered\n   * @param timestampProvider An object with a `now()` method that provides the current timestamp. This is used to\n   * calculate the amount of time something has been buffered.\n   */\n  constructor(\n    private _bufferSize = Infinity,\n    private _windowTime = Infinity,\n    private _timestampProvider: TimestampProvider = dateTimestampProvider\n  ) {\n    super();\n    this._infiniteTimeWindow = _windowTime === Infinity;\n    this._bufferSize = Math.max(1, _bufferSize);\n    this._windowTime = Math.max(1, _windowTime);\n  }\n\n  next(value: T): void {\n    const { isStopped, _buffer, _infiniteTimeWindow, _timestampProvider, _windowTime } = this;\n    if (!isStopped) {\n      _buffer.push(value);\n      !_infiniteTimeWindow && _buffer.push(_timestampProvider.now() + _windowTime);\n    }\n    this._trimBuffer();\n    super.next(value);\n  }\n\n  /** @internal */\n  protected _subscribe(subscriber: Subscriber<T>): Subscription {\n    this._throwIfClosed();\n    this._trimBuffer();\n\n    const subscription = this._innerSubscribe(subscriber);\n\n    const { _infiniteTimeWindow, _buffer } = this;\n    // We use a copy here, so reentrant code does not mutate our array while we're\n    // emitting it to a new subscriber.\n    const copy = _buffer.slice();\n    for (let i = 0; i < copy.length && !subscriber.closed; i += _infiniteTimeWindow ? 1 : 2) {\n      subscriber.next(copy[i] as T);\n    }\n\n    this._checkFinalizedStatuses(subscriber);\n\n    return subscription;\n  }\n\n  private _trimBuffer() {\n    const { _bufferSize, _timestampProvider, _buffer, _infiniteTimeWindow } = this;\n    // If we don't have an infinite buffer size, and we're over the length,\n    // use splice to truncate the old buffer values off. Note that we have to\n    // double the size for instances where we're not using an infinite time window\n    // because we're storing the values and the timestamps in the same array.\n    const adjustedBufferSize = (_infiniteTimeWindow ? 1 : 2) * _bufferSize;\n    _bufferSize < Infinity && adjustedBufferSize < _buffer.length && _buffer.splice(0, _buffer.length - adjustedBufferSize);\n\n    // Now, if we're not in an infinite time window, remove all values where the time is\n    // older than what is allowed.\n    if (!_infiniteTimeWindow) {\n      const now = _timestampProvider.now();\n      let last = 0;\n      // Search the array for the first timestamp that isn't expired and\n      // truncate the buffer up to that point.\n      for (let i = 1; i < _buffer.length && (_buffer[i] as number) <= now; i += 2) {\n        last = i;\n      }\n      last && _buffer.splice(0, last + 1);\n    }\n  }\n}\n", "import { Scheduler } from '../Scheduler';\nimport { Subscription } from '../Subscription';\nimport { SchedulerAction } from '../types';\n\n/**\n * A unit of work to be executed in a `scheduler`. An action is typically\n * created from within a {@link SchedulerLike} and an RxJS user does not need to concern\n * themselves about creating and manipulating an Action.\n *\n * ```ts\n * class Action<T> extends Subscription {\n *   new (scheduler: Scheduler, work: (state?: T) => void);\n *   schedule(state?: T, delay: number = 0): Subscription;\n * }\n * ```\n *\n * @class Action<T>\n */\nexport class Action<T> extends Subscription {\n  constructor(scheduler: Scheduler, work: (this: SchedulerAction<T>, state?: T) => void) {\n    super();\n  }\n  /**\n   * Schedules this action on its parent {@link SchedulerLike} for execution. May be passed\n   * some context object, `state`. May happen at some point in the future,\n   * according to the `delay` parameter, if specified.\n   * @param {T} [state] Some contextual data that the `work` function uses when\n   * called by the Scheduler.\n   * @param {number} [delay] Time to wait before executing the work, where the\n   * time unit is implicit and defined by the Scheduler.\n   * @return {void}\n   */\n  public schedule(state?: T, delay: number = 0): Subscription {\n    return this;\n  }\n}\n", "import type { TimerHandle } from './timerHandle';\ntype SetIntervalFunction = (handler: () => void, timeout?: number, ...args: any[]) => TimerHandle;\ntype ClearIntervalFunction = (handle: TimerHandle) => void;\n\ninterface IntervalProvider {\n  setInterval: SetIntervalFunction;\n  clearInterval: ClearIntervalFunction;\n  delegate:\n    | {\n        setInterval: SetIntervalFunction;\n        clearInterval: ClearIntervalFunction;\n      }\n    | undefined;\n}\n\nexport const intervalProvider: IntervalProvider = {\n  // When accessing the delegate, use the variable rather than `this` so that\n  // the functions can be called without being bound to the provider.\n  setInterval(handler: () => void, timeout?: number, ...args) {\n    const { delegate } = intervalProvider;\n    if (delegate?.setInterval) {\n      return delegate.setInterval(handler, timeout, ...args);\n    }\n    return setInterval(handler, timeout, ...args);\n  },\n  clearInterval(handle) {\n    const { delegate } = intervalProvider;\n    return (delegate?.clearInterval || clearInterval)(handle as any);\n  },\n  delegate: undefined,\n};\n", "import { Action } from './Action';\nimport { SchedulerAction } from '../types';\nimport { Subscription } from '../Subscription';\nimport { AsyncScheduler } from './AsyncScheduler';\nimport { intervalProvider } from './intervalProvider';\nimport { arrRemove } from '../util/arrRemove';\nimport { TimerHandle } from './timerHandle';\n\nexport class AsyncAction<T> extends Action<T> {\n  public id: TimerHandle | undefined;\n  public state?: T;\n  // @ts-ignore: Property has no initializer and is not definitely assigned\n  public delay: number;\n  protected pending: boolean = false;\n\n  constructor(protected scheduler: AsyncScheduler, protected work: (this: SchedulerAction<T>, state?: T) => void) {\n    super(scheduler, work);\n  }\n\n  public schedule(state?: T, delay: number = 0): Subscription {\n    if (this.closed) {\n      return this;\n    }\n\n    // Always replace the current state with the new state.\n    this.state = state;\n\n    const id = this.id;\n    const scheduler = this.scheduler;\n\n    //\n    // Important implementation note:\n    //\n    // Actions only execute once by default, unless rescheduled from within the\n    // scheduled callback. This allows us to implement single and repeat\n    // actions via the same code path, without adding API surface area, as well\n    // as mimic traditional recursion but across asynchronous boundaries.\n    //\n    // However, JS runtimes and timers distinguish between intervals achieved by\n    // serial `setTimeout` calls vs. a single `setInterval` call. An interval of\n    // serial `setTimeout` calls can be individually delayed, which delays\n    // scheduling the next `setTimeout`, and so on. `setInterval` attempts to\n    // guarantee the interval callback will be invoked more precisely to the\n    // interval period, regardless of load.\n    //\n    // Therefore, we use `setInterval` to schedule single and repeat actions.\n    // If the action reschedules itself with the same delay, the interval is not\n    // canceled. If the action doesn't reschedule, or reschedules with a\n    // different delay, the interval will be canceled after scheduled callback\n    // execution.\n    //\n    if (id != null) {\n      this.id = this.recycleAsyncId(scheduler, id, delay);\n    }\n\n    // Set the pending flag indicating that this action has been scheduled, or\n    // has recursively rescheduled itself.\n    this.pending = true;\n\n    this.delay = delay;\n    // If this action has already an async Id, don't request a new one.\n    this.id = this.id ?? this.requestAsyncId(scheduler, this.id, delay);\n\n    return this;\n  }\n\n  protected requestAsyncId(scheduler: AsyncScheduler, _id?: TimerHandle, delay: number = 0): TimerHandle {\n    return intervalProvider.setInterval(scheduler.flush.bind(scheduler, this), delay);\n  }\n\n  protected recycleAsyncId(_scheduler: AsyncScheduler, id?: TimerHandle, delay: number | null = 0): TimerHandle | undefined {\n    // If this action is rescheduled with the same delay time, don't clear the interval id.\n    if (delay != null && this.delay === delay && this.pending === false) {\n      return id;\n    }\n    // Otherwise, if the action's delay time is different from the current delay,\n    // or the action has been rescheduled before it's executed, clear the interval id\n    if (id != null) {\n      intervalProvider.clearInterval(id);\n    }\n\n    return undefined;\n  }\n\n  /**\n   * Immediately executes this action and the `work` it contains.\n   * @return {any}\n   */\n  public execute(state: T, delay: number): any {\n    if (this.closed) {\n      return new Error('executing a cancelled action');\n    }\n\n    this.pending = false;\n    const error = this._execute(state, delay);\n    if (error) {\n      return error;\n    } else if (this.pending === false && this.id != null) {\n      // Dequeue if the action didn't reschedule itself. Don't call\n      // unsubscribe(), because the action could reschedule later.\n      // For example:\n      // ```\n      // scheduler.schedule(function doWork(counter) {\n      //   /* ... I'm a busy worker bee ... */\n      //   var originalAction = this;\n      //   /* wait 100ms before rescheduling the action */\n      //   setTimeout(function () {\n      //     originalAction.schedule(counter + 1);\n      //   }, 100);\n      // }, 1000);\n      // ```\n      this.id = this.recycleAsyncId(this.scheduler, this.id, null);\n    }\n  }\n\n  protected _execute(state: T, _delay: number): any {\n    let errored: boolean = false;\n    let errorValue: any;\n    try {\n      this.work(state);\n    } catch (e) {\n      errored = true;\n      // HACK: Since code elsewhere is relying on the \"truthiness\" of the\n      // return here, we can't have it return \"\" or 0 or false.\n      // TODO: Clean this up when we refactor schedulers mid-version-8 or so.\n      errorValue = e ? e : new Error('Scheduled action threw falsy error');\n    }\n    if (errored) {\n      this.unsubscribe();\n      return errorValue;\n    }\n  }\n\n  unsubscribe() {\n    if (!this.closed) {\n      const { id, scheduler } = this;\n      const { actions } = scheduler;\n\n      this.work = this.state = this.scheduler = null!;\n      this.pending = false;\n\n      arrRemove(actions, this);\n      if (id != null) {\n        this.id = this.recycleAsyncId(scheduler, id, null);\n      }\n\n      this.delay = null!;\n      super.unsubscribe();\n    }\n  }\n}\n", "import { Action } from './scheduler/Action';\nimport { Subscription } from './Subscription';\nimport { SchedulerLike, SchedulerAction } from './types';\nimport { dateTimestampProvider } from './scheduler/dateTimestampProvider';\n\n/**\n * An execution context and a data structure to order tasks and schedule their\n * execution. Provides a notion of (potentially virtual) time, through the\n * `now()` getter method.\n *\n * Each unit of work in a Scheduler is called an `Action`.\n *\n * ```ts\n * class Scheduler {\n *   now(): number;\n *   schedule(work, delay?, state?): Subscription;\n * }\n * ```\n *\n * @class Scheduler\n * @deprecated Scheduler is an internal implementation detail of RxJS, and\n * should not be used directly. Rather, create your own class and implement\n * {@link SchedulerLike}. Will be made internal in v8.\n */\nexport class Scheduler implements SchedulerLike {\n  public static now: () => number = dateTimestampProvider.now;\n\n  constructor(private schedulerActionCtor: typeof Action, now: () => number = Scheduler.now) {\n    this.now = now;\n  }\n\n  /**\n   * A getter method that returns a number representing the current time\n   * (at the time this function was called) according to the scheduler's own\n   * internal clock.\n   * @return {number} A number that represents the current time. May or may not\n   * have a relation to wall-clock time. May or may not refer to a time unit\n   * (e.g. milliseconds).\n   */\n  public now: () => number;\n\n  /**\n   * Schedules a function, `work`, for execution. May happen at some point in\n   * the future, according to the `delay` parameter, if specified. May be passed\n   * some context object, `state`, which will be passed to the `work` function.\n   *\n   * The given arguments will be processed an stored as an Action object in a\n   * queue of actions.\n   *\n   * @param {function(state: ?T): ?Subscription} work A function representing a\n   * task, or some unit of work to be executed by the Scheduler.\n   * @param {number} [delay] Time to wait before executing the work, where the\n   * time unit is implicit and defined by the Scheduler itself.\n   * @param {T} [state] Some contextual data that the `work` function uses when\n   * called by the Scheduler.\n   * @return {Subscription} A subscription in order to be able to unsubscribe\n   * the scheduled work.\n   */\n  public schedule<T>(work: (this: SchedulerAction<T>, state?: T) => void, delay: number = 0, state?: T): Subscription {\n    return new this.schedulerActionCtor<T>(this, work).schedule(state, delay);\n  }\n}\n", "import { Scheduler } from '../Scheduler';\nimport { Action } from './Action';\nimport { AsyncAction } from './AsyncAction';\nimport { TimerHandle } from './timerHandle';\n\nexport class AsyncScheduler extends Scheduler {\n  public actions: Array<AsyncAction<any>> = [];\n  /**\n   * A flag to indicate whether the Scheduler is currently executing a batch of\n   * queued actions.\n   * @type {boolean}\n   * @internal\n   */\n  public _active: boolean = false;\n  /**\n   * An internal ID used to track the latest asynchronous task such as those\n   * coming from `setTimeout`, `setInterval`, `requestAnimationFrame`, and\n   * others.\n   * @type {any}\n   * @internal\n   */\n  public _scheduled: TimerHandle | undefined;\n\n  constructor(SchedulerAction: typeof Action, now: () => number = Scheduler.now) {\n    super(SchedulerAction, now);\n  }\n\n  public flush(action: AsyncAction<any>): void {\n    const { actions } = this;\n\n    if (this._active) {\n      actions.push(action);\n      return;\n    }\n\n    let error: any;\n    this._active = true;\n\n    do {\n      if ((error = action.execute(action.state, action.delay))) {\n        break;\n      }\n    } while ((action = actions.shift()!)); // exhaust the scheduler queue\n\n    this._active = false;\n\n    if (error) {\n      while ((action = actions.shift()!)) {\n        action.unsubscribe();\n      }\n      throw error;\n    }\n  }\n}\n", "import { AsyncAction } from './AsyncAction';\nimport { AsyncScheduler } from './AsyncScheduler';\n\n/**\n *\n * Async Scheduler\n *\n * <span class=\"informal\">Schedule task as if you used setTimeout(task, duration)</span>\n *\n * `async` scheduler schedules tasks asynchronously, by putting them on the JavaScript\n * event loop queue. It is best used to delay tasks in time or to schedule tasks repeating\n * in intervals.\n *\n * If you just want to \"defer\" task, that is to perform it right after currently\n * executing synchronous code ends (commonly achieved by `setTimeout(deferredTask, 0)`),\n * better choice will be the {@link asapScheduler} scheduler.\n *\n * ## Examples\n * Use async scheduler to delay task\n * ```ts\n * import { asyncScheduler } from 'rxjs';\n *\n * const task = () => console.log('it works!');\n *\n * asyncScheduler.schedule(task, 2000);\n *\n * // After 2 seconds logs:\n * // \"it works!\"\n * ```\n *\n * Use async scheduler to repeat task in intervals\n * ```ts\n * import { asyncScheduler } from 'rxjs';\n *\n * function task(state) {\n *   console.log(state);\n *   this.schedule(state + 1, 1000); // `this` references currently executing Action,\n *                                   // which we reschedule with new state and delay\n * }\n *\n * asyncScheduler.schedule(task, 3000, 0);\n *\n * // Logs:\n * // 0 after 3s\n * // 1 after 4s\n * // 2 after 5s\n * // 3 after 6s\n * ```\n */\n\nexport const asyncScheduler = new AsyncScheduler(AsyncAction);\n\n/**\n * @deprecated Renamed to {@link asyncScheduler}. Will be removed in v8.\n */\nexport const async = asyncScheduler;\n", "import { AsyncAction } from './AsyncAction';\nimport { AnimationFrameScheduler } from './AnimationFrameScheduler';\nimport { SchedulerAction } from '../types';\nimport { animationFrameProvider } from './animationFrameProvider';\nimport { TimerHandle } from './timerHandle';\n\nexport class AnimationFrameAction<T> extends AsyncAction<T> {\n  constructor(protected scheduler: AnimationFrameScheduler, protected work: (this: SchedulerAction<T>, state?: T) => void) {\n    super(scheduler, work);\n  }\n\n  protected requestAsyncId(scheduler: AnimationFrameScheduler, id?: TimerHandle, delay: number = 0): TimerHandle {\n    // If delay is greater than 0, request as an async action.\n    if (delay !== null && delay > 0) {\n      return super.requestAsyncId(scheduler, id, delay);\n    }\n    // Push the action to the end of the scheduler queue.\n    scheduler.actions.push(this);\n    // If an animation frame has already been requested, don't request another\n    // one. If an animation frame hasn't been requested yet, request one. Return\n    // the current animation frame request id.\n    return scheduler._scheduled || (scheduler._scheduled = animationFrameProvider.requestAnimationFrame(() => scheduler.flush(undefined)));\n  }\n\n  protected recycleAsyncId(scheduler: AnimationFrameScheduler, id?: TimerHandle, delay: number = 0): TimerHandle | undefined {\n    // If delay exists and is greater than 0, or if the delay is null (the\n    // action wasn't rescheduled) but was originally scheduled as an async\n    // action, then recycle as an async action.\n    if (delay != null ? delay > 0 : this.delay > 0) {\n      return super.recycleAsyncId(scheduler, id, delay);\n    }\n    // If the scheduler queue has no remaining actions with the same async id,\n    // cancel the requested animation frame and set the scheduled flag to\n    // undefined so the next AnimationFrameAction will request its own.\n    const { actions } = scheduler;\n    if (id != null && actions[actions.length - 1]?.id !== id) {\n      animationFrameProvider.cancelAnimationFrame(id as number);\n      scheduler._scheduled = undefined;\n    }\n    // Return undefined so the action knows to request a new async id if it's rescheduled.\n    return undefined;\n  }\n}\n", "import { AsyncAction } from './AsyncAction';\nimport { AsyncScheduler } from './AsyncScheduler';\n\nexport class AnimationFrameScheduler extends AsyncScheduler {\n  public flush(action?: AsyncAction<any>): void {\n    this._active = true;\n    // The async id that effects a call to flush is stored in _scheduled.\n    // Before executing an action, it's necessary to check the action's async\n    // id to determine whether it's supposed to be executed in the current\n    // flush.\n    // Previous implementations of this method used a count to determine this,\n    // but that was unsound, as actions that are unsubscribed - i.e. cancelled -\n    // are removed from the actions array and that can shift actions that are\n    // scheduled to be executed in a subsequent flush into positions at which\n    // they are executed within the current flush.\n    const flushId = this._scheduled;\n    this._scheduled = undefined;\n\n    const { actions } = this;\n    let error: any;\n    action = action || actions.shift()!;\n\n    do {\n      if ((error = action.execute(action.state, action.delay))) {\n        break;\n      }\n    } while ((action = actions[0]) && action.id === flushId && actions.shift());\n\n    this._active = false;\n\n    if (error) {\n      while ((action = actions[0]) && action.id === flushId && actions.shift()) {\n        action.unsubscribe();\n      }\n      throw error;\n    }\n  }\n}\n", "import { AnimationFrameAction } from './AnimationFrameAction';\nimport { AnimationFrameScheduler } from './AnimationFrameScheduler';\n\n/**\n *\n * Animation Frame Scheduler\n *\n * <span class=\"informal\">Perform task when `window.requestAnimationFrame` would fire</span>\n *\n * When `animationFrame` scheduler is used with delay, it will fall back to {@link asyncScheduler} scheduler\n * behaviour.\n *\n * Without delay, `animationFrame` scheduler can be used to create smooth browser animations.\n * It makes sure scheduled task will happen just before next browser content repaint,\n * thus performing animations as efficiently as possible.\n *\n * ## Example\n * Schedule div height animation\n * ```ts\n * // html: <div style=\"background: #0ff;\"></div>\n * import { animationFrameScheduler } from 'rxjs';\n *\n * const div = document.querySelector('div');\n *\n * animationFrameScheduler.schedule(function(height) {\n *   div.style.height = height + \"px\";\n *\n *   this.schedule(height + 1);  // `this` references currently executing Action,\n *                               // which we reschedule with new state\n * }, 0, 0);\n *\n * // You will see a div element growing in height\n * ```\n */\n\nexport const animationFrameScheduler = new AnimationFrameScheduler(AnimationFrameAction);\n\n/**\n * @deprecated Renamed to {@link animationFrameScheduler}. Will be removed in v8.\n */\nexport const animationFrame = animationFrameScheduler;\n", "import { Observable } from '../Observable';\nimport { SchedulerLike } from '../types';\n\n/**\n * A simple Observable that emits no items to the Observer and immediately\n * emits a complete notification.\n *\n * <span class=\"informal\">Just emits 'complete', and nothing else.</span>\n *\n * ![](empty.png)\n *\n * A simple Observable that only emits the complete notification. It can be used\n * for composing with other Observables, such as in a {@link mergeMap}.\n *\n * ## Examples\n *\n * Log complete notification\n *\n * ```ts\n * import { EMPTY } from 'rxjs';\n *\n * EMPTY.subscribe({\n *   next: () => console.log('Next'),\n *   complete: () => console.log('Complete!')\n * });\n *\n * // Outputs\n * // Complete!\n * ```\n *\n * Emit the number 7, then complete\n *\n * ```ts\n * import { EMPTY, startWith } from 'rxjs';\n *\n * const result = EMPTY.pipe(startWith(7));\n * result.subscribe(x => console.log(x));\n *\n * // Outputs\n * // 7\n * ```\n *\n * Map and flatten only odd numbers to the sequence `'a'`, `'b'`, `'c'`\n *\n * ```ts\n * import { interval, mergeMap, of, EMPTY } from 'rxjs';\n *\n * const interval$ = interval(1000);\n * const result = interval$.pipe(\n *   mergeMap(x => x % 2 === 1 ? of('a', 'b', 'c') : EMPTY),\n * );\n * result.subscribe(x => console.log(x));\n *\n * // Results in the following to the console:\n * // x is equal to the count on the interval, e.g. (0, 1, 2, 3, ...)\n * // x will occur every 1000ms\n * // if x % 2 is equal to 1, print a, b, c (each on its own)\n * // if x % 2 is not equal to 1, nothing will be output\n * ```\n *\n * @see {@link Observable}\n * @see {@link NEVER}\n * @see {@link of}\n * @see {@link throwError}\n */\nexport const EMPTY = new Observable<never>((subscriber) => subscriber.complete());\n\n/**\n * @param scheduler A {@link SchedulerLike} to use for scheduling\n * the emission of the complete notification.\n * @deprecated Replaced with the {@link EMPTY} constant or {@link scheduled} (e.g. `scheduled([], scheduler)`). Will be removed in v8.\n */\nexport function empty(scheduler?: SchedulerLike) {\n  return scheduler ? emptyScheduled(scheduler) : EMPTY;\n}\n\nfunction emptyScheduled(scheduler: SchedulerLike) {\n  return new Observable<never>((subscriber) => scheduler.schedule(() => subscriber.complete()));\n}\n", "import { SchedulerLike } from '../types';\nimport { isFunction } from './isFunction';\n\nexport function isScheduler(value: any): value is SchedulerLike {\n  return value && isFunction(value.schedule);\n}\n", "import { SchedulerLike } from '../types';\nimport { isFunction } from './isFunction';\nimport { isScheduler } from './isScheduler';\n\nfunction last<T>(arr: T[]): T | undefined {\n  return arr[arr.length - 1];\n}\n\nexport function popResultSelector(args: any[]): ((...args: unknown[]) => unknown) | undefined {\n  return isFunction(last(args)) ? args.pop() : undefined;\n}\n\nexport function popScheduler(args: any[]): SchedulerLike | undefined {\n  return isScheduler(last(args)) ? args.pop() : undefined;\n}\n\nexport function popNumber(args: any[], defaultValue: number): number {\n  return typeof last(args) === 'number' ? args.pop()! : defaultValue;\n}\n", "export const isArrayLike = (<T>(x: any): x is ArrayLike<T> => x && typeof x.length === 'number' && typeof x !== 'function');", "import { isFunction } from \"./isFunction\";\n\n/**\n * Tests to see if the object is \"thennable\".\n * @param value the object to test\n */\nexport function isPromise(value: any): value is PromiseLike<any> {\n  return isFunction(value?.then);\n}\n", "import { InteropObservable } from '../types';\nimport { observable as Symbol_observable } from '../symbol/observable';\nimport { isFunction } from './isFunction';\n\n/** Identifies an input as being Observable (but not necessary an Rx Observable) */\nexport function isInteropObservable(input: any): input is InteropObservable<any> {\n  return isFunction(input[Symbol_observable]);\n}\n", "import { isFunction } from './isFunction';\n\nexport function isAsyncIterable<T>(obj: any): obj is AsyncIterable<T> {\n  return Symbol.asyncIterator && isFunction(obj?.[Symbol.asyncIterator]);\n}\n", "/**\n * Creates the TypeError to throw if an invalid object is passed to `from` or `scheduled`.\n * @param input The object that was passed.\n */\nexport function createInvalidObservableTypeError(input: any) {\n  // TODO: We should create error codes that can be looked up, so this can be less verbose.\n  return new TypeError(\n    `You provided ${\n      input !== null && typeof input === 'object' ? 'an invalid object' : `'${input}'`\n    } where a stream was expected. You can provide an Observable, Promise, ReadableStream, Array, AsyncIterable, or Iterable.`\n  );\n}\n", "export function getSymbolIterator(): symbol {\n  if (typeof Symbol !== 'function' || !Symbol.iterator) {\n    return '@@iterator' as any;\n  }\n\n  return Symbol.iterator;\n}\n\nexport const iterator = getSymbolIterator();\n", "import { iterator as Symbol_iterator } from '../symbol/iterator';\nimport { isFunction } from './isFunction';\n\n/** Identifies an input as being an Iterable */\nexport function isIterable(input: any): input is Iterable<any> {\n  return isFunction(input?.[Symbol_iterator]);\n}\n", "import { ReadableStreamLike } from '../types';\nimport { isFunction } from './isFunction';\n\nexport async function* readableStreamLikeToAsyncGenerator<T>(readableStream: ReadableStreamLike<T>): AsyncGenerator<T> {\n  const reader = readableStream.getReader();\n  try {\n    while (true) {\n      const { value, done } = await reader.read();\n      if (done) {\n        return;\n      }\n      yield value!;\n    }\n  } finally {\n    reader.releaseLock();\n  }\n}\n\nexport function isReadableStreamLike<T>(obj: any): obj is ReadableStreamLike<T> {\n  // We don't want to use instanceof checks because they would return\n  // false for instances from another Realm, like an <iframe>.\n  return isFunction(obj?.getReader);\n}\n", "import { isArrayLike } from '../util/isArrayLike';\nimport { isPromise } from '../util/isPromise';\nimport { Observable } from '../Observable';\nimport { ObservableInput, ObservedValueOf, ReadableStreamLike } from '../types';\nimport { isInteropObservable } from '../util/isInteropObservable';\nimport { isAsyncIterable } from '../util/isAsyncIterable';\nimport { createInvalidObservableTypeError } from '../util/throwUnobservableError';\nimport { isIterable } from '../util/isIterable';\nimport { isReadableStreamLike, readableStreamLikeToAsyncGenerator } from '../util/isReadableStreamLike';\nimport { Subscriber } from '../Subscriber';\nimport { isFunction } from '../util/isFunction';\nimport { reportUnhandledError } from '../util/reportUnhandledError';\nimport { observable as Symbol_observable } from '../symbol/observable';\n\nexport function innerFrom<O extends ObservableInput<any>>(input: O): Observable<ObservedValueOf<O>>;\nexport function innerFrom<T>(input: ObservableInput<T>): Observable<T> {\n  if (input instanceof Observable) {\n    return input;\n  }\n  if (input != null) {\n    if (isInteropObservable(input)) {\n      return fromInteropObservable(input);\n    }\n    if (isArrayLike(input)) {\n      return fromArrayLike(input);\n    }\n    if (isPromise(input)) {\n      return fromPromise(input);\n    }\n    if (isAsyncIterable(input)) {\n      return fromAsyncIterable(input);\n    }\n    if (isIterable(input)) {\n      return fromIterable(input);\n    }\n    if (isReadableStreamLike(input)) {\n      return fromReadableStreamLike(input);\n    }\n  }\n\n  throw createInvalidObservableTypeError(input);\n}\n\n/**\n * Creates an RxJS Observable from an object that implements `Symbol.observable`.\n * @param obj An object that properly implements `Symbol.observable`.\n */\nexport function fromInteropObservable<T>(obj: any) {\n  return new Observable((subscriber: Subscriber<T>) => {\n    const obs = obj[Symbol_observable]();\n    if (isFunction(obs.subscribe)) {\n      return obs.subscribe(subscriber);\n    }\n    // Should be caught by observable subscribe function error handling.\n    throw new TypeError('Provided object does not correctly implement Symbol.observable');\n  });\n}\n\n/**\n * Synchronously emits the values of an array like and completes.\n * This is exported because there are creation functions and operators that need to\n * make direct use of the same logic, and there's no reason to make them run through\n * `from` conditionals because we *know* they're dealing with an array.\n * @param array The array to emit values from\n */\nexport function fromArrayLike<T>(array: ArrayLike<T>) {\n  return new Observable((subscriber: Subscriber<T>) => {\n    // Loop over the array and emit each value. Note two things here:\n    // 1. We're making sure that the subscriber is not closed on each loop.\n    //    This is so we don't continue looping over a very large array after\n    //    something like a `take`, `takeWhile`, or other synchronous unsubscription\n    //    has already unsubscribed.\n    // 2. In this form, reentrant code can alter that array we're looping over.\n    //    This is a known issue, but considered an edge case. The alternative would\n    //    be to copy the array before executing the loop, but this has\n    //    performance implications.\n    for (let i = 0; i < array.length && !subscriber.closed; i++) {\n      subscriber.next(array[i]);\n    }\n    subscriber.complete();\n  });\n}\n\nexport function fromPromise<T>(promise: PromiseLike<T>) {\n  return new Observable((subscriber: Subscriber<T>) => {\n    promise\n      .then(\n        (value) => {\n          if (!subscriber.closed) {\n            subscriber.next(value);\n            subscriber.complete();\n          }\n        },\n        (err: any) => subscriber.error(err)\n      )\n      .then(null, reportUnhandledError);\n  });\n}\n\nexport function fromIterable<T>(iterable: Iterable<T>) {\n  return new Observable((subscriber: Subscriber<T>) => {\n    for (const value of iterable) {\n      subscriber.next(value);\n      if (subscriber.closed) {\n        return;\n      }\n    }\n    subscriber.complete();\n  });\n}\n\nexport function fromAsyncIterable<T>(asyncIterable: AsyncIterable<T>) {\n  return new Observable((subscriber: Subscriber<T>) => {\n    process(asyncIterable, subscriber).catch((err) => subscriber.error(err));\n  });\n}\n\nexport function fromReadableStreamLike<T>(readableStream: ReadableStreamLike<T>) {\n  return fromAsyncIterable(readableStreamLikeToAsyncGenerator(readableStream));\n}\n\nasync function process<T>(asyncIterable: AsyncIterable<T>, subscriber: Subscriber<T>) {\n  for await (const value of asyncIterable) {\n    subscriber.next(value);\n    // A side-effect may have closed our subscriber,\n    // check before the next iteration.\n    if (subscriber.closed) {\n      return;\n    }\n  }\n  subscriber.complete();\n}\n", "import { Subscription } from '../Subscription';\nimport { SchedulerAction, SchedulerLike } from '../types';\n\nexport function executeSchedule(\n  parentSubscription: Subscription,\n  scheduler: SchedulerLike,\n  work: () => void,\n  delay: number,\n  repeat: true\n): void;\nexport function executeSchedule(\n  parentSubscription: Subscription,\n  scheduler: SchedulerLike,\n  work: () => void,\n  delay?: number,\n  repeat?: false\n): Subscription;\n\nexport function executeSchedule(\n  parentSubscription: Subscription,\n  scheduler: SchedulerLike,\n  work: () => void,\n  delay = 0,\n  repeat = false\n): Subscription | void {\n  const scheduleSubscription = scheduler.schedule(function (this: SchedulerAction<any>) {\n    work();\n    if (repeat) {\n      parentSubscription.add(this.schedule(null, delay));\n    } else {\n      this.unsubscribe();\n    }\n  }, delay);\n\n  parentSubscription.add(scheduleSubscription);\n\n  if (!repeat) {\n    // Because user-land scheduler implementations are unlikely to properly reuse\n    // Actions for repeat scheduling, we can't trust that the returned subscription\n    // will control repeat subscription scenarios. So we're trying to avoid using them\n    // incorrectly within this library.\n    return scheduleSubscription;\n  }\n}\n", "/** @prettier */\nimport { MonoTypeOperatorFunction, SchedulerLike } from '../types';\nimport { executeSchedule } from '../util/executeSchedule';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * Re-emits all notifications from source Observable with specified scheduler.\n *\n * <span class=\"informal\">Ensure a specific scheduler is used, from outside of an Observable.</span>\n *\n * `observeOn` is an operator that accepts a scheduler as a first parameter, which will be used to reschedule\n * notifications emitted by the source Observable. It might be useful, if you do not have control over\n * internal scheduler of a given Observable, but want to control when its values are emitted nevertheless.\n *\n * Returned Observable emits the same notifications (nexted values, complete and error events) as the source Observable,\n * but rescheduled with provided scheduler. Note that this doesn't mean that source Observables internal\n * scheduler will be replaced in any way. Original scheduler still will be used, but when the source Observable emits\n * notification, it will be immediately scheduled again - this time with scheduler passed to `observeOn`.\n * An anti-pattern would be calling `observeOn` on Observable that emits lots of values synchronously, to split\n * that emissions into asynchronous chunks. For this to happen, scheduler would have to be passed into the source\n * Observable directly (usually into the operator that creates it). `observeOn` simply delays notifications a\n * little bit more, to ensure that they are emitted at expected moments.\n *\n * As a matter of fact, `observeOn` accepts second parameter, which specifies in milliseconds with what delay notifications\n * will be emitted. The main difference between {@link delay} operator and `observeOn` is that `observeOn`\n * will delay all notifications - including error notifications - while `delay` will pass through error\n * from source Observable immediately when it is emitted. In general it is highly recommended to use `delay` operator\n * for any kind of delaying of values in the stream, while using `observeOn` to specify which scheduler should be used\n * for notification emissions in general.\n *\n * ## Example\n *\n * Ensure values in subscribe are called just before browser repaint\n *\n * ```ts\n * import { interval, observeOn, animationFrameScheduler } from 'rxjs';\n *\n * const someDiv = document.createElement('div');\n * someDiv.style.cssText = 'width: 200px;background: #09c';\n * document.body.appendChild(someDiv);\n * const intervals = interval(10);      // Intervals are scheduled\n *                                      // with async scheduler by default...\n * intervals.pipe(\n *   observeOn(animationFrameScheduler) // ...but we will observe on animationFrame\n * )                                    // scheduler to ensure smooth animation.\n * .subscribe(val => {\n *   someDiv.style.height = val + 'px';\n * });\n * ```\n *\n * @see {@link delay}\n *\n * @param scheduler Scheduler that will be used to reschedule notifications from source Observable.\n * @param delay Number of milliseconds that states with what delay every notification should be rescheduled.\n * @return A function that returns an Observable that emits the same\n * notifications as the source Observable, but with provided scheduler.\n */\nexport function observeOn<T>(scheduler: SchedulerLike, delay = 0): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value) => executeSchedule(subscriber, scheduler, () => subscriber.next(value), delay),\n        () => executeSchedule(subscriber, scheduler, () => subscriber.complete(), delay),\n        (err) => executeSchedule(subscriber, scheduler, () => subscriber.error(err), delay)\n      )\n    );\n  });\n}\n", "import { MonoTypeOperatorFunction, SchedulerLike } from '../types';\nimport { operate } from '../util/lift';\n\n/**\n * Asynchronously subscribes Observers to this Observable on the specified {@link SchedulerLike}.\n *\n * With `subscribeOn` you can decide what type of scheduler a specific Observable will be using when it is subscribed to.\n *\n * Schedulers control the speed and order of emissions to observers from an Observable stream.\n *\n * ![](subscribeOn.png)\n *\n * ## Example\n *\n * Given the following code:\n *\n * ```ts\n * import { of, merge } from 'rxjs';\n *\n * const a = of(1, 2, 3);\n * const b = of(4, 5, 6);\n *\n * merge(a, b).subscribe(console.log);\n *\n * // Outputs\n * // 1\n * // 2\n * // 3\n * // 4\n * // 5\n * // 6\n * ```\n *\n * Both Observable `a` and `b` will emit their values directly and synchronously once they are subscribed to.\n *\n * If we instead use the `subscribeOn` operator declaring that we want to use the {@link asyncScheduler} for values emitted by Observable `a`:\n *\n * ```ts\n * import { of, subscribeOn, asyncScheduler, merge } from 'rxjs';\n *\n * const a = of(1, 2, 3).pipe(subscribeOn(asyncScheduler));\n * const b = of(4, 5, 6);\n *\n * merge(a, b).subscribe(console.log);\n *\n * // Outputs\n * // 4\n * // 5\n * // 6\n * // 1\n * // 2\n * // 3\n * ```\n *\n * The reason for this is that Observable `b` emits its values directly and synchronously like before\n * but the emissions from `a` are scheduled on the event loop because we are now using the {@link asyncScheduler} for that specific Observable.\n *\n * @param scheduler The {@link SchedulerLike} to perform subscription actions on.\n * @param delay A delay to pass to the scheduler to delay subscriptions\n * @return A function that returns an Observable modified so that its\n * subscriptions happen on the specified {@link SchedulerLike}.\n */\nexport function subscribeOn<T>(scheduler: SchedulerLike, delay: number = 0): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    subscriber.add(scheduler.schedule(() => source.subscribe(subscriber), delay));\n  });\n}\n", "import { innerFrom } from '../observable/innerFrom';\nimport { observeOn } from '../operators/observeOn';\nimport { subscribeOn } from '../operators/subscribeOn';\nimport { InteropObservable, SchedulerLike } from '../types';\n\nexport function scheduleObservable<T>(input: InteropObservable<T>, scheduler: SchedulerLike) {\n  return innerFrom(input).pipe(subscribeOn(scheduler), observeOn(scheduler));\n}\n", "import { innerFrom } from '../observable/innerFrom';\nimport { observeOn } from '../operators/observeOn';\nimport { subscribeOn } from '../operators/subscribeOn';\nimport { SchedulerLike } from '../types';\n\nexport function schedulePromise<T>(input: PromiseLike<T>, scheduler: SchedulerLike) {\n  return innerFrom(input).pipe(subscribeOn(scheduler), observeOn(scheduler));\n}\n", "import { Observable } from '../Observable';\nimport { SchedulerLike } from '../types';\n\nexport function scheduleArray<T>(input: ArrayLike<T>, scheduler: SchedulerLike) {\n  return new Observable<T>((subscriber) => {\n    // The current array index.\n    let i = 0;\n    // Start iterating over the array like on a schedule.\n    return scheduler.schedule(function () {\n      if (i === input.length) {\n        // If we have hit the end of the array like in the\n        // previous job, we can complete.\n        subscriber.complete();\n      } else {\n        // Otherwise let's next the value at the current index,\n        // then increment our index.\n        subscriber.next(input[i++]);\n        // If the last emission didn't cause us to close the subscriber\n        // (via take or some side effect), reschedule the job and we'll\n        // make another pass.\n        if (!subscriber.closed) {\n          this.schedule();\n        }\n      }\n    });\n  });\n}\n", "import { Observable } from '../Observable';\nimport { SchedulerLike } from '../types';\nimport { iterator as Symbol_iterator } from '../symbol/iterator';\nimport { isFunction } from '../util/isFunction';\nimport { executeSchedule } from '../util/executeSchedule';\n\n/**\n * Used in {@link scheduled} to create an observable from an Iterable.\n * @param input The iterable to create an observable from\n * @param scheduler The scheduler to use\n */\nexport function scheduleIterable<T>(input: Iterable<T>, scheduler: SchedulerLike) {\n  return new Observable<T>((subscriber) => {\n    let iterator: Iterator<T, T>;\n\n    // Schedule the initial creation of the iterator from\n    // the iterable. This is so the code in the iterable is\n    // not called until the scheduled job fires.\n    executeSchedule(subscriber, scheduler, () => {\n      // Create the iterator.\n      iterator = (input as any)[Symbol_iterator]();\n\n      executeSchedule(\n        subscriber,\n        scheduler,\n        () => {\n          let value: T;\n          let done: boolean | undefined;\n          try {\n            // Pull the value out of the iterator\n            ({ value, done } = iterator.next());\n          } catch (err) {\n            // We got an error while pulling from the iterator\n            subscriber.error(err);\n            return;\n          }\n\n          if (done) {\n            // If it is \"done\" we just complete. This mimics the\n            // behavior of JavaScript's `for..of` consumption of\n            // iterables, which will not emit the value from an iterator\n            // result of `{ done: true: value: 'here' }`.\n            subscriber.complete();\n          } else {\n            // The iterable is not done, emit the value.\n            subscriber.next(value);\n          }\n        },\n        0,\n        true\n      );\n    });\n\n    // During finalization, if we see this iterator has a `return` method,\n    // then we know it is a Generator, and not just an Iterator. So we call\n    // the `return()` function. This will ensure that any `finally { }` blocks\n    // inside of the generator we can hit will be hit properly.\n    return () => isFunction(iterator?.return) && iterator.return();\n  });\n}\n", "import { SchedulerLike } from '../types';\nimport { Observable } from '../Observable';\nimport { executeSchedule } from '../util/executeSchedule';\n\nexport function scheduleAsyncIterable<T>(input: AsyncIterable<T>, scheduler: SchedulerLike) {\n  if (!input) {\n    throw new Error('Iterable cannot be null');\n  }\n  return new Observable<T>((subscriber) => {\n    executeSchedule(subscriber, scheduler, () => {\n      const iterator = input[Symbol.asyncIterator]();\n      executeSchedule(\n        subscriber,\n        scheduler,\n        () => {\n          iterator.next().then((result) => {\n            if (result.done) {\n              // This will remove the subscriptions from\n              // the parent subscription.\n              subscriber.complete();\n            } else {\n              subscriber.next(result.value);\n            }\n          });\n        },\n        0,\n        true\n      );\n    });\n  });\n}\n", "import { SchedulerLike, ReadableStreamLike } from '../types';\nimport { Observable } from '../Observable';\nimport { scheduleAsyncIterable } from './scheduleAsyncIterable';\nimport { readableStreamLikeToAsyncGenerator } from '../util/isReadableStreamLike';\n\nexport function scheduleReadableStreamLike<T>(input: ReadableStreamLike<T>, scheduler: SchedulerLike): Observable<T> {\n  return scheduleAsyncIterable(readableStreamLikeToAsyncGenerator(input), scheduler);\n}\n", "import { scheduleObservable } from './scheduleObservable';\nimport { schedulePromise } from './schedulePromise';\nimport { scheduleArray } from './scheduleArray';\nimport { scheduleIterable } from './scheduleIterable';\nimport { scheduleAsyncIterable } from './scheduleAsyncIterable';\nimport { isInteropObservable } from '../util/isInteropObservable';\nimport { isPromise } from '../util/isPromise';\nimport { isArrayLike } from '../util/isArrayLike';\nimport { isIterable } from '../util/isIterable';\nimport { ObservableInput, SchedulerLike } from '../types';\nimport { Observable } from '../Observable';\nimport { isAsyncIterable } from '../util/isAsyncIterable';\nimport { createInvalidObservableTypeError } from '../util/throwUnobservableError';\nimport { isReadableStreamLike } from '../util/isReadableStreamLike';\nimport { scheduleReadableStreamLike } from './scheduleReadableStreamLike';\n\n/**\n * Converts from a common {@link ObservableInput} type to an observable where subscription and emissions\n * are scheduled on the provided scheduler.\n *\n * @see {@link from}\n * @see {@link of}\n *\n * @param input The observable, array, promise, iterable, etc you would like to schedule\n * @param scheduler The scheduler to use to schedule the subscription and emissions from\n * the returned observable.\n */\nexport function scheduled<T>(input: ObservableInput<T>, scheduler: SchedulerLike): Observable<T> {\n  if (input != null) {\n    if (isInteropObservable(input)) {\n      return scheduleObservable(input, scheduler);\n    }\n    if (isArrayLike(input)) {\n      return scheduleArray(input, scheduler);\n    }\n    if (isPromise(input)) {\n      return schedulePromise(input, scheduler);\n    }\n    if (isAsyncIterable(input)) {\n      return scheduleAsyncIterable(input, scheduler);\n    }\n    if (isIterable(input)) {\n      return scheduleIterable(input, scheduler);\n    }\n    if (isReadableStreamLike(input)) {\n      return scheduleReadableStreamLike(input, scheduler);\n    }\n  }\n  throw createInvalidObservableTypeError(input);\n}\n", "import { Observable } from '../Observable';\nimport { ObservableInput, SchedulerLike, ObservedValueOf } from '../types';\nimport { scheduled } from '../scheduled/scheduled';\nimport { innerFrom } from './innerFrom';\n\nexport function from<O extends ObservableInput<any>>(input: O): Observable<ObservedValueOf<O>>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function from<O extends ObservableInput<any>>(input: O, scheduler: SchedulerLike | undefined): Observable<ObservedValueOf<O>>;\n\n/**\n * Creates an Observable from an Array, an array-like object, a Promise, an iterable object, or an Observable-like object.\n *\n * <span class=\"informal\">Converts almost anything to an Observable.</span>\n *\n * ![](from.png)\n *\n * `from` converts various other objects and data types into Observables. It also converts a Promise, an array-like, or an\n * <a href=\"https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Iteration_protocols#iterable\" target=\"_blank\">iterable</a>\n * object into an Observable that emits the items in that promise, array, or iterable. A String, in this context, is treated\n * as an array of characters. Observable-like objects (contains a function named with the ES2015 Symbol for Observable) can also be\n * converted through this operator.\n *\n * ## Examples\n *\n * Converts an array to an Observable\n *\n * ```ts\n * import { from } from 'rxjs';\n *\n * const array = [10, 20, 30];\n * const result = from(array);\n *\n * result.subscribe(x => console.log(x));\n *\n * // Logs:\n * // 10\n * // 20\n * // 30\n * ```\n *\n * Convert an infinite iterable (from a generator) to an Observable\n *\n * ```ts\n * import { from, take } from 'rxjs';\n *\n * function* generateDoubles(seed) {\n *    let i = seed;\n *    while (true) {\n *      yield i;\n *      i = 2 * i; // double it\n *    }\n * }\n *\n * const iterator = generateDoubles(3);\n * const result = from(iterator).pipe(take(10));\n *\n * result.subscribe(x => console.log(x));\n *\n * // Logs:\n * // 3\n * // 6\n * // 12\n * // 24\n * // 48\n * // 96\n * // 192\n * // 384\n * // 768\n * // 1536\n * ```\n *\n * With `asyncScheduler`\n *\n * ```ts\n * import { from, asyncScheduler } from 'rxjs';\n *\n * console.log('start');\n *\n * const array = [10, 20, 30];\n * const result = from(array, asyncScheduler);\n *\n * result.subscribe(x => console.log(x));\n *\n * console.log('end');\n *\n * // Logs:\n * // 'start'\n * // 'end'\n * // 10\n * // 20\n * // 30\n * ```\n *\n * @see {@link fromEvent}\n * @see {@link fromEventPattern}\n *\n * @param {ObservableInput<T>} A subscription object, a Promise, an Observable-like,\n * an Array, an iterable, or an array-like object to be converted.\n * @param {SchedulerLike} An optional {@link SchedulerLike} on which to schedule the emission of values.\n * @return {Observable<T>}\n */\nexport function from<T>(input: ObservableInput<T>, scheduler?: SchedulerLike): Observable<T> {\n  return scheduler ? scheduled(input, scheduler) : innerFrom(input);\n}\n", "import { SchedulerLike, ValueFromArray } from '../types';\nimport { Observable } from '../Observable';\nimport { popScheduler } from '../util/args';\nimport { from } from './from';\n\n// Devs are more likely to pass null or undefined than they are a scheduler\n// without accompanying values. To make things easier for (naughty) devs who\n// use the `strictNullChecks: false` TypeScript compiler option, these\n// overloads with explicit null and undefined values are included.\n\nexport function of(value: null): Observable<null>;\nexport function of(value: undefined): Observable<undefined>;\n\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function of(scheduler: SchedulerLike): Observable<never>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function of<A extends readonly unknown[]>(...valuesAndScheduler: [...A, SchedulerLike]): Observable<ValueFromArray<A>>;\n\nexport function of(): Observable<never>;\n/** @deprecated Do not specify explicit type parameters. Signatures with type parameters that cannot be inferred will be removed in v8. */\nexport function of<T>(): Observable<T>;\nexport function of<T>(value: T): Observable<T>;\nexport function of<A extends readonly unknown[]>(...values: A): Observable<ValueFromArray<A>>;\n\n/**\n * Converts the arguments to an observable sequence.\n *\n * <span class=\"informal\">Each argument becomes a `next` notification.</span>\n *\n * ![](of.png)\n *\n * Unlike {@link from}, it does not do any flattening and emits each argument in whole\n * as a separate `next` notification.\n *\n * ## Examples\n *\n * Emit the values `10, 20, 30`\n *\n * ```ts\n * import { of } from 'rxjs';\n *\n * of(10, 20, 30)\n *   .subscribe({\n *     next: value => console.log('next:', value),\n *     error: err => console.log('error:', err),\n *     complete: () => console.log('the end'),\n *   });\n *\n * // Outputs\n * // next: 10\n * // next: 20\n * // next: 30\n * // the end\n * ```\n *\n * Emit the array `[1, 2, 3]`\n *\n * ```ts\n * import { of } from 'rxjs';\n *\n * of([1, 2, 3])\n *   .subscribe({\n *     next: value => console.log('next:', value),\n *     error: err => console.log('error:', err),\n *     complete: () => console.log('the end'),\n *   });\n *\n * // Outputs\n * // next: [1, 2, 3]\n * // the end\n * ```\n *\n * @see {@link from}\n * @see {@link range}\n *\n * @param {...T} values A comma separated list of arguments you want to be emitted\n * @return {Observable} An Observable that emits the arguments\n * described above and then completes.\n */\nexport function of<T>(...args: Array<T | SchedulerLike>): Observable<T> {\n  const scheduler = popScheduler(args);\n  return from(args as T[], scheduler);\n}\n", "import { Observable } from '../Observable';\nimport { Subscriber } from '../Subscriber';\nimport { SchedulerLike } from '../types';\nimport { isFunction } from '../util/isFunction';\n\n/**\n * Creates an observable that will create an error instance and push it to the consumer as an error\n * immediately upon subscription.\n *\n * <span class=\"informal\">Just errors and does nothing else</span>\n *\n * ![](throw.png)\n *\n * This creation function is useful for creating an observable that will create an error and error every\n * time it is subscribed to. Generally, inside of most operators when you might want to return an errored\n * observable, this is unnecessary. In most cases, such as in the inner return of {@link concatMap},\n * {@link mergeMap}, {@link defer}, and many others, you can simply throw the error, and RxJS will pick\n * that up and notify the consumer of the error.\n *\n * ## Example\n *\n * Create a simple observable that will create a new error with a timestamp and log it\n * and the message every time you subscribe to it\n *\n * ```ts\n * import { throwError } from 'rxjs';\n *\n * let errorCount = 0;\n *\n * const errorWithTimestamp$ = throwError(() => {\n *   const error: any = new Error(`This is error number ${ ++errorCount }`);\n *   error.timestamp = Date.now();\n *   return error;\n * });\n *\n * errorWithTimestamp$.subscribe({\n *   error: err => console.log(err.timestamp, err.message)\n * });\n *\n * errorWithTimestamp$.subscribe({\n *   error: err => console.log(err.timestamp, err.message)\n * });\n *\n * // Logs the timestamp and a new error message for each subscription\n * ```\n *\n * ### Unnecessary usage\n *\n * Using `throwError` inside of an operator or creation function\n * with a callback, is usually not necessary\n *\n * ```ts\n * import { of, concatMap, timer, throwError } from 'rxjs';\n *\n * const delays$ = of(1000, 2000, Infinity, 3000);\n *\n * delays$.pipe(\n *   concatMap(ms => {\n *     if (ms < 10000) {\n *       return timer(ms);\n *     } else {\n *       // This is probably overkill.\n *       return throwError(() => new Error(`Invalid time ${ ms }`));\n *     }\n *   })\n * )\n * .subscribe({\n *   next: console.log,\n *   error: console.error\n * });\n * ```\n *\n * You can just throw the error instead\n *\n * ```ts\n * import { of, concatMap, timer } from 'rxjs';\n *\n * const delays$ = of(1000, 2000, Infinity, 3000);\n *\n * delays$.pipe(\n *   concatMap(ms => {\n *     if (ms < 10000) {\n *       return timer(ms);\n *     } else {\n *       // Cleaner and easier to read for most folks.\n *       throw new Error(`Invalid time ${ ms }`);\n *     }\n *   })\n * )\n * .subscribe({\n *   next: console.log,\n *   error: console.error\n * });\n * ```\n *\n * @param errorFactory A factory function that will create the error instance that is pushed.\n */\nexport function throwError(errorFactory: () => any): Observable<never>;\n\n/**\n * Returns an observable that will error with the specified error immediately upon subscription.\n *\n * @param error The error instance to emit\n * @deprecated Support for passing an error value will be removed in v8. Instead, pass a factory function to `throwError(() => new Error('test'))`. This is\n * because it will create the error at the moment it should be created and capture a more appropriate stack trace. If\n * for some reason you need to create the error ahead of time, you can still do that: `const err = new Error('test'); throwError(() => err);`.\n */\nexport function throwError(error: any): Observable<never>;\n\n/**\n * Notifies the consumer of an error using a given scheduler by scheduling it at delay `0` upon subscription.\n *\n * @param errorOrErrorFactory An error instance or error factory\n * @param scheduler A scheduler to use to schedule the error notification\n * @deprecated The `scheduler` parameter will be removed in v8.\n * Use `throwError` in combination with {@link observeOn}: `throwError(() => new Error('test')).pipe(observeOn(scheduler));`.\n * Details: https://rxjs.dev/deprecations/scheduler-argument\n */\nexport function throwError(errorOrErrorFactory: any, scheduler: SchedulerLike): Observable<never>;\n\nexport function throwError(errorOrErrorFactory: any, scheduler?: SchedulerLike): Observable<never> {\n  const errorFactory = isFunction(errorOrErrorFactory) ? errorOrErrorFactory : () => errorOrErrorFactory;\n  const init = (subscriber: Subscriber<never>) => subscriber.error(errorFactory());\n  return new Observable(scheduler ? (subscriber) => scheduler.schedule(init as any, 0, subscriber) : init);\n}\n", "import { createErrorClass } from './createErrorClass';\n\nexport interface EmptyError extends Error {}\n\nexport interface EmptyErrorCtor {\n  /**\n   * @deprecated Internal implementation detail. Do not construct error instances.\n   * Cannot be tagged as internal: https://github.com/ReactiveX/rxjs/issues/6269\n   */\n  new (): EmptyError;\n}\n\n/**\n * An error thrown when an Observable or a sequence was queried but has no\n * elements.\n *\n * @see {@link first}\n * @see {@link last}\n * @see {@link single}\n * @see {@link firstValueFrom}\n * @see {@link lastValueFrom}\n *\n * @class EmptyError\n */\nexport const EmptyError: EmptyErrorCtor = createErrorClass((_super) => function EmptyErrorImpl(this: any) {\n  _super(this);\n  this.name = 'EmptyError';\n  this.message = 'no elements in sequence';\n});\n", "/**\n * Checks to see if a value is not only a `Date` object,\n * but a *valid* `Date` object that can be converted to a\n * number. For example, `new Date('blah')` is indeed an\n * `instanceof Date`, however it cannot be converted to a\n * number.\n */\nexport function isValidDate(value: any): value is Date {\n  return value instanceof Date && !isNaN(value as any);\n}\n", "import { OperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\nexport function map<T, R>(project: (value: T, index: number) => R): OperatorFunction<T, R>;\n/** @deprecated Use a closure instead of a `thisArg`. Signatures accepting a `thisArg` will be removed in v8. */\nexport function map<T, R, A>(project: (this: A, value: T, index: number) => R, thisArg: A): OperatorFunction<T, R>;\n\n/**\n * Applies a given `project` function to each value emitted by the source\n * Observable, and emits the resulting values as an Observable.\n *\n * <span class=\"informal\">Like [Array.prototype.map()](https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array/map),\n * it passes each source value through a transformation function to get\n * corresponding output values.</span>\n *\n * ![](map.png)\n *\n * Similar to the well known `Array.prototype.map` function, this operator\n * applies a projection to each value and emits that projection in the output\n * Observable.\n *\n * ## Example\n *\n * Map every click to the `clientX` position of that click\n *\n * ```ts\n * import { fromEvent, map } from 'rxjs';\n *\n * const clicks = fromEvent<PointerEvent>(document, 'click');\n * const positions = clicks.pipe(map(ev => ev.clientX));\n *\n * positions.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link mapTo}\n * @see {@link pluck}\n *\n * @param {function(value: T, index: number): R} project The function to apply\n * to each `value` emitted by the source Observable. The `index` parameter is\n * the number `i` for the i-th emission that has happened since the\n * subscription, starting from the number `0`.\n * @param {any} [thisArg] An optional argument to define what `this` is in the\n * `project` function.\n * @return A function that returns an Observable that emits the values from the\n * source Observable transformed by the given `project` function.\n */\nexport function map<T, R>(project: (value: T, index: number) => R, thisArg?: any): OperatorFunction<T, R> {\n  return operate((source, subscriber) => {\n    // The index of the value from the source. Used with projection.\n    let index = 0;\n    // Subscribe to the source, all errors and completions are sent along\n    // to the consumer.\n    source.subscribe(\n      createOperatorSubscriber(subscriber, (value: T) => {\n        // Call the projection function with the appropriate this context,\n        // and send the resulting value to the consumer.\n        subscriber.next(project.call(thisArg, value, index++));\n      })\n    );\n  });\n}\n", "import { OperatorFunction } from \"../types\";\nimport { map } from \"../operators/map\";\n\nconst { isArray } = Array;\n\nfunction callOrApply<T, R>(fn: ((...values: T[]) => R), args: T|T[]): R {\n    return isArray(args) ? fn(...args) : fn(args);\n}\n\n/**\n * Used in several -- mostly deprecated -- situations where we need to \n * apply a list of arguments or a single argument to a result selector.\n */\nexport function mapOneOrManyArgs<T, R>(fn: ((...values: T[]) => R)): OperatorFunction<T|T[], R> {\n    return map(args => callOrApply(fn, args))\n}", "const { isArray } = Array;\nconst { getPrototypeOf, prototype: objectProto, keys: getKeys } = Object;\n\n/**\n * Used in functions where either a list of arguments, a single array of arguments, or a\n * dictionary of arguments can be returned. Returns an object with an `args` property with\n * the arguments in an array, if it is a dictionary, it will also return the `keys` in another\n * property.\n */\nexport function argsArgArrayOrObject<T, O extends Record<string, T>>(args: T[] | [O] | [T[]]): { args: T[]; keys: string[] | null } {\n  if (args.length === 1) {\n    const first = args[0];\n    if (isArray(first)) {\n      return { args: first, keys: null };\n    }\n    if (isPOJO(first)) {\n      const keys = getKeys(first);\n      return {\n        args: keys.map((key) => first[key]),\n        keys,\n      };\n    }\n  }\n\n  return { args: args as T[], keys: null };\n}\n\nfunction isPOJO(obj: any): obj is object {\n  return obj && typeof obj === 'object' && getPrototypeOf(obj) === objectProto;\n}\n", "export function createObject(keys: string[], values: any[]) {\n  return keys.reduce((result, key, i) => ((result[key] = values[i]), result), {} as any);\n}\n", "import { Observable } from '../Observable';\nimport { ObservableInput, SchedulerLike, ObservedValueOf, ObservableInputTuple } from '../types';\nimport { argsArgArrayOrObject } from '../util/argsArgArrayOrObject';\nimport { Subscriber } from '../Subscriber';\nimport { from } from './from';\nimport { identity } from '../util/identity';\nimport { Subscription } from '../Subscription';\nimport { mapOneOrManyArgs } from '../util/mapOneOrManyArgs';\nimport { popResultSelector, popScheduler } from '../util/args';\nimport { createObject } from '../util/createObject';\nimport { createOperatorSubscriber } from '../operators/OperatorSubscriber';\nimport { AnyCatcher } from '../AnyCatcher';\nimport { executeSchedule } from '../util/executeSchedule';\n\n// combineLatest(any)\n// We put this first because we need to catch cases where the user has supplied\n// _exactly `any`_ as the argument. Since `any` literally matches _anything_,\n// we don't want it to randomly hit one of the other type signatures below,\n// as we have no idea at build-time what type we should be returning when given an any.\n\n/**\n * You have passed `any` here, we can't figure out if it is\n * an array or an object, so you're getting `unknown`. Use better types.\n * @param arg Something typed as `any`\n */\nexport function combineLatest<T extends AnyCatcher>(arg: T): Observable<unknown>;\n\n// combineLatest([a, b, c])\nexport function combineLatest(sources: []): Observable<never>;\nexport function combineLatest<A extends readonly unknown[]>(sources: readonly [...ObservableInputTuple<A>]): Observable<A>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `combineLatestAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function combineLatest<A extends readonly unknown[], R>(\n  sources: readonly [...ObservableInputTuple<A>],\n  resultSelector: (...values: A) => R,\n  scheduler: SchedulerLike\n): Observable<R>;\nexport function combineLatest<A extends readonly unknown[], R>(\n  sources: readonly [...ObservableInputTuple<A>],\n  resultSelector: (...values: A) => R\n): Observable<R>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `combineLatestAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function combineLatest<A extends readonly unknown[]>(\n  sources: readonly [...ObservableInputTuple<A>],\n  scheduler: SchedulerLike\n): Observable<A>;\n\n// combineLatest(a, b, c)\n/** @deprecated Pass an array of sources instead. The rest-parameters signature will be removed in v8. Details: https://rxjs.dev/deprecations/array-argument */\nexport function combineLatest<A extends readonly unknown[]>(...sources: [...ObservableInputTuple<A>]): Observable<A>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `combineLatestAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function combineLatest<A extends readonly unknown[], R>(\n  ...sourcesAndResultSelectorAndScheduler: [...ObservableInputTuple<A>, (...values: A) => R, SchedulerLike]\n): Observable<R>;\n/** @deprecated Pass an array of sources instead. The rest-parameters signature will be removed in v8. Details: https://rxjs.dev/deprecations/array-argument */\nexport function combineLatest<A extends readonly unknown[], R>(\n  ...sourcesAndResultSelector: [...ObservableInputTuple<A>, (...values: A) => R]\n): Observable<R>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `combineLatestAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function combineLatest<A extends readonly unknown[]>(\n  ...sourcesAndScheduler: [...ObservableInputTuple<A>, SchedulerLike]\n): Observable<A>;\n\n// combineLatest({a, b, c})\nexport function combineLatest(sourcesObject: { [K in any]: never }): Observable<never>;\nexport function combineLatest<T extends Record<string, ObservableInput<any>>>(\n  sourcesObject: T\n): Observable<{ [K in keyof T]: ObservedValueOf<T[K]> }>;\n\n/**\n * Combines multiple Observables to create an Observable whose values are\n * calculated from the latest values of each of its input Observables.\n *\n * <span class=\"informal\">Whenever any input Observable emits a value, it\n * computes a formula using the latest values from all the inputs, then emits\n * the output of that formula.</span>\n *\n * ![](combineLatest.png)\n *\n * `combineLatest` combines the values from all the Observables passed in the\n * observables array. This is done by subscribing to each Observable in order and,\n * whenever any Observable emits, collecting an array of the most recent\n * values from each Observable. So if you pass `n` Observables to this operator,\n * the returned Observable will always emit an array of `n` values, in an order\n * corresponding to the order of the passed Observables (the value from the first Observable\n * will be at index 0 of the array and so on).\n *\n * Static version of `combineLatest` accepts an array of Observables. Note that an array of\n * Observables is a good choice, if you don't know beforehand how many Observables\n * you will combine. Passing an empty array will result in an Observable that\n * completes immediately.\n *\n * To ensure the output array always has the same length, `combineLatest` will\n * actually wait for all input Observables to emit at least once,\n * before it starts emitting results. This means if some Observable emits\n * values before other Observables started emitting, all these values but the last\n * will be lost. On the other hand, if some Observable does not emit a value but\n * completes, resulting Observable will complete at the same moment without\n * emitting anything, since it will now be impossible to include a value from the\n * completed Observable in the resulting array. Also, if some input Observable does\n * not emit any value and never completes, `combineLatest` will also never emit\n * and never complete, since, again, it will wait for all streams to emit some\n * value.\n *\n * If at least one Observable was passed to `combineLatest` and all passed Observables\n * emitted something, the resulting Observable will complete when all combined\n * streams complete. So even if some Observable completes, the result of\n * `combineLatest` will still emit values when other Observables do. In case\n * of a completed Observable, its value from now on will always be the last\n * emitted value. On the other hand, if any Observable errors, `combineLatest`\n * will error immediately as well, and all other Observables will be unsubscribed.\n *\n * ## Examples\n *\n * Combine two timer Observables\n *\n * ```ts\n * import { timer, combineLatest } from 'rxjs';\n *\n * const firstTimer = timer(0, 1000); // emit 0, 1, 2... after every second, starting from now\n * const secondTimer = timer(500, 1000); // emit 0, 1, 2... after every second, starting 0,5s from now\n * const combinedTimers = combineLatest([firstTimer, secondTimer]);\n * combinedTimers.subscribe(value => console.log(value));\n * // Logs\n * // [0, 0] after 0.5s\n * // [1, 0] after 1s\n * // [1, 1] after 1.5s\n * // [2, 1] after 2s\n * ```\n *\n * Combine a dictionary of Observables\n *\n * ```ts\n * import { of, delay, startWith, combineLatest } from 'rxjs';\n *\n * const observables = {\n *   a: of(1).pipe(delay(1000), startWith(0)),\n *   b: of(5).pipe(delay(5000), startWith(0)),\n *   c: of(10).pipe(delay(10000), startWith(0))\n * };\n * const combined = combineLatest(observables);\n * combined.subscribe(value => console.log(value));\n * // Logs\n * // { a: 0, b: 0, c: 0 } immediately\n * // { a: 1, b: 0, c: 0 } after 1s\n * // { a: 1, b: 5, c: 0 } after 5s\n * // { a: 1, b: 5, c: 10 } after 10s\n * ```\n *\n * Combine an array of Observables\n *\n * ```ts\n * import { of, delay, startWith, combineLatest } from 'rxjs';\n *\n * const observables = [1, 5, 10].map(\n *   n => of(n).pipe(\n *     delay(n * 1000), // emit 0 and then emit n after n seconds\n *     startWith(0)\n *   )\n * );\n * const combined = combineLatest(observables);\n * combined.subscribe(value => console.log(value));\n * // Logs\n * // [0, 0, 0] immediately\n * // [1, 0, 0] after 1s\n * // [1, 5, 0] after 5s\n * // [1, 5, 10] after 10s\n * ```\n *\n * Use map operator to dynamically calculate the Body-Mass Index\n *\n * ```ts\n * import { of, combineLatest, map } from 'rxjs';\n *\n * const weight = of(70, 72, 76, 79, 75);\n * const height = of(1.76, 1.77, 1.78);\n * const bmi = combineLatest([weight, height]).pipe(\n *   map(([w, h]) => w / (h * h)),\n * );\n * bmi.subscribe(x => console.log('BMI is ' + x));\n *\n * // With output to console:\n * // BMI is 24.212293388429753\n * // BMI is 23.93948099205209\n * // BMI is 23.671253629592222\n * ```\n *\n * @see {@link combineLatestAll}\n * @see {@link merge}\n * @see {@link withLatestFrom}\n *\n * @param {ObservableInput} [observables] An array of input Observables to combine with each other.\n * An array of Observables must be given as the first argument.\n * @param {function} [project] An optional function to project the values from\n * the combined latest values into a new value on the output Observable.\n * @param {SchedulerLike} [scheduler=null] The {@link SchedulerLike} to use for subscribing to\n * each input Observable.\n * @return {Observable} An Observable of projected values from the most recent\n * values from each input Observable, or an array of the most recent values from\n * each input Observable.\n */\nexport function combineLatest<O extends ObservableInput<any>, R>(...args: any[]): Observable<R> | Observable<ObservedValueOf<O>[]> {\n  const scheduler = popScheduler(args);\n  const resultSelector = popResultSelector(args);\n\n  const { args: observables, keys } = argsArgArrayOrObject(args);\n\n  if (observables.length === 0) {\n    // If no observables are passed, or someone has passed an empty array\n    // of observables, or even an empty object POJO, we need to just\n    // complete (EMPTY), but we have to honor the scheduler provided if any.\n    return from([], scheduler as any);\n  }\n\n  const result = new Observable<ObservedValueOf<O>[]>(\n    combineLatestInit(\n      observables as ObservableInput<ObservedValueOf<O>>[],\n      scheduler,\n      keys\n        ? // A handler for scrubbing the array of args into a dictionary.\n          (values) => createObject(keys, values)\n        : // A passthrough to just return the array\n          identity\n    )\n  );\n\n  return resultSelector ? (result.pipe(mapOneOrManyArgs(resultSelector)) as Observable<R>) : result;\n}\n\nexport function combineLatestInit(\n  observables: ObservableInput<any>[],\n  scheduler?: SchedulerLike,\n  valueTransform: (values: any[]) => any = identity\n) {\n  return (subscriber: Subscriber<any>) => {\n    // The outer subscription. We're capturing this in a function\n    // because we may have to schedule it.\n    maybeSchedule(\n      scheduler,\n      () => {\n        const { length } = observables;\n        // A store for the values each observable has emitted so far. We match observable to value on index.\n        const values = new Array(length);\n        // The number of currently active subscriptions, as they complete, we decrement this number to see if\n        // we are all done combining values, so we can complete the result.\n        let active = length;\n        // The number of inner sources that still haven't emitted the first value\n        // We need to track this because all sources need to emit one value in order\n        // to start emitting values.\n        let remainingFirstValues = length;\n        // The loop to kick off subscription. We're keying everything on index `i` to relate the observables passed\n        // in to the slot in the output array or the key in the array of keys in the output dictionary.\n        for (let i = 0; i < length; i++) {\n          maybeSchedule(\n            scheduler,\n            () => {\n              const source = from(observables[i], scheduler as any);\n              let hasFirstValue = false;\n              source.subscribe(\n                createOperatorSubscriber(\n                  subscriber,\n                  (value) => {\n                    // When we get a value, record it in our set of values.\n                    values[i] = value;\n                    if (!hasFirstValue) {\n                      // If this is our first value, record that.\n                      hasFirstValue = true;\n                      remainingFirstValues--;\n                    }\n                    if (!remainingFirstValues) {\n                      // We're not waiting for any more\n                      // first values, so we can emit!\n                      subscriber.next(valueTransform(values.slice()));\n                    }\n                  },\n                  () => {\n                    if (!--active) {\n                      // We only complete the result if we have no more active\n                      // inner observables.\n                      subscriber.complete();\n                    }\n                  }\n                )\n              );\n            },\n            subscriber\n          );\n        }\n      },\n      subscriber\n    );\n  };\n}\n\n/**\n * A small utility to handle the couple of locations where we want to schedule if a scheduler was provided,\n * but we don't if there was no scheduler.\n */\nfunction maybeSchedule(scheduler: SchedulerLike | undefined, execute: () => void, subscription: Subscription) {\n  if (scheduler) {\n    executeSchedule(subscription, scheduler, execute);\n  } else {\n    execute();\n  }\n}\n", "import { Observable } from '../Observable';\nimport { innerFrom } from '../observable/innerFrom';\nimport { Subscriber } from '../Subscriber';\nimport { ObservableInput, SchedulerLike } from '../types';\nimport { executeSchedule } from '../util/executeSchedule';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * A process embodying the general \"merge\" strategy. This is used in\n * `mergeMap` and `mergeScan` because the logic is otherwise nearly identical.\n * @param source The original source observable\n * @param subscriber The consumer subscriber\n * @param project The projection function to get our inner sources\n * @param concurrent The number of concurrent inner subscriptions\n * @param onBeforeNext Additional logic to apply before nexting to our consumer\n * @param expand If `true` this will perform an \"expand\" strategy, which differs only\n * in that it recurses, and the inner subscription must be schedule-able.\n * @param innerSubScheduler A scheduler to use to schedule inner subscriptions,\n * this is to support the expand strategy, mostly, and should be deprecated\n */\nexport function mergeInternals<T, R>(\n  source: Observable<T>,\n  subscriber: Subscriber<R>,\n  project: (value: T, index: number) => ObservableInput<R>,\n  concurrent: number,\n  onBeforeNext?: (innerValue: R) => void,\n  expand?: boolean,\n  innerSubScheduler?: SchedulerLike,\n  additionalFinalizer?: () => void\n) {\n  // Buffered values, in the event of going over our concurrency limit\n  const buffer: T[] = [];\n  // The number of active inner subscriptions.\n  let active = 0;\n  // An index to pass to our accumulator function\n  let index = 0;\n  // Whether or not the outer source has completed.\n  let isComplete = false;\n\n  /**\n   * Checks to see if we can complete our result or not.\n   */\n  const checkComplete = () => {\n    // If the outer has completed, and nothing is left in the buffer,\n    // and we don't have any active inner subscriptions, then we can\n    // Emit the state and complete.\n    if (isComplete && !buffer.length && !active) {\n      subscriber.complete();\n    }\n  };\n\n  // If we're under our concurrency limit, just start the inner subscription, otherwise buffer and wait.\n  const outerNext = (value: T) => (active < concurrent ? doInnerSub(value) : buffer.push(value));\n\n  const doInnerSub = (value: T) => {\n    // If we're expanding, we need to emit the outer values and the inner values\n    // as the inners will \"become outers\" in a way as they are recursively fed\n    // back to the projection mechanism.\n    expand && subscriber.next(value as any);\n\n    // Increment the number of active subscriptions so we can track it\n    // against our concurrency limit later.\n    active++;\n\n    // A flag used to show that the inner observable completed.\n    // This is checked during finalization to see if we should\n    // move to the next item in the buffer, if there is on.\n    let innerComplete = false;\n\n    // Start our inner subscription.\n    innerFrom(project(value, index++)).subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (innerValue) => {\n          // `mergeScan` has additional handling here. For example\n          // taking the inner value and updating state.\n          onBeforeNext?.(innerValue);\n\n          if (expand) {\n            // If we're expanding, then just recurse back to our outer\n            // handler. It will emit the value first thing.\n            outerNext(innerValue as any);\n          } else {\n            // Otherwise, emit the inner value.\n            subscriber.next(innerValue);\n          }\n        },\n        () => {\n          // Flag that we have completed, so we know to check the buffer\n          // during finalization.\n          innerComplete = true;\n        },\n        // Errors are passed to the destination.\n        undefined,\n        () => {\n          // During finalization, if the inner completed (it wasn't errored or\n          // cancelled), then we want to try the next item in the buffer if\n          // there is one.\n          if (innerComplete) {\n            // We have to wrap this in a try/catch because it happens during\n            // finalization, possibly asynchronously, and we want to pass\n            // any errors that happen (like in a projection function) to\n            // the outer Subscriber.\n            try {\n              // INNER SOURCE COMPLETE\n              // Decrement the active count to ensure that the next time\n              // we try to call `doInnerSub`, the number is accurate.\n              active--;\n              // If we have more values in the buffer, try to process those\n              // Note that this call will increment `active` ahead of the\n              // next conditional, if there were any more inner subscriptions\n              // to start.\n              while (buffer.length && active < concurrent) {\n                const bufferedValue = buffer.shift()!;\n                // Particularly for `expand`, we need to check to see if a scheduler was provided\n                // for when we want to start our inner subscription. Otherwise, we just start\n                // are next inner subscription.\n                if (innerSubScheduler) {\n                  executeSchedule(subscriber, innerSubScheduler, () => doInnerSub(bufferedValue));\n                } else {\n                  doInnerSub(bufferedValue);\n                }\n              }\n              // Check to see if we can complete, and complete if so.\n              checkComplete();\n            } catch (err) {\n              subscriber.error(err);\n            }\n          }\n        }\n      )\n    );\n  };\n\n  // Subscribe to our source observable.\n  source.subscribe(\n    createOperatorSubscriber(subscriber, outerNext, () => {\n      // Outer completed, make a note of it, and check to see if we can complete everything.\n      isComplete = true;\n      checkComplete();\n    })\n  );\n\n  // Additional finalization (for when the destination is torn down).\n  // Other finalization is added implicitly via subscription above.\n  return () => {\n    additionalFinalizer?.();\n  };\n}\n", "import { ObservableInput, OperatorFunction, ObservedValueOf } from '../types';\nimport { map } from './map';\nimport { innerFrom } from '../observable/innerFrom';\nimport { operate } from '../util/lift';\nimport { mergeInternals } from './mergeInternals';\nimport { isFunction } from '../util/isFunction';\n\n/* tslint:disable:max-line-length */\nexport function mergeMap<T, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O,\n  concurrent?: number\n): OperatorFunction<T, ObservedValueOf<O>>;\n/** @deprecated The `resultSelector` parameter will be removed in v8. Use an inner `map` instead. Details: https://rxjs.dev/deprecations/resultSelector */\nexport function mergeMap<T, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O,\n  resultSelector: undefined,\n  concurrent?: number\n): OperatorFunction<T, ObservedValueOf<O>>;\n/** @deprecated The `resultSelector` parameter will be removed in v8. Use an inner `map` instead. Details: https://rxjs.dev/deprecations/resultSelector */\nexport function mergeMap<T, R, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O,\n  resultSelector: (outerValue: T, innerValue: ObservedValueOf<O>, outerIndex: number, innerIndex: number) => R,\n  concurrent?: number\n): OperatorFunction<T, R>;\n/* tslint:enable:max-line-length */\n\n/**\n * Projects each source value to an Observable which is merged in the output\n * Observable.\n *\n * <span class=\"informal\">Maps each value to an Observable, then flattens all of\n * these inner Observables using {@link mergeAll}.</span>\n *\n * ![](mergeMap.png)\n *\n * Returns an Observable that emits items based on applying a function that you\n * supply to each item emitted by the source Observable, where that function\n * returns an Observable, and then merging those resulting Observables and\n * emitting the results of this merger.\n *\n * ## Example\n *\n * Map and flatten each letter to an Observable ticking every 1 second\n *\n * ```ts\n * import { of, mergeMap, interval, map } from 'rxjs';\n *\n * const letters = of('a', 'b', 'c');\n * const result = letters.pipe(\n *   mergeMap(x => interval(1000).pipe(map(i => x + i)))\n * );\n *\n * result.subscribe(x => console.log(x));\n *\n * // Results in the following:\n * // a0\n * // b0\n * // c0\n * // a1\n * // b1\n * // c1\n * // continues to list a, b, c every second with respective ascending integers\n * ```\n *\n * @see {@link concatMap}\n * @see {@link exhaustMap}\n * @see {@link merge}\n * @see {@link mergeAll}\n * @see {@link mergeMapTo}\n * @see {@link mergeScan}\n * @see {@link switchMap}\n *\n * @param {function(value: T, ?index: number): ObservableInput} project A function\n * that, when applied to an item emitted by the source Observable, returns an\n * Observable.\n * @param {number} [concurrent=Infinity] Maximum number of input\n * Observables being subscribed to concurrently.\n * @return A function that returns an Observable that emits the result of\n * applying the projection function (and the optional deprecated\n * `resultSelector`) to each item emitted by the source Observable and merging\n * the results of the Observables obtained from this transformation.\n */\nexport function mergeMap<T, R, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O,\n  resultSelector?: ((outerValue: T, innerValue: ObservedValueOf<O>, outerIndex: number, innerIndex: number) => R) | number,\n  concurrent: number = Infinity\n): OperatorFunction<T, ObservedValueOf<O> | R> {\n  if (isFunction(resultSelector)) {\n    // DEPRECATED PATH\n    return mergeMap((a, i) => map((b: any, ii: number) => resultSelector(a, b, i, ii))(innerFrom(project(a, i))), concurrent);\n  } else if (typeof resultSelector === 'number') {\n    concurrent = resultSelector;\n  }\n\n  return operate((source, subscriber) => mergeInternals(source, subscriber, project, concurrent));\n}\n", "import { mergeMap } from './mergeMap';\nimport { identity } from '../util/identity';\nimport { OperatorFunction, ObservableInput, ObservedValueOf } from '../types';\n\n/**\n * Converts a higher-order Observable into a first-order Observable which\n * concurrently delivers all values that are emitted on the inner Observables.\n *\n * <span class=\"informal\">Flattens an Observable-of-Observables.</span>\n *\n * ![](mergeAll.png)\n *\n * `mergeAll` subscribes to an Observable that emits Observables, also known as\n * a higher-order Observable. Each time it observes one of these emitted inner\n * Observables, it subscribes to that and delivers all the values from the\n * inner Observable on the output Observable. The output Observable only\n * completes once all inner Observables have completed. Any error delivered by\n * a inner Observable will be immediately emitted on the output Observable.\n *\n * ## Examples\n *\n * Spawn a new interval Observable for each click event, and blend their outputs as one Observable\n *\n * ```ts\n * import { fromEvent, map, interval, mergeAll } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const higherOrder = clicks.pipe(map(() => interval(1000)));\n * const firstOrder = higherOrder.pipe(mergeAll());\n *\n * firstOrder.subscribe(x => console.log(x));\n * ```\n *\n * Count from 0 to 9 every second for each click, but only allow 2 concurrent timers\n *\n * ```ts\n * import { fromEvent, map, interval, take, mergeAll } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const higherOrder = clicks.pipe(\n *   map(() => interval(1000).pipe(take(10)))\n * );\n * const firstOrder = higherOrder.pipe(mergeAll(2));\n *\n * firstOrder.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link combineLatestAll}\n * @see {@link concatAll}\n * @see {@link exhaustAll}\n * @see {@link merge}\n * @see {@link mergeMap}\n * @see {@link mergeMapTo}\n * @see {@link mergeScan}\n * @see {@link switchAll}\n * @see {@link switchMap}\n * @see {@link zipAll}\n *\n * @param {number} [concurrent=Infinity] Maximum number of inner\n * Observables being subscribed to concurrently.\n * @return A function that returns an Observable that emits values coming from\n * all the inner Observables emitted by the source Observable.\n */\nexport function mergeAll<O extends ObservableInput<any>>(concurrent: number = Infinity): OperatorFunction<O, ObservedValueOf<O>> {\n  return mergeMap(identity, concurrent);\n}\n", "import { mergeAll } from './mergeAll';\nimport { OperatorFunction, ObservableInput, ObservedValueOf } from '../types';\n\n/**\n * Converts a higher-order Observable into a first-order Observable by\n * concatenating the inner Observables in order.\n *\n * <span class=\"informal\">Flattens an Observable-of-Observables by putting one\n * inner Observable after the other.</span>\n *\n * ![](concatAll.svg)\n *\n * Joins every Observable emitted by the source (a higher-order Observable), in\n * a serial fashion. It subscribes to each inner Observable only after the\n * previous inner Observable has completed, and merges all of their values into\n * the returned observable.\n *\n * __Warning:__ If the source Observable emits Observables quickly and\n * endlessly, and the inner Observables it emits generally complete slower than\n * the source emits, you can run into memory issues as the incoming Observables\n * collect in an unbounded buffer.\n *\n * Note: `concatAll` is equivalent to `mergeAll` with concurrency parameter set\n * to `1`.\n *\n * ## Example\n *\n * For each click event, tick every second from 0 to 3, with no concurrency\n *\n * ```ts\n * import { fromEvent, map, interval, take, concatAll } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const higherOrder = clicks.pipe(\n *   map(() => interval(1000).pipe(take(4)))\n * );\n * const firstOrder = higherOrder.pipe(concatAll());\n * firstOrder.subscribe(x => console.log(x));\n *\n * // Results in the following:\n * // (results are not concurrent)\n * // For every click on the \"document\" it will emit values 0 to 3 spaced\n * // on a 1000ms interval\n * // one click = 1000ms-> 0 -1000ms-> 1 -1000ms-> 2 -1000ms-> 3\n * ```\n *\n * @see {@link combineLatestAll}\n * @see {@link concat}\n * @see {@link concatMap}\n * @see {@link concatMapTo}\n * @see {@link exhaustAll}\n * @see {@link mergeAll}\n * @see {@link switchAll}\n * @see {@link switchMap}\n * @see {@link zipAll}\n *\n * @return A function that returns an Observable emitting values from all the\n * inner Observables concatenated.\n */\nexport function concatAll<O extends ObservableInput<any>>(): OperatorFunction<O, ObservedValueOf<O>> {\n  return mergeAll(1);\n}\n", "import { Observable } from '../Observable';\nimport { ObservableInputTuple, SchedulerLike } from '../types';\nimport { concatAll } from '../operators/concatAll';\nimport { popScheduler } from '../util/args';\nimport { from } from './from';\n\nexport function concat<T extends readonly unknown[]>(...inputs: [...ObservableInputTuple<T>]): Observable<T[number]>;\nexport function concat<T extends readonly unknown[]>(\n  ...inputsAndScheduler: [...ObservableInputTuple<T>, SchedulerLike]\n): Observable<T[number]>;\n\n/**\n * Creates an output Observable which sequentially emits all values from the first given\n * Observable and then moves on to the next.\n *\n * <span class=\"informal\">Concatenates multiple Observables together by\n * sequentially emitting their values, one Observable after the other.</span>\n *\n * ![](concat.png)\n *\n * `concat` joins multiple Observables together, by subscribing to them one at a time and\n * merging their results into the output Observable. You can pass either an array of\n * Observables, or put them directly as arguments. Passing an empty array will result\n * in Observable that completes immediately.\n *\n * `concat` will subscribe to first input Observable and emit all its values, without\n * changing or affecting them in any way. When that Observable completes, it will\n * subscribe to then next Observable passed and, again, emit its values. This will be\n * repeated, until the operator runs out of Observables. When last input Observable completes,\n * `concat` will complete as well. At any given moment only one Observable passed to operator\n * emits values. If you would like to emit values from passed Observables concurrently, check out\n * {@link merge} instead, especially with optional `concurrent` parameter. As a matter of fact,\n * `concat` is an equivalent of `merge` operator with `concurrent` parameter set to `1`.\n *\n * Note that if some input Observable never completes, `concat` will also never complete\n * and Observables following the one that did not complete will never be subscribed. On the other\n * hand, if some Observable simply completes immediately after it is subscribed, it will be\n * invisible for `concat`, which will just move on to the next Observable.\n *\n * If any Observable in chain errors, instead of passing control to the next Observable,\n * `concat` will error immediately as well. Observables that would be subscribed after\n * the one that emitted error, never will.\n *\n * If you pass to `concat` the same Observable many times, its stream of values\n * will be \"replayed\" on every subscription, which means you can repeat given Observable\n * as many times as you like. If passing the same Observable to `concat` 1000 times becomes tedious,\n * you can always use {@link repeat}.\n *\n * ## Examples\n *\n * Concatenate a timer counting from 0 to 3 with a synchronous sequence from 1 to 10\n *\n * ```ts\n * import { interval, take, range, concat } from 'rxjs';\n *\n * const timer = interval(1000).pipe(take(4));\n * const sequence = range(1, 10);\n * const result = concat(timer, sequence);\n * result.subscribe(x => console.log(x));\n *\n * // results in:\n * // 0 -1000ms-> 1 -1000ms-> 2 -1000ms-> 3 -immediate-> 1 ... 10\n * ```\n *\n * Concatenate 3 Observables\n *\n * ```ts\n * import { interval, take, concat } from 'rxjs';\n *\n * const timer1 = interval(1000).pipe(take(10));\n * const timer2 = interval(2000).pipe(take(6));\n * const timer3 = interval(500).pipe(take(10));\n *\n * const result = concat(timer1, timer2, timer3);\n * result.subscribe(x => console.log(x));\n *\n * // results in the following:\n * // (Prints to console sequentially)\n * // -1000ms-> 0 -1000ms-> 1 -1000ms-> ... 9\n * // -2000ms-> 0 -2000ms-> 1 -2000ms-> ... 5\n * // -500ms-> 0 -500ms-> 1 -500ms-> ... 9\n * ```\n *\n * Concatenate the same Observable to repeat it\n *\n * ```ts\n * import { interval, take, concat } from 'rxjs';\n *\n * const timer = interval(1000).pipe(take(2));\n *\n * concat(timer, timer) // concatenating the same Observable!\n *   .subscribe({\n *     next: value => console.log(value),\n *     complete: () => console.log('...and it is done!')\n *   });\n *\n * // Logs:\n * // 0 after 1s\n * // 1 after 2s\n * // 0 after 3s\n * // 1 after 4s\n * // '...and it is done!' also after 4s\n * ```\n *\n * @see {@link concatAll}\n * @see {@link concatMap}\n * @see {@link concatMapTo}\n * @see {@link startWith}\n * @see {@link endWith}\n *\n * @param args Input Observables to concatenate.\n */\nexport function concat(...args: any[]): Observable<unknown> {\n  return concatAll()(from(args, popScheduler(args)));\n}\n", "import { Observable } from '../Observable';\nimport { ObservedValueOf, ObservableInput } from '../types';\nimport { innerFrom } from './innerFrom';\n\n/**\n * Creates an Observable that, on subscribe, calls an Observable factory to\n * make an Observable for each new Observer.\n *\n * <span class=\"informal\">Creates the Observable lazily, that is, only when it\n * is subscribed.\n * </span>\n *\n * ![](defer.png)\n *\n * `defer` allows you to create an Observable only when the Observer\n * subscribes. It waits until an Observer subscribes to it, calls the given\n * factory function to get an Observable -- where a factory function typically\n * generates a new Observable -- and subscribes the Observer to this Observable.\n * In case the factory function returns a falsy value, then EMPTY is used as\n * Observable instead. Last but not least, an exception during the factory\n * function call is transferred to the Observer by calling `error`.\n *\n * ## Example\n *\n * Subscribe to either an Observable of clicks or an Observable of interval, at random\n *\n * ```ts\n * import { defer, fromEvent, interval } from 'rxjs';\n *\n * const clicksOrInterval = defer(() => {\n *   return Math.random() > 0.5\n *     ? fromEvent(document, 'click')\n *     : interval(1000);\n * });\n * clicksOrInterval.subscribe(x => console.log(x));\n *\n * // Results in the following behavior:\n * // If the result of Math.random() is greater than 0.5 it will listen\n * // for clicks anywhere on the \"document\"; when document is clicked it\n * // will log a MouseEvent object to the console. If the result is less\n * // than 0.5 it will emit ascending numbers, one every second(1000ms).\n * ```\n *\n * @see {@link Observable}\n *\n * @param {function(): ObservableInput} observableFactory The Observable\n * factory function to invoke for each Observer that subscribes to the output\n * Observable. May also return a Promise, which will be converted on the fly\n * to an Observable.\n * @return {Observable} An Observable whose Observers' subscriptions trigger\n * an invocation of the given Observable factory function.\n */\nexport function defer<R extends ObservableInput<any>>(observableFactory: () => R): Observable<ObservedValueOf<R>> {\n  return new Observable<ObservedValueOf<R>>((subscriber) => {\n    innerFrom(observableFactory()).subscribe(subscriber);\n  });\n}\n", "import { innerFrom } from '../observable/innerFrom';\nimport { Observable } from '../Observable';\nimport { mergeMap } from '../operators/mergeMap';\nimport { isArrayLike } from '../util/isArrayLike';\nimport { isFunction } from '../util/isFunction';\nimport { mapOneOrManyArgs } from '../util/mapOneOrManyArgs';\n\n// These constants are used to create handler registry functions using array mapping below.\nconst nodeEventEmitterMethods = ['addListener', 'removeListener'] as const;\nconst eventTargetMethods = ['addEventListener', 'removeEventListener'] as const;\nconst jqueryMethods = ['on', 'off'] as const;\n\nexport interface NodeStyleEventEmitter {\n  addListener(eventName: string | symbol, handler: NodeEventHandler): this;\n  removeListener(eventName: string | symbol, handler: NodeEventHandler): this;\n}\n\nexport type NodeEventHandler = (...args: any[]) => void;\n\n// For APIs that implement `addListener` and `removeListener` methods that may\n// not use the same arguments or return EventEmitter values\n// such as React Native\nexport interface NodeCompatibleEventEmitter {\n  addListener(eventName: string, handler: NodeEventHandler): void | {};\n  removeListener(eventName: string, handler: NodeEventHandler): void | {};\n}\n\n// Use handler types like those in @types/jquery. See:\n// https://github.com/DefinitelyTyped/DefinitelyTyped/blob/847731ba1d7fa6db6b911c0e43aa0afe596e7723/types/jquery/misc.d.ts#L6395\nexport interface JQueryStyleEventEmitter<TContext, T> {\n  on(eventName: string, handler: (this: TContext, t: T, ...args: any[]) => any): void;\n  off(eventName: string, handler: (this: TContext, t: T, ...args: any[]) => any): void;\n}\n\nexport interface EventListenerObject<E> {\n  handleEvent(evt: E): void;\n}\n\nexport interface HasEventTargetAddRemove<E> {\n  addEventListener(\n    type: string,\n    listener: ((evt: E) => void) | EventListenerObject<E> | null,\n    options?: boolean | AddEventListenerOptions\n  ): void;\n  removeEventListener(\n    type: string,\n    listener: ((evt: E) => void) | EventListenerObject<E> | null,\n    options?: EventListenerOptions | boolean\n  ): void;\n}\n\nexport interface EventListenerOptions {\n  capture?: boolean;\n  passive?: boolean;\n  once?: boolean;\n}\n\nexport interface AddEventListenerOptions extends EventListenerOptions {\n  once?: boolean;\n  passive?: boolean;\n}\n\nexport function fromEvent<T>(target: HasEventTargetAddRemove<T> | ArrayLike<HasEventTargetAddRemove<T>>, eventName: string): Observable<T>;\nexport function fromEvent<T, R>(\n  target: HasEventTargetAddRemove<T> | ArrayLike<HasEventTargetAddRemove<T>>,\n  eventName: string,\n  resultSelector: (event: T) => R\n): Observable<R>;\nexport function fromEvent<T>(\n  target: HasEventTargetAddRemove<T> | ArrayLike<HasEventTargetAddRemove<T>>,\n  eventName: string,\n  options: EventListenerOptions\n): Observable<T>;\nexport function fromEvent<T, R>(\n  target: HasEventTargetAddRemove<T> | ArrayLike<HasEventTargetAddRemove<T>>,\n  eventName: string,\n  options: EventListenerOptions,\n  resultSelector: (event: T) => R\n): Observable<R>;\n\nexport function fromEvent(target: NodeStyleEventEmitter | ArrayLike<NodeStyleEventEmitter>, eventName: string): Observable<unknown>;\n/** @deprecated Do not specify explicit type parameters. Signatures with type parameters that cannot be inferred will be removed in v8. */\nexport function fromEvent<T>(target: NodeStyleEventEmitter | ArrayLike<NodeStyleEventEmitter>, eventName: string): Observable<T>;\nexport function fromEvent<R>(\n  target: NodeStyleEventEmitter | ArrayLike<NodeStyleEventEmitter>,\n  eventName: string,\n  resultSelector: (...args: any[]) => R\n): Observable<R>;\n\nexport function fromEvent(\n  target: NodeCompatibleEventEmitter | ArrayLike<NodeCompatibleEventEmitter>,\n  eventName: string\n): Observable<unknown>;\n/** @deprecated Do not specify explicit type parameters. Signatures with type parameters that cannot be inferred will be removed in v8. */\nexport function fromEvent<T>(target: NodeCompatibleEventEmitter | ArrayLike<NodeCompatibleEventEmitter>, eventName: string): Observable<T>;\nexport function fromEvent<R>(\n  target: NodeCompatibleEventEmitter | ArrayLike<NodeCompatibleEventEmitter>,\n  eventName: string,\n  resultSelector: (...args: any[]) => R\n): Observable<R>;\n\nexport function fromEvent<T>(\n  target: JQueryStyleEventEmitter<any, T> | ArrayLike<JQueryStyleEventEmitter<any, T>>,\n  eventName: string\n): Observable<T>;\nexport function fromEvent<T, R>(\n  target: JQueryStyleEventEmitter<any, T> | ArrayLike<JQueryStyleEventEmitter<any, T>>,\n  eventName: string,\n  resultSelector: (value: T, ...args: any[]) => R\n): Observable<R>;\n\n/**\n * Creates an Observable that emits events of a specific type coming from the\n * given event target.\n *\n * <span class=\"informal\">Creates an Observable from DOM events, or Node.js\n * EventEmitter events or others.</span>\n *\n * ![](fromEvent.png)\n *\n * `fromEvent` accepts as a first argument event target, which is an object with methods\n * for registering event handler functions. As a second argument it takes string that indicates\n * type of event we want to listen for. `fromEvent` supports selected types of event targets,\n * which are described in detail below. If your event target does not match any of the ones listed,\n * you should use {@link fromEventPattern}, which can be used on arbitrary APIs.\n * When it comes to APIs supported by `fromEvent`, their methods for adding and removing event\n * handler functions have different names, but they all accept a string describing event type\n * and function itself, which will be called whenever said event happens.\n *\n * Every time resulting Observable is subscribed, event handler function will be registered\n * to event target on given event type. When that event fires, value\n * passed as a first argument to registered function will be emitted by output Observable.\n * When Observable is unsubscribed, function will be unregistered from event target.\n *\n * Note that if event target calls registered function with more than one argument, second\n * and following arguments will not appear in resulting stream. In order to get access to them,\n * you can pass to `fromEvent` optional project function, which will be called with all arguments\n * passed to event handler. Output Observable will then emit value returned by project function,\n * instead of the usual value.\n *\n * Remember that event targets listed below are checked via duck typing. It means that\n * no matter what kind of object you have and no matter what environment you work in,\n * you can safely use `fromEvent` on that object if it exposes described methods (provided\n * of course they behave as was described above). So for example if Node.js library exposes\n * event target which has the same method names as DOM EventTarget, `fromEvent` is still\n * a good choice.\n *\n * If the API you use is more callback then event handler oriented (subscribed\n * callback function fires only once and thus there is no need to manually\n * unregister it), you should use {@link bindCallback} or {@link bindNodeCallback}\n * instead.\n *\n * `fromEvent` supports following types of event targets:\n *\n * **DOM EventTarget**\n *\n * This is an object with `addEventListener` and `removeEventListener` methods.\n *\n * In the browser, `addEventListener` accepts - apart from event type string and event\n * handler function arguments - optional third parameter, which is either an object or boolean,\n * both used for additional configuration how and when passed function will be called. When\n * `fromEvent` is used with event target of that type, you can provide this values\n * as third parameter as well.\n *\n * **Node.js EventEmitter**\n *\n * An object with `addListener` and `removeListener` methods.\n *\n * **JQuery-style event target**\n *\n * An object with `on` and `off` methods\n *\n * **DOM NodeList**\n *\n * List of DOM Nodes, returned for example by `document.querySelectorAll` or `Node.childNodes`.\n *\n * Although this collection is not event target in itself, `fromEvent` will iterate over all Nodes\n * it contains and install event handler function in every of them. When returned Observable\n * is unsubscribed, function will be removed from all Nodes.\n *\n * **DOM HtmlCollection**\n *\n * Just as in case of NodeList it is a collection of DOM nodes. Here as well event handler function is\n * installed and removed in each of elements.\n *\n *\n * ## Examples\n *\n * Emit clicks happening on the DOM document\n *\n * ```ts\n * import { fromEvent } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * clicks.subscribe(x => console.log(x));\n *\n * // Results in:\n * // MouseEvent object logged to console every time a click\n * // occurs on the document.\n * ```\n *\n * Use `addEventListener` with capture option\n *\n * ```ts\n * import { fromEvent } from 'rxjs';\n *\n * const div = document.createElement('div');\n * div.style.cssText = 'width: 200px; height: 200px; background: #09c;';\n * document.body.appendChild(div);\n *\n * // note optional configuration parameter which will be passed to addEventListener\n * const clicksInDocument = fromEvent(document, 'click', { capture: true });\n * const clicksInDiv = fromEvent(div, 'click');\n *\n * clicksInDocument.subscribe(() => console.log('document'));\n * clicksInDiv.subscribe(() => console.log('div'));\n *\n * // By default events bubble UP in DOM tree, so normally\n * // when we would click on div in document\n * // \"div\" would be logged first and then \"document\".\n * // Since we specified optional `capture` option, document\n * // will catch event when it goes DOWN DOM tree, so console\n * // will log \"document\" and then \"div\".\n * ```\n *\n * @see {@link bindCallback}\n * @see {@link bindNodeCallback}\n * @see {@link fromEventPattern}\n *\n * @param {FromEventTarget<T>} target The DOM EventTarget, Node.js\n * EventEmitter, JQuery-like event target, NodeList or HTMLCollection to attach the event handler to.\n * @param {string} eventName The event name of interest, being emitted by the\n * `target`.\n * @param {EventListenerOptions} [options] Options to pass through to addEventListener\n * @return {Observable<T>}\n */\nexport function fromEvent<T>(\n  target: any,\n  eventName: string,\n  options?: EventListenerOptions | ((...args: any[]) => T),\n  resultSelector?: (...args: any[]) => T\n): Observable<T> {\n  if (isFunction(options)) {\n    resultSelector = options;\n    options = undefined;\n  }\n  if (resultSelector) {\n    return fromEvent<T>(target, eventName, options as EventListenerOptions).pipe(mapOneOrManyArgs(resultSelector));\n  }\n\n  // Figure out our add and remove methods. In order to do this,\n  // we are going to analyze the target in a preferred order, if\n  // the target matches a given signature, we take the two \"add\" and \"remove\"\n  // method names and apply them to a map to create opposite versions of the\n  // same function. This is because they all operate in duplicate pairs,\n  // `addListener(name, handler)`, `removeListener(name, handler)`, for example.\n  // The call only differs by method name, as to whether or not you're adding or removing.\n  const [add, remove] =\n    // If it is an EventTarget, we need to use a slightly different method than the other two patterns.\n    isEventTarget(target)\n      ? eventTargetMethods.map((methodName) => (handler: any) => target[methodName](eventName, handler, options as EventListenerOptions))\n      : // In all other cases, the call pattern is identical with the exception of the method names.\n      isNodeStyleEventEmitter(target)\n      ? nodeEventEmitterMethods.map(toCommonHandlerRegistry(target, eventName))\n      : isJQueryStyleEventEmitter(target)\n      ? jqueryMethods.map(toCommonHandlerRegistry(target, eventName))\n      : [];\n\n  // If add is falsy, it's because we didn't match a pattern above.\n  // Check to see if it is an ArrayLike, because if it is, we want to\n  // try to apply fromEvent to all of it's items. We do this check last,\n  // because there are may be some types that are both ArrayLike *and* implement\n  // event registry points, and we'd rather delegate to that when possible.\n  if (!add) {\n    if (isArrayLike(target)) {\n      return mergeMap((subTarget: any) => fromEvent(subTarget, eventName, options as EventListenerOptions))(\n        innerFrom(target)\n      ) as Observable<T>;\n    }\n  }\n\n  // If add is falsy and we made it here, it's because we didn't\n  // match any valid target objects above.\n  if (!add) {\n    throw new TypeError('Invalid event target');\n  }\n\n  return new Observable<T>((subscriber) => {\n    // The handler we are going to register. Forwards the event object, by itself, or\n    // an array of arguments to the event handler, if there is more than one argument,\n    // to the consumer.\n    const handler = (...args: any[]) => subscriber.next(1 < args.length ? args : args[0]);\n    // Do the work of adding the handler to the target.\n    add(handler);\n    // When we finalize, we want to remove the handler and free up memory.\n    return () => remove!(handler);\n  });\n}\n\n/**\n * Used to create `add` and `remove` functions to register and unregister event handlers\n * from a target in the most common handler pattern, where there are only two arguments.\n * (e.g.  `on(name, fn)`, `off(name, fn)`, `addListener(name, fn)`, or `removeListener(name, fn)`)\n * @param target The target we're calling methods on\n * @param eventName The event name for the event we're creating register or unregister functions for\n */\nfunction toCommonHandlerRegistry(target: any, eventName: string) {\n  return (methodName: string) => (handler: any) => target[methodName](eventName, handler);\n}\n\n/**\n * Checks to see if the target implements the required node-style EventEmitter methods\n * for adding and removing event handlers.\n * @param target the object to check\n */\nfunction isNodeStyleEventEmitter(target: any): target is NodeStyleEventEmitter {\n  return isFunction(target.addListener) && isFunction(target.removeListener);\n}\n\n/**\n * Checks to see if the target implements the required jQuery-style EventEmitter methods\n * for adding and removing event handlers.\n * @param target the object to check\n */\nfunction isJQueryStyleEventEmitter(target: any): target is JQueryStyleEventEmitter<any, any> {\n  return isFunction(target.on) && isFunction(target.off);\n}\n\n/**\n * Checks to see if the target implements the required EventTarget methods\n * for adding and removing event handlers.\n * @param target the object to check\n */\nfunction isEventTarget(target: any): target is HasEventTargetAddRemove<any> {\n  return isFunction(target.addEventListener) && isFunction(target.removeEventListener);\n}\n", "import { Observable } from '../Observable';\nimport { isFunction } from '../util/isFunction';\nimport { NodeEventHandler } from './fromEvent';\nimport { mapOneOrManyArgs } from '../util/mapOneOrManyArgs';\n\n/* tslint:disable:max-line-length */\nexport function fromEventPattern<T>(\n  addHandler: (handler: NodeEventHandler) => any,\n  removeHandler?: (handler: NodeEventHandler, signal?: any) => void\n): Observable<T>;\nexport function fromEventPattern<T>(\n  addHandler: (handler: NodeEventHandler) => any,\n  removeHandler?: (handler: NodeEventHandler, signal?: any) => void,\n  resultSelector?: (...args: any[]) => T\n): Observable<T>;\n/* tslint:enable:max-line-length */\n\n/**\n * Creates an Observable from an arbitrary API for registering event handlers.\n *\n * <span class=\"informal\">When that method for adding event handler was something {@link fromEvent}\n * was not prepared for.</span>\n *\n * ![](fromEventPattern.png)\n *\n * `fromEventPattern` allows you to convert into an Observable any API that supports registering handler functions\n * for events. It is similar to {@link fromEvent}, but far\n * more flexible. In fact, all use cases of {@link fromEvent} could be easily handled by\n * `fromEventPattern` (although in slightly more verbose way).\n *\n * This operator accepts as a first argument an `addHandler` function, which will be injected with\n * handler parameter. That handler is actually an event handler function that you now can pass\n * to API expecting it. `addHandler` will be called whenever Observable\n * returned by the operator is subscribed, so registering handler in API will not\n * necessarily happen when `fromEventPattern` is called.\n *\n * After registration, every time an event that we listen to happens,\n * Observable returned by `fromEventPattern` will emit value that event handler\n * function was called with. Note that if event handler was called with more\n * than one argument, second and following arguments will not appear in the Observable.\n *\n * If API you are using allows to unregister event handlers as well, you can pass to `fromEventPattern`\n * another function - `removeHandler` - as a second parameter. It will be injected\n * with the same handler function as before, which now you can use to unregister\n * it from the API. `removeHandler` will be called when consumer of resulting Observable\n * unsubscribes from it.\n *\n * In some APIs unregistering is actually handled differently. Method registering an event handler\n * returns some kind of token, which is later used to identify which function should\n * be unregistered or it itself has method that unregisters event handler.\n * If that is the case with your API, make sure token returned\n * by registering method is returned by `addHandler`. Then it will be passed\n * as a second argument to `removeHandler`, where you will be able to use it.\n *\n * If you need access to all event handler parameters (not only the first one),\n * or you need to transform them in any way, you can call `fromEventPattern` with optional\n * third parameter - project function which will accept all arguments passed to\n * event handler when it is called. Whatever is returned from project function will appear on\n * resulting stream instead of usual event handlers first argument. This means\n * that default project can be thought of as function that takes its first parameter\n * and ignores the rest.\n *\n * ## Examples\n *\n * Emits clicks happening on the DOM document\n *\n * ```ts\n * import { fromEventPattern } from 'rxjs';\n *\n * function addClickHandler(handler) {\n *   document.addEventListener('click', handler);\n * }\n *\n * function removeClickHandler(handler) {\n *   document.removeEventListener('click', handler);\n * }\n *\n * const clicks = fromEventPattern(\n *   addClickHandler,\n *   removeClickHandler\n * );\n * clicks.subscribe(x => console.log(x));\n *\n * // Whenever you click anywhere in the browser, DOM MouseEvent\n * // object will be logged.\n * ```\n *\n * Use with API that returns cancellation token\n *\n * ```ts\n * import { fromEventPattern } from 'rxjs';\n *\n * const token = someAPI.registerEventHandler(function() {});\n * someAPI.unregisterEventHandler(token); // this APIs cancellation method accepts\n *                                        // not handler itself, but special token.\n *\n * const someAPIObservable = fromEventPattern(\n *   function(handler) { return someAPI.registerEventHandler(handler); }, // Note that we return the token here...\n *   function(handler, token) { someAPI.unregisterEventHandler(token); }  // ...to then use it here.\n * );\n * ```\n *\n * Use with project function\n *\n * ```ts\n * import { fromEventPattern } from 'rxjs';\n *\n * someAPI.registerEventHandler((eventType, eventMessage) => {\n *   console.log(eventType, eventMessage); // Logs 'EVENT_TYPE' 'EVENT_MESSAGE' to console.\n * });\n *\n * const someAPIObservable = fromEventPattern(\n *   handler => someAPI.registerEventHandler(handler),\n *   handler => someAPI.unregisterEventHandler(handler)\n *   (eventType, eventMessage) => eventType + ' --- ' + eventMessage // without that function only 'EVENT_TYPE'\n * );                                                                // would be emitted by the Observable\n *\n * someAPIObservable.subscribe(value => console.log(value));\n *\n * // Logs:\n * // 'EVENT_TYPE --- EVENT_MESSAGE'\n * ```\n *\n * @see {@link fromEvent}\n * @see {@link bindCallback}\n * @see {@link bindNodeCallback}\n *\n * @param {function(handler: Function): any} addHandler A function that takes\n * a `handler` function as argument and attaches it somehow to the actual\n * source of events.\n * @param {function(handler: Function, token?: any): void} [removeHandler] A function that\n * takes a `handler` function as an argument and removes it from the event source. If `addHandler`\n * returns some kind of token, `removeHandler` function will have it as a second parameter.\n * @param {function(...args: any): T} [project] A function to\n * transform results. It takes the arguments from the event handler and\n * should return a single value.\n * @return {Observable<T>} Observable which, when an event happens, emits first parameter\n * passed to registered event handler. Alternatively it emits whatever project function returns\n * at that moment.\n */\nexport function fromEventPattern<T>(\n  addHandler: (handler: NodeEventHandler) => any,\n  removeHandler?: (handler: NodeEventHandler, signal?: any) => void,\n  resultSelector?: (...args: any[]) => T\n): Observable<T | T[]> {\n  if (resultSelector) {\n    return fromEventPattern<T>(addHandler, removeHandler).pipe(mapOneOrManyArgs(resultSelector));\n  }\n\n  return new Observable<T | T[]>((subscriber) => {\n    const handler = (...e: T[]) => subscriber.next(e.length === 1 ? e[0] : e);\n    const retValue = addHandler(handler);\n    return isFunction(removeHandler) ? () => removeHandler(handler, retValue) : undefined;\n  });\n}\n", "import { Observable } from '../Observable';\nimport { SchedulerLike } from '../types';\nimport { async as asyncScheduler } from '../scheduler/async';\nimport { isScheduler } from '../util/isScheduler';\nimport { isValidDate } from '../util/isDate';\n\n/**\n * Creates an observable that will wait for a specified time period, or exact date, before\n * emitting the number 0.\n *\n * <span class=\"informal\">Used to emit a notification after a delay.</span>\n *\n * This observable is useful for creating delays in code, or racing against other values\n * for ad-hoc timeouts.\n *\n * The `delay` is specified by default in milliseconds, however providing a custom scheduler could\n * create a different behavior.\n *\n * ## Examples\n *\n * Wait 3 seconds and start another observable\n *\n * You might want to use `timer` to delay subscription to an\n * observable by a set amount of time. Here we use a timer with\n * {@link concatMapTo} or {@link concatMap} in order to wait\n * a few seconds and start a subscription to a source.\n *\n * ```ts\n * import { of, timer, concatMap } from 'rxjs';\n *\n * // This could be any observable\n * const source = of(1, 2, 3);\n *\n * timer(3000)\n *   .pipe(concatMap(() => source))\n *   .subscribe(console.log);\n * ```\n *\n * Take all values until the start of the next minute\n *\n * Using a `Date` as the trigger for the first emission, you can\n * do things like wait until midnight to fire an event, or in this case,\n * wait until a new minute starts (chosen so the example wouldn't take\n * too long to run) in order to stop watching a stream. Leveraging\n * {@link takeUntil}.\n *\n * ```ts\n * import { interval, takeUntil, timer } from 'rxjs';\n *\n * // Build a Date object that marks the\n * // next minute.\n * const currentDate = new Date();\n * const startOfNextMinute = new Date(\n *   currentDate.getFullYear(),\n *   currentDate.getMonth(),\n *   currentDate.getDate(),\n *   currentDate.getHours(),\n *   currentDate.getMinutes() + 1\n * );\n *\n * // This could be any observable stream\n * const source = interval(1000);\n *\n * const result = source.pipe(\n *   takeUntil(timer(startOfNextMinute))\n * );\n *\n * result.subscribe(console.log);\n * ```\n *\n * ### Known Limitations\n *\n * - The {@link asyncScheduler} uses `setTimeout` which has limitations for how far in the future it can be scheduled.\n *\n * - If a `scheduler` is provided that returns a timestamp other than an epoch from `now()`, and\n * a `Date` object is passed to the `dueTime` argument, the calculation for when the first emission\n * should occur will be incorrect. In this case, it would be best to do your own calculations\n * ahead of time, and pass a `number` in as the `dueTime`.\n *\n * @param due If a `number`, the amount of time in milliseconds to wait before emitting.\n * If a `Date`, the exact time at which to emit.\n * @param scheduler The scheduler to use to schedule the delay. Defaults to {@link asyncScheduler}.\n */\nexport function timer(due: number | Date, scheduler?: SchedulerLike): Observable<0>;\n\n/**\n * Creates an observable that starts an interval after a specified delay, emitting incrementing numbers -- starting at `0` --\n * on each interval after words.\n *\n * The `delay` and `intervalDuration` are specified by default in milliseconds, however providing a custom scheduler could\n * create a different behavior.\n *\n * ## Example\n *\n * ### Start an interval that starts right away\n *\n * Since {@link interval} waits for the passed delay before starting,\n * sometimes that's not ideal. You may want to start an interval immediately.\n * `timer` works well for this. Here we have both side-by-side so you can\n * see them in comparison.\n *\n * Note that this observable will never complete.\n *\n * ```ts\n * import { timer, interval } from 'rxjs';\n *\n * timer(0, 1000).subscribe(n => console.log('timer', n));\n * interval(1000).subscribe(n => console.log('interval', n));\n * ```\n *\n * ### Known Limitations\n *\n * - The {@link asyncScheduler} uses `setTimeout` which has limitations for how far in the future it can be scheduled.\n *\n * - If a `scheduler` is provided that returns a timestamp other than an epoch from `now()`, and\n * a `Date` object is passed to the `dueTime` argument, the calculation for when the first emission\n * should occur will be incorrect. In this case, it would be best to do your own calculations\n * ahead of time, and pass a `number` in as the `startDue`.\n * @param startDue If a `number`, is the time to wait before starting the interval.\n * If a `Date`, is the exact time at which to start the interval.\n * @param intervalDuration The delay between each value emitted in the interval. Passing a\n * negative number here will result in immediate completion after the first value is emitted, as though\n * no `intervalDuration` was passed at all.\n * @param scheduler The scheduler to use to schedule the delay. Defaults to {@link asyncScheduler}.\n */\nexport function timer(startDue: number | Date, intervalDuration: number, scheduler?: SchedulerLike): Observable<number>;\n\n/**\n * @deprecated The signature allowing `undefined` to be passed for `intervalDuration` will be removed in v8. Use the `timer(dueTime, scheduler?)` signature instead.\n */\nexport function timer(dueTime: number | Date, unused: undefined, scheduler?: SchedulerLike): Observable<0>;\n\nexport function timer(\n  dueTime: number | Date = 0,\n  intervalOrScheduler?: number | SchedulerLike,\n  scheduler: SchedulerLike = asyncScheduler\n): Observable<number> {\n  // Since negative intervalDuration is treated as though no\n  // interval was specified at all, we start with a negative number.\n  let intervalDuration = -1;\n\n  if (intervalOrScheduler != null) {\n    // If we have a second argument, and it's a scheduler,\n    // override the scheduler we had defaulted. Otherwise,\n    // it must be an interval.\n    if (isScheduler(intervalOrScheduler)) {\n      scheduler = intervalOrScheduler;\n    } else {\n      // Note that this *could* be negative, in which case\n      // it's like not passing an intervalDuration at all.\n      intervalDuration = intervalOrScheduler;\n    }\n  }\n\n  return new Observable((subscriber) => {\n    // If a valid date is passed, calculate how long to wait before\n    // executing the first value... otherwise, if it's a number just schedule\n    // that many milliseconds (or scheduler-specified unit size) in the future.\n    let due = isValidDate(dueTime) ? +dueTime - scheduler!.now() : dueTime;\n\n    if (due < 0) {\n      // Ensure we don't schedule in the future.\n      due = 0;\n    }\n\n    // The incrementing value we emit.\n    let n = 0;\n\n    // Start the timer.\n    return scheduler.schedule(function () {\n      if (!subscriber.closed) {\n        // Emit the next value and increment.\n        subscriber.next(n++);\n\n        if (0 <= intervalDuration) {\n          // If we have a interval after the initial timer,\n          // reschedule with the period.\n          this.schedule(undefined, intervalDuration);\n        } else {\n          // We didn't have an interval. So just complete.\n          subscriber.complete();\n        }\n      }\n    }, due);\n  });\n}\n", "import { Observable } from '../Observable';\nimport { ObservableInput, ObservableInputTuple, SchedulerLike } from '../types';\nimport { mergeAll } from '../operators/mergeAll';\nimport { innerFrom } from './innerFrom';\nimport { EMPTY } from './empty';\nimport { popNumber, popScheduler } from '../util/args';\nimport { from } from './from';\n\nexport function merge<A extends readonly unknown[]>(...sources: [...ObservableInputTuple<A>]): Observable<A[number]>;\nexport function merge<A extends readonly unknown[]>(...sourcesAndConcurrency: [...ObservableInputTuple<A>, number?]): Observable<A[number]>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `mergeAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function merge<A extends readonly unknown[]>(\n  ...sourcesAndScheduler: [...ObservableInputTuple<A>, SchedulerLike?]\n): Observable<A[number]>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `mergeAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function merge<A extends readonly unknown[]>(\n  ...sourcesAndConcurrencyAndScheduler: [...ObservableInputTuple<A>, number?, SchedulerLike?]\n): Observable<A[number]>;\n\n/**\n * Creates an output Observable which concurrently emits all values from every\n * given input Observable.\n *\n * <span class=\"informal\">Flattens multiple Observables together by blending\n * their values into one Observable.</span>\n *\n * ![](merge.png)\n *\n * `merge` subscribes to each given input Observable (as arguments), and simply\n * forwards (without doing any transformation) all the values from all the input\n * Observables to the output Observable. The output Observable only completes\n * once all input Observables have completed. Any error delivered by an input\n * Observable will be immediately emitted on the output Observable.\n *\n * ## Examples\n *\n * Merge together two Observables: 1s interval and clicks\n *\n * ```ts\n * import { merge, fromEvent, interval } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const timer = interval(1000);\n * const clicksOrTimer = merge(clicks, timer);\n * clicksOrTimer.subscribe(x => console.log(x));\n *\n * // Results in the following:\n * // timer will emit ascending values, one every second(1000ms) to console\n * // clicks logs MouseEvents to console every time the \"document\" is clicked\n * // Since the two streams are merged you see these happening\n * // as they occur.\n * ```\n *\n * Merge together 3 Observables, but run only 2 concurrently\n *\n * ```ts\n * import { interval, take, merge } from 'rxjs';\n *\n * const timer1 = interval(1000).pipe(take(10));\n * const timer2 = interval(2000).pipe(take(6));\n * const timer3 = interval(500).pipe(take(10));\n *\n * const concurrent = 2; // the argument\n * const merged = merge(timer1, timer2, timer3, concurrent);\n * merged.subscribe(x => console.log(x));\n *\n * // Results in the following:\n * // - First timer1 and timer2 will run concurrently\n * // - timer1 will emit a value every 1000ms for 10 iterations\n * // - timer2 will emit a value every 2000ms for 6 iterations\n * // - after timer1 hits its max iteration, timer2 will\n * //   continue, and timer3 will start to run concurrently with timer2\n * // - when timer2 hits its max iteration it terminates, and\n * //   timer3 will continue to emit a value every 500ms until it is complete\n * ```\n *\n * @see {@link mergeAll}\n * @see {@link mergeMap}\n * @see {@link mergeMapTo}\n * @see {@link mergeScan}\n *\n * @param {...ObservableInput} observables Input Observables to merge together.\n * @param {number} [concurrent=Infinity] Maximum number of input\n * Observables being subscribed to concurrently.\n * @param {SchedulerLike} [scheduler=null] The {@link SchedulerLike} to use for managing\n * concurrency of input Observables.\n * @return {Observable} an Observable that emits items that are the result of\n * every input Observable.\n */\nexport function merge(...args: (ObservableInput<unknown> | number | SchedulerLike)[]): Observable<unknown> {\n  const scheduler = popScheduler(args);\n  const concurrent = popNumber(args, Infinity);\n  const sources = args as ObservableInput<unknown>[];\n  return !sources.length\n    ? // No source provided\n      EMPTY\n    : sources.length === 1\n    ? // One source? Just return it.\n      innerFrom(sources[0])\n    : // Merge all sources\n      mergeAll(concurrent)(from(sources, scheduler));\n}\n", "import { Observable } from '../Observable';\nimport { noop } from '../util/noop';\n\n/**\n * An Observable that emits no items to the Observer and never completes.\n *\n * ![](never.png)\n *\n * A simple Observable that emits neither values nor errors nor the completion\n * notification. It can be used for testing purposes or for composing with other\n * Observables. Please note that by never emitting a complete notification, this\n * Observable keeps the subscription from being disposed automatically.\n * Subscriptions need to be manually disposed.\n *\n * ##  Example\n *\n * Emit the number 7, then never emit anything else (not even complete)\n *\n * ```ts\n * import { NEVER, startWith } from 'rxjs';\n *\n * const info = () => console.log('Will not be called');\n *\n * const result = NEVER.pipe(startWith(7));\n * result.subscribe({\n *   next: x => console.log(x),\n *   error: info,\n *   complete: info\n * });\n * ```\n *\n * @see {@link Observable}\n * @see {@link EMPTY}\n * @see {@link of}\n * @see {@link throwError}\n */\nexport const NEVER = new Observable<never>(noop);\n\n/**\n * @deprecated Replaced with the {@link NEVER} constant. Will be removed in v8.\n */\nexport function never() {\n  return NEVER;\n}\n", "const { isArray } = Array;\n\n/**\n * Used in operators and functions that accept either a list of arguments, or an array of arguments\n * as a single argument.\n */\nexport function argsOrArgArray<T>(args: (T | T[])[]): T[] {\n  return args.length === 1 && isArray(args[0]) ? args[0] : (args as T[]);\n}\n", "import { OperatorFunction, MonoTypeOperatorFunction, TruthyTypesOf } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/** @deprecated Use a closure instead of a `thisArg`. Signatures accepting a `thisArg` will be removed in v8. */\nexport function filter<T, S extends T, A>(predicate: (this: A, value: T, index: number) => value is S, thisArg: A): OperatorFunction<T, S>;\nexport function filter<T, S extends T>(predicate: (value: T, index: number) => value is S): OperatorFunction<T, S>;\nexport function filter<T>(predicate: BooleanConstructor): OperatorFunction<T, TruthyTypesOf<T>>;\n/** @deprecated Use a closure instead of a `thisArg`. Signatures accepting a `thisArg` will be removed in v8. */\nexport function filter<T, A>(predicate: (this: A, value: T, index: number) => boolean, thisArg: A): MonoTypeOperatorFunction<T>;\nexport function filter<T>(predicate: (value: T, index: number) => boolean): MonoTypeOperatorFunction<T>;\n\n/**\n * Filter items emitted by the source Observable by only emitting those that\n * satisfy a specified predicate.\n *\n * <span class=\"informal\">Like\n * [Array.prototype.filter()](https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array/filter),\n * it only emits a value from the source if it passes a criterion function.</span>\n *\n * ![](filter.png)\n *\n * Similar to the well-known `Array.prototype.filter` method, this operator\n * takes values from the source Observable, passes them through a `predicate`\n * function and only emits those values that yielded `true`.\n *\n * ## Example\n *\n * Emit only click events whose target was a DIV element\n *\n * ```ts\n * import { fromEvent, filter } from 'rxjs';\n *\n * const div = document.createElement('div');\n * div.style.cssText = 'width: 200px; height: 200px; background: #09c;';\n * document.body.appendChild(div);\n *\n * const clicks = fromEvent(document, 'click');\n * const clicksOnDivs = clicks.pipe(filter(ev => (<HTMLElement>ev.target).tagName === 'DIV'));\n * clicksOnDivs.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link distinct}\n * @see {@link distinctUntilChanged}\n * @see {@link distinctUntilKeyChanged}\n * @see {@link ignoreElements}\n * @see {@link partition}\n * @see {@link skip}\n *\n * @param predicate A function that\n * evaluates each value emitted by the source Observable. If it returns `true`,\n * the value is emitted, if `false` the value is not passed to the output\n * Observable. The `index` parameter is the number `i` for the i-th source\n * emission that has happened since the subscription, starting from the number\n * `0`.\n * @param thisArg An optional argument to determine the value of `this`\n * in the `predicate` function.\n * @return A function that returns an Observable that emits items from the\n * source Observable that satisfy the specified `predicate`.\n */\nexport function filter<T>(predicate: (value: T, index: number) => boolean, thisArg?: any): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    // An index passed to our predicate function on each call.\n    let index = 0;\n\n    // Subscribe to the source, all errors and completions are\n    // forwarded to the consumer.\n    source.subscribe(\n      // Call the predicate with the appropriate `this` context,\n      // if the predicate returns `true`, then send the value\n      // to the consumer.\n      createOperatorSubscriber(subscriber, (value) => predicate.call(thisArg, value, index++) && subscriber.next(value))\n    );\n  });\n}\n", "import { Observable } from '../Observable';\nimport { ObservableInputTuple } from '../types';\nimport { innerFrom } from './innerFrom';\nimport { argsOrArgArray } from '../util/argsOrArgArray';\nimport { EMPTY } from './empty';\nimport { createOperatorSubscriber } from '../operators/OperatorSubscriber';\nimport { popResultSelector } from '../util/args';\n\nexport function zip<A extends readonly unknown[]>(sources: [...ObservableInputTuple<A>]): Observable<A>;\nexport function zip<A extends readonly unknown[], R>(\n  sources: [...ObservableInputTuple<A>],\n  resultSelector: (...values: A) => R\n): Observable<R>;\nexport function zip<A extends readonly unknown[]>(...sources: [...ObservableInputTuple<A>]): Observable<A>;\nexport function zip<A extends readonly unknown[], R>(\n  ...sourcesAndResultSelector: [...ObservableInputTuple<A>, (...values: A) => R]\n): Observable<R>;\n\n/**\n * Combines multiple Observables to create an Observable whose values are calculated from the values, in order, of each\n * of its input Observables.\n *\n * If the last parameter is a function, this function is used to compute the created value from the input values.\n * Otherwise, an array of the input values is returned.\n *\n * ## Example\n *\n * Combine age and name from different sources\n *\n * ```ts\n * import { of, zip, map } from 'rxjs';\n *\n * const age$ = of(27, 25, 29);\n * const name$ = of('Foo', 'Bar', 'Beer');\n * const isDev$ = of(true, true, false);\n *\n * zip(age$, name$, isDev$).pipe(\n *   map(([age, name, isDev]) => ({ age, name, isDev }))\n * )\n * .subscribe(x => console.log(x));\n *\n * // Outputs\n * // { age: 27, name: 'Foo', isDev: true }\n * // { age: 25, name: 'Bar', isDev: true }\n * // { age: 29, name: 'Beer', isDev: false }\n * ```\n *\n * @param sources\n * @return {Observable<R>}\n */\nexport function zip(...args: unknown[]): Observable<unknown> {\n  const resultSelector = popResultSelector(args);\n\n  const sources = argsOrArgArray(args) as Observable<unknown>[];\n\n  return sources.length\n    ? new Observable<unknown[]>((subscriber) => {\n        // A collection of buffers of values from each source.\n        // Keyed by the same index with which the sources were passed in.\n        let buffers: unknown[][] = sources.map(() => []);\n\n        // An array of flags of whether or not the sources have completed.\n        // This is used to check to see if we should complete the result.\n        // Keyed by the same index with which the sources were passed in.\n        let completed = sources.map(() => false);\n\n        // When everything is done, release the arrays above.\n        subscriber.add(() => {\n          buffers = completed = null!;\n        });\n\n        // Loop over our sources and subscribe to each one. The index `i` is\n        // especially important here, because we use it in closures below to\n        // access the related buffers and completion properties\n        for (let sourceIndex = 0; !subscriber.closed && sourceIndex < sources.length; sourceIndex++) {\n          innerFrom(sources[sourceIndex]).subscribe(\n            createOperatorSubscriber(\n              subscriber,\n              (value) => {\n                buffers[sourceIndex].push(value);\n                // if every buffer has at least one value in it, then we\n                // can shift out the oldest value from each buffer and emit\n                // them as an array.\n                if (buffers.every((buffer) => buffer.length)) {\n                  const result: any = buffers.map((buffer) => buffer.shift()!);\n                  // Emit the array. If theres' a result selector, use that.\n                  subscriber.next(resultSelector ? resultSelector(...result) : result);\n                  // If any one of the sources is both complete and has an empty buffer\n                  // then we complete the result. This is because we cannot possibly have\n                  // any more values to zip together.\n                  if (buffers.some((buffer, i) => !buffer.length && completed[i])) {\n                    subscriber.complete();\n                  }\n                }\n              },\n              () => {\n                // This source completed. Mark it as complete so we can check it later\n                // if we have to.\n                completed[sourceIndex] = true;\n                // But, if this complete source has nothing in its buffer, then we\n                // can complete the result, because we can't possibly have any more\n                // values from this to zip together with the other values.\n                !buffers[sourceIndex].length && subscriber.complete();\n              }\n            )\n          );\n        }\n\n        // When everything is done, release the arrays above.\n        return () => {\n          buffers = completed = null!;\n        };\n      })\n    : EMPTY;\n}\n", "import { Subscriber } from '../Subscriber';\nimport { MonoTypeOperatorFunction, ObservableInput } from '../types';\n\nimport { operate } from '../util/lift';\nimport { innerFrom } from '../observable/innerFrom';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * Ignores source values for a duration determined by another Observable, then\n * emits the most recent value from the source Observable, then repeats this\n * process.\n *\n * <span class=\"informal\">It's like {@link auditTime}, but the silencing\n * duration is determined by a second Observable.</span>\n *\n * ![](audit.svg)\n *\n * `audit` is similar to `throttle`, but emits the last value from the silenced\n * time window, instead of the first value. `audit` emits the most recent value\n * from the source Observable on the output Observable as soon as its internal\n * timer becomes disabled, and ignores source values while the timer is enabled.\n * Initially, the timer is disabled. As soon as the first source value arrives,\n * the timer is enabled by calling the `durationSelector` function with the\n * source value, which returns the \"duration\" Observable. When the duration\n * Observable emits a value, the timer is disabled, then the most\n * recent source value is emitted on the output Observable, and this process\n * repeats for the next source value.\n *\n * ## Example\n *\n * Emit clicks at a rate of at most one click per second\n *\n * ```ts\n * import { fromEvent, audit, interval } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(audit(ev => interval(1000)));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link auditTime}\n * @see {@link debounce}\n * @see {@link delayWhen}\n * @see {@link sample}\n * @see {@link throttle}\n *\n * @param durationSelector A function\n * that receives a value from the source Observable, for computing the silencing\n * duration, returned as an Observable or a Promise.\n * @return A function that returns an Observable that performs rate-limiting of\n * emissions from the source Observable.\n */\nexport function audit<T>(durationSelector: (value: T) => ObservableInput<any>): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    let hasValue = false;\n    let lastValue: T | null = null;\n    let durationSubscriber: Subscriber<any> | null = null;\n    let isComplete = false;\n\n    const endDuration = () => {\n      durationSubscriber?.unsubscribe();\n      durationSubscriber = null;\n      if (hasValue) {\n        hasValue = false;\n        const value = lastValue!;\n        lastValue = null;\n        subscriber.next(value);\n      }\n      isComplete && subscriber.complete();\n    };\n\n    const cleanupDuration = () => {\n      durationSubscriber = null;\n      isComplete && subscriber.complete();\n    };\n\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value) => {\n          hasValue = true;\n          lastValue = value;\n          if (!durationSubscriber) {\n            innerFrom(durationSelector(value)).subscribe(\n              (durationSubscriber = createOperatorSubscriber(subscriber, endDuration, cleanupDuration))\n            );\n          }\n        },\n        () => {\n          isComplete = true;\n          (!hasValue || !durationSubscriber || durationSubscriber.closed) && subscriber.complete();\n        }\n      )\n    );\n  });\n}\n", "import { asyncScheduler } from '../scheduler/async';\nimport { audit } from './audit';\nimport { timer } from '../observable/timer';\nimport { MonoTypeOperatorFunction, SchedulerLike } from '../types';\n\n/**\n * Ignores source values for `duration` milliseconds, then emits the most recent\n * value from the source Observable, then repeats this process.\n *\n * <span class=\"informal\">When it sees a source value, it ignores that plus\n * the next ones for `duration` milliseconds, and then it emits the most recent\n * value from the source.</span>\n *\n * ![](auditTime.png)\n *\n * `auditTime` is similar to `throttleTime`, but emits the last value from the\n * silenced time window, instead of the first value. `auditTime` emits the most\n * recent value from the source Observable on the output Observable as soon as\n * its internal timer becomes disabled, and ignores source values while the\n * timer is enabled. Initially, the timer is disabled. As soon as the first\n * source value arrives, the timer is enabled. After `duration` milliseconds (or\n * the time unit determined internally by the optional `scheduler`) has passed,\n * the timer is disabled, then the most recent source value is emitted on the\n * output Observable, and this process repeats for the next source value.\n * Optionally takes a {@link SchedulerLike} for managing timers.\n *\n * ## Example\n *\n * Emit clicks at a rate of at most one click per second\n *\n * ```ts\n * import { fromEvent, auditTime } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(auditTime(1000));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link audit}\n * @see {@link debounceTime}\n * @see {@link delay}\n * @see {@link sampleTime}\n * @see {@link throttleTime}\n *\n * @param {number} duration Time to wait before emitting the most recent source\n * value, measured in milliseconds or the time unit determined internally\n * by the optional `scheduler`.\n * @param {SchedulerLike} [scheduler=async] The {@link SchedulerLike} to use for\n * managing the timers that handle the rate-limiting behavior.\n * @return A function that returns an Observable that performs rate-limiting of\n * emissions from the source Observable.\n */\nexport function auditTime<T>(duration: number, scheduler: SchedulerLike = asyncScheduler): MonoTypeOperatorFunction<T> {\n  return audit(() => timer(duration, scheduler));\n}\n", "import { OperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { arrRemove } from '../util/arrRemove';\n\n/**\n * Buffers the source Observable values until the size hits the maximum\n * `bufferSize` given.\n *\n * <span class=\"informal\">Collects values from the past as an array, and emits\n * that array only when its size reaches `bufferSize`.</span>\n *\n * ![](bufferCount.png)\n *\n * Buffers a number of values from the source Observable by `bufferSize` then\n * emits the buffer and clears it, and starts a new buffer each\n * `startBufferEvery` values. If `startBufferEvery` is not provided or is\n * `null`, then new buffers are started immediately at the start of the source\n * and when each buffer closes and is emitted.\n *\n * ## Examples\n *\n * Emit the last two click events as an array\n *\n * ```ts\n * import { fromEvent, bufferCount } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const buffered = clicks.pipe(bufferCount(2));\n * buffered.subscribe(x => console.log(x));\n * ```\n *\n * On every click, emit the last two click events as an array\n *\n * ```ts\n * import { fromEvent, bufferCount } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const buffered = clicks.pipe(bufferCount(2, 1));\n * buffered.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link buffer}\n * @see {@link bufferTime}\n * @see {@link bufferToggle}\n * @see {@link bufferWhen}\n * @see {@link pairwise}\n * @see {@link windowCount}\n *\n * @param {number} bufferSize The maximum size of the buffer emitted.\n * @param {number} [startBufferEvery] Interval at which to start a new buffer.\n * For example if `startBufferEvery` is `2`, then a new buffer will be started\n * on every other value from the source. A new buffer is started at the\n * beginning of the source by default.\n * @return A function that returns an Observable of arrays of buffered values.\n */\nexport function bufferCount<T>(bufferSize: number, startBufferEvery: number | null = null): OperatorFunction<T, T[]> {\n  // If no `startBufferEvery` value was supplied, then we're\n  // opening and closing on the bufferSize itself.\n  startBufferEvery = startBufferEvery ?? bufferSize;\n\n  return operate((source, subscriber) => {\n    let buffers: T[][] = [];\n    let count = 0;\n\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value) => {\n          let toEmit: T[][] | null = null;\n\n          // Check to see if we need to start a buffer.\n          // This will start one at the first value, and then\n          // a new one every N after that.\n          if (count++ % startBufferEvery! === 0) {\n            buffers.push([]);\n          }\n\n          // Push our value into our active buffers.\n          for (const buffer of buffers) {\n            buffer.push(value);\n            // Check to see if we're over the bufferSize\n            // if we are, record it so we can emit it later.\n            // If we emitted it now and removed it, it would\n            // mutate the `buffers` array while we're looping\n            // over it.\n            if (bufferSize <= buffer.length) {\n              toEmit = toEmit ?? [];\n              toEmit.push(buffer);\n            }\n          }\n\n          if (toEmit) {\n            // We have found some buffers that are over the\n            // `bufferSize`. Emit them, and remove them from our\n            // buffers list.\n            for (const buffer of toEmit) {\n              arrRemove(buffers, buffer);\n              subscriber.next(buffer);\n            }\n          }\n        },\n        () => {\n          // When the source completes, emit all of our\n          // active buffers.\n          for (const buffer of buffers) {\n            subscriber.next(buffer);\n          }\n          subscriber.complete();\n        },\n        // Pass all errors through to consumer.\n        undefined,\n        () => {\n          // Clean up our memory when we finalize\n          buffers = null!;\n        }\n      )\n    );\n  });\n}\n", "import { Observable } from '../Observable';\n\nimport { ObservableInput, OperatorFunction, ObservedValueOf } from '../types';\nimport { Subscription } from '../Subscription';\nimport { innerFrom } from '../observable/innerFrom';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { operate } from '../util/lift';\n\n/* tslint:disable:max-line-length */\nexport function catchError<T, O extends ObservableInput<any>>(\n  selector: (err: any, caught: Observable<T>) => O\n): OperatorFunction<T, T | ObservedValueOf<O>>;\n/* tslint:enable:max-line-length */\n\n/**\n * Catches errors on the observable to be handled by returning a new observable or throwing an error.\n *\n * <span class=\"informal\">\n * It only listens to the error channel and ignores notifications.\n * Handles errors from the source observable, and maps them to a new observable.\n * The error may also be rethrown, or a new error can be thrown to emit an error from the result.\n * </span>\n *\n * ![](catch.png)\n *\n * This operator handles errors, but forwards along all other events to the resulting observable.\n * If the source observable terminates with an error, it will map that error to a new observable,\n * subscribe to it, and forward all of its events to the resulting observable.\n *\n * ## Examples\n *\n * Continue with a different Observable when there's an error\n *\n * ```ts\n * import { of, map, catchError } from 'rxjs';\n *\n * of(1, 2, 3, 4, 5)\n *   .pipe(\n *     map(n => {\n *       if (n === 4) {\n *         throw 'four!';\n *       }\n *       return n;\n *     }),\n *     catchError(err => of('I', 'II', 'III', 'IV', 'V'))\n *   )\n *   .subscribe(x => console.log(x));\n *   // 1, 2, 3, I, II, III, IV, V\n * ```\n *\n * Retry the caught source Observable again in case of error, similar to `retry()` operator\n *\n * ```ts\n * import { of, map, catchError, take } from 'rxjs';\n *\n * of(1, 2, 3, 4, 5)\n *   .pipe(\n *     map(n => {\n *       if (n === 4) {\n *         throw 'four!';\n *       }\n *       return n;\n *     }),\n *     catchError((err, caught) => caught),\n *     take(30)\n *   )\n *   .subscribe(x => console.log(x));\n *   // 1, 2, 3, 1, 2, 3, ...\n * ```\n *\n * Throw a new error when the source Observable throws an error\n *\n * ```ts\n * import { of, map, catchError } from 'rxjs';\n *\n * of(1, 2, 3, 4, 5)\n *   .pipe(\n *     map(n => {\n *       if (n === 4) {\n *         throw 'four!';\n *       }\n *       return n;\n *     }),\n *     catchError(err => {\n *       throw 'error in source. Details: ' + err;\n *     })\n *   )\n *   .subscribe({\n *     next: x => console.log(x),\n *     error: err => console.log(err)\n *   });\n *   // 1, 2, 3, error in source. Details: four!\n * ```\n *\n * @see {@link onErrorResumeNext}\n * @see {@link repeat}\n * @see {@link repeatWhen}\n * @see {@link retry }\n * @see {@link retryWhen}\n *\n * @param {function} selector a function that takes as arguments `err`, which is the error, and `caught`, which\n * is the source observable, in case you'd like to \"retry\" that observable by returning it again. Whatever observable\n * is returned by the `selector` will be used to continue the observable chain.\n * @return A function that returns an Observable that originates from either\n * the source or the Observable returned by the `selector` function.\n */\nexport function catchError<T, O extends ObservableInput<any>>(\n  selector: (err: any, caught: Observable<T>) => O\n): OperatorFunction<T, T | ObservedValueOf<O>> {\n  return operate((source, subscriber) => {\n    let innerSub: Subscription | null = null;\n    let syncUnsub = false;\n    let handledResult: Observable<ObservedValueOf<O>>;\n\n    innerSub = source.subscribe(\n      createOperatorSubscriber(subscriber, undefined, undefined, (err) => {\n        handledResult = innerFrom(selector(err, catchError(selector)(source)));\n        if (innerSub) {\n          innerSub.unsubscribe();\n          innerSub = null;\n          handledResult.subscribe(subscriber);\n        } else {\n          // We don't have an innerSub yet, that means the error was synchronous\n          // because the subscribe call hasn't returned yet.\n          syncUnsub = true;\n        }\n      })\n    );\n\n    if (syncUnsub) {\n      // We have a synchronous error, we need to make sure to\n      // finalize right away. This ensures that callbacks in the `finalize` operator are called\n      // at the right time, and that finalization occurs at the expected\n      // time between the source error and the subscription to the\n      // next observable.\n      innerSub.unsubscribe();\n      innerSub = null;\n      handledResult!.subscribe(subscriber);\n    }\n  });\n}\n", "import { Observable } from '../Observable';\nimport { Subscriber } from '../Subscriber';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * A basic scan operation. This is used for `scan` and `reduce`.\n * @param accumulator The accumulator to use\n * @param seed The seed value for the state to accumulate\n * @param hasSeed Whether or not a seed was provided\n * @param emitOnNext Whether or not to emit the state on next\n * @param emitBeforeComplete Whether or not to emit the before completion\n */\n\nexport function scanInternals<V, A, S>(\n  accumulator: (acc: V | A | S, value: V, index: number) => A,\n  seed: S,\n  hasSeed: boolean,\n  emitOnNext: boolean,\n  emitBeforeComplete?: undefined | true\n) {\n  return (source: Observable<V>, subscriber: Subscriber<any>) => {\n    // Whether or not we have state yet. This will only be\n    // false before the first value arrives if we didn't get\n    // a seed value.\n    let hasState = hasSeed;\n    // The state that we're tracking, starting with the seed,\n    // if there is one, and then updated by the return value\n    // from the accumulator on each emission.\n    let state: any = seed;\n    // An index to pass to the accumulator function.\n    let index = 0;\n\n    // Subscribe to our source. All errors and completions are passed through.\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value) => {\n          // Always increment the index.\n          const i = index++;\n          // Set the state\n          state = hasState\n            ? // We already have state, so we can get the new state from the accumulator\n              accumulator(state, value, i)\n            : // We didn't have state yet, a seed value was not provided, so\n\n              // we set the state to the first value, and mark that we have state now\n              ((hasState = true), value);\n\n          // Maybe send it to the consumer.\n          emitOnNext && subscriber.next(state);\n        },\n        // If an onComplete was given, call it, otherwise\n        // just pass through the complete notification to the consumer.\n        emitBeforeComplete &&\n          (() => {\n            hasState && subscriber.next(state);\n            subscriber.complete();\n          })\n      )\n    );\n  };\n}\n", "import { combineLatestInit } from '../observable/combineLatest';\nimport { ObservableInput, ObservableInputTuple, OperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { argsOrArgArray } from '../util/argsOrArgArray';\nimport { mapOneOrManyArgs } from '../util/mapOneOrManyArgs';\nimport { pipe } from '../util/pipe';\nimport { popResultSelector } from '../util/args';\n\n/** @deprecated Replaced with {@link combineLatestWith}. Will be removed in v8. */\nexport function combineLatest<T, A extends readonly unknown[], R>(\n  sources: [...ObservableInputTuple<A>],\n  project: (...values: [T, ...A]) => R\n): OperatorFunction<T, R>;\n/** @deprecated Replaced with {@link combineLatestWith}. Will be removed in v8. */\nexport function combineLatest<T, A extends readonly unknown[], R>(sources: [...ObservableInputTuple<A>]): OperatorFunction<T, [T, ...A]>;\n\n/** @deprecated Replaced with {@link combineLatestWith}. Will be removed in v8. */\nexport function combineLatest<T, A extends readonly unknown[], R>(\n  ...sourcesAndProject: [...ObservableInputTuple<A>, (...values: [T, ...A]) => R]\n): OperatorFunction<T, R>;\n/** @deprecated Replaced with {@link combineLatestWith}. Will be removed in v8. */\nexport function combineLatest<T, A extends readonly unknown[], R>(...sources: [...ObservableInputTuple<A>]): OperatorFunction<T, [T, ...A]>;\n\n/**\n * @deprecated Replaced with {@link combineLatestWith}. Will be removed in v8.\n */\nexport function combineLatest<T, R>(...args: (ObservableInput<any> | ((...values: any[]) => R))[]): OperatorFunction<T, unknown> {\n  const resultSelector = popResultSelector(args);\n  return resultSelector\n    ? pipe(combineLatest(...(args as Array<ObservableInput<any>>)), mapOneOrManyArgs(resultSelector))\n    : operate((source, subscriber) => {\n        combineLatestInit([source, ...argsOrArgArray(args)])(subscriber);\n      });\n}\n", "import { ObservableInputTuple, OperatorFunction, Cons } from '../types';\nimport { combineLatest } from './combineLatest';\n\n/**\n * Create an observable that combines the latest values from all passed observables and the source\n * into arrays and emits them.\n *\n * Returns an observable, that when subscribed to, will subscribe to the source observable and all\n * sources provided as arguments. Once all sources emit at least one value, all of the latest values\n * will be emitted as an array. After that, every time any source emits a value, all of the latest values\n * will be emitted as an array.\n *\n * This is a useful operator for eagerly calculating values based off of changed inputs.\n *\n * ## Example\n *\n * Simple concatenation of values from two inputs\n *\n * ```ts\n * import { fromEvent, combineLatestWith, map } from 'rxjs';\n *\n * // Setup: Add two inputs to the page\n * const input1 = document.createElement('input');\n * document.body.appendChild(input1);\n * const input2 = document.createElement('input');\n * document.body.appendChild(input2);\n *\n * // Get streams of changes\n * const input1Changes$ = fromEvent(input1, 'change');\n * const input2Changes$ = fromEvent(input2, 'change');\n *\n * // Combine the changes by adding them together\n * input1Changes$.pipe(\n *   combineLatestWith(input2Changes$),\n *   map(([e1, e2]) => (<HTMLInputElement>e1.target).value + ' - ' + (<HTMLInputElement>e2.target).value)\n * )\n * .subscribe(x => console.log(x));\n * ```\n *\n * @param otherSources the other sources to subscribe to.\n * @return A function that returns an Observable that emits the latest\n * emissions from both source and provided Observables.\n */\nexport function combineLatestWith<T, A extends readonly unknown[]>(\n  ...otherSources: [...ObservableInputTuple<A>]\n): OperatorFunction<T, Cons<T, A>> {\n  return combineLatest(...otherSources);\n}\n", "import { asyncScheduler } from '../scheduler/async';\nimport { Subscription } from '../Subscription';\nimport { MonoTypeOperatorFunction, SchedulerAction, SchedulerLike } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * Emits a notification from the source Observable only after a particular time span\n * has passed without another source emission.\n *\n * <span class=\"informal\">It's like {@link delay}, but passes only the most\n * recent notification from each burst of emissions.</span>\n *\n * ![](debounceTime.png)\n *\n * `debounceTime` delays notifications emitted by the source Observable, but drops\n * previous pending delayed emissions if a new notification arrives on the source\n * Observable. This operator keeps track of the most recent notification from the\n * source Observable, and emits that only when `dueTime` has passed\n * without any other notification appearing on the source Observable. If a new value\n * appears before `dueTime` silence occurs, the previous notification will be dropped\n * and will not be emitted and a new `dueTime` is scheduled.\n * If the completing event happens during `dueTime` the last cached notification\n * is emitted before the completion event is forwarded to the output observable.\n * If the error event happens during `dueTime` or after it only the error event is\n * forwarded to the output observable. The cache notification is not emitted in this case.\n *\n * This is a rate-limiting operator, because it is impossible for more than one\n * notification to be emitted in any time window of duration `dueTime`, but it is also\n * a delay-like operator since output emissions do not occur at the same time as\n * they did on the source Observable. Optionally takes a {@link SchedulerLike} for\n * managing timers.\n *\n * ## Example\n *\n * Emit the most recent click after a burst of clicks\n *\n * ```ts\n * import { fromEvent, debounceTime } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(debounceTime(1000));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link audit}\n * @see {@link auditTime}\n * @see {@link debounce}\n * @see {@link sample}\n * @see {@link sampleTime}\n * @see {@link throttle}\n * @see {@link throttleTime}\n *\n * @param {number} dueTime The timeout duration in milliseconds (or the time\n * unit determined internally by the optional `scheduler`) for the window of\n * time required to wait for emission silence before emitting the most recent\n * source value.\n * @param {SchedulerLike} [scheduler=async] The {@link SchedulerLike} to use for\n * managing the timers that handle the timeout for each value.\n * @return A function that returns an Observable that delays the emissions of\n * the source Observable by the specified `dueTime`, and may drop some values\n * if they occur too frequently.\n */\nexport function debounceTime<T>(dueTime: number, scheduler: SchedulerLike = asyncScheduler): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    let activeTask: Subscription | null = null;\n    let lastValue: T | null = null;\n    let lastTime: number | null = null;\n\n    const emit = () => {\n      if (activeTask) {\n        // We have a value! Free up memory first, then emit the value.\n        activeTask.unsubscribe();\n        activeTask = null;\n        const value = lastValue!;\n        lastValue = null;\n        subscriber.next(value);\n      }\n    };\n    function emitWhenIdle(this: SchedulerAction<unknown>) {\n      // This is called `dueTime` after the first value\n      // but we might have received new values during this window!\n\n      const targetTime = lastTime! + dueTime;\n      const now = scheduler.now();\n      if (now < targetTime) {\n        // On that case, re-schedule to the new target\n        activeTask = this.schedule(undefined, targetTime - now);\n        subscriber.add(activeTask);\n        return;\n      }\n\n      emit();\n    }\n\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value: T) => {\n          lastValue = value;\n          lastTime = scheduler.now();\n\n          // Only set up a task if it's not already up\n          if (!activeTask) {\n            activeTask = scheduler.schedule(emitWhenIdle, dueTime);\n            subscriber.add(activeTask);\n          }\n        },\n        () => {\n          // Source completed.\n          // Emit any pending debounced values then complete\n          emit();\n          subscriber.complete();\n        },\n        // Pass all errors through to consumer.\n        undefined,\n        () => {\n          // Finalization.\n          lastValue = activeTask = null;\n        }\n      )\n    );\n  });\n}\n", "import { OperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * Emits a given value if the source Observable completes without emitting any\n * `next` value, otherwise mirrors the source Observable.\n *\n * <span class=\"informal\">If the source Observable turns out to be empty, then\n * this operator will emit a default value.</span>\n *\n * ![](defaultIfEmpty.png)\n *\n * `defaultIfEmpty` emits the values emitted by the source Observable or a\n * specified default value if the source Observable is empty (completes without\n * having emitted any `next` value).\n *\n * ## Example\n *\n * If no clicks happen in 5 seconds, then emit 'no clicks'\n *\n * ```ts\n * import { fromEvent, takeUntil, interval, defaultIfEmpty } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const clicksBeforeFive = clicks.pipe(takeUntil(interval(5000)));\n * const result = clicksBeforeFive.pipe(defaultIfEmpty('no clicks'));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link empty}\n * @see {@link last}\n *\n * @param defaultValue The default value used if the source\n * Observable is empty.\n * @return A function that returns an Observable that emits either the\n * specified `defaultValue` if the source Observable emits no items, or the\n * values emitted by the source Observable.\n */\nexport function defaultIfEmpty<T, R>(defaultValue: R): OperatorFunction<T, T | R> {\n  return operate((source, subscriber) => {\n    let hasValue = false;\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value) => {\n          hasValue = true;\n          subscriber.next(value);\n        },\n        () => {\n          if (!hasValue) {\n            subscriber.next(defaultValue!);\n          }\n          subscriber.complete();\n        }\n      )\n    );\n  });\n}\n", "import { MonoTypeOperatorFunction } from '../types';\nimport { EMPTY } from '../observable/empty';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * Emits only the first `count` values emitted by the source Observable.\n *\n * <span class=\"informal\">Takes the first `count` values from the source, then\n * completes.</span>\n *\n * ![](take.png)\n *\n * `take` returns an Observable that emits only the first `count` values emitted\n * by the source Observable. If the source emits fewer than `count` values then\n * all of its values are emitted. After that, it completes, regardless if the\n * source completes.\n *\n * ## Example\n *\n * Take the first 5 seconds of an infinite 1-second interval Observable\n *\n * ```ts\n * import { interval, take } from 'rxjs';\n *\n * const intervalCount = interval(1000);\n * const takeFive = intervalCount.pipe(take(5));\n * takeFive.subscribe(x => console.log(x));\n *\n * // Logs:\n * // 0\n * // 1\n * // 2\n * // 3\n * // 4\n * ```\n *\n * @see {@link takeLast}\n * @see {@link takeUntil}\n * @see {@link takeWhile}\n * @see {@link skip}\n *\n * @param count The maximum number of `next` values to emit.\n * @return A function that returns an Observable that emits only the first\n * `count` values emitted by the source Observable, or all of the values from\n * the source if the source emits fewer than `count` values.\n */\nexport function take<T>(count: number): MonoTypeOperatorFunction<T> {\n  return count <= 0\n    ? // If we are taking no values, that's empty.\n      () => EMPTY\n    : operate((source, subscriber) => {\n        let seen = 0;\n        source.subscribe(\n          createOperatorSubscriber(subscriber, (value) => {\n            // Increment the number of values we have seen,\n            // then check it against the allowed count to see\n            // if we are still letting values through.\n            if (++seen <= count) {\n              subscriber.next(value);\n              // If we have met or passed our allowed count,\n              // we need to complete. We have to do <= here,\n              // because re-entrant code will increment `seen` twice.\n              if (count <= seen) {\n                subscriber.complete();\n              }\n            }\n          })\n        );\n      });\n}\n", "import { OperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { noop } from '../util/noop';\n\n/**\n * Ignores all items emitted by the source Observable and only passes calls of `complete` or `error`.\n *\n * ![](ignoreElements.png)\n *\n * The `ignoreElements` operator suppresses all items emitted by the source Observable,\n * but allows its termination notification (either `error` or `complete`) to pass through unchanged.\n *\n * If you do not care about the items being emitted by an Observable, but you do want to be notified\n * when it completes or when it terminates with an error, you can apply the `ignoreElements` operator\n * to the Observable, which will ensure that it will never call its observers\u2019 `next` handlers.\n *\n * ## Example\n *\n * Ignore all `next` emissions from the source\n *\n * ```ts\n * import { of, ignoreElements } from 'rxjs';\n *\n * of('you', 'talking', 'to', 'me')\n *   .pipe(ignoreElements())\n *   .subscribe({\n *     next: word => console.log(word),\n *     error: err => console.log('error:', err),\n *     complete: () => console.log('the end'),\n *   });\n *\n * // result:\n * // 'the end'\n * ```\n *\n * @return A function that returns an empty Observable that only calls\n * `complete` or `error`, based on which one is called by the source\n * Observable.\n */\nexport function ignoreElements(): OperatorFunction<unknown, never> {\n  return operate((source, subscriber) => {\n    source.subscribe(createOperatorSubscriber(subscriber, noop));\n  });\n}\n", "import { OperatorFunction } from '../types';\nimport { map } from './map';\n\n/** @deprecated To be removed in v9. Use {@link map} instead: `map(() => value)`. */\nexport function mapTo<R>(value: R): OperatorFunction<unknown, R>;\n/**\n * @deprecated Do not specify explicit type parameters. Signatures with type parameters\n * that cannot be inferred will be removed in v8. `mapTo` itself will be removed in v9,\n * use {@link map} instead: `map(() => value)`.\n * */\nexport function mapTo<T, R>(value: R): OperatorFunction<T, R>;\n\n/**\n * Emits the given constant value on the output Observable every time the source\n * Observable emits a value.\n *\n * <span class=\"informal\">Like {@link map}, but it maps every source value to\n * the same output value every time.</span>\n *\n * ![](mapTo.png)\n *\n * Takes a constant `value` as argument, and emits that whenever the source\n * Observable emits a value. In other words, ignores the actual source value,\n * and simply uses the emission moment to know when to emit the given `value`.\n *\n * ## Example\n *\n * Map every click to the string `'Hi'`\n *\n * ```ts\n * import { fromEvent, mapTo } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const greetings = clicks.pipe(mapTo('Hi'));\n *\n * greetings.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link map}\n *\n * @param value The value to map each source value to.\n * @return A function that returns an Observable that emits the given `value`\n * every time the source Observable emits.\n * @deprecated To be removed in v9. Use {@link map} instead: `map(() => value)`.\n */\nexport function mapTo<R>(value: R): OperatorFunction<unknown, R> {\n  return map(() => value);\n}\n", "import { Observable } from '../Observable';\nimport { MonoTypeOperatorFunction, ObservableInput } from '../types';\nimport { concat } from '../observable/concat';\nimport { take } from './take';\nimport { ignoreElements } from './ignoreElements';\nimport { mapTo } from './mapTo';\nimport { mergeMap } from './mergeMap';\nimport { innerFrom } from '../observable/innerFrom';\n\n/** @deprecated The `subscriptionDelay` parameter will be removed in v8. */\nexport function delayWhen<T>(\n  delayDurationSelector: (value: T, index: number) => ObservableInput<any>,\n  subscriptionDelay: Observable<any>\n): MonoTypeOperatorFunction<T>;\nexport function delayWhen<T>(delayDurationSelector: (value: T, index: number) => ObservableInput<any>): MonoTypeOperatorFunction<T>;\n\n/**\n * Delays the emission of items from the source Observable by a given time span\n * determined by the emissions of another Observable.\n *\n * <span class=\"informal\">It's like {@link delay}, but the time span of the\n * delay duration is determined by a second Observable.</span>\n *\n * ![](delayWhen.png)\n *\n * `delayWhen` operator shifts each emitted value from the source Observable by\n * a time span determined by another Observable. When the source emits a value,\n * the `delayDurationSelector` function is called with the value emitted from\n * the source Observable as the first argument to the `delayDurationSelector`.\n * The `delayDurationSelector` function should return an {@link ObservableInput},\n * that is internally converted to an Observable that is called the \"duration\"\n * Observable.\n *\n * The source value is emitted on the output Observable only when the \"duration\"\n * Observable emits ({@link guide/glossary-and-semantics#next next}s) any value.\n * Upon that, the \"duration\" Observable gets unsubscribed.\n *\n * Before RxJS V7, the {@link guide/glossary-and-semantics#complete completion}\n * of the \"duration\" Observable would have been triggering the emission of the\n * source value to the output Observable, but with RxJS V7, this is not the case\n * anymore.\n *\n * Only next notifications (from the \"duration\" Observable) trigger values from\n * the source Observable to be passed to the output Observable. If the \"duration\"\n * Observable only emits the complete notification (without next), the value\n * emitted by the source Observable will never get to the output Observable - it\n * will be swallowed. If the \"duration\" Observable errors, the error will be\n * propagated to the output Observable.\n *\n * Optionally, `delayWhen` takes a second argument, `subscriptionDelay`, which\n * is an Observable. When `subscriptionDelay` emits its first value or\n * completes, the source Observable is subscribed to and starts behaving like\n * described in the previous paragraph. If `subscriptionDelay` is not provided,\n * `delayWhen` will subscribe to the source Observable as soon as the output\n * Observable is subscribed.\n *\n * ## Example\n *\n * Delay each click by a random amount of time, between 0 and 5 seconds\n *\n * ```ts\n * import { fromEvent, delayWhen, interval } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const delayedClicks = clicks.pipe(\n *   delayWhen(() => interval(Math.random() * 5000))\n * );\n * delayedClicks.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link delay}\n * @see {@link throttle}\n * @see {@link throttleTime}\n * @see {@link debounce}\n * @see {@link debounceTime}\n * @see {@link sample}\n * @see {@link sampleTime}\n * @see {@link audit}\n * @see {@link auditTime}\n *\n * @param delayDurationSelector A function that returns an `ObservableInput` for\n * each `value` emitted by the source Observable, which is then used to delay the\n * emission of that `value` on the output Observable until the `ObservableInput`\n * returned from this function emits a next value. When called, beside `value`,\n * this function receives a zero-based `index` of the emission order.\n * @param subscriptionDelay An Observable that triggers the subscription to the\n * source Observable once it emits any value.\n * @return A function that returns an Observable that delays the emissions of\n * the source Observable by an amount of time specified by the Observable\n * returned by `delayDurationSelector`.\n */\nexport function delayWhen<T>(\n  delayDurationSelector: (value: T, index: number) => ObservableInput<any>,\n  subscriptionDelay?: Observable<any>\n): MonoTypeOperatorFunction<T> {\n  if (subscriptionDelay) {\n    // DEPRECATED PATH\n    return (source: Observable<T>) =>\n      concat(subscriptionDelay.pipe(take(1), ignoreElements()), source.pipe(delayWhen(delayDurationSelector)));\n  }\n\n  return mergeMap((value, index) => innerFrom(delayDurationSelector(value, index)).pipe(take(1), mapTo(value)));\n}\n", "import { asyncScheduler } from '../scheduler/async';\nimport { MonoTypeOperatorFunction, SchedulerLike } from '../types';\nimport { delayWhen } from './delayWhen';\nimport { timer } from '../observable/timer';\n\n/**\n * Delays the emission of items from the source Observable by a given timeout or\n * until a given Date.\n *\n * <span class=\"informal\">Time shifts each item by some specified amount of\n * milliseconds.</span>\n *\n * ![](delay.svg)\n *\n * If the delay argument is a Number, this operator time shifts the source\n * Observable by that amount of time expressed in milliseconds. The relative\n * time intervals between the values are preserved.\n *\n * If the delay argument is a Date, this operator time shifts the start of the\n * Observable execution until the given date occurs.\n *\n * ## Examples\n *\n * Delay each click by one second\n *\n * ```ts\n * import { fromEvent, delay } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const delayedClicks = clicks.pipe(delay(1000)); // each click emitted after 1 second\n * delayedClicks.subscribe(x => console.log(x));\n * ```\n *\n * Delay all clicks until a future date happens\n *\n * ```ts\n * import { fromEvent, delay } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const date = new Date('March 15, 2050 12:00:00'); // in the future\n * const delayedClicks = clicks.pipe(delay(date)); // click emitted only after that date\n * delayedClicks.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link delayWhen}\n * @see {@link throttle}\n * @see {@link throttleTime}\n * @see {@link debounce}\n * @see {@link debounceTime}\n * @see {@link sample}\n * @see {@link sampleTime}\n * @see {@link audit}\n * @see {@link auditTime}\n *\n * @param {number|Date} due The delay duration in milliseconds (a `number`) or\n * a `Date` until which the emission of the source items is delayed.\n * @param {SchedulerLike} [scheduler=async] The {@link SchedulerLike} to use for\n * managing the timers that handle the time-shift for each item.\n * @return A function that returns an Observable that delays the emissions of\n * the source Observable by the specified timeout or Date.\n */\nexport function delay<T>(due: number | Date, scheduler: SchedulerLike = asyncScheduler): MonoTypeOperatorFunction<T> {\n  const duration = timer(due, scheduler);\n  return delayWhen(() => duration);\n}\n", "import { MonoTypeOperatorFunction } from '../types';\nimport { identity } from '../util/identity';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\nexport function distinctUntilChanged<T>(comparator?: (previous: T, current: T) => boolean): MonoTypeOperatorFunction<T>;\nexport function distinctUntilChanged<T, K>(\n  comparator: (previous: K, current: K) => boolean,\n  keySelector: (value: T) => K\n): MonoTypeOperatorFunction<T>;\n\n/**\n * Returns a result {@link Observable} that emits all values pushed by the source observable if they\n * are distinct in comparison to the last value the result observable emitted.\n *\n * When provided without parameters or with the first parameter (`{@link distinctUntilChanged#comparator comparator}`),\n * it behaves like this:\n *\n * 1. It will always emit the first value from the source.\n * 2. For all subsequent values pushed by the source, they will be compared to the previously emitted values\n *    using the provided `comparator` or an `===` equality check.\n * 3. If the value pushed by the source is determined to be unequal by this check, that value is emitted and\n *    becomes the new \"previously emitted value\" internally.\n *\n * When the second parameter (`{@link distinctUntilChanged#keySelector keySelector}`) is provided, the behavior\n * changes:\n *\n * 1. It will always emit the first value from the source.\n * 2. The `keySelector` will be run against all values, including the first value.\n * 3. For all values after the first, the selected key will be compared against the key selected from\n *    the previously emitted value using the `comparator`.\n * 4. If the keys are determined to be unequal by this check, the value (not the key), is emitted\n *    and the selected key from that value is saved for future comparisons against other keys.\n *\n * ## Examples\n *\n * A very basic example with no `{@link distinctUntilChanged#comparator comparator}`. Note that `1` is emitted more than once,\n * because it's distinct in comparison to the _previously emitted_ value,\n * not in comparison to _all other emitted values_.\n *\n * ```ts\n * import { of, distinctUntilChanged } from 'rxjs';\n *\n * of(1, 1, 1, 2, 2, 2, 1, 1, 3, 3)\n *   .pipe(distinctUntilChanged())\n *   .subscribe(console.log);\n * // Logs: 1, 2, 1, 3\n * ```\n *\n * With a `{@link distinctUntilChanged#comparator comparator}`, you can do custom comparisons. Let's say\n * you only want to emit a value when all of its components have\n * changed:\n *\n * ```ts\n * import { of, distinctUntilChanged } from 'rxjs';\n *\n * const totallyDifferentBuilds$ = of(\n *   { engineVersion: '1.1.0', transmissionVersion: '1.2.0' },\n *   { engineVersion: '1.1.0', transmissionVersion: '1.4.0' },\n *   { engineVersion: '1.3.0', transmissionVersion: '1.4.0' },\n *   { engineVersion: '1.3.0', transmissionVersion: '1.5.0' },\n *   { engineVersion: '2.0.0', transmissionVersion: '1.5.0' }\n * ).pipe(\n *   distinctUntilChanged((prev, curr) => {\n *     return (\n *       prev.engineVersion === curr.engineVersion ||\n *       prev.transmissionVersion === curr.transmissionVersion\n *     );\n *   })\n * );\n *\n * totallyDifferentBuilds$.subscribe(console.log);\n *\n * // Logs:\n * // { engineVersion: '1.1.0', transmissionVersion: '1.2.0' }\n * // { engineVersion: '1.3.0', transmissionVersion: '1.4.0' }\n * // { engineVersion: '2.0.0', transmissionVersion: '1.5.0' }\n * ```\n *\n * You can also provide a custom `{@link distinctUntilChanged#comparator comparator}` to check that emitted\n * changes are only in one direction. Let's say you only want to get\n * the next record temperature:\n *\n * ```ts\n * import { of, distinctUntilChanged } from 'rxjs';\n *\n * const temps$ = of(30, 31, 20, 34, 33, 29, 35, 20);\n *\n * const recordHighs$ = temps$.pipe(\n *   distinctUntilChanged((prevHigh, temp) => {\n *     // If the current temp is less than\n *     // or the same as the previous record,\n *     // the record hasn't changed.\n *     return temp <= prevHigh;\n *   })\n * );\n *\n * recordHighs$.subscribe(console.log);\n * // Logs: 30, 31, 34, 35\n * ```\n *\n * Selecting update events only when the `updatedBy` field shows\n * the account changed hands.\n *\n * ```ts\n * import { of, distinctUntilChanged } from 'rxjs';\n *\n * // A stream of updates to a given account\n * const accountUpdates$ = of(\n *   { updatedBy: 'blesh', data: [] },\n *   { updatedBy: 'blesh', data: [] },\n *   { updatedBy: 'ncjamieson', data: [] },\n *   { updatedBy: 'ncjamieson', data: [] },\n *   { updatedBy: 'blesh', data: [] }\n * );\n *\n * // We only want the events where it changed hands\n * const changedHands$ = accountUpdates$.pipe(\n *   distinctUntilChanged(undefined, update => update.updatedBy)\n * );\n *\n * changedHands$.subscribe(console.log);\n * // Logs:\n * // { updatedBy: 'blesh', data: Array[0] }\n * // { updatedBy: 'ncjamieson', data: Array[0] }\n * // { updatedBy: 'blesh', data: Array[0] }\n * ```\n *\n * @see {@link distinct}\n * @see {@link distinctUntilKeyChanged}\n *\n * @param comparator A function used to compare the previous and current keys for\n * equality. Defaults to a `===` check.\n * @param keySelector Used to select a key value to be passed to the `comparator`.\n *\n * @return A function that returns an Observable that emits items from the\n * source Observable with distinct values.\n */\nexport function distinctUntilChanged<T, K>(\n  comparator?: (previous: K, current: K) => boolean,\n  keySelector: (value: T) => K = identity as (value: T) => K\n): MonoTypeOperatorFunction<T> {\n  // We've been allowing `null` do be passed as the `compare`, so we can't do\n  // a default value for the parameter, because that will only work\n  // for `undefined`.\n  comparator = comparator ?? defaultCompare;\n\n  return operate((source, subscriber) => {\n    // The previous key, used to compare against keys selected\n    // from new arrivals to determine \"distinctiveness\".\n    let previousKey: K;\n    // Whether or not this is the first value we've gotten.\n    let first = true;\n\n    source.subscribe(\n      createOperatorSubscriber(subscriber, (value) => {\n        // We always call the key selector.\n        const currentKey = keySelector(value);\n\n        // If it's the first value, we always emit it.\n        // Otherwise, we compare this key to the previous key, and\n        // if the comparer returns false, we emit.\n        if (first || !comparator!(previousKey, currentKey)) {\n          // Update our state *before* we emit the value\n          // as emission can be the source of re-entrant code\n          // in functional libraries like this. We only really\n          // need to do this if it's the first value, or if the\n          // key we're tracking in previous needs to change.\n          first = false;\n          previousKey = currentKey;\n\n          // Emit the value!\n          subscriber.next(value);\n        }\n      })\n    );\n  });\n}\n\nfunction defaultCompare(a: any, b: any) {\n  return a === b;\n}\n", "import { distinctUntilChanged } from './distinctUntilChanged';\nimport { MonoTypeOperatorFunction } from '../types';\n\n/* tslint:disable:max-line-length */\nexport function distinctUntilKeyChanged<T>(key: keyof T): MonoTypeOperatorFunction<T>;\nexport function distinctUntilKeyChanged<T, K extends keyof T>(key: K, compare: (x: T[K], y: T[K]) => boolean): MonoTypeOperatorFunction<T>;\n/* tslint:enable:max-line-length */\n\n/**\n * Returns an Observable that emits all items emitted by the source Observable that are distinct by comparison from the previous item,\n * using a property accessed by using the key provided to check if the two items are distinct.\n *\n * If a comparator function is provided, then it will be called for each item to test for whether or not that value should be emitted.\n *\n * If a comparator function is not provided, an equality check is used by default.\n *\n * ## Examples\n *\n * An example comparing the name of persons\n *\n * ```ts\n * import { of, distinctUntilKeyChanged } from 'rxjs';\n *\n * of(\n *   { age: 4, name: 'Foo' },\n *   { age: 7, name: 'Bar' },\n *   { age: 5, name: 'Foo' },\n *   { age: 6, name: 'Foo' }\n * ).pipe(\n *   distinctUntilKeyChanged('name')\n * )\n * .subscribe(x => console.log(x));\n *\n * // displays:\n * // { age: 4, name: 'Foo' }\n * // { age: 7, name: 'Bar' }\n * // { age: 5, name: 'Foo' }\n * ```\n *\n * An example comparing the first letters of the name\n *\n * ```ts\n * import { of, distinctUntilKeyChanged } from 'rxjs';\n *\n * of(\n *   { age: 4, name: 'Foo1' },\n *   { age: 7, name: 'Bar' },\n *   { age: 5, name: 'Foo2' },\n *   { age: 6, name: 'Foo3' }\n * ).pipe(\n *   distinctUntilKeyChanged('name', (x, y) => x.substring(0, 3) === y.substring(0, 3))\n * )\n * .subscribe(x => console.log(x));\n *\n * // displays:\n * // { age: 4, name: 'Foo1' }\n * // { age: 7, name: 'Bar' }\n * // { age: 5, name: 'Foo2' }\n * ```\n *\n * @see {@link distinct}\n * @see {@link distinctUntilChanged}\n *\n * @param {string} key String key for object property lookup on each item.\n * @param {function} [compare] Optional comparison function called to test if an item is distinct from the previous item in the source.\n * @return A function that returns an Observable that emits items from the\n * source Observable with distinct values based on the key specified.\n */\nexport function distinctUntilKeyChanged<T, K extends keyof T>(key: K, compare?: (x: T[K], y: T[K]) => boolean): MonoTypeOperatorFunction<T> {\n  return distinctUntilChanged((x: T, y: T) => compare ? compare(x[key], y[key]) : x[key] === y[key]);\n}\n", "import { EmptyError } from '../util/EmptyError';\nimport { MonoTypeOperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * If the source observable completes without emitting a value, it will emit\n * an error. The error will be created at that time by the optional\n * `errorFactory` argument, otherwise, the error will be {@link EmptyError}.\n *\n * ![](throwIfEmpty.png)\n *\n * ## Example\n *\n * Throw an error if the document wasn't clicked within 1 second\n *\n * ```ts\n * import { fromEvent, takeUntil, timer, throwIfEmpty } from 'rxjs';\n *\n * const click$ = fromEvent(document, 'click');\n *\n * click$.pipe(\n *   takeUntil(timer(1000)),\n *   throwIfEmpty(() => new Error('The document was not clicked within 1 second'))\n * )\n * .subscribe({\n *   next() {\n *    console.log('The document was clicked');\n *   },\n *   error(err) {\n *     console.error(err.message);\n *   }\n * });\n * ```\n *\n * @param errorFactory A factory function called to produce the\n * error to be thrown when the source observable completes without emitting a\n * value.\n * @return A function that returns an Observable that throws an error if the\n * source Observable completed without emitting.\n */\nexport function throwIfEmpty<T>(errorFactory: () => any = defaultErrorFactory): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    let hasValue = false;\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value) => {\n          hasValue = true;\n          subscriber.next(value);\n        },\n        () => (hasValue ? subscriber.complete() : subscriber.error(errorFactory()))\n      )\n    );\n  });\n}\n\nfunction defaultErrorFactory() {\n  return new EmptyError();\n}\n", "/** prettier */\nimport { Observable } from '../Observable';\nimport { concat } from '../observable/concat';\nimport { of } from '../observable/of';\nimport { MonoTypeOperatorFunction, SchedulerLike, OperatorFunction, ValueFromArray } from '../types';\n\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `concatAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function endWith<T>(scheduler: SchedulerLike): MonoTypeOperatorFunction<T>;\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `concatAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function endWith<T, A extends unknown[] = T[]>(\n  ...valuesAndScheduler: [...A, SchedulerLike]\n): OperatorFunction<T, T | ValueFromArray<A>>;\n\nexport function endWith<T, A extends unknown[] = T[]>(...values: A): OperatorFunction<T, T | ValueFromArray<A>>;\n\n/**\n * Returns an observable that will emit all values from the source, then synchronously emit\n * the provided value(s) immediately after the source completes.\n *\n * NOTE: Passing a last argument of a Scheduler is _deprecated_, and may result in incorrect\n * types in TypeScript.\n *\n * This is useful for knowing when an observable ends. Particularly when paired with an\n * operator like {@link takeUntil}\n *\n * ![](endWith.png)\n *\n * ## Example\n *\n * Emit values to know when an interval starts and stops. The interval will\n * stop when a user clicks anywhere on the document.\n *\n * ```ts\n * import { interval, map, fromEvent, startWith, takeUntil, endWith } from 'rxjs';\n *\n * const ticker$ = interval(5000).pipe(\n *   map(() => 'tick')\n * );\n *\n * const documentClicks$ = fromEvent(document, 'click');\n *\n * ticker$.pipe(\n *   startWith('interval started'),\n *   takeUntil(documentClicks$),\n *   endWith('interval ended by click')\n * )\n * .subscribe(x => console.log(x));\n *\n * // Result (assuming a user clicks after 15 seconds)\n * // 'interval started'\n * // 'tick'\n * // 'tick'\n * // 'tick'\n * // 'interval ended by click'\n * ```\n *\n * @see {@link startWith}\n * @see {@link concat}\n * @see {@link takeUntil}\n *\n * @param values Items you want the modified Observable to emit last.\n * @return A function that returns an Observable that emits all values from the\n * source, then synchronously emits the provided value(s) immediately after the\n * source completes.\n */\nexport function endWith<T>(...values: Array<T | SchedulerLike>): MonoTypeOperatorFunction<T> {\n  return (source: Observable<T>) => concat(source, of(...values)) as Observable<T>;\n}\n", "import { MonoTypeOperatorFunction } from '../types';\nimport { operate } from '../util/lift';\n\n/**\n * Returns an Observable that mirrors the source Observable, but will call a specified function when\n * the source terminates on complete or error.\n * The specified function will also be called when the subscriber explicitly unsubscribes.\n *\n * ## Examples\n *\n * Execute callback function when the observable completes\n *\n * ```ts\n * import { interval, take, finalize } from 'rxjs';\n *\n * // emit value in sequence every 1 second\n * const source = interval(1000);\n * const example = source.pipe(\n *   take(5), //take only the first 5 values\n *   finalize(() => console.log('Sequence complete')) // Execute when the observable completes\n * );\n * const subscribe = example.subscribe(val => console.log(val));\n *\n * // results:\n * // 0\n * // 1\n * // 2\n * // 3\n * // 4\n * // 'Sequence complete'\n * ```\n *\n * Execute callback function when the subscriber explicitly unsubscribes\n *\n * ```ts\n * import { interval, finalize, tap, noop, timer } from 'rxjs';\n *\n * const source = interval(100).pipe(\n *   finalize(() => console.log('[finalize] Called')),\n *   tap({\n *     next: () => console.log('[next] Called'),\n *     error: () => console.log('[error] Not called'),\n *     complete: () => console.log('[tap complete] Not called')\n *   })\n * );\n *\n * const sub = source.subscribe({\n *   next: x => console.log(x),\n *   error: noop,\n *   complete: () => console.log('[complete] Not called')\n * });\n *\n * timer(150).subscribe(() => sub.unsubscribe());\n *\n * // results:\n * // '[next] Called'\n * // 0\n * // '[finalize] Called'\n * ```\n *\n * @param {function} callback Function to be called when source terminates.\n * @return A function that returns an Observable that mirrors the source, but\n * will call the specified function on termination.\n */\nexport function finalize<T>(callback: () => void): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    // TODO: This try/finally was only added for `useDeprecatedSynchronousErrorHandling`.\n    // REMOVE THIS WHEN THAT HOT GARBAGE IS REMOVED IN V8.\n    try {\n      source.subscribe(subscriber);\n    } finally {\n      subscriber.add(callback);\n    }\n  });\n}\n", "import { Observable } from '../Observable';\nimport { EmptyError } from '../util/EmptyError';\nimport { OperatorFunction, TruthyTypesOf } from '../types';\nimport { filter } from './filter';\nimport { take } from './take';\nimport { defaultIfEmpty } from './defaultIfEmpty';\nimport { throwIfEmpty } from './throwIfEmpty';\nimport { identity } from '../util/identity';\n\nexport function first<T, D = T>(predicate?: null, defaultValue?: D): OperatorFunction<T, T | D>;\nexport function first<T>(predicate: BooleanConstructor): OperatorFunction<T, TruthyTypesOf<T>>;\nexport function first<T, D>(predicate: BooleanConstructor, defaultValue: D): OperatorFunction<T, TruthyTypesOf<T> | D>;\nexport function first<T, S extends T>(\n  predicate: (value: T, index: number, source: Observable<T>) => value is S,\n  defaultValue?: S\n): OperatorFunction<T, S>;\nexport function first<T, S extends T, D>(\n  predicate: (value: T, index: number, source: Observable<T>) => value is S,\n  defaultValue: D\n): OperatorFunction<T, S | D>;\nexport function first<T, D = T>(\n  predicate: (value: T, index: number, source: Observable<T>) => boolean,\n  defaultValue?: D\n): OperatorFunction<T, T | D>;\n\n/**\n * Emits only the first value (or the first value that meets some condition)\n * emitted by the source Observable.\n *\n * <span class=\"informal\">Emits only the first value. Or emits only the first\n * value that passes some test.</span>\n *\n * ![](first.png)\n *\n * If called with no arguments, `first` emits the first value of the source\n * Observable, then completes. If called with a `predicate` function, `first`\n * emits the first value of the source that matches the specified condition. Throws an error if\n * `defaultValue` was not provided and a matching element is not found.\n *\n * ## Examples\n *\n * Emit only the first click that happens on the DOM\n *\n * ```ts\n * import { fromEvent, first } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(first());\n * result.subscribe(x => console.log(x));\n * ```\n *\n * Emits the first click that happens on a DIV\n *\n * ```ts\n * import { fromEvent, first } from 'rxjs';\n *\n * const div = document.createElement('div');\n * div.style.cssText = 'width: 200px; height: 200px; background: #09c;';\n * document.body.appendChild(div);\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(first(ev => (<HTMLElement>ev.target).tagName === 'DIV'));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link filter}\n * @see {@link find}\n * @see {@link take}\n *\n * @throws {EmptyError} Delivers an EmptyError to the Observer's `error`\n * callback if the Observable completes before any `next` notification was sent.\n * This is how `first()` is different from {@link take}(1) which completes instead.\n *\n * @param {function(value: T, index: number, source: Observable<T>): boolean} [predicate]\n * An optional function called with each item to test for condition matching.\n * @param {D} [defaultValue] The default value emitted in case no valid value\n * was found on the source.\n * @return A function that returns an Observable that emits the first item that\n * matches the condition.\n */\nexport function first<T, D>(\n  predicate?: ((value: T, index: number, source: Observable<T>) => boolean) | null,\n  defaultValue?: D\n): OperatorFunction<T, T | D> {\n  const hasDefaultValue = arguments.length >= 2;\n  return (source: Observable<T>) =>\n    source.pipe(\n      predicate ? filter((v, i) => predicate(v, i, source)) : identity,\n      take(1),\n      hasDefaultValue ? defaultIfEmpty(defaultValue!) : throwIfEmpty(() => new EmptyError())\n    );\n}\n", "import { ObservableInput, ObservableInputTuple, OperatorFunction, SchedulerLike } from '../types';\nimport { operate } from '../util/lift';\nimport { argsOrArgArray } from '../util/argsOrArgArray';\nimport { mergeAll } from './mergeAll';\nimport { popNumber, popScheduler } from '../util/args';\nimport { from } from '../observable/from';\n\n/** @deprecated Replaced with {@link mergeWith}. Will be removed in v8. */\nexport function merge<T, A extends readonly unknown[]>(...sources: [...ObservableInputTuple<A>]): OperatorFunction<T, T | A[number]>;\n/** @deprecated Replaced with {@link mergeWith}. Will be removed in v8. */\nexport function merge<T, A extends readonly unknown[]>(\n  ...sourcesAndConcurrency: [...ObservableInputTuple<A>, number]\n): OperatorFunction<T, T | A[number]>;\n/** @deprecated Replaced with {@link mergeWith}. Will be removed in v8. */\nexport function merge<T, A extends readonly unknown[]>(\n  ...sourcesAndScheduler: [...ObservableInputTuple<A>, SchedulerLike]\n): OperatorFunction<T, T | A[number]>;\n/** @deprecated Replaced with {@link mergeWith}. Will be removed in v8. */\nexport function merge<T, A extends readonly unknown[]>(\n  ...sourcesAndConcurrencyAndScheduler: [...ObservableInputTuple<A>, number, SchedulerLike]\n): OperatorFunction<T, T | A[number]>;\n\nexport function merge<T>(...args: unknown[]): OperatorFunction<T, unknown> {\n  const scheduler = popScheduler(args);\n  const concurrent = popNumber(args, Infinity);\n  args = argsOrArgArray(args);\n\n  return operate((source, subscriber) => {\n    mergeAll(concurrent)(from([source, ...(args as ObservableInput<T>[])], scheduler)).subscribe(subscriber);\n  });\n}\n", "import { ObservableInputTuple, OperatorFunction } from '../types';\nimport { merge } from './merge';\n\n/**\n * Merge the values from all observables to a single observable result.\n *\n * Creates an observable, that when subscribed to, subscribes to the source\n * observable, and all other sources provided as arguments. All values from\n * every source are emitted from the resulting subscription.\n *\n * When all sources complete, the resulting observable will complete.\n *\n * When any source errors, the resulting observable will error.\n *\n * ## Example\n *\n * Joining all outputs from multiple user input event streams\n *\n * ```ts\n * import { fromEvent, map, mergeWith } from 'rxjs';\n *\n * const clicks$ = fromEvent(document, 'click').pipe(map(() => 'click'));\n * const mousemoves$ = fromEvent(document, 'mousemove').pipe(map(() => 'mousemove'));\n * const dblclicks$ = fromEvent(document, 'dblclick').pipe(map(() => 'dblclick'));\n *\n * mousemoves$\n *   .pipe(mergeWith(clicks$, dblclicks$))\n *   .subscribe(x => console.log(x));\n *\n * // result (assuming user interactions)\n * // 'mousemove'\n * // 'mousemove'\n * // 'mousemove'\n * // 'click'\n * // 'click'\n * // 'dblclick'\n * ```\n *\n * @see {@link merge}\n *\n * @param otherSources the sources to combine the current source with.\n * @return A function that returns an Observable that merges the values from\n * all given Observables.\n */\nexport function mergeWith<T, A extends readonly unknown[]>(\n  ...otherSources: [...ObservableInputTuple<A>]\n): OperatorFunction<T, T | A[number]> {\n  return merge(...otherSources);\n}\n", "import { Subscription } from '../Subscription';\nimport { EMPTY } from '../observable/empty';\nimport { operate } from '../util/lift';\nimport { MonoTypeOperatorFunction, ObservableInput } from '../types';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { innerFrom } from '../observable/innerFrom';\nimport { timer } from '../observable/timer';\n\nexport interface RepeatConfig {\n  /**\n   * The number of times to repeat the source. Defaults to `Infinity`.\n   */\n  count?: number;\n\n  /**\n   * If a `number`, will delay the repeat of the source by that number of milliseconds.\n   * If a function, it will provide the number of times the source has been subscribed to,\n   * and the return value should be a valid observable input that will notify when the source\n   * should be repeated. If the notifier observable is empty, the result will complete.\n   */\n  delay?: number | ((count: number) => ObservableInput<any>);\n}\n\n/**\n * Returns an Observable that will resubscribe to the source stream when the source stream completes.\n *\n * <span class=\"informal\">Repeats all values emitted on the source. It's like {@link retry}, but for non error cases.</span>\n *\n * ![](repeat.png)\n *\n * Repeat will output values from a source until the source completes, then it will resubscribe to the\n * source a specified number of times, with a specified delay. Repeat can be particularly useful in\n * combination with closing operators like {@link take}, {@link takeUntil}, {@link first}, or {@link takeWhile},\n * as it can be used to restart a source again from scratch.\n *\n * Repeat is very similar to {@link retry}, where {@link retry} will resubscribe to the source in the error case, but\n * `repeat` will resubscribe if the source completes.\n *\n * Note that `repeat` will _not_ catch errors. Use {@link retry} for that.\n *\n * - `repeat(0)` returns an empty observable\n * - `repeat()` will repeat forever\n * - `repeat({ delay: 200 })` will repeat forever, with a delay of 200ms between repetitions.\n * - `repeat({ count: 2, delay: 400 })` will repeat twice, with a delay of 400ms between repetitions.\n * - `repeat({ delay: (count) => timer(count * 1000) })` will repeat forever, but will have a delay that grows by one second for each repetition.\n *\n * ## Example\n *\n * Repeat a message stream\n *\n * ```ts\n * import { of, repeat } from 'rxjs';\n *\n * const source = of('Repeat message');\n * const result = source.pipe(repeat(3));\n *\n * result.subscribe(x => console.log(x));\n *\n * // Results\n * // 'Repeat message'\n * // 'Repeat message'\n * // 'Repeat message'\n * ```\n *\n * Repeat 3 values, 2 times\n *\n * ```ts\n * import { interval, take, repeat } from 'rxjs';\n *\n * const source = interval(1000);\n * const result = source.pipe(take(3), repeat(2));\n *\n * result.subscribe(x => console.log(x));\n *\n * // Results every second\n * // 0\n * // 1\n * // 2\n * // 0\n * // 1\n * // 2\n * ```\n *\n * Defining two complex repeats with delays on the same source.\n * Note that the second repeat cannot be called until the first\n * repeat as exhausted it's count.\n *\n * ```ts\n * import { defer, of, repeat } from 'rxjs';\n *\n * const source = defer(() => {\n *    return of(`Hello, it is ${new Date()}`)\n * });\n *\n * source.pipe(\n *    // Repeat 3 times with a delay of 1 second between repetitions\n *    repeat({\n *      count: 3,\n *      delay: 1000,\n *    }),\n *\n *    // *Then* repeat forever, but with an exponential step-back\n *    // maxing out at 1 minute.\n *    repeat({\n *      delay: (count) => timer(Math.min(60000, 2 ^ count * 1000))\n *    })\n * )\n * ```\n *\n * @see {@link repeatWhen}\n * @see {@link retry}\n *\n * @param count The number of times the source Observable items are repeated, a count of 0 will yield\n * an empty Observable.\n */\nexport function repeat<T>(countOrConfig?: number | RepeatConfig): MonoTypeOperatorFunction<T> {\n  let count = Infinity;\n  let delay: RepeatConfig['delay'];\n\n  if (countOrConfig != null) {\n    if (typeof countOrConfig === 'object') {\n      ({ count = Infinity, delay } = countOrConfig);\n    } else {\n      count = countOrConfig;\n    }\n  }\n\n  return count <= 0\n    ? () => EMPTY\n    : operate((source, subscriber) => {\n        let soFar = 0;\n        let sourceSub: Subscription | null;\n\n        const resubscribe = () => {\n          sourceSub?.unsubscribe();\n          sourceSub = null;\n          if (delay != null) {\n            const notifier = typeof delay === 'number' ? timer(delay) : innerFrom(delay(soFar));\n            const notifierSubscriber = createOperatorSubscriber(subscriber, () => {\n              notifierSubscriber.unsubscribe();\n              subscribeToSource();\n            });\n            notifier.subscribe(notifierSubscriber);\n          } else {\n            subscribeToSource();\n          }\n        };\n\n        const subscribeToSource = () => {\n          let syncUnsub = false;\n          sourceSub = source.subscribe(\n            createOperatorSubscriber(subscriber, undefined, () => {\n              if (++soFar < count) {\n                if (sourceSub) {\n                  resubscribe();\n                } else {\n                  syncUnsub = true;\n                }\n              } else {\n                subscriber.complete();\n              }\n            })\n          );\n\n          if (syncUnsub) {\n            resubscribe();\n          }\n        };\n\n        subscribeToSource();\n      });\n}\n", "import { innerFrom } from '../observable/innerFrom';\nimport { MonoTypeOperatorFunction, ObservableInput } from '../types';\nimport { operate } from '../util/lift';\nimport { noop } from '../util/noop';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/**\n * Emits the most recently emitted value from the source Observable whenever\n * another Observable, the `notifier`, emits.\n *\n * <span class=\"informal\">It's like {@link sampleTime}, but samples whenever\n * the `notifier` `ObservableInput` emits something.</span>\n *\n * ![](sample.png)\n *\n * Whenever the `notifier` `ObservableInput` emits a value, `sample`\n * looks at the source Observable and emits whichever value it has most recently\n * emitted since the previous sampling, unless the source has not emitted\n * anything since the previous sampling. The `notifier` is subscribed to as soon\n * as the output Observable is subscribed.\n *\n * ## Example\n *\n * On every click, sample the most recent `seconds` timer\n *\n * ```ts\n * import { fromEvent, interval, sample } from 'rxjs';\n *\n * const seconds = interval(1000);\n * const clicks = fromEvent(document, 'click');\n * const result = seconds.pipe(sample(clicks));\n *\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link audit}\n * @see {@link debounce}\n * @see {@link sampleTime}\n * @see {@link throttle}\n *\n * @param notifier The `ObservableInput` to use for sampling the\n * source Observable.\n * @return A function that returns an Observable that emits the results of\n * sampling the values emitted by the source Observable whenever the notifier\n * Observable emits value or completes.\n */\nexport function sample<T>(notifier: ObservableInput<any>): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    let hasValue = false;\n    let lastValue: T | null = null;\n    source.subscribe(\n      createOperatorSubscriber(subscriber, (value) => {\n        hasValue = true;\n        lastValue = value;\n      })\n    );\n    innerFrom(notifier).subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        () => {\n          if (hasValue) {\n            hasValue = false;\n            const value = lastValue!;\n            lastValue = null;\n            subscriber.next(value);\n          }\n        },\n        noop\n      )\n    );\n  });\n}\n", "import { OperatorFunction } from '../types';\nimport { operate } from '../util/lift';\nimport { scanInternals } from './scanInternals';\n\nexport function scan<V, A = V>(accumulator: (acc: A | V, value: V, index: number) => A): OperatorFunction<V, V | A>;\nexport function scan<V, A>(accumulator: (acc: A, value: V, index: number) => A, seed: A): OperatorFunction<V, A>;\nexport function scan<V, A, S>(accumulator: (acc: A | S, value: V, index: number) => A, seed: S): OperatorFunction<V, A>;\n\n// TODO: link to a \"redux pattern\" section in the guide (location TBD)\n\n/**\n * Useful for encapsulating and managing state. Applies an accumulator (or \"reducer function\")\n * to each value from the source after an initial state is established -- either via\n * a `seed` value (second argument), or from the first value from the source.\n *\n * <span class=\"informal\">It's like {@link reduce}, but emits the current\n * accumulation state after each update</span>\n *\n * ![](scan.png)\n *\n * This operator maintains an internal state and emits it after processing each value as follows:\n *\n * 1. First value arrives\n *   - If a `seed` value was supplied (as the second argument to `scan`), let `state = seed` and `value = firstValue`.\n *   - If NO `seed` value was supplied (no second argument), let `state = firstValue` and go to 3.\n * 2. Let `state = accumulator(state, value)`.\n *   - If an error is thrown by `accumulator`, notify the consumer of an error. The process ends.\n * 3. Emit `state`.\n * 4. Next value arrives, let `value = nextValue`, go to 2.\n *\n * ## Examples\n *\n * An average of previous numbers. This example shows how\n * not providing a `seed` can prime the stream with the\n * first value from the source.\n *\n * ```ts\n * import { of, scan, map } from 'rxjs';\n *\n * const numbers$ = of(1, 2, 3);\n *\n * numbers$\n *   .pipe(\n *     // Get the sum of the numbers coming in.\n *     scan((total, n) => total + n),\n *     // Get the average by dividing the sum by the total number\n *     // received so far (which is 1 more than the zero-based index).\n *     map((sum, index) => sum / (index + 1))\n *   )\n *   .subscribe(console.log);\n * ```\n *\n * The Fibonacci sequence. This example shows how you can use\n * a seed to prime accumulation process. Also... you know... Fibonacci.\n * So important to like, computers and stuff that its whiteboarded\n * in job interviews. Now you can show them the Rx version! (Please don't, haha)\n *\n * ```ts\n * import { interval, scan, map, startWith } from 'rxjs';\n *\n * const firstTwoFibs = [0, 1];\n * // An endless stream of Fibonacci numbers.\n * const fibonacci$ = interval(1000).pipe(\n *   // Scan to get the fibonacci numbers (after 0, 1)\n *   scan(([a, b]) => [b, a + b], firstTwoFibs),\n *   // Get the second number in the tuple, it's the one you calculated\n *   map(([, n]) => n),\n *   // Start with our first two digits :)\n *   startWith(...firstTwoFibs)\n * );\n *\n * fibonacci$.subscribe(console.log);\n * ```\n *\n * @see {@link expand}\n * @see {@link mergeScan}\n * @see {@link reduce}\n * @see {@link switchScan}\n *\n * @param accumulator A \"reducer function\". This will be called for each value after an initial state is\n * acquired.\n * @param seed The initial state. If this is not provided, the first value from the source will\n * be used as the initial state, and emitted without going through the accumulator. All subsequent values\n * will be processed by the accumulator function. If this is provided, all values will go through\n * the accumulator function.\n * @return A function that returns an Observable of the accumulated values.\n */\nexport function scan<V, A, S>(accumulator: (acc: V | A | S, value: V, index: number) => A, seed?: S): OperatorFunction<V, V | A> {\n  // providing a seed of `undefined` *should* be valid and trigger\n  // hasSeed! so don't use `seed !== undefined` checks!\n  // For this reason, we have to check it here at the original call site\n  // otherwise inside Operator/Subscriber we won't know if `undefined`\n  // means they didn't provide anything or if they literally provided `undefined`\n  return operate(scanInternals(accumulator, seed as S, arguments.length >= 2, true));\n}\n", "import { innerFrom } from '../observable/innerFrom';\nimport { Subject } from '../Subject';\nimport { SafeSubscriber } from '../Subscriber';\nimport { Subscription } from '../Subscription';\nimport { MonoTypeOperatorFunction, SubjectLike, ObservableInput } from '../types';\nimport { operate } from '../util/lift';\n\nexport interface ShareConfig<T> {\n  /**\n   * The factory used to create the subject that will connect the source observable to\n   * multicast consumers.\n   */\n  connector?: () => SubjectLike<T>;\n  /**\n   * If `true`, the resulting observable will reset internal state on error from source and return to a \"cold\" state. This\n   * allows the resulting observable to be \"retried\" in the event of an error.\n   * If `false`, when an error comes from the source it will push the error into the connecting subject, and the subject\n   * will remain the connecting subject, meaning the resulting observable will not go \"cold\" again, and subsequent retries\n   * or resubscriptions will resubscribe to that same subject. In all cases, RxJS subjects will emit the same error again, however\n   * {@link ReplaySubject} will also push its buffered values before pushing the error.\n   * It is also possible to pass a notifier factory returning an `ObservableInput` instead which grants more fine-grained\n   * control over how and when the reset should happen. This allows behaviors like conditional or delayed resets.\n   */\n  resetOnError?: boolean | ((error: any) => ObservableInput<any>);\n  /**\n   * If `true`, the resulting observable will reset internal state on completion from source and return to a \"cold\" state. This\n   * allows the resulting observable to be \"repeated\" after it is done.\n   * If `false`, when the source completes, it will push the completion through the connecting subject, and the subject\n   * will remain the connecting subject, meaning the resulting observable will not go \"cold\" again, and subsequent repeats\n   * or resubscriptions will resubscribe to that same subject.\n   * It is also possible to pass a notifier factory returning an `ObservableInput` instead which grants more fine-grained\n   * control over how and when the reset should happen. This allows behaviors like conditional or delayed resets.\n   */\n  resetOnComplete?: boolean | (() => ObservableInput<any>);\n  /**\n   * If `true`, when the number of subscribers to the resulting observable reaches zero due to those subscribers unsubscribing, the\n   * internal state will be reset and the resulting observable will return to a \"cold\" state. This means that the next\n   * time the resulting observable is subscribed to, a new subject will be created and the source will be subscribed to\n   * again.\n   * If `false`, when the number of subscribers to the resulting observable reaches zero due to unsubscription, the subject\n   * will remain connected to the source, and new subscriptions to the result will be connected through that same subject.\n   * It is also possible to pass a notifier factory returning an `ObservableInput` instead which grants more fine-grained\n   * control over how and when the reset should happen. This allows behaviors like conditional or delayed resets.\n   */\n  resetOnRefCountZero?: boolean | (() => ObservableInput<any>);\n}\n\nexport function share<T>(): MonoTypeOperatorFunction<T>;\n\nexport function share<T>(options: ShareConfig<T>): MonoTypeOperatorFunction<T>;\n\n/**\n * Returns a new Observable that multicasts (shares) the original Observable. As long as there is at least one\n * Subscriber this Observable will be subscribed and emitting data. When all subscribers have unsubscribed it will\n * unsubscribe from the source Observable. Because the Observable is multicasting it makes the stream `hot`.\n * This is an alias for `multicast(() => new Subject()), refCount()`.\n *\n * The subscription to the underlying source Observable can be reset (unsubscribe and resubscribe for new subscribers),\n * if the subscriber count to the shared observable drops to 0, or if the source Observable errors or completes. It is\n * possible to use notifier factories for the resets to allow for behaviors like conditional or delayed resets. Please\n * note that resetting on error or complete of the source Observable does not behave like a transparent retry or restart\n * of the source because the error or complete will be forwarded to all subscribers and their subscription will be\n * closed. Only new subscribers after a reset on error or complete happened will cause a fresh subscription to the\n * source. To achieve transparent retries or restarts pipe the source through appropriate operators before sharing.\n *\n * ![](share.png)\n *\n * ## Example\n *\n * Generate new multicast Observable from the `source` Observable value\n *\n * ```ts\n * import { interval, tap, map, take, share } from 'rxjs';\n *\n * const source = interval(1000).pipe(\n *   tap(x => console.log('Processing: ', x)),\n *   map(x => x * x),\n *   take(6),\n *   share()\n * );\n *\n * source.subscribe(x => console.log('subscription 1: ', x));\n * source.subscribe(x => console.log('subscription 2: ', x));\n *\n * // Logs:\n * // Processing: 0\n * // subscription 1: 0\n * // subscription 2: 0\n * // Processing: 1\n * // subscription 1: 1\n * // subscription 2: 1\n * // Processing: 2\n * // subscription 1: 4\n * // subscription 2: 4\n * // Processing: 3\n * // subscription 1: 9\n * // subscription 2: 9\n * // Processing: 4\n * // subscription 1: 16\n * // subscription 2: 16\n * // Processing: 5\n * // subscription 1: 25\n * // subscription 2: 25\n * ```\n *\n * ## Example with notifier factory: Delayed reset\n *\n * ```ts\n * import { interval, take, share, timer } from 'rxjs';\n *\n * const source = interval(1000).pipe(\n *   take(3),\n *   share({\n *     resetOnRefCountZero: () => timer(1000)\n *   })\n * );\n *\n * const subscriptionOne = source.subscribe(x => console.log('subscription 1: ', x));\n * setTimeout(() => subscriptionOne.unsubscribe(), 1300);\n *\n * setTimeout(() => source.subscribe(x => console.log('subscription 2: ', x)), 1700);\n *\n * setTimeout(() => source.subscribe(x => console.log('subscription 3: ', x)), 5000);\n *\n * // Logs:\n * // subscription 1:  0\n * // (subscription 1 unsubscribes here)\n * // (subscription 2 subscribes here ~400ms later, source was not reset)\n * // subscription 2:  1\n * // subscription 2:  2\n * // (subscription 2 unsubscribes here)\n * // (subscription 3 subscribes here ~2000ms later, source did reset before)\n * // subscription 3:  0\n * // subscription 3:  1\n * // subscription 3:  2\n * ```\n *\n * @see {@link shareReplay}\n *\n * @return A function that returns an Observable that mirrors the source.\n */\nexport function share<T>(options: ShareConfig<T> = {}): MonoTypeOperatorFunction<T> {\n  const { connector = () => new Subject<T>(), resetOnError = true, resetOnComplete = true, resetOnRefCountZero = true } = options;\n  // It's necessary to use a wrapper here, as the _operator_ must be\n  // referentially transparent. Otherwise, it cannot be used in calls to the\n  // static `pipe` function - to create a partial pipeline.\n  //\n  // The _operator function_ - the function returned by the _operator_ - will\n  // not be referentially transparent - as it shares its source - but the\n  // _operator function_ is called when the complete pipeline is composed via a\n  // call to a source observable's `pipe` method - not when the static `pipe`\n  // function is called.\n  return (wrapperSource) => {\n    let connection: SafeSubscriber<T> | undefined;\n    let resetConnection: Subscription | undefined;\n    let subject: SubjectLike<T> | undefined;\n    let refCount = 0;\n    let hasCompleted = false;\n    let hasErrored = false;\n\n    const cancelReset = () => {\n      resetConnection?.unsubscribe();\n      resetConnection = undefined;\n    };\n    // Used to reset the internal state to a \"cold\"\n    // state, as though it had never been subscribed to.\n    const reset = () => {\n      cancelReset();\n      connection = subject = undefined;\n      hasCompleted = hasErrored = false;\n    };\n    const resetAndUnsubscribe = () => {\n      // We need to capture the connection before\n      // we reset (if we need to reset).\n      const conn = connection;\n      reset();\n      conn?.unsubscribe();\n    };\n\n    return operate<T, T>((source, subscriber) => {\n      refCount++;\n      if (!hasErrored && !hasCompleted) {\n        cancelReset();\n      }\n\n      // Create the subject if we don't have one yet. Grab a local reference to\n      // it as well, which avoids non-null assertions when using it and, if we\n      // connect to it now, then error/complete need a reference after it was\n      // reset.\n      const dest = (subject = subject ?? connector());\n\n      // Add the finalization directly to the subscriber - instead of returning it -\n      // so that the handling of the subscriber's unsubscription will be wired\n      // up _before_ the subscription to the source occurs. This is done so that\n      // the assignment to the source connection's `closed` property will be seen\n      // by synchronous firehose sources.\n      subscriber.add(() => {\n        refCount--;\n\n        // If we're resetting on refCount === 0, and it's 0, we only want to do\n        // that on \"unsubscribe\", really. Resetting on error or completion is a different\n        // configuration.\n        if (refCount === 0 && !hasErrored && !hasCompleted) {\n          resetConnection = handleReset(resetAndUnsubscribe, resetOnRefCountZero);\n        }\n      });\n\n      // The following line adds the subscription to the subscriber passed.\n      // Basically, `subscriber === dest.subscribe(subscriber)` is `true`.\n      dest.subscribe(subscriber);\n\n      if (\n        !connection &&\n        // Check this shareReplay is still activate - it can be reset to 0\n        // and be \"unsubscribed\" _before_ it actually subscribes.\n        // If we were to subscribe then, it'd leak and get stuck.\n        refCount > 0\n      ) {\n        // We need to create a subscriber here - rather than pass an observer and\n        // assign the returned subscription to connection - because it's possible\n        // for reentrant subscriptions to the shared observable to occur and in\n        // those situations we want connection to be already-assigned so that we\n        // don't create another connection to the source.\n        connection = new SafeSubscriber({\n          next: (value) => dest.next(value),\n          error: (err) => {\n            hasErrored = true;\n            cancelReset();\n            resetConnection = handleReset(reset, resetOnError, err);\n            dest.error(err);\n          },\n          complete: () => {\n            hasCompleted = true;\n            cancelReset();\n            resetConnection = handleReset(reset, resetOnComplete);\n            dest.complete();\n          },\n        });\n        innerFrom(source).subscribe(connection);\n      }\n    })(wrapperSource);\n  };\n}\n\nfunction handleReset<T extends unknown[] = never[]>(\n  reset: () => void,\n  on: boolean | ((...args: T) => ObservableInput<any>),\n  ...args: T\n): Subscription | undefined {\n  if (on === true) {\n    reset();\n    return;\n  }\n\n  if (on === false) {\n    return;\n  }\n\n  const onSubscriber = new SafeSubscriber({\n    next: () => {\n      onSubscriber.unsubscribe();\n      reset();\n    },\n  });\n\n  return innerFrom(on(...args)).subscribe(onSubscriber);\n}\n", "import { ReplaySubject } from '../ReplaySubject';\nimport { MonoTypeOperatorFunction, SchedulerLike } from '../types';\nimport { share } from './share';\n\nexport interface ShareReplayConfig {\n  bufferSize?: number;\n  windowTime?: number;\n  refCount: boolean;\n  scheduler?: SchedulerLike;\n}\n\nexport function shareReplay<T>(config: ShareReplayConfig): MonoTypeOperatorFunction<T>;\nexport function shareReplay<T>(bufferSize?: number, windowTime?: number, scheduler?: SchedulerLike): MonoTypeOperatorFunction<T>;\n\n/**\n * Share source and replay specified number of emissions on subscription.\n *\n * This operator is a specialization of `replay` that connects to a source observable\n * and multicasts through a `ReplaySubject` constructed with the specified arguments.\n * A successfully completed source will stay cached in the `shareReplay`ed observable forever,\n * but an errored source can be retried.\n *\n * ## Why use `shareReplay`?\n *\n * You generally want to use `shareReplay` when you have side-effects or taxing computations\n * that you do not wish to be executed amongst multiple subscribers.\n * It may also be valuable in situations where you know you will have late subscribers to\n * a stream that need access to previously emitted values.\n * This ability to replay values on subscription is what differentiates {@link share} and `shareReplay`.\n *\n * ## Reference counting\n *\n * By default `shareReplay` will use `refCount` of false, meaning that it will _not_ unsubscribe the\n * source when the reference counter drops to zero, i.e. the inner `ReplaySubject` will _not_ be unsubscribed\n * (and potentially run for ever).\n * This is the default as it is expected that `shareReplay` is often used to keep around expensive to setup\n * observables which we want to keep running instead of having to do the expensive setup again.\n *\n * As of RXJS version 6.4.0 a new overload signature was added to allow for manual control over what\n * happens when the operators internal reference counter drops to zero.\n * If `refCount` is true, the source will be unsubscribed from once the reference count drops to zero, i.e.\n * the inner `ReplaySubject` will be unsubscribed. All new subscribers will receive value emissions from a\n * new `ReplaySubject` which in turn will cause a new subscription to the source observable.\n *\n * ## Examples\n *\n * Example with a third subscriber coming late to the party\n *\n * ```ts\n * import { interval, take, shareReplay } from 'rxjs';\n *\n * const shared$ = interval(2000).pipe(\n *   take(6),\n *   shareReplay(3)\n * );\n *\n * shared$.subscribe(x => console.log('sub A: ', x));\n * shared$.subscribe(y => console.log('sub B: ', y));\n *\n * setTimeout(() => {\n *   shared$.subscribe(y => console.log('sub C: ', y));\n * }, 11000);\n *\n * // Logs:\n * // (after ~2000 ms)\n * // sub A: 0\n * // sub B: 0\n * // (after ~4000 ms)\n * // sub A: 1\n * // sub B: 1\n * // (after ~6000 ms)\n * // sub A: 2\n * // sub B: 2\n * // (after ~8000 ms)\n * // sub A: 3\n * // sub B: 3\n * // (after ~10000 ms)\n * // sub A: 4\n * // sub B: 4\n * // (after ~11000 ms, sub C gets the last 3 values)\n * // sub C: 2\n * // sub C: 3\n * // sub C: 4\n * // (after ~12000 ms)\n * // sub A: 5\n * // sub B: 5\n * // sub C: 5\n * ```\n *\n * Example for `refCount` usage\n *\n * ```ts\n * import { Observable, tap, interval, shareReplay, take } from 'rxjs';\n *\n * const log = <T>(name: string, source: Observable<T>) => source.pipe(\n *   tap({\n *     subscribe: () => console.log(`${ name }: subscribed`),\n *     next: value => console.log(`${ name }: ${ value }`),\n *     complete: () => console.log(`${ name }: completed`),\n *     finalize: () => console.log(`${ name }: unsubscribed`)\n *   })\n * );\n *\n * const obs$ = log('source', interval(1000));\n *\n * const shared$ = log('shared', obs$.pipe(\n *   shareReplay({ bufferSize: 1, refCount: true }),\n *   take(2)\n * ));\n *\n * shared$.subscribe(x => console.log('sub A: ', x));\n * shared$.subscribe(y => console.log('sub B: ', y));\n *\n * // PRINTS:\n * // shared: subscribed <-- reference count = 1\n * // source: subscribed\n * // shared: subscribed <-- reference count = 2\n * // source: 0\n * // shared: 0\n * // sub A: 0\n * // shared: 0\n * // sub B: 0\n * // source: 1\n * // shared: 1\n * // sub A: 1\n * // shared: completed <-- take(2) completes the subscription for sub A\n * // shared: unsubscribed <-- reference count = 1\n * // shared: 1\n * // sub B: 1\n * // shared: completed <-- take(2) completes the subscription for sub B\n * // shared: unsubscribed <-- reference count = 0\n * // source: unsubscribed <-- replaySubject unsubscribes from source observable because the reference count dropped to 0 and refCount is true\n *\n * // In case of refCount being false, the unsubscribe is never called on the source and the source would keep on emitting, even if no subscribers\n * // are listening.\n * // source: 2\n * // source: 3\n * // source: 4\n * // ...\n * ```\n *\n * @see {@link publish}\n * @see {@link share}\n * @see {@link publishReplay}\n *\n * @param configOrBufferSize Maximum element count of the replay buffer or {@link ShareReplayConfig configuration}\n * object.\n * @param windowTime Maximum time length of the replay buffer in milliseconds.\n * @param scheduler Scheduler where connected observers within the selector function\n * will be invoked on.\n * @return A function that returns an Observable sequence that contains the\n * elements of a sequence produced by multicasting the source sequence within a\n * selector function.\n */\nexport function shareReplay<T>(\n  configOrBufferSize?: ShareReplayConfig | number,\n  windowTime?: number,\n  scheduler?: SchedulerLike\n): MonoTypeOperatorFunction<T> {\n  let bufferSize: number;\n  let refCount = false;\n  if (configOrBufferSize && typeof configOrBufferSize === 'object') {\n    ({ bufferSize = Infinity, windowTime = Infinity, refCount = false, scheduler } = configOrBufferSize);\n  } else {\n    bufferSize = (configOrBufferSize ?? Infinity) as number;\n  }\n  return share<T>({\n    connector: () => new ReplaySubject(bufferSize, windowTime, scheduler),\n    resetOnError: true,\n    resetOnComplete: false,\n    resetOnRefCountZero: refCount,\n  });\n}\n", "import { MonoTypeOperatorFunction } from '../types';\nimport { filter } from './filter';\n\n/**\n * Returns an Observable that skips the first `count` items emitted by the source Observable.\n *\n * ![](skip.png)\n *\n * Skips the values until the sent notifications are equal or less than provided skip count. It raises\n * an error if skip count is equal or more than the actual number of emits and source raises an error.\n *\n * ## Example\n *\n * Skip the values before the emission\n *\n * ```ts\n * import { interval, skip } from 'rxjs';\n *\n * // emit every half second\n * const source = interval(500);\n * // skip the first 10 emitted values\n * const result = source.pipe(skip(10));\n *\n * result.subscribe(value => console.log(value));\n * // output: 10...11...12...13...\n * ```\n *\n * @see {@link last}\n * @see {@link skipWhile}\n * @see {@link skipUntil}\n * @see {@link skipLast}\n *\n * @param {Number} count - The number of times, items emitted by source Observable should be skipped.\n * @return A function that returns an Observable that skips the first `count`\n * values emitted by the source Observable.\n */\nexport function skip<T>(count: number): MonoTypeOperatorFunction<T> {\n  return filter((_, index) => count <= index);\n}\n", "import { MonoTypeOperatorFunction, ObservableInput } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { innerFrom } from '../observable/innerFrom';\nimport { noop } from '../util/noop';\n\n/**\n * Returns an Observable that skips items emitted by the source Observable until a second Observable emits an item.\n *\n * The `skipUntil` operator causes the observable stream to skip the emission of values until the passed in observable\n * emits the first value. This can be particularly useful in combination with user interactions, responses of HTTP\n * requests or waiting for specific times to pass by.\n *\n * ![](skipUntil.png)\n *\n * Internally, the `skipUntil` operator subscribes to the passed in `notifier` `ObservableInput` (which gets converted\n * to an Observable) in order to recognize the emission of its first value. When `notifier` emits next, the operator\n * unsubscribes from it and starts emitting the values of the *source* observable until it completes or errors. It\n * will never let the *source* observable emit any values if the `notifier` completes or throws an error without\n * emitting a value before.\n *\n * ## Example\n *\n * In the following example, all emitted values of the interval observable are skipped until the user clicks anywhere\n * within the page\n *\n * ```ts\n * import { interval, fromEvent, skipUntil } from 'rxjs';\n *\n * const intervalObservable = interval(1000);\n * const click = fromEvent(document, 'click');\n *\n * const emitAfterClick = intervalObservable.pipe(\n *   skipUntil(click)\n * );\n * // clicked at 4.6s. output: 5...6...7...8........ or\n * // clicked at 7.3s. output: 8...9...10..11.......\n * emitAfterClick.subscribe(value => console.log(value));\n * ```\n *\n * @see {@link last}\n * @see {@link skip}\n * @see {@link skipWhile}\n * @see {@link skipLast}\n *\n * @param notifier An `ObservableInput` that has to emit an item before the source Observable elements begin to\n * be mirrored by the resulting Observable.\n * @return A function that returns an Observable that skips items from the\n * source Observable until the `notifier` Observable emits an item, then emits the\n * remaining items.\n */\nexport function skipUntil<T>(notifier: ObservableInput<any>): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    let taking = false;\n\n    const skipSubscriber = createOperatorSubscriber(\n      subscriber,\n      () => {\n        skipSubscriber?.unsubscribe();\n        taking = true;\n      },\n      noop\n    );\n\n    innerFrom(notifier).subscribe(skipSubscriber);\n\n    source.subscribe(createOperatorSubscriber(subscriber, (value) => taking && subscriber.next(value)));\n  });\n}\n", "import { concat } from '../observable/concat';\nimport { OperatorFunction, SchedulerLike, ValueFromArray } from '../types';\nimport { popScheduler } from '../util/args';\nimport { operate } from '../util/lift';\n\n// Devs are more likely to pass null or undefined than they are a scheduler\n// without accompanying values. To make things easier for (naughty) devs who\n// use the `strictNullChecks: false` TypeScript compiler option, these\n// overloads with explicit null and undefined values are included.\n\nexport function startWith<T>(value: null): OperatorFunction<T, T | null>;\nexport function startWith<T>(value: undefined): OperatorFunction<T, T | undefined>;\n\n/** @deprecated The `scheduler` parameter will be removed in v8. Use `scheduled` and `concatAll`. Details: https://rxjs.dev/deprecations/scheduler-argument */\nexport function startWith<T, A extends readonly unknown[] = T[]>(\n  ...valuesAndScheduler: [...A, SchedulerLike]\n): OperatorFunction<T, T | ValueFromArray<A>>;\nexport function startWith<T, A extends readonly unknown[] = T[]>(...values: A): OperatorFunction<T, T | ValueFromArray<A>>;\n\n/**\n * Returns an observable that, at the moment of subscription, will synchronously emit all\n * values provided to this operator, then subscribe to the source and mirror all of its emissions\n * to subscribers.\n *\n * This is a useful way to know when subscription has occurred on an existing observable.\n *\n * <span class=\"informal\">First emits its arguments in order, and then any\n * emissions from the source.</span>\n *\n * ![](startWith.png)\n *\n * ## Examples\n *\n * Emit a value when a timer starts.\n *\n * ```ts\n * import { timer, map, startWith } from 'rxjs';\n *\n * timer(1000)\n *   .pipe(\n *     map(() => 'timer emit'),\n *     startWith('timer start')\n *   )\n *   .subscribe(x => console.log(x));\n *\n * // results:\n * // 'timer start'\n * // 'timer emit'\n * ```\n *\n * @param values Items you want the modified Observable to emit first.\n * @return A function that returns an Observable that synchronously emits\n * provided values before subscribing to the source Observable.\n *\n * @see {@link endWith}\n * @see {@link finalize}\n * @see {@link concat}\n */\nexport function startWith<T, D>(...values: D[]): OperatorFunction<T, T | D> {\n  const scheduler = popScheduler(values);\n  return operate((source, subscriber) => {\n    // Here we can't pass `undefined` as a scheduler, because if we did, the\n    // code inside of `concat` would be confused by the `undefined`, and treat it\n    // like an invalid observable. So we have to split it two different ways.\n    (scheduler ? concat(values, source, scheduler) : concat(values, source)).subscribe(subscriber);\n  });\n}\n", "import { Subscriber } from '../Subscriber';\nimport { ObservableInput, OperatorFunction, ObservedValueOf } from '../types';\nimport { innerFrom } from '../observable/innerFrom';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\n/* tslint:disable:max-line-length */\nexport function switchMap<T, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O\n): OperatorFunction<T, ObservedValueOf<O>>;\n/** @deprecated The `resultSelector` parameter will be removed in v8. Use an inner `map` instead. Details: https://rxjs.dev/deprecations/resultSelector */\nexport function switchMap<T, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O,\n  resultSelector: undefined\n): OperatorFunction<T, ObservedValueOf<O>>;\n/** @deprecated The `resultSelector` parameter will be removed in v8. Use an inner `map` instead. Details: https://rxjs.dev/deprecations/resultSelector */\nexport function switchMap<T, R, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O,\n  resultSelector: (outerValue: T, innerValue: ObservedValueOf<O>, outerIndex: number, innerIndex: number) => R\n): OperatorFunction<T, R>;\n/* tslint:enable:max-line-length */\n\n/**\n * Projects each source value to an Observable which is merged in the output\n * Observable, emitting values only from the most recently projected Observable.\n *\n * <span class=\"informal\">Maps each value to an Observable, then flattens all of\n * these inner Observables using {@link switchAll}.</span>\n *\n * ![](switchMap.png)\n *\n * Returns an Observable that emits items based on applying a function that you\n * supply to each item emitted by the source Observable, where that function\n * returns an (so-called \"inner\") Observable. Each time it observes one of these\n * inner Observables, the output Observable begins emitting the items emitted by\n * that inner Observable. When a new inner Observable is emitted, `switchMap`\n * stops emitting items from the earlier-emitted inner Observable and begins\n * emitting items from the new one. It continues to behave like this for\n * subsequent inner Observables.\n *\n * ## Example\n *\n * Generate new Observable according to source Observable values\n *\n * ```ts\n * import { of, switchMap } from 'rxjs';\n *\n * const switched = of(1, 2, 3).pipe(switchMap(x => of(x, x ** 2, x ** 3)));\n * switched.subscribe(x => console.log(x));\n * // outputs\n * // 1\n * // 1\n * // 1\n * // 2\n * // 4\n * // 8\n * // 3\n * // 9\n * // 27\n * ```\n *\n * Restart an interval Observable on every click event\n *\n * ```ts\n * import { fromEvent, switchMap, interval } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(switchMap(() => interval(1000)));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link concatMap}\n * @see {@link exhaustMap}\n * @see {@link mergeMap}\n * @see {@link switchAll}\n * @see {@link switchMapTo}\n *\n * @param {function(value: T, index: number): ObservableInput} project A function\n * that, when applied to an item emitted by the source Observable, returns an\n * Observable.\n * @return A function that returns an Observable that emits the result of\n * applying the projection function (and the optional deprecated\n * `resultSelector`) to each item emitted by the source Observable and taking\n * only the values from the most recently projected inner Observable.\n */\nexport function switchMap<T, R, O extends ObservableInput<any>>(\n  project: (value: T, index: number) => O,\n  resultSelector?: (outerValue: T, innerValue: ObservedValueOf<O>, outerIndex: number, innerIndex: number) => R\n): OperatorFunction<T, ObservedValueOf<O> | R> {\n  return operate((source, subscriber) => {\n    let innerSubscriber: Subscriber<ObservedValueOf<O>> | null = null;\n    let index = 0;\n    // Whether or not the source subscription has completed\n    let isComplete = false;\n\n    // We only complete the result if the source is complete AND we don't have an active inner subscription.\n    // This is called both when the source completes and when the inners complete.\n    const checkComplete = () => isComplete && !innerSubscriber && subscriber.complete();\n\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        (value) => {\n          // Cancel the previous inner subscription if there was one\n          innerSubscriber?.unsubscribe();\n          let innerIndex = 0;\n          const outerIndex = index++;\n          // Start the next inner subscription\n          innerFrom(project(value, outerIndex)).subscribe(\n            (innerSubscriber = createOperatorSubscriber(\n              subscriber,\n              // When we get a new inner value, next it through. Note that this is\n              // handling the deprecate result selector here. This is because with this architecture\n              // it ends up being smaller than using the map operator.\n              (innerValue) => subscriber.next(resultSelector ? resultSelector(value, innerValue, outerIndex, innerIndex++) : innerValue),\n              () => {\n                // The inner has completed. Null out the inner subscriber to\n                // free up memory and to signal that we have no inner subscription\n                // currently.\n                innerSubscriber = null!;\n                checkComplete();\n              }\n            ))\n          );\n        },\n        () => {\n          isComplete = true;\n          checkComplete();\n        }\n      )\n    );\n  });\n}\n", "import { MonoTypeOperatorFunction, ObservableInput } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { innerFrom } from '../observable/innerFrom';\nimport { noop } from '../util/noop';\n\n/**\n * Emits the values emitted by the source Observable until a `notifier`\n * Observable emits a value.\n *\n * <span class=\"informal\">Lets values pass until a second Observable,\n * `notifier`, emits a value. Then, it completes.</span>\n *\n * ![](takeUntil.png)\n *\n * `takeUntil` subscribes and begins mirroring the source Observable. It also\n * monitors a second Observable, `notifier` that you provide. If the `notifier`\n * emits a value, the output Observable stops mirroring the source Observable\n * and completes. If the `notifier` doesn't emit any value and completes\n * then `takeUntil` will pass all values.\n *\n * ## Example\n *\n * Tick every second until the first click happens\n *\n * ```ts\n * import { interval, fromEvent, takeUntil } from 'rxjs';\n *\n * const source = interval(1000);\n * const clicks = fromEvent(document, 'click');\n * const result = source.pipe(takeUntil(clicks));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link take}\n * @see {@link takeLast}\n * @see {@link takeWhile}\n * @see {@link skip}\n *\n * @param {Observable} notifier The Observable whose first emitted value will\n * cause the output Observable of `takeUntil` to stop emitting values from the\n * source Observable.\n * @return A function that returns an Observable that emits the values from the\n * source Observable until `notifier` emits its first value.\n */\nexport function takeUntil<T>(notifier: ObservableInput<any>): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    innerFrom(notifier).subscribe(createOperatorSubscriber(subscriber, () => subscriber.complete(), noop));\n    !subscriber.closed && source.subscribe(subscriber);\n  });\n}\n", "import { OperatorFunction, MonoTypeOperatorFunction, TruthyTypesOf } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\n\nexport function takeWhile<T>(predicate: BooleanConstructor, inclusive: true): MonoTypeOperatorFunction<T>;\nexport function takeWhile<T>(predicate: BooleanConstructor, inclusive: false): OperatorFunction<T, TruthyTypesOf<T>>;\nexport function takeWhile<T>(predicate: BooleanConstructor): OperatorFunction<T, TruthyTypesOf<T>>;\nexport function takeWhile<T, S extends T>(predicate: (value: T, index: number) => value is S): OperatorFunction<T, S>;\nexport function takeWhile<T, S extends T>(predicate: (value: T, index: number) => value is S, inclusive: false): OperatorFunction<T, S>;\nexport function takeWhile<T>(predicate: (value: T, index: number) => boolean, inclusive?: boolean): MonoTypeOperatorFunction<T>;\n\n/**\n * Emits values emitted by the source Observable so long as each value satisfies\n * the given `predicate`, and then completes as soon as this `predicate` is not\n * satisfied.\n *\n * <span class=\"informal\">Takes values from the source only while they pass the\n * condition given. When the first value does not satisfy, it completes.</span>\n *\n * ![](takeWhile.png)\n *\n * `takeWhile` subscribes and begins mirroring the source Observable. Each value\n * emitted on the source is given to the `predicate` function which returns a\n * boolean, representing a condition to be satisfied by the source values. The\n * output Observable emits the source values until such time as the `predicate`\n * returns false, at which point `takeWhile` stops mirroring the source\n * Observable and completes the output Observable.\n *\n * ## Example\n *\n * Emit click events only while the clientX property is greater than 200\n *\n * ```ts\n * import { fromEvent, takeWhile } from 'rxjs';\n *\n * const clicks = fromEvent<PointerEvent>(document, 'click');\n * const result = clicks.pipe(takeWhile(ev => ev.clientX > 200));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link take}\n * @see {@link takeLast}\n * @see {@link takeUntil}\n * @see {@link skip}\n *\n * @param {function(value: T, index: number): boolean} predicate A function that\n * evaluates a value emitted by the source Observable and returns a boolean.\n * Also takes the (zero-based) index as the second argument.\n * @param {boolean} inclusive When set to `true` the value that caused\n * `predicate` to return `false` will also be emitted.\n * @return A function that returns an Observable that emits values from the\n * source Observable so long as each value satisfies the condition defined by\n * the `predicate`, then completes.\n */\nexport function takeWhile<T>(predicate: (value: T, index: number) => boolean, inclusive = false): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    let index = 0;\n    source.subscribe(\n      createOperatorSubscriber(subscriber, (value) => {\n        const result = predicate(value, index++);\n        (result || inclusive) && subscriber.next(value);\n        !result && subscriber.complete();\n      })\n    );\n  });\n}\n", "import { MonoTypeOperatorFunction, Observer } from '../types';\nimport { isFunction } from '../util/isFunction';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { identity } from '../util/identity';\n\n/**\n * An extension to the {@link Observer} interface used only by the {@link tap} operator.\n *\n * It provides a useful set of callbacks a user can register to do side-effects in\n * cases other than what the usual {@link Observer} callbacks are\n * ({@link guide/glossary-and-semantics#next next},\n * {@link guide/glossary-and-semantics#error error} and/or\n * {@link guide/glossary-and-semantics#complete complete}).\n *\n * ## Example\n *\n * ```ts\n * import { fromEvent, switchMap, tap, interval, take } from 'rxjs';\n *\n * const source$ = fromEvent(document, 'click');\n * const result$ = source$.pipe(\n *   switchMap((_, i) => i % 2 === 0\n *     ? fromEvent(document, 'mousemove').pipe(\n *         tap({\n *           subscribe: () => console.log('Subscribed to the mouse move events after click #' + i),\n *           unsubscribe: () => console.log('Mouse move events #' + i + ' unsubscribed'),\n *           finalize: () => console.log('Mouse move events #' + i + ' finalized')\n *         })\n *       )\n *     : interval(1_000).pipe(\n *         take(5),\n *         tap({\n *           subscribe: () => console.log('Subscribed to the 1-second interval events after click #' + i),\n *           unsubscribe: () => console.log('1-second interval events #' + i + ' unsubscribed'),\n *           finalize: () => console.log('1-second interval events #' + i + ' finalized')\n *         })\n *       )\n *   )\n * );\n *\n * const subscription = result$.subscribe({\n *   next: console.log\n * });\n *\n * setTimeout(() => {\n *   console.log('Unsubscribe after 60 seconds');\n *   subscription.unsubscribe();\n * }, 60_000);\n * ```\n */\nexport interface TapObserver<T> extends Observer<T> {\n  /**\n   * The callback that `tap` operator invokes at the moment when the source Observable\n   * gets subscribed to.\n   */\n  subscribe: () => void;\n  /**\n   * The callback that `tap` operator invokes when an explicit\n   * {@link guide/glossary-and-semantics#unsubscription unsubscribe} happens. It won't get invoked on\n   * `error` or `complete` events.\n   */\n  unsubscribe: () => void;\n  /**\n   * The callback that `tap` operator invokes when any kind of\n   * {@link guide/glossary-and-semantics#finalization finalization} happens - either when\n   * the source Observable `error`s or `complete`s or when it gets explicitly unsubscribed\n   * by the user. There is no difference in using this callback or the {@link finalize}\n   * operator, but if you're already using `tap` operator, you can use this callback\n   * instead. You'd get the same result in either case.\n   */\n  finalize: () => void;\n}\nexport function tap<T>(observerOrNext?: Partial<TapObserver<T>> | ((value: T) => void)): MonoTypeOperatorFunction<T>;\n/** @deprecated Instead of passing separate callback arguments, use an observer argument. Signatures taking separate callback arguments will be removed in v8. Details: https://rxjs.dev/deprecations/subscribe-arguments */\nexport function tap<T>(\n  next?: ((value: T) => void) | null,\n  error?: ((error: any) => void) | null,\n  complete?: (() => void) | null\n): MonoTypeOperatorFunction<T>;\n\n/**\n * Used to perform side-effects for notifications from the source observable\n *\n * <span class=\"informal\">Used when you want to affect outside state with a notification without altering the notification</span>\n *\n * ![](tap.png)\n *\n * Tap is designed to allow the developer a designated place to perform side effects. While you _could_ perform side-effects\n * inside of a `map` or a `mergeMap`, that would make their mapping functions impure, which isn't always a big deal, but will\n * make it so you can't do things like memoize those functions. The `tap` operator is designed solely for such side-effects to\n * help you remove side-effects from other operations.\n *\n * For any notification, next, error, or complete, `tap` will call the appropriate callback you have provided to it, via a function\n * reference, or a partial observer, then pass that notification down the stream.\n *\n * The observable returned by `tap` is an exact mirror of the source, with one exception: Any error that occurs -- synchronously -- in a handler\n * provided to `tap` will be emitted as an error from the returned observable.\n *\n * > Be careful! You can mutate objects as they pass through the `tap` operator's handlers.\n *\n * The most common use of `tap` is actually for debugging. You can place a `tap(console.log)` anywhere\n * in your observable `pipe`, log out the notifications as they are emitted by the source returned by the previous\n * operation.\n *\n * ## Examples\n *\n * Check a random number before it is handled. Below is an observable that will use a random number between 0 and 1,\n * and emit `'big'` or `'small'` depending on the size of that number. But we wanted to log what the original number\n * was, so we have added a `tap(console.log)`.\n *\n * ```ts\n * import { of, tap, map } from 'rxjs';\n *\n * of(Math.random()).pipe(\n *   tap(console.log),\n *   map(n => n > 0.5 ? 'big' : 'small')\n * ).subscribe(console.log);\n * ```\n *\n * Using `tap` to analyze a value and force an error. Below is an observable where in our system we only\n * want to emit numbers 3 or less we get from another source. We can force our observable to error\n * using `tap`.\n *\n * ```ts\n * import { of, tap } from 'rxjs';\n *\n * const source = of(1, 2, 3, 4, 5);\n *\n * source.pipe(\n *   tap(n => {\n *     if (n > 3) {\n *       throw new TypeError(`Value ${ n } is greater than 3`);\n *     }\n *   })\n * )\n * .subscribe({ next: console.log, error: err => console.log(err.message) });\n * ```\n *\n * We want to know when an observable completes before moving on to the next observable. The system\n * below will emit a random series of `'X'` characters from 3 different observables in sequence. The\n * only way we know when one observable completes and moves to the next one, in this case, is because\n * we have added a `tap` with the side effect of logging to console.\n *\n * ```ts\n * import { of, concatMap, interval, take, map, tap } from 'rxjs';\n *\n * of(1, 2, 3).pipe(\n *   concatMap(n => interval(1000).pipe(\n *     take(Math.round(Math.random() * 10)),\n *     map(() => 'X'),\n *     tap({ complete: () => console.log(`Done with ${ n }`) })\n *   ))\n * )\n * .subscribe(console.log);\n * ```\n *\n * @see {@link finalize}\n * @see {@link TapObserver}\n *\n * @param observerOrNext A next handler or partial observer\n * @param error An error handler\n * @param complete A completion handler\n * @return A function that returns an Observable identical to the source, but\n * runs the specified Observer or callback(s) for each item.\n */\nexport function tap<T>(\n  observerOrNext?: Partial<TapObserver<T>> | ((value: T) => void) | null,\n  error?: ((e: any) => void) | null,\n  complete?: (() => void) | null\n): MonoTypeOperatorFunction<T> {\n  // We have to check to see not only if next is a function,\n  // but if error or complete were passed. This is because someone\n  // could technically call tap like `tap(null, fn)` or `tap(null, null, fn)`.\n  const tapObserver =\n    isFunction(observerOrNext) || error || complete\n      ? // tslint:disable-next-line: no-object-literal-type-assertion\n        ({ next: observerOrNext as Exclude<typeof observerOrNext, Partial<TapObserver<T>>>, error, complete } as Partial<TapObserver<T>>)\n      : observerOrNext;\n\n  return tapObserver\n    ? operate((source, subscriber) => {\n        tapObserver.subscribe?.();\n        let isUnsub = true;\n        source.subscribe(\n          createOperatorSubscriber(\n            subscriber,\n            (value) => {\n              tapObserver.next?.(value);\n              subscriber.next(value);\n            },\n            () => {\n              isUnsub = false;\n              tapObserver.complete?.();\n              subscriber.complete();\n            },\n            (err) => {\n              isUnsub = false;\n              tapObserver.error?.(err);\n              subscriber.error(err);\n            },\n            () => {\n              if (isUnsub) {\n                tapObserver.unsubscribe?.();\n              }\n              tapObserver.finalize?.();\n            }\n          )\n        );\n      })\n    : // Tap was called with no valid tap observer or handler\n      // (e.g. `tap(null, null, null)` or `tap(null)` or `tap()`)\n      // so we're going to just mirror the source.\n      identity;\n}\n", "import { Subscription } from '../Subscription';\n\nimport { MonoTypeOperatorFunction, ObservableInput } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { innerFrom } from '../observable/innerFrom';\n\n/**\n * An object interface used by {@link throttle} or {@link throttleTime} that ensure\n * configuration options of these operators.\n *\n * @see {@link throttle}\n * @see {@link throttleTime}\n */\nexport interface ThrottleConfig {\n  /**\n   * If `true`, the resulting Observable will emit the first value from the source\n   * Observable at the **start** of the \"throttling\" process (when starting an\n   * internal timer that prevents other emissions from the source to pass through).\n   * If `false`, it will not emit the first value from the source Observable at the\n   * start of the \"throttling\" process.\n   *\n   * If not provided, defaults to: `true`.\n   */\n  leading?: boolean;\n  /**\n   * If `true`, the resulting Observable will emit the last value from the source\n   * Observable at the **end** of the \"throttling\" process (when ending an internal\n   * timer that prevents other emissions from the source to pass through).\n   * If `false`, it will not emit the last value from the source Observable at the\n   * end of the \"throttling\" process.\n   *\n   * If not provided, defaults to: `false`.\n   */\n  trailing?: boolean;\n}\n\n/**\n * Emits a value from the source Observable, then ignores subsequent source\n * values for a duration determined by another Observable, then repeats this\n * process.\n *\n * <span class=\"informal\">It's like {@link throttleTime}, but the silencing\n * duration is determined by a second Observable.</span>\n *\n * ![](throttle.svg)\n *\n * `throttle` emits the source Observable values on the output Observable\n * when its internal timer is disabled, and ignores source values when the timer\n * is enabled. Initially, the timer is disabled. As soon as the first source\n * value arrives, it is forwarded to the output Observable, and then the timer\n * is enabled by calling the `durationSelector` function with the source value,\n * which returns the \"duration\" Observable. When the duration Observable emits a\n * value, the timer is disabled, and this process repeats for the\n * next source value.\n *\n * ## Example\n *\n * Emit clicks at a rate of at most one click per second\n *\n * ```ts\n * import { fromEvent, throttle, interval } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(throttle(() => interval(1000)));\n *\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link audit}\n * @see {@link debounce}\n * @see {@link delayWhen}\n * @see {@link sample}\n * @see {@link throttleTime}\n *\n * @param durationSelector A function that receives a value from the source\n * Observable, for computing the silencing duration for each source value,\n * returned as an `ObservableInput`.\n * @param config A configuration object to define `leading` and `trailing`\n * behavior. Defaults to `{ leading: true, trailing: false }`.\n * @return A function that returns an Observable that performs the throttle\n * operation to limit the rate of emissions from the source.\n */\nexport function throttle<T>(durationSelector: (value: T) => ObservableInput<any>, config?: ThrottleConfig): MonoTypeOperatorFunction<T> {\n  return operate((source, subscriber) => {\n    const { leading = true, trailing = false } = config ?? {};\n    let hasValue = false;\n    let sendValue: T | null = null;\n    let throttled: Subscription | null = null;\n    let isComplete = false;\n\n    const endThrottling = () => {\n      throttled?.unsubscribe();\n      throttled = null;\n      if (trailing) {\n        send();\n        isComplete && subscriber.complete();\n      }\n    };\n\n    const cleanupThrottling = () => {\n      throttled = null;\n      isComplete && subscriber.complete();\n    };\n\n    const startThrottle = (value: T) =>\n      (throttled = innerFrom(durationSelector(value)).subscribe(createOperatorSubscriber(subscriber, endThrottling, cleanupThrottling)));\n\n    const send = () => {\n      if (hasValue) {\n        // Ensure we clear out our value and hasValue flag\n        // before we emit, otherwise reentrant code can cause\n        // issues here.\n        hasValue = false;\n        const value = sendValue!;\n        sendValue = null;\n        // Emit the value.\n        subscriber.next(value);\n        !isComplete && startThrottle(value);\n      }\n    };\n\n    source.subscribe(\n      createOperatorSubscriber(\n        subscriber,\n        // Regarding the presence of throttled.closed in the following\n        // conditions, if a synchronous duration selector is specified - weird,\n        // but legal - an already-closed subscription will be assigned to\n        // throttled, so the subscription's closed property needs to be checked,\n        // too.\n        (value) => {\n          hasValue = true;\n          sendValue = value;\n          !(throttled && !throttled.closed) && (leading ? send() : startThrottle(value));\n        },\n        () => {\n          isComplete = true;\n          !(trailing && hasValue && throttled && !throttled.closed) && subscriber.complete();\n        }\n      )\n    );\n  });\n}\n", "import { asyncScheduler } from '../scheduler/async';\nimport { throttle, ThrottleConfig } from './throttle';\nimport { MonoTypeOperatorFunction, SchedulerLike } from '../types';\nimport { timer } from '../observable/timer';\n\n/**\n * Emits a value from the source Observable, then ignores subsequent source\n * values for `duration` milliseconds, then repeats this process.\n *\n * <span class=\"informal\">Lets a value pass, then ignores source values for the\n * next `duration` milliseconds.</span>\n *\n * ![](throttleTime.png)\n *\n * `throttleTime` emits the source Observable values on the output Observable\n * when its internal timer is disabled, and ignores source values when the timer\n * is enabled. Initially, the timer is disabled. As soon as the first source\n * value arrives, it is forwarded to the output Observable, and then the timer\n * is enabled. After `duration` milliseconds (or the time unit determined\n * internally by the optional `scheduler`) has passed, the timer is disabled,\n * and this process repeats for the next source value. Optionally takes a\n * {@link SchedulerLike} for managing timers.\n *\n * ## Examples\n *\n * ### Limit click rate\n *\n * Emit clicks at a rate of at most one click per second\n *\n * ```ts\n * import { fromEvent, throttleTime } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const result = clicks.pipe(throttleTime(1000));\n *\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link auditTime}\n * @see {@link debounceTime}\n * @see {@link delay}\n * @see {@link sampleTime}\n * @see {@link throttle}\n *\n * @param duration Time to wait before emitting another value after\n * emitting the last value, measured in milliseconds or the time unit determined\n * internally by the optional `scheduler`.\n * @param scheduler The {@link SchedulerLike} to use for\n * managing the timers that handle the throttling. Defaults to {@link asyncScheduler}.\n * @param config A configuration object to define `leading` and\n * `trailing` behavior. Defaults to `{ leading: true, trailing: false }`.\n * @return A function that returns an Observable that performs the throttle\n * operation to limit the rate of emissions from the source.\n */\nexport function throttleTime<T>(\n  duration: number,\n  scheduler: SchedulerLike = asyncScheduler,\n  config?: ThrottleConfig\n): MonoTypeOperatorFunction<T> {\n  const duration$ = timer(duration, scheduler);\n  return throttle(() => duration$, config);\n}\n", "import { OperatorFunction, ObservableInputTuple } from '../types';\nimport { operate } from '../util/lift';\nimport { createOperatorSubscriber } from './OperatorSubscriber';\nimport { innerFrom } from '../observable/innerFrom';\nimport { identity } from '../util/identity';\nimport { noop } from '../util/noop';\nimport { popResultSelector } from '../util/args';\n\nexport function withLatestFrom<T, O extends unknown[]>(...inputs: [...ObservableInputTuple<O>]): OperatorFunction<T, [T, ...O]>;\n\nexport function withLatestFrom<T, O extends unknown[], R>(\n  ...inputs: [...ObservableInputTuple<O>, (...value: [T, ...O]) => R]\n): OperatorFunction<T, R>;\n\n/**\n * Combines the source Observable with other Observables to create an Observable\n * whose values are calculated from the latest values of each, only when the\n * source emits.\n *\n * <span class=\"informal\">Whenever the source Observable emits a value, it\n * computes a formula using that value plus the latest values from other input\n * Observables, then emits the output of that formula.</span>\n *\n * ![](withLatestFrom.png)\n *\n * `withLatestFrom` combines each value from the source Observable (the\n * instance) with the latest values from the other input Observables only when\n * the source emits a value, optionally using a `project` function to determine\n * the value to be emitted on the output Observable. All input Observables must\n * emit at least one value before the output Observable will emit a value.\n *\n * ## Example\n *\n * On every click event, emit an array with the latest timer event plus the click event\n *\n * ```ts\n * import { fromEvent, interval, withLatestFrom } from 'rxjs';\n *\n * const clicks = fromEvent(document, 'click');\n * const timer = interval(1000);\n * const result = clicks.pipe(withLatestFrom(timer));\n * result.subscribe(x => console.log(x));\n * ```\n *\n * @see {@link combineLatest}\n *\n * @param {ObservableInput} other An input Observable to combine with the source\n * Observable. More than one input Observables may be given as argument.\n * @param {Function} [project] Projection function for combining values\n * together. Receives all values in order of the Observables passed, where the\n * first parameter is a value from the source Observable. (e.g.\n * `a.pipe(withLatestFrom(b, c), map(([a1, b1, c1]) => a1 + b1 + c1))`). If this is not\n * passed, arrays will be emitted on the output Observable.\n * @return A function that returns an Observable of projected values from the\n * most recent values from each input Observable, or an array of the most\n * recent values from each input Observable.\n */\nexport function withLatestFrom<T, R>(...inputs: any[]): OperatorFunction<T, R | any[]> {\n  const project = popResultSelector(inputs) as ((...args: any[]) => R) | undefined;\n\n  return operate((source, subscriber) => {\n    const len = inputs.length;\n    const otherValues = new Array(len);\n    // An array of whether or not the other sources have emitted. Matched with them by index.\n    // TODO: At somepoint, we should investigate the performance implications here, and look\n    // into using a `Set()` and checking the `size` to see if we're ready.\n    let hasValue = inputs.map(() => false);\n    // Flipped true when we have at least one value from all other sources and\n    // we are ready to start emitting values.\n    let ready = false;\n\n    // Other sources. Note that here we are not checking `subscriber.closed`,\n    // this causes all inputs to be subscribed to, even if nothing can be emitted\n    // from them. This is an important distinction because subscription constitutes\n    // a side-effect.\n    for (let i = 0; i < len; i++) {\n      innerFrom(inputs[i]).subscribe(\n        createOperatorSubscriber(\n          subscriber,\n          (value) => {\n            otherValues[i] = value;\n            if (!ready && !hasValue[i]) {\n              // If we're not ready yet, flag to show this observable has emitted.\n              hasValue[i] = true;\n              // Intentionally terse code.\n              // If all of our other observables have emitted, set `ready` to `true`,\n              // so we know we can start emitting values, then clean up the `hasValue` array,\n              // because we don't need it anymore.\n              (ready = hasValue.every(identity)) && (hasValue = null!);\n            }\n          },\n          // Completing one of the other sources has\n          // no bearing on the completion of our result.\n          noop\n        )\n      );\n    }\n\n    // Source subscription\n    source.subscribe(\n      createOperatorSubscriber(subscriber, (value) => {\n        if (ready) {\n          // We have at least one value from the other sources. Go ahead and emit.\n          const values = [value, ...otherValues];\n          subscriber.next(project ? project(...values) : values);\n        }\n      })\n    );\n  });\n}\n", "import { zip as zipStatic } from '../observable/zip';\nimport { ObservableInput, ObservableInputTuple, OperatorFunction, Cons } from '../types';\nimport { operate } from '../util/lift';\n\n/** @deprecated Replaced with {@link zipWith}. Will be removed in v8. */\nexport function zip<T, A extends readonly unknown[]>(otherInputs: [...ObservableInputTuple<A>]): OperatorFunction<T, Cons<T, A>>;\n/** @deprecated Replaced with {@link zipWith}. Will be removed in v8. */\nexport function zip<T, A extends readonly unknown[], R>(\n  otherInputsAndProject: [...ObservableInputTuple<A>],\n  project: (...values: Cons<T, A>) => R\n): OperatorFunction<T, R>;\n/** @deprecated Replaced with {@link zipWith}. Will be removed in v8. */\nexport function zip<T, A extends readonly unknown[]>(...otherInputs: [...ObservableInputTuple<A>]): OperatorFunction<T, Cons<T, A>>;\n/** @deprecated Replaced with {@link zipWith}. Will be removed in v8. */\nexport function zip<T, A extends readonly unknown[], R>(\n  ...otherInputsAndProject: [...ObservableInputTuple<A>, (...values: Cons<T, A>) => R]\n): OperatorFunction<T, R>;\n\n/**\n * @deprecated Replaced with {@link zipWith}. Will be removed in v8.\n */\nexport function zip<T, R>(...sources: Array<ObservableInput<any> | ((...values: Array<any>) => R)>): OperatorFunction<T, any> {\n  return operate((source, subscriber) => {\n    zipStatic(source as ObservableInput<any>, ...(sources as Array<ObservableInput<any>>)).subscribe(subscriber);\n  });\n}\n", "import { ObservableInputTuple, OperatorFunction, Cons } from '../types';\nimport { zip } from './zip';\n\n/**\n * Subscribes to the source, and the observable inputs provided as arguments, and combines their values, by index, into arrays.\n *\n * What is meant by \"combine by index\": The first value from each will be made into a single array, then emitted,\n * then the second value from each will be combined into a single array and emitted, then the third value\n * from each will be combined into a single array and emitted, and so on.\n *\n * This will continue until it is no longer able to combine values of the same index into an array.\n *\n * After the last value from any one completed source is emitted in an array, the resulting observable will complete,\n * as there is no way to continue \"zipping\" values together by index.\n *\n * Use-cases for this operator are limited. There are memory concerns if one of the streams is emitting\n * values at a much faster rate than the others. Usage should likely be limited to streams that emit\n * at a similar pace, or finite streams of known length.\n *\n * In many cases, authors want `combineLatestWith` and not `zipWith`.\n *\n * @param otherInputs other observable inputs to collate values from.\n * @return A function that returns an Observable that emits items by index\n * combined from the source Observable and provided Observables, in form of an\n * array.\n */\nexport function zipWith<T, A extends readonly unknown[]>(...otherInputs: [...ObservableInputTuple<A>]): OperatorFunction<T, Cons<T, A>> {\n  return zip(...otherInputs);\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  ReplaySubject,\n  Subject,\n  fromEvent\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch document\n *\n * Documents are implemented as subjects, so all downstream observables are\n * automatically updated when a new document is emitted.\n *\n * @returns Document subject\n */\nexport function watchDocument(): Subject<Document> {\n  const document$ = new ReplaySubject<Document>(1)\n  fromEvent(document, \"DOMContentLoaded\", { once: true })\n    .subscribe(() => document$.next(document))\n\n  /* Return document */\n  return document$\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve all elements matching the query selector\n *\n * @template T - Element type\n *\n * @param selector - Query selector\n * @param node - Node of reference\n *\n * @returns Elements\n */\nexport function getElements<T extends keyof HTMLElementTagNameMap>(\n  selector: T, node?: ParentNode\n): HTMLElementTagNameMap[T][]\n\nexport function getElements<T extends HTMLElement>(\n  selector: string, node?: ParentNode\n): T[]\n\nexport function getElements<T extends HTMLElement>(\n  selector: string, node: ParentNode = document\n): T[] {\n  return Array.from(node.querySelectorAll<T>(selector))\n}\n\n/**\n * Retrieve an element matching a query selector or throw a reference error\n *\n * Note that this function assumes that the element is present. If unsure if an\n * element is existent, use the `getOptionalElement` function instead.\n *\n * @template T - Element type\n *\n * @param selector - Query selector\n * @param node - Node of reference\n *\n * @returns Element\n */\nexport function getElement<T extends keyof HTMLElementTagNameMap>(\n  selector: T, node?: ParentNode\n): HTMLElementTagNameMap[T]\n\nexport function getElement<T extends HTMLElement>(\n  selector: string, node?: ParentNode\n): T\n\nexport function getElement<T extends HTMLElement>(\n  selector: string, node: ParentNode = document\n): T {\n  const el = getOptionalElement<T>(selector, node)\n  if (typeof el === \"undefined\")\n    throw new ReferenceError(\n      `Missing element: expected \"${selector}\" to be present`\n    )\n\n  /* Return element */\n  return el\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Retrieve an optional element matching the query selector\n *\n * @template T - Element type\n *\n * @param selector - Query selector\n * @param node - Node of reference\n *\n * @returns Element or nothing\n */\nexport function getOptionalElement<T extends keyof HTMLElementTagNameMap>(\n  selector: T, node?: ParentNode\n): HTMLElementTagNameMap[T] | undefined\n\nexport function getOptionalElement<T extends HTMLElement>(\n  selector: string, node?: ParentNode\n): T | undefined\n\nexport function getOptionalElement<T extends HTMLElement>(\n  selector: string, node: ParentNode = document\n): T | undefined {\n  return node.querySelector<T>(selector) || undefined\n}\n\n/**\n * Retrieve the currently active element\n *\n * @returns Element or nothing\n */\nexport function getActiveElement(): HTMLElement | undefined {\n  return document.activeElement instanceof HTMLElement\n    ? document.activeElement || undefined\n    : undefined\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  debounceTime,\n  distinctUntilChanged,\n  fromEvent,\n  map,\n  merge,\n  shareReplay,\n  startWith\n} from \"rxjs\"\n\nimport { getActiveElement } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Focus observable\n *\n * Previously, this observer used `focus` and `blur` events to determine whether\n * an element is focused, but this doesn't work if there are focusable elements\n * within the elements itself. A better solutions are `focusin` and `focusout`\n * events, which bubble up the tree and allow for more fine-grained control.\n *\n * `debounceTime` is necessary, because when a focus change happens inside an\n * element, the observable would first emit `false` and then `true` again.\n */\nconst observer$ = merge(\n  fromEvent(document.body, \"focusin\"),\n  fromEvent(document.body, \"focusout\")\n)\n  .pipe(\n    debounceTime(1),\n    startWith(undefined),\n    map(() => getActiveElement() || document.body),\n    shareReplay(1)\n  )\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch element focus\n *\n * @param el - Element\n *\n * @returns Element focus observable\n */\nexport function watchElementFocus(\n  el: HTMLElement\n): Observable<boolean> {\n  return observer$\n    .pipe(\n      map(active => el.contains(active)),\n      distinctUntilChanged()\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  animationFrameScheduler,\n  auditTime,\n  fromEvent,\n  map,\n  merge,\n  startWith\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Element offset\n */\nexport interface ElementOffset {\n  x: number                            /* Horizontal offset */\n  y: number                            /* Vertical offset */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve element offset\n *\n * @param el - Element\n *\n * @returns Element offset\n */\nexport function getElementOffset(\n  el: HTMLElement\n): ElementOffset {\n  return {\n    x: el.offsetLeft,\n    y: el.offsetTop\n  }\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch element offset\n *\n * @param el - Element\n *\n * @returns Element offset observable\n */\nexport function watchElementOffset(\n  el: HTMLElement\n): Observable<ElementOffset> {\n  return merge(\n    fromEvent(window, \"load\"),\n    fromEvent(window, \"resize\")\n  )\n    .pipe(\n      auditTime(0, animationFrameScheduler),\n      map(() => getElementOffset(el)),\n      startWith(getElementOffset(el))\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  animationFrameScheduler,\n  auditTime,\n  fromEvent,\n  map,\n  merge,\n  startWith\n} from \"rxjs\"\n\nimport { ElementOffset } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve element content offset (= scroll offset)\n *\n * @param el - Element\n *\n * @returns Element content offset\n */\nexport function getElementContentOffset(\n  el: HTMLElement\n): ElementOffset {\n  return {\n    x: el.scrollLeft,\n    y: el.scrollTop\n  }\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch element content offset\n *\n * @param el - Element\n *\n * @returns Element content offset observable\n */\nexport function watchElementContentOffset(\n  el: HTMLElement\n): Observable<ElementOffset> {\n  return merge(\n    fromEvent(el, \"scroll\"),\n    fromEvent(window, \"resize\")\n  )\n    .pipe(\n      auditTime(0, animationFrameScheduler),\n      map(() => getElementContentOffset(el)),\n      startWith(getElementContentOffset(el))\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { JSX as JSXInternal } from \"preact\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * HTML attributes\n */\ntype Attributes =\n  & JSXInternal.HTMLAttributes\n  & JSXInternal.SVGAttributes\n  & Record<string, any>\n\n/**\n * Child element\n */\ntype Child =\n  | ChildNode\n  | HTMLElement\n  | Text\n  | string\n  | number\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Append a child node to an element\n *\n * @param el - Element\n * @param child - Child node(s)\n */\nfunction appendChild(el: HTMLElement, child: Child | Child[]): void {\n\n  /* Handle primitive types (including raw HTML) */\n  if (typeof child === \"string\" || typeof child === \"number\") {\n    el.innerHTML += child.toString()\n\n  /* Handle nodes */\n  } else if (child instanceof Node) {\n    el.appendChild(child)\n\n  /* Handle nested children */\n  } else if (Array.isArray(child)) {\n    for (const node of child)\n      appendChild(el, node)\n  }\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * JSX factory\n *\n * @template T - Element type\n *\n * @param tag - HTML tag\n * @param attributes - HTML attributes\n * @param children - Child elements\n *\n * @returns Element\n */\nexport function h<T extends keyof HTMLElementTagNameMap>(\n  tag: T, attributes?: Attributes | null, ...children: Child[]\n): HTMLElementTagNameMap[T]\n\nexport function h<T extends h.JSX.Element>(\n  tag: string, attributes?: Attributes | null, ...children: Child[]\n): T\n\nexport function h<T extends h.JSX.Element>(\n  tag: string, attributes?: Attributes | null, ...children: Child[]\n): T {\n  const el = document.createElement(tag)\n\n  /* Set attributes, if any */\n  if (attributes)\n    for (const attr of Object.keys(attributes)) {\n      if (typeof attributes[attr] === \"undefined\")\n        continue\n\n      /* Set default attribute or boolean */\n      if (typeof attributes[attr] !== \"boolean\")\n        el.setAttribute(attr, attributes[attr])\n      else\n        el.setAttribute(attr, \"\")\n    }\n\n  /* Append child nodes */\n  for (const child of children)\n    appendChild(el, child)\n\n  /* Return element */\n  return el as T\n}\n\n/* ----------------------------------------------------------------------------\n * Namespace\n * ------------------------------------------------------------------------- */\n\nexport declare namespace h {\n  namespace JSX {\n    type Element = HTMLElement\n    type IntrinsicElements = JSXInternal.IntrinsicElements\n  }\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Round a number for display with repository facts\n *\n * This is a reverse-engineered version of GitHub's weird rounding algorithm\n * for stars, forks and all other numbers. While all numbers below `1,000` are\n * returned as-is, bigger numbers are converted to fixed numbers:\n *\n * - `1,049` => `1k`\n * - `1,050` => `1.1k`\n * - `1,949` => `1.9k`\n * - `1,950` => `2k`\n *\n * @param value - Original value\n *\n * @returns Rounded value\n */\nexport function round(value: number): string {\n  if (value > 999) {\n    const digits = +((value - 950) % 1000 > 99)\n    return `${((value + 0.000001) / 1000).toFixed(digits)}k`\n  } else {\n    return value.toString()\n  }\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  defer,\n  finalize,\n  fromEvent,\n  map,\n  merge,\n  switchMap,\n  take,\n  throwError\n} from \"rxjs\"\n\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Create and load a `script` element\n *\n * This function returns an observable that will emit when the script was\n * successfully loaded, or throw an error if it wasn't.\n *\n * @param src - Script URL\n *\n * @returns Script observable\n */\nexport function watchScript(src: string): Observable<void> {\n  const script = h(\"script\", { src })\n  return defer(() => {\n    document.head.appendChild(script)\n    return merge(\n      fromEvent(script, \"load\"),\n      fromEvent(script, \"error\")\n        .pipe(\n          switchMap(() => (\n            throwError(() => new ReferenceError(`Invalid script: ${src}`))\n          ))\n        )\n    )\n      .pipe(\n        map(() => undefined),\n        finalize(() => document.head.removeChild(script)),\n        take(1)\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  NEVER,\n  Observable,\n  Subject,\n  defer,\n  filter,\n  finalize,\n  map,\n  merge,\n  of,\n  shareReplay,\n  startWith,\n  switchMap,\n  tap\n} from \"rxjs\"\n\nimport { watchScript } from \"../../../script\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Element offset\n */\nexport interface ElementSize {\n  width: number                        /* Element width */\n  height: number                       /* Element height */\n}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Resize observer entry subject\n */\nconst entry$ = new Subject<ResizeObserverEntry>()\n\n/**\n * Resize observer observable\n *\n * This observable will create a `ResizeObserver` on the first subscription\n * and will automatically terminate it when there are no more subscribers.\n * It's quite important to centralize observation in a single `ResizeObserver`,\n * as the performance difference can be quite dramatic, as the link shows.\n *\n * If the browser doesn't have a `ResizeObserver` implementation available, a\n * polyfill is automatically downloaded from unpkg.com. This is also compatible\n * with the built-in privacy plugin, which will download the polyfill and put\n * it alongside the built site for self-hosting.\n *\n * @see https://bit.ly/3iIYfEm - Google Groups on performance\n */\nconst observer$ = defer(() => (\n  typeof ResizeObserver === \"undefined\"\n    ? watchScript(\"https://unpkg.com/resize-observer-polyfill\")\n    : of(undefined)\n))\n  .pipe(\n    map(() => new ResizeObserver(entries => {\n      for (const entry of entries)\n        entry$.next(entry)\n    })),\n    switchMap(observer => merge(NEVER, of(observer))\n      .pipe(\n        finalize(() => observer.disconnect())\n      )\n    ),\n    shareReplay(1)\n  )\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve element size\n *\n * @param el - Element\n *\n * @returns Element size\n */\nexport function getElementSize(\n  el: HTMLElement\n): ElementSize {\n  return {\n    width:  el.offsetWidth,\n    height: el.offsetHeight\n  }\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch element size\n *\n * This function returns an observable that subscribes to a single internal\n * instance of `ResizeObserver` upon subscription, and emit resize events until\n * termination. Note that this function should not be called with the same\n * element twice, as the first unsubscription will terminate observation.\n *\n * Sadly, we can't use the `DOMRect` objects returned by the observer, because\n * we need the emitted values to be consistent with `getElementSize`, which will\n * return the used values (rounded) and not actual values (unrounded). Thus, we\n * use the `offset*` properties. See the linked GitHub issue.\n *\n * @see https://bit.ly/3m0k3he - GitHub issue\n *\n * @param el - Element\n *\n * @returns Element size observable\n */\nexport function watchElementSize(\n  el: HTMLElement\n): Observable<ElementSize> {\n  return observer$\n    .pipe(\n      tap(observer => observer.observe(el)),\n      switchMap(observer => entry$\n        .pipe(\n          filter(({ target }) => target === el),\n          finalize(() => observer.unobserve(el)),\n          map(() => getElementSize(el))\n        )\n      ),\n      startWith(getElementSize(el))\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { ElementSize } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve element content size (= scroll width and height)\n *\n * @param el - Element\n *\n * @returns Element content size\n */\nexport function getElementContentSize(\n  el: HTMLElement\n): ElementSize {\n  return {\n    width:  el.scrollWidth,\n    height: el.scrollHeight\n  }\n}\n\n/**\n * Retrieve the overflowing container of an element, if any\n *\n * @param el - Element\n *\n * @returns Overflowing container or nothing\n */\nexport function getElementContainer(\n  el: HTMLElement\n): HTMLElement | undefined {\n  let parent = el.parentElement\n  while (parent)\n    if (\n      el.scrollWidth  <= parent.scrollWidth &&\n      el.scrollHeight <= parent.scrollHeight\n    )\n      parent = (el = parent).parentElement\n    else\n      break\n\n  /* Return overflowing container */\n  return parent ? el : undefined\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  NEVER,\n  Observable,\n  Subject,\n  defer,\n  distinctUntilChanged,\n  filter,\n  finalize,\n  map,\n  merge,\n  of,\n  shareReplay,\n  switchMap,\n  tap\n} from \"rxjs\"\n\nimport {\n  getElementContentSize,\n  getElementSize,\n  watchElementContentOffset\n} from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Intersection observer entry subject\n */\nconst entry$ = new Subject<IntersectionObserverEntry>()\n\n/**\n * Intersection observer observable\n *\n * This observable will create an `IntersectionObserver` on first subscription\n * and will automatically terminate it when there are no more subscribers.\n *\n * @see https://bit.ly/3iIYfEm - Google Groups on performance\n */\nconst observer$ = defer(() => of(\n  new IntersectionObserver(entries => {\n    for (const entry of entries)\n      entry$.next(entry)\n  }, {\n    threshold: 0\n  })\n))\n  .pipe(\n    switchMap(observer => merge(NEVER, of(observer))\n      .pipe(\n        finalize(() => observer.disconnect())\n      )\n    ),\n    shareReplay(1)\n  )\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch element visibility\n *\n * @param el - Element\n *\n * @returns Element visibility observable\n */\nexport function watchElementVisibility(\n  el: HTMLElement\n): Observable<boolean> {\n  return observer$\n    .pipe(\n      tap(observer => observer.observe(el)),\n      switchMap(observer => entry$\n        .pipe(\n          filter(({ target }) => target === el),\n          finalize(() => observer.unobserve(el)),\n          map(({ isIntersecting }) => isIntersecting)\n        )\n      )\n    )\n}\n\n/**\n * Watch element boundary\n *\n * This function returns an observable which emits whether the bottom content\n * boundary (= scroll offset) of an element is within a certain threshold.\n *\n * @param el - Element\n * @param threshold - Threshold\n *\n * @returns Element boundary observable\n */\nexport function watchElementBoundary(\n  el: HTMLElement, threshold = 16\n): Observable<boolean> {\n  return watchElementContentOffset(el)\n    .pipe(\n      map(({ y }) => {\n        const visible = getElementSize(el)\n        const content = getElementContentSize(el)\n        return y >= (\n          content.height - visible.height - threshold\n        )\n      }),\n      distinctUntilChanged()\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  fromEvent,\n  map,\n  startWith\n} from \"rxjs\"\n\nimport { getElement } from \"../element\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Toggle\n */\nexport type Toggle =\n  | \"drawer\"                           /* Toggle for drawer */\n  | \"search\"                           /* Toggle for search */\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Toggle map\n */\nconst toggles: Record<Toggle, HTMLInputElement> = {\n  drawer: getElement(\"[data-md-toggle=drawer]\"),\n  search: getElement(\"[data-md-toggle=search]\")\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve the value of a toggle\n *\n * @param name - Toggle\n *\n * @returns Toggle value\n */\nexport function getToggle(name: Toggle): boolean {\n  return toggles[name].checked\n}\n\n/**\n * Set toggle\n *\n * Simulating a click event seems to be the most cross-browser compatible way\n * of changing the value while also emitting a `change` event. Before, Material\n * used `CustomEvent` to programmatically change the value of a toggle, but this\n * is a much simpler and cleaner solution which doesn't require a polyfill.\n *\n * @param name - Toggle\n * @param value - Toggle value\n */\nexport function setToggle(name: Toggle, value: boolean): void {\n  if (toggles[name].checked !== value)\n    toggles[name].click()\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch toggle\n *\n * @param name - Toggle\n *\n * @returns Toggle value observable\n */\nexport function watchToggle(name: Toggle): Observable<boolean> {\n  const el = toggles[name]\n  return fromEvent(el, \"change\")\n    .pipe(\n      map(() => el.checked),\n      startWith(el.checked)\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  filter,\n  fromEvent,\n  map,\n  merge,\n  share,\n  startWith,\n  switchMap\n} from \"rxjs\"\n\nimport { getActiveElement } from \"../element\"\nimport { getToggle } from \"../toggle\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Keyboard mode\n */\nexport type KeyboardMode =\n  | \"global\"                           /* Global */\n  | \"search\"                           /* Search is open */\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Keyboard\n */\nexport interface Keyboard {\n  mode: KeyboardMode                   /* Keyboard mode */\n  type: string                         /* Key type */\n  claim(): void                        /* Key claim */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Check whether an element may receive keyboard input\n *\n * @param el - Element\n * @param type - Key type\n *\n * @returns Test result\n */\nfunction isSusceptibleToKeyboard(\n  el: HTMLElement, type: string\n): boolean {\n  switch (el.constructor) {\n\n    /* Input elements */\n    case HTMLInputElement:\n      /* @ts-expect-error - omit unnecessary type cast */\n      if (el.type === \"radio\")\n        return /^Arrow/.test(type)\n      else\n        return true\n\n    /* Select element and textarea */\n    case HTMLSelectElement:\n    case HTMLTextAreaElement:\n      return true\n\n    /* Everything else */\n    default:\n      return el.isContentEditable\n  }\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch composition events\n *\n * @returns Composition observable\n */\nexport function watchComposition(): Observable<boolean> {\n  return merge(\n    fromEvent(window, \"compositionstart\").pipe(map(() => true)),\n    fromEvent(window, \"compositionend\").pipe(map(() => false))\n  )\n    .pipe(\n      startWith(false)\n    )\n}\n\n/**\n * Watch keyboard\n *\n * @returns Keyboard observable\n */\nexport function watchKeyboard(): Observable<Keyboard> {\n  const keyboard$ = fromEvent<KeyboardEvent>(window, \"keydown\")\n    .pipe(\n      filter(ev => !(ev.metaKey || ev.ctrlKey)),\n      map(ev => ({\n        mode: getToggle(\"search\") ? \"search\" : \"global\",\n        type: ev.key,\n        claim() {\n          ev.preventDefault()\n          ev.stopPropagation()\n        }\n      } as Keyboard)),\n      filter(({ mode, type }) => {\n        if (mode === \"global\") {\n          const active = getActiveElement()\n          if (typeof active !== \"undefined\")\n            return !isSusceptibleToKeyboard(active, type)\n        }\n        return true\n      }),\n      share()\n    )\n\n  /* Don't emit during composition events - see https://bit.ly/3te3Wl8 */\n  return watchComposition()\n    .pipe(\n      switchMap(active => !active ? keyboard$ : EMPTY)\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { Subject } from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve location\n *\n * This function returns a `URL` object (and not `Location`) to normalize the\n * typings across the application. Furthermore, locations need to be tracked\n * without setting them and `Location` is a singleton which represents the\n * current location.\n *\n * @returns URL\n */\nexport function getLocation(): URL {\n  return new URL(location.href)\n}\n\n/**\n * Set location\n *\n * If instant navigation is enabled, this function creates a temporary anchor\n * element, sets the `href` attribute, appends it to the body, clicks it, and\n * then removes it again. The event will bubble up the DOM and trigger be\n * intercepted by the instant loading business logic.\n *\n * Note that we must append and remove the anchor element, or the event will\n * not bubble up the DOM, making it impossible to intercept it.\n *\n * @param url - URL to navigate to\n * @param navigate - Force navigation\n */\nexport function setLocation(\n  url: URL | HTMLLinkElement, navigate = false\n): void {\n  if (feature(\"navigation.instant\") && !navigate) {\n    const el = h(\"a\", { href: url.href })\n    document.body.appendChild(el)\n    el.click()\n    el.remove()\n\n  // If we're not using instant navigation, and the page should not be reloaded\n  // just instruct the browser to navigate to the given URL\n  } else {\n    location.href = url.href\n  }\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch location\n *\n * @returns Location subject\n */\nexport function watchLocation(): Subject<URL> {\n  return new Subject<URL>()\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  filter,\n  fromEvent,\n  map,\n  merge,\n  shareReplay,\n  startWith\n} from \"rxjs\"\n\nimport { getOptionalElement } from \"~/browser\"\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve location hash\n *\n * @returns Location hash\n */\nexport function getLocationHash(): string {\n  return location.hash.slice(1)\n}\n\n/**\n * Set location hash\n *\n * Setting a new fragment identifier via `location.hash` will have no effect\n * if the value doesn't change. When a new fragment identifier is set, we want\n * the browser to target the respective element at all times, which is why we\n * use this dirty little trick.\n *\n * @param hash - Location hash\n */\nexport function setLocationHash(hash: string): void {\n  const el = h(\"a\", { href: hash })\n  el.addEventListener(\"click\", ev => ev.stopPropagation())\n  el.click()\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch location hash\n *\n * @param location$ - Location observable\n *\n * @returns Location hash observable\n */\nexport function watchLocationHash(\n  location$: Observable<URL>\n): Observable<string> {\n  return merge(\n    fromEvent<HashChangeEvent>(window, \"hashchange\"),\n    location$\n  )\n    .pipe(\n      map(getLocationHash),\n      startWith(getLocationHash()),\n      filter(hash => hash.length > 0),\n      shareReplay(1)\n    )\n}\n\n/**\n * Watch location target\n *\n * @param location$ - Location observable\n *\n * @returns Location target observable\n */\nexport function watchLocationTarget(\n  location$: Observable<URL>\n): Observable<HTMLElement> {\n  return watchLocationHash(location$)\n    .pipe(\n      map(id => getOptionalElement(`[id=\"${id}\"]`)!),\n      filter(el => typeof el !== \"undefined\")\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  fromEvent,\n  fromEventPattern,\n  map,\n  merge,\n  startWith,\n  switchMap\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch media query\n *\n * Note that although `MediaQueryList.addListener` is deprecated we have to\n * use it, because it's the only way to ensure proper downward compatibility.\n *\n * @see https://bit.ly/3dUBH2m - GitHub issue\n *\n * @param query - Media query\n *\n * @returns Media observable\n */\nexport function watchMedia(query: string): Observable<boolean> {\n  const media = matchMedia(query)\n  return fromEventPattern<boolean>(next => (\n    media.addListener(() => next(media.matches))\n  ))\n    .pipe(\n      startWith(media.matches)\n    )\n}\n\n/**\n * Watch print mode\n *\n * @returns Print observable\n */\nexport function watchPrint(): Observable<boolean> {\n  const media = matchMedia(\"print\")\n  return merge(\n    fromEvent(window, \"beforeprint\").pipe(map(() => true)),\n    fromEvent(window, \"afterprint\").pipe(map(() => false))\n  )\n    .pipe(\n      startWith(media.matches)\n    )\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Toggle an observable with a media observable\n *\n * @template T - Data type\n *\n * @param query$ - Media observable\n * @param factory - Observable factory\n *\n * @returns Toggled observable\n */\nexport function at<T>(\n  query$: Observable<boolean>, factory: () => Observable<T>\n): Observable<T> {\n  return query$\n    .pipe(\n      switchMap(active => active ? factory() : EMPTY)\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  map,\n  shareReplay,\n  switchMap\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Options\n */\ninterface Options {\n  progress$?: Subject<number>          // Progress subject\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch the given URL\n *\n * If the request fails (e.g. when dispatched from `file://` locations), the\n * observable will complete without emitting a value.\n *\n * @param url - Request URL\n * @param options - Options\n *\n * @returns Response observable\n */\nexport function request(\n  url: URL | string, options?: Options\n): Observable<Blob> {\n  return new Observable<Blob>(observer => {\n    const req = new XMLHttpRequest()\n    req.open(\"GET\",  `${url}`)\n    req.responseType = \"blob\"\n\n    // Handle response\n    req.addEventListener(\"load\", () => {\n      if (req.status >= 200 && req.status < 300) {\n        observer.next(req.response)\n        observer.complete()\n      } else {\n        observer.error(new Error(req.statusText))\n      }\n    })\n\n    // Handle network errors\n    req.addEventListener(\"error\", () => {\n      observer.error(new Error(\"Network Error\"))\n    })\n\n    // Handle aborted requests\n    req.addEventListener(\"abort\", () => {\n      observer.error(new Error(\"Request aborted\"))\n    })\n\n    // Handle download progress\n    if (typeof options?.progress$ !== \"undefined\") {\n      req.addEventListener(\"progress\", event => {\n        if (event.lengthComputable) {\n          options.progress$!.next((event.loaded / event.total) * 100)\n        } else { // https://bugs.chromium.org/p/chromium/issues/detail?id=463622\n          const totalFromHeader = Number(req.getResponseHeader(\"Content-Length\")) || 0\n          options.progress$!.next((event.loaded / totalFromHeader) * 100)\n        }\n      })\n\n      // Immediately set progress to 5% to indicate that we're loading\n      options.progress$.next(5)\n    }\n\n    // Send request\n    req.send()\n  })\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Fetch JSON from the given URL\n *\n * @template T - Data type\n *\n * @param url - Request URL\n * @param options - Options\n *\n * @returns Data observable\n */\nexport function requestJSON<T>(\n  url: URL | string, options?: Options\n): Observable<T> {\n  return request(url, options)\n    .pipe(\n      switchMap(res => res.text()),\n      map(body => JSON.parse(body) as T),\n      shareReplay(1)\n    )\n}\n\n/**\n * Fetch XML from the given URL\n *\n * @param url - Request URL\n * @param options - Options\n *\n * @returns Data observable\n */\nexport function requestXML(\n  url: URL | string, options?: Options\n): Observable<Document> {\n  const dom = new DOMParser()\n  return request(url, options)\n    .pipe(\n      switchMap(res => res.text()),\n      map(res => dom.parseFromString(res, \"text/xml\")),\n      shareReplay(1)\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  fromEvent,\n  map,\n  merge,\n  startWith\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Viewport offset\n */\nexport interface ViewportOffset {\n  x: number                            /* Horizontal offset */\n  y: number                            /* Vertical offset */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve viewport offset\n *\n * On iOS Safari, viewport offset can be negative due to overflow scrolling.\n * As this may induce strange behaviors downstream, we'll just limit it to 0.\n *\n * @returns Viewport offset\n */\nexport function getViewportOffset(): ViewportOffset {\n  return {\n    x: Math.max(0, scrollX),\n    y: Math.max(0, scrollY)\n  }\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch viewport offset\n *\n * @returns Viewport offset observable\n */\nexport function watchViewportOffset(): Observable<ViewportOffset> {\n  return merge(\n    fromEvent(window, \"scroll\", { passive: true }),\n    fromEvent(window, \"resize\", { passive: true })\n  )\n    .pipe(\n      map(getViewportOffset),\n      startWith(getViewportOffset())\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  fromEvent,\n  map,\n  startWith\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Viewport size\n */\nexport interface ViewportSize {\n  width: number                        /* Viewport width */\n  height: number                       /* Viewport height */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve viewport size\n *\n * @returns Viewport size\n */\nexport function getViewportSize(): ViewportSize {\n  return {\n    width:  innerWidth,\n    height: innerHeight\n  }\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Watch viewport size\n *\n * @returns Viewport size observable\n */\nexport function watchViewportSize(): Observable<ViewportSize> {\n  return fromEvent(window, \"resize\", { passive: true })\n    .pipe(\n      map(getViewportSize),\n      startWith(getViewportSize())\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  combineLatest,\n  map,\n  shareReplay\n} from \"rxjs\"\n\nimport {\n  ViewportOffset,\n  watchViewportOffset\n} from \"../offset\"\nimport {\n  ViewportSize,\n  watchViewportSize\n} from \"../size\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Viewport\n */\nexport interface Viewport {\n  offset: ViewportOffset               /* Viewport offset */\n  size: ViewportSize                   /* Viewport size */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch viewport\n *\n * @returns Viewport observable\n */\nexport function watchViewport(): Observable<Viewport> {\n  return combineLatest([\n    watchViewportOffset(),\n    watchViewportSize()\n  ])\n    .pipe(\n      map(([offset, size]) => ({ offset, size })),\n      shareReplay(1)\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  combineLatest,\n  distinctUntilKeyChanged,\n  map\n} from \"rxjs\"\n\nimport { Header } from \"~/components\"\n\nimport { getElementOffset } from \"../../element\"\nimport { Viewport } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch viewport relative to element\n *\n * @param el - Element\n * @param options - Options\n *\n * @returns Viewport observable\n */\nexport function watchViewportAt(\n  el: HTMLElement, { viewport$, header$ }: WatchOptions\n): Observable<Viewport> {\n  const size$ = viewport$\n    .pipe(\n      distinctUntilKeyChanged(\"size\")\n    )\n\n  /* Compute element offset */\n  const offset$ = combineLatest([size$, header$])\n    .pipe(\n      map(() => getElementOffset(el))\n    )\n\n  /* Compute relative viewport, return hot observable */\n  return combineLatest([header$, viewport$, offset$])\n    .pipe(\n      map(([{ height }, { offset, size }, { x, y }]) => ({\n        offset: {\n          x: offset.x - x,\n          y: offset.y - y + height\n        },\n        size\n      }))\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  endWith,\n  fromEvent,\n  ignoreElements,\n  mergeWith,\n  share,\n  takeUntil\n} from \"rxjs\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Worker message\n */\nexport interface WorkerMessage {\n  type: unknown                        /* Message type */\n  data?: unknown                       /* Message data */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Create an observable for receiving from a web worker\n *\n * @template T - Data type\n *\n * @param worker - Web worker\n *\n * @returns Message observable\n */\nfunction recv<T>(worker: Worker): Observable<T> {\n  return fromEvent<MessageEvent<T>, T>(worker, \"message\", ev => ev.data)\n}\n\n/**\n * Create a subject for sending to a web worker\n *\n * @template T - Data type\n *\n * @param worker - Web worker\n *\n * @returns Message subject\n */\nfunction send<T>(worker: Worker): Subject<T> {\n  const send$ = new Subject<T>()\n  send$.subscribe(data => worker.postMessage(data))\n\n  /* Return message subject */\n  return send$\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Create a bidirectional communication channel to a web worker\n *\n * @template T - Data type\n *\n * @param url - Worker URL\n * @param worker - Worker\n *\n * @returns Worker subject\n */\nexport function watchWorker<T extends WorkerMessage>(\n  url: string, worker = new Worker(url)\n): Subject<T> {\n  const recv$ = recv<T>(worker)\n  const send$ = send<T>(worker)\n\n  /* Create worker subject and forward messages */\n  const worker$ = new Subject<T>()\n  worker$.subscribe(send$)\n\n  /* Return worker subject */\n  const done$ = send$.pipe(ignoreElements(), endWith(true))\n  return worker$\n    .pipe(\n      ignoreElements(),\n      mergeWith(recv$.pipe(takeUntil(done$))),\n      share()\n    ) as Subject<T>\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { getElement, getLocation } from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Feature flag\n */\nexport type Flag =\n  | \"announce.dismiss\"                 /* Dismissable announcement bar */\n  | \"content.code.annotate\"            /* Code annotations */\n  | \"content.code.copy\"                /* Code copy button */\n  | \"content.lazy\"                     /* Lazy content elements */\n  | \"content.tabs.link\"                /* Link content tabs */\n  | \"header.autohide\"                  /* Hide header */\n  | \"navigation.expand\"                /* Automatic expansion */\n  | \"navigation.indexes\"               /* Section pages */\n  | \"navigation.instant\"               /* Instant navigation */\n  | \"navigation.instant.progress\"      /* Instant navigation progress */\n  | \"navigation.sections\"              /* Section navigation */\n  | \"navigation.tabs\"                  /* Tabs navigation */\n  | \"navigation.tabs.sticky\"           /* Tabs navigation (sticky) */\n  | \"navigation.top\"                   /* Back-to-top button */\n  | \"navigation.tracking\"              /* Anchor tracking */\n  | \"search.highlight\"                 /* Search highlighting */\n  | \"search.share\"                     /* Search sharing */\n  | \"search.suggest\"                   /* Search suggestions */\n  | \"toc.follow\"                       /* Following table of contents */\n  | \"toc.integrate\"                    /* Integrated table of contents */\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Translation\n */\nexport type Translation =\n  | \"clipboard.copy\"                   /* Copy to clipboard */\n  | \"clipboard.copied\"                 /* Copied to clipboard */\n  | \"search.result.placeholder\"        /* Type to start searching */\n  | \"search.result.none\"               /* No matching documents */\n  | \"search.result.one\"                /* 1 matching document */\n  | \"search.result.other\"              /* # matching documents */\n  | \"search.result.more.one\"           /* 1 more on this page */\n  | \"search.result.more.other\"         /* # more on this page */\n  | \"search.result.term.missing\"       /* Missing */\n  | \"select.version\"                   /* Version selector */\n\n/**\n * Translations\n */\nexport type Translations =\n  Record<Translation, string>\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Versioning\n */\nexport interface Versioning {\n  provider: \"mike\"                     /* Version provider */\n  default?: string | string[]          /* Default version */\n}\n\n/**\n * Configuration\n */\nexport interface Config {\n  base: string                         /* Base URL */\n  features: Flag[]                     /* Feature flags */\n  translations: Translations           /* Translations */\n  search: string                       /* Search worker URL */\n  tags?: Record<string, string>        /* Tags mapping */\n  version?: Versioning                 /* Versioning */\n}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve global configuration and make base URL absolute\n */\nconst script = getElement(\"#__config\")\nconst config: Config = JSON.parse(script.textContent!)\nconfig.base = `${new URL(config.base, getLocation())}`\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve global configuration\n *\n * @returns Global configuration\n */\nexport function configuration(): Config {\n  return config\n}\n\n/**\n * Check whether a feature flag is enabled\n *\n * @param flag - Feature flag\n *\n * @returns Test result\n */\nexport function feature(flag: Flag): boolean {\n  return config.features.includes(flag)\n}\n\n/**\n * Retrieve the translation for the given key\n *\n * @param key - Key to be translated\n * @param value - Positional value, if any\n *\n * @returns Translation\n */\nexport function translation(\n  key: Translation, value?: string | number\n): string {\n  return typeof value !== \"undefined\"\n    ? config.translations[key].replace(\"#\", value.toString())\n    : config.translations[key]\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { getElement, getElements } from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Component type\n */\nexport type ComponentType =\n  | \"announce\"                         /* Announcement bar */\n  | \"container\"                        /* Container */\n  | \"consent\"                          /* Consent */\n  | \"content\"                          /* Content */\n  | \"dialog\"                           /* Dialog */\n  | \"header\"                           /* Header */\n  | \"header-title\"                     /* Header title */\n  | \"header-topic\"                     /* Header topic */\n  | \"main\"                             /* Main area */\n  | \"outdated\"                         /* Version warning */\n  | \"palette\"                          /* Color palette */\n  | \"progress\"                         /* Progress indicator */\n  | \"search\"                           /* Search */\n  | \"search-query\"                     /* Search input */\n  | \"search-result\"                    /* Search results */\n  | \"search-share\"                     /* Search sharing */\n  | \"search-suggest\"                   /* Search suggestions */\n  | \"sidebar\"                          /* Sidebar */\n  | \"skip\"                             /* Skip link */\n  | \"source\"                           /* Repository information */\n  | \"tabs\"                             /* Navigation tabs */\n  | \"toc\"                              /* Table of contents */\n  | \"top\"                              /* Back-to-top button */\n\n/**\n * Component\n *\n * @template T - Component type\n * @template U - Reference type\n */\nexport type Component<\n  T extends {} = {},\n  U extends HTMLElement = HTMLElement\n> =\n  T & {\n    ref: U                             /* Component reference */\n  }\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Component type map\n */\ninterface ComponentTypeMap {\n  \"announce\": HTMLElement              /* Announcement bar */\n  \"container\": HTMLElement             /* Container */\n  \"consent\": HTMLElement               /* Consent */\n  \"content\": HTMLElement               /* Content */\n  \"dialog\": HTMLElement                /* Dialog */\n  \"header\": HTMLElement                /* Header */\n  \"header-title\": HTMLElement          /* Header title */\n  \"header-topic\": HTMLElement          /* Header topic */\n  \"main\": HTMLElement                  /* Main area */\n  \"outdated\": HTMLElement              /* Version warning */\n  \"palette\": HTMLElement               /* Color palette */\n  \"progress\": HTMLElement              /* Progress indicator */\n  \"search\": HTMLElement                /* Search */\n  \"search-query\": HTMLInputElement     /* Search input */\n  \"search-result\": HTMLElement         /* Search results */\n  \"search-share\": HTMLAnchorElement    /* Search sharing */\n  \"search-suggest\": HTMLElement        /* Search suggestions */\n  \"sidebar\": HTMLElement               /* Sidebar */\n  \"skip\": HTMLAnchorElement            /* Skip link */\n  \"source\": HTMLAnchorElement          /* Repository information */\n  \"tabs\": HTMLElement                  /* Navigation tabs */\n  \"toc\": HTMLElement                   /* Table of contents */\n  \"top\": HTMLAnchorElement             /* Back-to-top button */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve the element for a given component or throw a reference error\n *\n * @template T - Component type\n *\n * @param type - Component type\n * @param node - Node of reference\n *\n * @returns Element\n */\nexport function getComponentElement<T extends ComponentType>(\n  type: T, node: ParentNode = document\n): ComponentTypeMap[T] {\n  return getElement(`[data-md-component=${type}]`, node)\n}\n\n/**\n * Retrieve all elements for a given component\n *\n * @template T - Component type\n *\n * @param type - Component type\n * @param node - Node of reference\n *\n * @returns Elements\n */\nexport function getComponentElements<T extends ComponentType>(\n  type: T, node: ParentNode = document\n): ComponentTypeMap[T][] {\n  return getElements(`[data-md-component=${type}]`, node)\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  defer,\n  finalize,\n  fromEvent,\n  map,\n  tap\n} from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport { getElement } from \"~/browser\"\n\nimport { Component } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Announcement bar\n */\nexport interface Announce {\n  hash: number                        /* Content hash */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch announcement bar\n *\n * @param el - Announcement bar element\n *\n * @returns Announcement bar observable\n */\nexport function watchAnnounce(\n  el: HTMLElement\n): Observable<Announce> {\n  const button = getElement(\".md-typeset > :first-child\", el)\n  return fromEvent(button, \"click\", { once: true })\n    .pipe(\n      map(() => getElement(\".md-typeset\", el)),\n      map(content => ({ hash: __md_hash(content.innerHTML) }))\n    )\n}\n\n/**\n * Mount announcement bar\n *\n * @param el - Announcement bar element\n *\n * @returns Announcement bar component observable\n */\nexport function mountAnnounce(\n  el: HTMLElement\n): Observable<Component<Announce>> {\n  if (!feature(\"announce.dismiss\") || !el.childElementCount)\n    return EMPTY\n\n  /* Support instant navigation - see https://t.ly/3FTme */\n  if (!el.hidden) {\n    const content = getElement(\".md-typeset\", el)\n    if (__md_hash(content.innerHTML) === __md_get(\"__announce\"))\n      el.hidden = true\n  }\n\n  /* Mount component on subscription */\n  return defer(() => {\n    const push$ = new Subject<Announce>()\n    push$.subscribe(({ hash }) => {\n      el.hidden = true\n\n      /* Persist preference in local storage */\n      __md_set<number>(\"__announce\", hash)\n    })\n\n    /* Create and return component */\n    return watchAnnounce(el)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  finalize,\n  map,\n  tap\n} from \"rxjs\"\n\nimport { Component } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Consent\n */\nexport interface Consent {\n  hidden: boolean                      /* Consent is hidden */\n}\n\n/**\n * Consent defaults\n */\nexport interface ConsentDefaults {\n  analytics?: boolean                  /* Consent for Analytics */\n  github?: boolean                     /* Consent for GitHub */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  target$: Observable<HTMLElement>     /* Target observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  target$: Observable<HTMLElement>     /* Target observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch consent\n *\n * @param el - Consent element\n * @param options - Options\n *\n * @returns Consent observable\n */\nexport function watchConsent(\n  el: HTMLElement, { target$ }: WatchOptions\n): Observable<Consent> {\n  return target$\n    .pipe(\n      map(target => ({ hidden: target !== el }))\n    )\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Mount consent\n *\n * @param el - Consent element\n * @param options - Options\n *\n * @returns Consent component observable\n */\nexport function mountConsent(\n  el: HTMLElement, options: MountOptions\n): Observable<Component<Consent>> {\n  const internal$ = new Subject<Consent>()\n  internal$.subscribe(({ hidden }) => {\n    el.hidden = hidden\n  })\n\n  /* Create and return component */\n  return watchConsent(el, options)\n    .pipe(\n      tap(state => internal$.next(state)),\n      finalize(() => internal$.complete()),\n      map(state => ({ ref: el, ...state }))\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  animationFrameScheduler,\n  auditTime,\n  combineLatest,\n  debounceTime,\n  defer,\n  delay,\n  endWith,\n  filter,\n  finalize,\n  fromEvent,\n  ignoreElements,\n  map,\n  merge,\n  switchMap,\n  take,\n  takeUntil,\n  tap,\n  throttleTime,\n  withLatestFrom\n} from \"rxjs\"\n\nimport {\n  ElementOffset,\n  getActiveElement,\n  getElementSize,\n  watchElementContentOffset,\n  watchElementFocus,\n  watchElementOffset,\n  watchElementVisibility\n} from \"~/browser\"\n\nimport { Component } from \"../../../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Annotation\n */\nexport interface Annotation {\n  active: boolean                      /* Annotation is active */\n  offset: ElementOffset                /* Annotation offset */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  target$: Observable<HTMLElement>     /* Location target observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch annotation\n *\n * @param el - Annotation element\n * @param container - Containing element\n *\n * @returns Annotation observable\n */\nexport function watchAnnotation(\n  el: HTMLElement, container: HTMLElement\n): Observable<Annotation> {\n  const offset$ = defer(() => combineLatest([\n    watchElementOffset(el),\n    watchElementContentOffset(container)\n  ]))\n    .pipe(\n      map(([{ x, y }, scroll]): ElementOffset => {\n        const { width, height } = getElementSize(el)\n        return ({\n          x: x - scroll.x + width  / 2,\n          y: y - scroll.y + height / 2\n        })\n      })\n    )\n\n  /* Actively watch annotation on focus */\n  return watchElementFocus(el)\n    .pipe(\n      switchMap(active => offset$\n        .pipe(\n          map(offset => ({ active, offset })),\n          take(+!active || Infinity)\n        )\n      )\n    )\n}\n\n/**\n * Mount annotation\n *\n * @param el - Annotation element\n * @param container - Containing element\n * @param options - Options\n *\n * @returns Annotation component observable\n */\nexport function mountAnnotation(\n  el: HTMLElement, container: HTMLElement, { target$ }: MountOptions\n): Observable<Component<Annotation>> {\n  const [tooltip, index] = Array.from(el.children)\n\n  /* Mount component on subscription */\n  return defer(() => {\n    const push$ = new Subject<Annotation>()\n    const done$ = push$.pipe(ignoreElements(), endWith(true))\n    push$.subscribe({\n\n      /* Handle emission */\n      next({ offset }) {\n        el.style.setProperty(\"--md-tooltip-x\", `${offset.x}px`)\n        el.style.setProperty(\"--md-tooltip-y\", `${offset.y}px`)\n      },\n\n      /* Handle complete */\n      complete() {\n        el.style.removeProperty(\"--md-tooltip-x\")\n        el.style.removeProperty(\"--md-tooltip-y\")\n      }\n    })\n\n    /* Start animation only when annotation is visible */\n    watchElementVisibility(el)\n      .pipe(\n        takeUntil(done$)\n      )\n        .subscribe(visible => {\n          el.toggleAttribute(\"data-md-visible\", visible)\n        })\n\n    /* Toggle tooltip presence to mitigate empty lines when copying */\n    merge(\n      push$.pipe(filter(({ active }) => active)),\n      push$.pipe(debounceTime(250), filter(({ active }) => !active))\n    )\n      .subscribe({\n\n        /* Handle emission */\n        next({ active }) {\n          if (active)\n            el.prepend(tooltip)\n          else\n            tooltip.remove()\n        },\n\n        /* Handle complete */\n        complete() {\n          el.prepend(tooltip)\n        }\n      })\n\n    /* Toggle tooltip visibility */\n    push$\n      .pipe(\n        auditTime(16, animationFrameScheduler)\n      )\n        .subscribe(({ active }) => {\n          tooltip.classList.toggle(\"md-tooltip--active\", active)\n        })\n\n    /* Track relative origin of tooltip */\n    push$\n      .pipe(\n        throttleTime(125, animationFrameScheduler),\n        filter(() => !!el.offsetParent),\n        map(() => el.offsetParent!.getBoundingClientRect()),\n        map(({ x }) => x)\n      )\n        .subscribe({\n\n          /* Handle emission */\n          next(origin) {\n            if (origin)\n              el.style.setProperty(\"--md-tooltip-0\", `${-origin}px`)\n            else\n              el.style.removeProperty(\"--md-tooltip-0\")\n          },\n\n          /* Handle complete */\n          complete() {\n            el.style.removeProperty(\"--md-tooltip-0\")\n          }\n        })\n\n    /* Allow to copy link without scrolling to anchor */\n    fromEvent<MouseEvent>(index, \"click\")\n      .pipe(\n        takeUntil(done$),\n        filter(ev => !(ev.metaKey || ev.ctrlKey))\n      )\n        .subscribe(ev => {\n          ev.stopPropagation()\n          ev.preventDefault()\n        })\n\n    /* Allow to open link in new tab or blur on close */\n    fromEvent<MouseEvent>(index, \"mousedown\")\n      .pipe(\n        takeUntil(done$),\n        withLatestFrom(push$)\n      )\n        .subscribe(([ev, { active }]) => {\n\n          /* Open in new tab */\n          if (ev.button !== 0 || ev.metaKey || ev.ctrlKey) {\n            ev.preventDefault()\n\n          /* Close annotation */\n          } else if (active) {\n            ev.preventDefault()\n\n            /* Focus parent annotation, if any */\n            const parent = el.parentElement!.closest(\".md-annotation\")\n            if (parent instanceof HTMLElement)\n              parent.focus()\n            else\n              getActiveElement()?.blur()\n          }\n        })\n\n    /* Open and focus annotation on location target */\n    target$\n      .pipe(\n        takeUntil(done$),\n        filter(target => target === tooltip),\n        delay(125)\n      )\n        .subscribe(() => el.focus())\n\n    /* Create and return component */\n    return watchAnnotation(el, container)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render a tooltip\n *\n * @param id - Tooltip identifier\n *\n * @returns Element\n */\nexport function renderTooltip(id?: string): HTMLElement {\n  return (\n    <div class=\"md-tooltip\" id={id}>\n      <div class=\"md-tooltip__inner md-typeset\"></div>\n    </div>\n  )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { h } from \"~/utilities\"\n\nimport { renderTooltip } from \"../tooltip\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render an annotation\n *\n * @param id - Annotation identifier\n * @param prefix - Tooltip identifier prefix\n *\n * @returns Element\n */\nexport function renderAnnotation(\n  id: string | number, prefix?: string\n): HTMLElement {\n  prefix = prefix ? `${prefix}_annotation_${id}` : undefined\n\n  /* Render tooltip with anchor, if given */\n  if (prefix) {\n    const anchor = prefix ? `#${prefix}` : undefined\n    return (\n      <aside class=\"md-annotation\" tabIndex={0}>\n        {renderTooltip(prefix)}\n        <a href={anchor} class=\"md-annotation__index\" tabIndex={-1}>\n          <span data-md-annotation-id={id}></span>\n        </a>\n      </aside>\n    )\n  } else {\n    return (\n      <aside class=\"md-annotation\" tabIndex={0}>\n        {renderTooltip(prefix)}\n        <span class=\"md-annotation__index\" tabIndex={-1}>\n          <span data-md-annotation-id={id}></span>\n        </span>\n      </aside>\n    )\n  }\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { translation } from \"~/_\"\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render a 'copy-to-clipboard' button\n *\n * @param id - Unique identifier\n *\n * @returns Element\n */\nexport function renderClipboardButton(id: string): HTMLElement {\n  return (\n    <button\n      class=\"md-clipboard md-icon\"\n      title={translation(\"clipboard.copy\")}\n      data-clipboard-target={`#${id} > code`}\n    ></button>\n  )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { ComponentChild } from \"preact\"\n\nimport { configuration, feature, translation } from \"~/_\"\nimport { SearchItem } from \"~/integrations/search\"\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Render flag\n */\nconst enum Flag {\n  TEASER = 1,                          /* Render teaser */\n  PARENT = 2                           /* Render as parent */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper function\n * ------------------------------------------------------------------------- */\n\n/**\n * Render a search document\n *\n * @param document - Search document\n * @param flag - Render flags\n *\n * @returns Element\n */\nfunction renderSearchDocument(\n  document: SearchItem, flag: Flag\n): HTMLElement {\n  const parent = flag & Flag.PARENT\n  const teaser = flag & Flag.TEASER\n\n  /* Render missing query terms */\n  const missing = Object.keys(document.terms)\n    .filter(key => !document.terms[key])\n    .reduce<ComponentChild[]>((list, key) => [\n      ...list, <del>{key}</del>, \" \"\n    ], [])\n    .slice(0, -1)\n\n  /* Assemble query string for highlighting */\n  const config = configuration()\n  const url = new URL(document.location, config.base)\n  if (feature(\"search.highlight\"))\n    url.searchParams.set(\"h\", Object.entries(document.terms)\n      .filter(([, match]) => match)\n      .reduce((highlight, [value]) => `${highlight} ${value}`.trim(), \"\")\n    )\n\n  /* Render article or section, depending on flags */\n  const { tags } = configuration()\n  return (\n    <a href={`${url}`} class=\"md-search-result__link\" tabIndex={-1}>\n      <article\n        class=\"md-search-result__article md-typeset\"\n        data-md-score={document.score.toFixed(2)}\n      >\n        {parent > 0 && <div class=\"md-search-result__icon md-icon\"></div>}\n        {parent > 0 && <h1>{document.title}</h1>}\n        {parent <= 0 && <h2>{document.title}</h2>}\n        {teaser > 0 && document.text.length > 0 &&\n          document.text\n        }\n        {document.tags && document.tags.map(tag => {\n          const type = tags\n            ? tag in tags\n              ? `md-tag-icon md-tag--${tags[tag]}`\n              : \"md-tag-icon\"\n            : \"\"\n          return (\n            <span class={`md-tag ${type}`}>{tag}</span>\n          )\n        })}\n        {teaser > 0 && missing.length > 0 &&\n          <p class=\"md-search-result__terms\">\n            {translation(\"search.result.term.missing\")}: {...missing}\n          </p>\n        }\n      </article>\n    </a>\n  )\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render a search result\n *\n * @param result - Search result\n *\n * @returns Element\n */\nexport function renderSearchResultItem(\n  result: SearchItem[]\n): HTMLElement {\n  const threshold = result[0].score\n  const docs = [...result]\n\n  const config = configuration()\n\n  /* Find and extract parent article */\n  const parent = docs.findIndex(doc => {\n    const l = `${new URL(doc.location, config.base)}` // @todo hacky\n    return !l.includes(\"#\")\n  })\n  const [article] = docs.splice(parent, 1)\n\n  /* Determine last index above threshold */\n  let index = docs.findIndex(doc => doc.score < threshold)\n  if (index === -1)\n    index = docs.length\n\n  /* Partition sections */\n  const best = docs.slice(0, index)\n  const more = docs.slice(index)\n\n  /* Render children */\n  const children = [\n    renderSearchDocument(article, Flag.PARENT | +(!parent && index === 0)),\n    ...best.map(section => renderSearchDocument(section, Flag.TEASER)),\n    ...more.length ? [\n      <details class=\"md-search-result__more\">\n        <summary tabIndex={-1}>\n          <div>\n            {more.length > 0 && more.length === 1\n              ? translation(\"search.result.more.one\")\n              : translation(\"search.result.more.other\", more.length)\n            }\n          </div>\n        </summary>\n        {...more.map(section => renderSearchDocument(section, Flag.TEASER))}\n      </details>\n    ] : []\n  ]\n\n  /* Render search result */\n  return (\n    <li class=\"md-search-result__item\">\n      {children}\n    </li>\n  )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { SourceFacts } from \"~/components\"\nimport { h, round } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render repository facts\n *\n * @param facts - Repository facts\n *\n * @returns Element\n */\nexport function renderSourceFacts(facts: SourceFacts): HTMLElement {\n  return (\n    <ul class=\"md-source__facts\">\n      {Object.entries(facts).map(([key, value]) => (\n        <li class={`md-source__fact md-source__fact--${key}`}>\n          {typeof value === \"number\" ? round(value) : value}\n        </li>\n      ))}\n    </ul>\n  )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Tabbed control type\n */\ntype TabbedControlType =\n  | \"prev\"\n  | \"next\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render control for content tabs\n *\n * @param type - Control type\n *\n * @returns Element\n */\nexport function renderTabbedControl(\n  type: TabbedControlType\n): HTMLElement {\n  const classes = `tabbed-control tabbed-control--${type}`\n  return (\n    <div class={classes} hidden>\n      <button class=\"tabbed-button\" tabIndex={-1} aria-hidden=\"true\"></button>\n    </div>\n  )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render a table inside a wrapper to improve scrolling on mobile\n *\n * @param table - Table element\n *\n * @returns Element\n */\nexport function renderTable(table: HTMLElement): HTMLElement {\n  return (\n    <div class=\"md-typeset__scrollwrap\">\n      <div class=\"md-typeset__table\">\n        {table}\n      </div>\n    </div>\n  )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { configuration, translation } from \"~/_\"\nimport { h } from \"~/utilities\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Version\n */\nexport interface Version {\n  version: string                      /* Version identifier */\n  title: string                        /* Version title */\n  aliases: string[]                    /* Version aliases */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render a version\n *\n * @param version - Version\n *\n * @returns Element\n */\nfunction renderVersion(version: Version): HTMLElement {\n  const config = configuration()\n\n  /* Ensure trailing slash - see https://bit.ly/3rL5u3f */\n  const url = new URL(`../${version.version}/`, config.base)\n  return (\n    <li class=\"md-version__item\">\n      <a href={`${url}`} class=\"md-version__link\">\n        {version.title}\n      </a>\n    </li>\n  )\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Render a version selector\n *\n * @param versions - Versions\n * @param active - Active version\n *\n * @returns Element\n */\nexport function renderVersionSelector(\n  versions: Version[], active: Version\n): HTMLElement {\n  return (\n    <div class=\"md-version\">\n      <button\n        class=\"md-version__current\"\n        aria-label={translation(\"select.version\")}\n      >\n        {active.title}\n      </button>\n      <ul class=\"md-version__list\">\n        {versions.map(renderVersion)}\n      </ul>\n    </div>\n  )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  defer,\n  endWith,\n  finalize,\n  ignoreElements,\n  merge,\n  share,\n  takeUntil\n} from \"rxjs\"\n\nimport {\n  getElement,\n  getElements,\n  getOptionalElement\n} from \"~/browser\"\nimport { renderAnnotation } from \"~/templates\"\n\nimport { Component } from \"../../../_\"\nimport {\n  Annotation,\n  mountAnnotation\n} from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  target$: Observable<HTMLElement>     /* Location target observable */\n  print$: Observable<boolean>          /* Media print observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Find all annotation hosts in the containing element\n *\n * @param container - Containing element\n *\n * @returns Annotation hosts\n */\nfunction findHosts(container: HTMLElement): HTMLElement[] {\n  return container.tagName === \"CODE\"\n    ? getElements(\".c, .c1, .cm\", container)\n    : [container]\n}\n\n/**\n * Find all annotation markers in the containing element\n *\n * @param container - Containing element\n *\n * @returns Annotation markers\n */\nfunction findMarkers(container: HTMLElement): Text[] {\n  const markers: Text[] = []\n  for (const el of findHosts(container)) {\n    const nodes: Text[] = []\n\n    /* Find all text nodes in current element */\n    const it = document.createNodeIterator(el, NodeFilter.SHOW_TEXT)\n    for (let node = it.nextNode(); node; node = it.nextNode())\n      nodes.push(node as Text)\n\n    /* Find all markers in each text node */\n    for (let text of nodes) {\n      let match: RegExpExecArray | null\n\n      /* Split text at marker and add to list */\n      while ((match = /(\\(\\d+\\))(!)?/.exec(text.textContent!))) {\n        const [, id, force] = match\n        if (typeof force === \"undefined\") {\n          const marker = text.splitText(match.index)\n          text = marker.splitText(id.length)\n          markers.push(marker)\n\n        /* Replace entire text with marker */\n        } else {\n          text.textContent = id\n          markers.push(text)\n          break\n        }\n      }\n    }\n  }\n  return markers\n}\n\n/**\n * Swap the child nodes of two elements\n *\n * @param source - Source element\n * @param target - Target element\n */\nfunction swap(source: HTMLElement, target: HTMLElement): void {\n  target.append(...Array.from(source.childNodes))\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount annotation list\n *\n * This function analyzes the containing code block and checks for markers\n * referring to elements in the given annotation list. If no markers are found,\n * the list is left untouched. Otherwise, list elements are rendered as\n * annotations inside the code block.\n *\n * @param el - Annotation list element\n * @param container - Containing element\n * @param options - Options\n *\n * @returns Annotation component observable\n */\nexport function mountAnnotationList(\n  el: HTMLElement, container: HTMLElement, { target$, print$ }: MountOptions\n): Observable<Component<Annotation>> {\n\n  /* Compute prefix for tooltip anchors */\n  const parent = container.closest(\"[id]\")\n  const prefix = parent?.id\n\n  /* Find and replace all markers with empty annotations */\n  const annotations = new Map<string, HTMLElement>()\n  for (const marker of findMarkers(container)) {\n    const [, id] = marker.textContent!.match(/\\((\\d+)\\)/)!\n    if (getOptionalElement(`:scope > li:nth-child(${id})`, el)) {\n      annotations.set(id, renderAnnotation(id, prefix))\n      marker.replaceWith(annotations.get(id)!)\n    }\n  }\n\n  /* Keep list if there are no annotations to render */\n  if (annotations.size === 0)\n    return EMPTY\n\n  /* Mount component on subscription */\n  return defer(() => {\n    const push$ = new Subject()\n    const done$ = push$.pipe(ignoreElements(), endWith(true))\n\n    /* Retrieve container pairs for swapping */\n    const pairs: [HTMLElement, HTMLElement][] = []\n    for (const [id, annotation] of annotations)\n      pairs.push([\n        getElement(\".md-typeset\", annotation),\n        getElement(`:scope > li:nth-child(${id})`, el)\n      ])\n\n    /* Handle print mode - see https://bit.ly/3rgPdpt */\n    print$.pipe(takeUntil(done$))\n      .subscribe(active => {\n        el.hidden = !active\n\n        /* Add class to discern list element */\n        el.classList.toggle(\"md-annotation-list\", active)\n\n        /* Show annotations in code block or list (print) */\n        for (const [inner, child] of pairs)\n          if (!active)\n            swap(child, inner)\n          else\n            swap(inner, child)\n      })\n\n    /* Create and return component */\n    return merge(...[...annotations]\n      .map(([, annotation]) => (\n        mountAnnotation(annotation, container, { target$ })\n      ))\n    )\n      .pipe(\n        finalize(() => push$.complete()),\n        share()\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { EMPTY, Observable, defer } from \"rxjs\"\n\nimport { Component } from \"../../../_\"\nimport { Annotation } from \"../_\"\nimport { mountAnnotationList } from \"../list\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  target$: Observable<HTMLElement>     /* Location target observable */\n  print$: Observable<boolean>          /* Media print observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Find list element directly following a block\n *\n * @param el - Annotation block element\n *\n * @returns List element or nothing\n */\nfunction findList(el: HTMLElement): HTMLElement | undefined {\n  if (el.nextElementSibling) {\n    const sibling = el.nextElementSibling as HTMLElement\n    if (sibling.tagName === \"OL\")\n      return sibling\n\n    /* Skip empty paragraphs - see https://bit.ly/3r4ZJ2O */\n    else if (sibling.tagName === \"P\" && !sibling.children.length)\n      return findList(sibling)\n  }\n\n  /* Everything else */\n  return undefined\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount annotation block\n *\n * @param el - Annotation block element\n * @param options - Options\n *\n * @returns Annotation component observable\n */\nexport function mountAnnotationBlock(\n  el: HTMLElement, options: MountOptions\n): Observable<Component<Annotation>> {\n  return defer(() => {\n    const list = findList(el)\n    return typeof list !== \"undefined\"\n      ? mountAnnotationList(list, el, options)\n      : EMPTY\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport ClipboardJS from \"clipboard\"\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  defer,\n  distinctUntilChanged,\n  distinctUntilKeyChanged,\n  filter,\n  finalize,\n  map,\n  mergeWith,\n  switchMap,\n  take,\n  tap\n} from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport {\n  getElementContentSize,\n  watchElementSize,\n  watchElementVisibility\n} from \"~/browser\"\nimport { renderClipboardButton } from \"~/templates\"\n\nimport { Component } from \"../../../_\"\nimport {\n  Annotation,\n  mountAnnotationList\n} from \"../../annotation\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Code block\n */\nexport interface CodeBlock {\n  scrollable: boolean                  /* Code block overflows */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  target$: Observable<HTMLElement>     /* Location target observable */\n  print$: Observable<boolean>          /* Media print observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Global sequence number for code blocks\n */\nlet sequence = 0\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Find candidate list element directly following a code block\n *\n * @param el - Code block element\n *\n * @returns List element or nothing\n */\nfunction findCandidateList(el: HTMLElement): HTMLElement | undefined {\n  if (el.nextElementSibling) {\n    const sibling = el.nextElementSibling as HTMLElement\n    if (sibling.tagName === \"OL\")\n      return sibling\n\n    /* Skip empty paragraphs - see https://bit.ly/3r4ZJ2O */\n    else if (sibling.tagName === \"P\" && !sibling.children.length)\n      return findCandidateList(sibling)\n  }\n\n  /* Everything else */\n  return undefined\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch code block\n *\n * This function monitors size changes of the viewport, as well as switches of\n * content tabs with embedded code blocks, as both may trigger overflow.\n *\n * @param el - Code block element\n *\n * @returns Code block observable\n */\nexport function watchCodeBlock(\n  el: HTMLElement\n): Observable<CodeBlock> {\n  return watchElementSize(el)\n    .pipe(\n      map(({ width }) => {\n        const content = getElementContentSize(el)\n        return {\n          scrollable: content.width > width\n        }\n      }),\n      distinctUntilKeyChanged(\"scrollable\")\n    )\n}\n\n/**\n * Mount code block\n *\n * This function ensures that an overflowing code block is focusable through\n * keyboard, so it can be scrolled without a mouse to improve on accessibility.\n * Furthermore, if code annotations are enabled, they are mounted if and only\n * if the code block is currently visible, e.g., not in a hidden content tab.\n *\n * Note that code blocks may be mounted eagerly or lazily. If they're mounted\n * lazily (on first visibility), code annotation anchor links will not work,\n * as they are evaluated on initial page load, and code annotations in general\n * might feel a little bumpier.\n *\n * @param el - Code block element\n * @param options - Options\n *\n * @returns Code block and annotation component observable\n */\nexport function mountCodeBlock(\n  el: HTMLElement, options: MountOptions\n): Observable<Component<CodeBlock | Annotation>> {\n  const { matches: hover } = matchMedia(\"(hover)\")\n\n  /* Defer mounting of code block - see https://bit.ly/3vHVoVD */\n  const factory$ = defer(() => {\n    const push$ = new Subject<CodeBlock>()\n    push$.subscribe(({ scrollable }) => {\n      if (scrollable && hover)\n        el.setAttribute(\"tabindex\", \"0\")\n      else\n        el.removeAttribute(\"tabindex\")\n    })\n\n    /* Render button for Clipboard.js integration */\n    if (ClipboardJS.isSupported()) {\n      if (el.closest(\".copy\") || (\n        feature(\"content.code.copy\") && !el.closest(\".no-copy\")\n      )) {\n        const parent = el.closest(\"pre\")!\n        parent.id = `__code_${sequence++}`\n        parent.insertBefore(\n          renderClipboardButton(parent.id),\n          el\n        )\n      }\n    }\n\n    /* Handle code annotations */\n    const container = el.closest(\".highlight\")\n    if (container instanceof HTMLElement) {\n      const list = findCandidateList(container)\n\n      /* Mount code annotations, if enabled */\n      if (typeof list !== \"undefined\" && (\n        container.classList.contains(\"annotate\") ||\n        feature(\"content.code.annotate\")\n      )) {\n        const annotations$ = mountAnnotationList(list, el, options)\n\n        /* Create and return component */\n        return watchCodeBlock(el)\n          .pipe(\n            tap(state => push$.next(state)),\n            finalize(() => push$.complete()),\n            map(state => ({ ref: el, ...state })),\n            mergeWith(\n              watchElementSize(container)\n                .pipe(\n                  map(({ width, height }) => width && height),\n                  distinctUntilChanged(),\n                  switchMap(active => active ? annotations$ : EMPTY)\n                )\n            )\n          )\n      }\n    }\n\n    /* Create and return component */\n    return watchCodeBlock(el)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n\n  /* Mount code block lazily */\n  if (feature(\"content.lazy\"))\n    return watchElementVisibility(el)\n      .pipe(\n        filter(visible => visible),\n        take(1),\n        switchMap(() => factory$)\n      )\n\n  /* Mount code block */\n  return factory$\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  defer,\n  filter,\n  finalize,\n  map,\n  merge,\n  tap\n} from \"rxjs\"\n\nimport { Component } from \"../../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Details\n */\nexport interface Details {\n  action: \"open\" | \"close\"             /* Details state */\n  reveal?: boolean                     /* Details is revealed */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  target$: Observable<HTMLElement>     /* Location target observable */\n  print$: Observable<boolean>          /* Media print observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  target$: Observable<HTMLElement>     /* Location target observable */\n  print$: Observable<boolean>          /* Media print observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch details\n *\n * @param el - Details element\n * @param options - Options\n *\n * @returns Details observable\n */\nexport function watchDetails(\n  el: HTMLDetailsElement, { target$, print$ }: WatchOptions\n): Observable<Details> {\n  let open = true\n  return merge(\n\n    /* Open and focus details on location target */\n    target$\n      .pipe(\n        map(target => target.closest(\"details:not([open])\")!),\n        filter(details => el === details),\n        map(() => ({\n          action: \"open\", reveal: true\n        }) as Details)\n      ),\n\n    /* Open details on print and close afterwards */\n    print$\n      .pipe(\n        filter(active => active || !open),\n        tap(() => open = el.open),\n        map(active => ({\n          action: active ? \"open\" : \"close\"\n        }) as Details)\n      )\n  )\n}\n\n/**\n * Mount details\n *\n * This function ensures that `details` tags are opened on anchor jumps and\n * prior to printing, so the whole content of the page is visible.\n *\n * @param el - Details element\n * @param options - Options\n *\n * @returns Details component observable\n */\nexport function mountDetails(\n  el: HTMLDetailsElement, options: MountOptions\n): Observable<Component<Details>> {\n  return defer(() => {\n    const push$ = new Subject<Details>()\n    push$.subscribe(({ action, reveal }) => {\n      el.toggleAttribute(\"open\", action === \"open\")\n      if (reveal)\n        el.scrollIntoView()\n    })\n\n    /* Create and return component */\n    return watchDetails(el, options)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", ".node circle,.node ellipse,.node path,.node polygon,.node rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}marker{fill:var(--md-mermaid-edge-color)!important}.edgeLabel .label rect{fill:#0000}.label{color:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.label foreignObject{line-height:normal;overflow:visible}.label div .edgeLabel{color:var(--md-mermaid-label-fg-color)}.edgeLabel,.edgeLabel rect,.label div .edgeLabel{background-color:var(--md-mermaid-label-bg-color)}.edgeLabel,.edgeLabel rect{fill:var(--md-mermaid-label-bg-color);color:var(--md-mermaid-edge-color)}.edgePath .path,.flowchart-link{stroke:var(--md-mermaid-edge-color);stroke-width:.05rem}.edgePath .arrowheadPath{fill:var(--md-mermaid-edge-color);stroke:none}.cluster rect{fill:var(--md-default-fg-color--lightest);stroke:var(--md-default-fg-color--lighter)}.cluster span{color:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}g #flowchart-circleEnd,g #flowchart-circleStart,g #flowchart-crossEnd,g #flowchart-crossStart,g #flowchart-pointEnd,g #flowchart-pointStart{stroke:none}g.classGroup line,g.classGroup rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}g.classGroup text{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.classLabel .box{fill:var(--md-mermaid-label-bg-color);background-color:var(--md-mermaid-label-bg-color);opacity:1}.classLabel .label{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.node .divider{stroke:var(--md-mermaid-node-fg-color)}.relation{stroke:var(--md-mermaid-edge-color)}.cardinality{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.cardinality text{fill:inherit!important}defs #classDiagram-compositionEnd,defs #classDiagram-compositionStart,defs #classDiagram-dependencyEnd,defs #classDiagram-dependencyStart,defs #classDiagram-extensionEnd,defs #classDiagram-extensionStart{fill:var(--md-mermaid-edge-color)!important;stroke:var(--md-mermaid-edge-color)!important}defs #classDiagram-aggregationEnd,defs #classDiagram-aggregationStart{fill:var(--md-mermaid-label-bg-color)!important;stroke:var(--md-mermaid-edge-color)!important}g.stateGroup rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}g.stateGroup .state-title{fill:var(--md-mermaid-label-fg-color)!important;font-family:var(--md-mermaid-font-family)}g.stateGroup .composit{fill:var(--md-mermaid-label-bg-color)}.nodeLabel{color:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.node circle.state-end,.node circle.state-start,.start-state{fill:var(--md-mermaid-edge-color);stroke:none}.end-state-inner,.end-state-outer{fill:var(--md-mermaid-edge-color)}.end-state-inner,.node circle.state-end{stroke:var(--md-mermaid-label-bg-color)}.transition{stroke:var(--md-mermaid-edge-color)}[id^=state-fork] rect,[id^=state-join] rect{fill:var(--md-mermaid-edge-color)!important;stroke:none!important}.statediagram-cluster.statediagram-cluster .inner{fill:var(--md-default-bg-color)}.statediagram-cluster rect{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}.statediagram-state rect.divider{fill:var(--md-default-fg-color--lightest);stroke:var(--md-default-fg-color--lighter)}defs #statediagram-barbEnd{stroke:var(--md-mermaid-edge-color)}.attributeBoxEven,.attributeBoxOdd{fill:var(--md-mermaid-node-bg-color);stroke:var(--md-mermaid-node-fg-color)}.entityBox{fill:var(--md-mermaid-label-bg-color);stroke:var(--md-mermaid-node-fg-color)}.entityLabel{fill:var(--md-mermaid-label-fg-color);font-family:var(--md-mermaid-font-family)}.relationshipLabelBox{fill:var(--md-mermaid-label-bg-color);fill-opacity:1;background-color:var(--md-mermaid-label-bg-color);opacity:1}.relationshipLabel{fill:var(--md-mermaid-label-fg-color)}.relationshipLine{stroke:var(--md-mermaid-edge-color)}defs #ONE_OR_MORE_END *,defs #ONE_OR_MORE_START *,defs #ONLY_ONE_END *,defs #ONLY_ONE_START *,defs #ZERO_OR_MORE_END *,defs #ZERO_OR_MORE_START *,defs #ZERO_OR_ONE_END *,defs #ZERO_OR_ONE_START *{stroke:var(--md-mermaid-edge-color)!important}defs #ZERO_OR_MORE_END circle,defs #ZERO_OR_MORE_START circle{fill:var(--md-mermaid-label-bg-color)}.actor{fill:var(--md-mermaid-sequence-actor-bg-color);stroke:var(--md-mermaid-sequence-actor-border-color)}text.actor>tspan{fill:var(--md-mermaid-sequence-actor-fg-color);font-family:var(--md-mermaid-font-family)}line{stroke:var(--md-mermaid-sequence-actor-line-color)}.actor-man circle,.actor-man line{fill:var(--md-mermaid-sequence-actorman-bg-color);stroke:var(--md-mermaid-sequence-actorman-line-color)}.messageLine0,.messageLine1{stroke:var(--md-mermaid-sequence-message-line-color)}.note{fill:var(--md-mermaid-sequence-note-bg-color);stroke:var(--md-mermaid-sequence-note-border-color)}.loopText,.loopText>tspan,.messageText,.noteText>tspan{stroke:none;font-family:var(--md-mermaid-font-family)!important}.messageText{fill:var(--md-mermaid-sequence-message-fg-color)}.loopText,.loopText>tspan{fill:var(--md-mermaid-sequence-loop-fg-color)}.noteText>tspan{fill:var(--md-mermaid-sequence-note-fg-color)}#arrowhead path{fill:var(--md-mermaid-sequence-message-line-color);stroke:none}.loopLine{fill:var(--md-mermaid-sequence-loop-bg-color);stroke:var(--md-mermaid-sequence-loop-border-color)}.labelBox{fill:var(--md-mermaid-sequence-label-bg-color);stroke:none}.labelText,.labelText>span{fill:var(--md-mermaid-sequence-label-fg-color);font-family:var(--md-mermaid-font-family)}.sequenceNumber{fill:var(--md-mermaid-sequence-number-fg-color)}rect.rect{fill:var(--md-mermaid-sequence-box-bg-color);stroke:none}rect.rect+text.text{fill:var(--md-mermaid-sequence-box-fg-color)}defs #sequencenumber{fill:var(--md-mermaid-sequence-number-bg-color)!important}", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  map,\n  of,\n  shareReplay,\n  tap\n} from \"rxjs\"\n\nimport { watchScript } from \"~/browser\"\nimport { h } from \"~/utilities\"\n\nimport { Component } from \"../../_\"\n\nimport themeCSS from \"./index.css\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mermaid diagram\n */\nexport interface Mermaid {}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Mermaid instance observable\n */\nlet mermaid$: Observable<void>\n\n/**\n * Global sequence number for diagrams\n */\nlet sequence = 0\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch Mermaid script\n *\n * @returns Mermaid scripts observable\n */\nfunction fetchScripts(): Observable<void> {\n  return typeof mermaid === \"undefined\" || mermaid instanceof Element\n    ? watchScript(\"https://unpkg.com/mermaid@9.4.3/dist/mermaid.min.js\")\n    : of(undefined)\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount Mermaid diagram\n *\n * @param el - Code block element\n *\n * @returns Mermaid diagram component observable\n */\nexport function mountMermaid(\n  el: HTMLElement\n): Observable<Component<Mermaid>> {\n  el.classList.remove(\"mermaid\") // Hack: mitigate https://bit.ly/3CiN6Du\n  mermaid$ ||= fetchScripts()\n    .pipe(\n      tap(() => mermaid.initialize({\n        startOnLoad: false,\n        themeCSS,\n        sequence: {\n          actorFontSize: \"16px\", // Hack: mitigate https://bit.ly/3y0NEi3\n          messageFontSize: \"16px\",\n          noteFontSize: \"16px\"\n        }\n      })),\n      map(() => undefined),\n      shareReplay(1)\n    )\n\n  /* Render diagram */\n  mermaid$.subscribe(() => {\n    el.classList.add(\"mermaid\") // Hack: mitigate https://bit.ly/3CiN6Du\n    const id = `__mermaid_${sequence++}`\n\n    /* Create host element to replace code block */\n    const host = h(\"div\", { class: \"mermaid\" })\n    const text = el.textContent\n\n    /* Render and inject diagram */\n    mermaid.mermaidAPI.render(id, text, (svg: string, fn: Function) => {\n\n      /* Create a shadow root and inject diagram */\n      const shadow = host.attachShadow({ mode: \"closed\" })\n      shadow.innerHTML = svg\n\n      /* Replace code block with diagram and bind functions */\n      el.replaceWith(host)\n      fn?.(shadow)\n    })\n  })\n\n  /* Create and return component */\n  return mermaid$\n    .pipe(\n      map(() => ({ ref: el }))\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { Observable, of } from \"rxjs\"\n\nimport { renderTable } from \"~/templates\"\nimport { h } from \"~/utilities\"\n\nimport { Component } from \"../../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Data table\n */\nexport interface DataTable {}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Sentinel for replacement\n */\nconst sentinel = h(\"table\")\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount data table\n *\n * This function wraps a data table in another scrollable container, so it can\n * be smoothly scrolled on smaller screen sizes and won't break the layout.\n *\n * @param el - Data table element\n *\n * @returns Data table component observable\n */\nexport function mountDataTable(\n  el: HTMLElement\n): Observable<Component<DataTable>> {\n  el.replaceWith(sentinel)\n  sentinel.replaceWith(renderTable(el))\n\n  /* Create and return component */\n  return of({ ref: el })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  animationFrameScheduler,\n  asyncScheduler,\n  auditTime,\n  combineLatest,\n  defer,\n  endWith,\n  finalize,\n  fromEvent,\n  ignoreElements,\n  map,\n  merge,\n  skip,\n  startWith,\n  subscribeOn,\n  takeUntil,\n  tap,\n  withLatestFrom\n} from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport {\n  Viewport,\n  getElement,\n  getElementContentOffset,\n  getElementContentSize,\n  getElementOffset,\n  getElementSize,\n  getElements,\n  watchElementContentOffset,\n  watchElementSize\n} from \"~/browser\"\nimport { renderTabbedControl } from \"~/templates\"\n\nimport { Component } from \"../../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Content tabs\n */\nexport interface ContentTabs {\n  active: HTMLLabelElement             /* Active tab label */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch content tabs\n *\n * @param el - Content tabs element\n *\n * @returns Content tabs observable\n */\nexport function watchContentTabs(\n  el: HTMLElement\n): Observable<ContentTabs> {\n  const inputs = getElements<HTMLInputElement>(\":scope > input\", el)\n  const initial = inputs.find(input => input.checked) || inputs[0]\n  return merge(...inputs.map(input => fromEvent(input, \"change\")\n    .pipe(\n      map(() => getElement<HTMLLabelElement>(`label[for=\"${input.id}\"]`))\n    )\n  ))\n    .pipe(\n      startWith(getElement<HTMLLabelElement>(`label[for=\"${initial.id}\"]`)),\n      map(active => ({ active }))\n    )\n}\n\n/**\n * Mount content tabs\n *\n * This function scrolls the active tab into view. While this functionality is\n * provided by browsers as part of `scrollInfoView`, browsers will always also\n * scroll the vertical axis, which we do not want. Thus, we decided to provide\n * this functionality ourselves.\n *\n * @param el - Content tabs element\n * @param options - Options\n *\n * @returns Content tabs component observable\n */\nexport function mountContentTabs(\n  el: HTMLElement, { viewport$ }: MountOptions\n): Observable<Component<ContentTabs>> {\n\n  /* Render content tab previous button for pagination */\n  const prev = renderTabbedControl(\"prev\")\n  el.append(prev)\n\n  /* Render content tab next button for pagination */\n  const next = renderTabbedControl(\"next\")\n  el.append(next)\n\n  /* Mount component on subscription */\n  const container = getElement(\".tabbed-labels\", el)\n  return defer(() => {\n    const push$ = new Subject<ContentTabs>()\n    const done$ = push$.pipe(ignoreElements(), endWith(true))\n    combineLatest([push$, watchElementSize(el)])\n      .pipe(\n        auditTime(1, animationFrameScheduler),\n        takeUntil(done$)\n      )\n        .subscribe({\n\n          /* Handle emission */\n          next([{ active }, size]) {\n            const offset = getElementOffset(active)\n            const { width } = getElementSize(active)\n\n            /* Set tab indicator offset and width */\n            el.style.setProperty(\"--md-indicator-x\", `${offset.x}px`)\n            el.style.setProperty(\"--md-indicator-width\", `${width}px`)\n\n            /* Scroll container to active content tab */\n            const content = getElementContentOffset(container)\n            if (\n              offset.x         < content.x              ||\n              offset.x + width > content.x + size.width\n            )\n              container.scrollTo({\n                left: Math.max(0, offset.x - 16),\n                behavior: \"smooth\"\n              })\n          },\n\n          /* Handle complete */\n          complete() {\n            el.style.removeProperty(\"--md-indicator-x\")\n            el.style.removeProperty(\"--md-indicator-width\")\n          }\n        })\n\n    /* Hide content tab buttons on borders */\n    combineLatest([\n      watchElementContentOffset(container),\n      watchElementSize(container)\n    ])\n      .pipe(\n        takeUntil(done$)\n      )\n        .subscribe(([offset, size]) => {\n          const content = getElementContentSize(container)\n          prev.hidden = offset.x < 16\n          next.hidden = offset.x > content.width - size.width - 16\n        })\n\n    /* Paginate content tab container on click */\n    merge(\n      fromEvent(prev, \"click\").pipe(map(() => -1)),\n      fromEvent(next, \"click\").pipe(map(() => +1))\n    )\n      .pipe(\n        takeUntil(done$)\n      )\n        .subscribe(direction => {\n          const { width } = getElementSize(container)\n          container.scrollBy({\n            left: width * direction,\n            behavior: \"smooth\"\n          })\n        })\n\n    /* Set up linking of content tabs, if enabled */\n    if (feature(\"content.tabs.link\"))\n      push$.pipe(\n        skip(1),\n        withLatestFrom(viewport$)\n      )\n        .subscribe(([{ active }, { offset }]) => {\n          const tab = active.innerText.trim()\n          if (active.hasAttribute(\"data-md-switching\")) {\n            active.removeAttribute(\"data-md-switching\")\n\n          /* Determine viewport offset of active tab */\n          } else {\n            const y = el.offsetTop - offset.y\n\n            /* Passively activate other tabs */\n            for (const set of getElements(\"[data-tabs]\"))\n              for (const input of getElements<HTMLInputElement>(\n                \":scope > input\", set\n              )) {\n                const label = getElement(`label[for=\"${input.id}\"]`)\n                if (\n                  label !== active &&\n                  label.innerText.trim() === tab\n                ) {\n                  label.setAttribute(\"data-md-switching\", \"\")\n                  input.click()\n                  break\n                }\n              }\n\n            /* Bring active tab into view */\n            window.scrollTo({\n              top: el.offsetTop - y\n            })\n\n            /* Persist active tabs in local storage */\n            const tabs = __md_get<string[]>(\"__tabs\") || []\n            __md_set(\"__tabs\", [...new Set([tab, ...tabs])])\n          }\n        })\n\n    /* Pause media (audio, video) on switch - see https://bit.ly/3Bk6cel */\n    push$.pipe(takeUntil(done$))\n      .subscribe(() => {\n        for (const media of getElements<HTMLAudioElement>(\"audio, video\", el))\n          media.pause()\n      })\n\n    /* Create and return component */\n    return watchContentTabs(el)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n    .pipe(\n      subscribeOn(asyncScheduler)\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { Observable, merge } from \"rxjs\"\n\nimport { Viewport, getElements } from \"~/browser\"\n\nimport { Component } from \"../../_\"\nimport {\n  Annotation,\n  mountAnnotationBlock\n} from \"../annotation\"\nimport {\n  CodeBlock,\n  mountCodeBlock\n} from \"../code\"\nimport {\n  Details,\n  mountDetails\n} from \"../details\"\nimport {\n  Mermaid,\n  mountMermaid\n} from \"../mermaid\"\nimport {\n  DataTable,\n  mountDataTable\n} from \"../table\"\nimport {\n  ContentTabs,\n  mountContentTabs\n} from \"../tabs\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Content\n */\nexport type Content =\n  | Annotation\n  | CodeBlock\n  | ContentTabs\n  | DataTable\n  | Details\n  | Mermaid\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  target$: Observable<HTMLElement>     /* Location target observable */\n  print$: Observable<boolean>          /* Media print observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount content\n *\n * This function mounts all components that are found in the content of the\n * actual article, including code blocks, data tables and details.\n *\n * @param el - Content element\n * @param options - Options\n *\n * @returns Content component observable\n */\nexport function mountContent(\n  el: HTMLElement, { viewport$, target$, print$ }: MountOptions\n): Observable<Component<Content>> {\n  return merge(\n\n    /* Annotations */\n    ...getElements(\".annotate:not(.highlight)\", el)\n      .map(child => mountAnnotationBlock(child, { target$, print$ })),\n\n    /* Code blocks */\n    ...getElements(\"pre:not(.mermaid) > code\", el)\n      .map(child => mountCodeBlock(child, { target$, print$ })),\n\n    /* Mermaid diagrams */\n    ...getElements(\"pre.mermaid\", el)\n      .map(child => mountMermaid(child)),\n\n    /* Data tables */\n    ...getElements(\"table:not([class])\", el)\n      .map(child => mountDataTable(child)),\n\n    /* Details */\n    ...getElements(\"details\", el)\n      .map(child => mountDetails(child, { target$, print$ })),\n\n    /* Content tabs */\n    ...getElements(\"[data-tabs]\", el)\n      .map(child => mountContentTabs(child, { viewport$ }))\n  )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  defer,\n  delay,\n  finalize,\n  map,\n  merge,\n  of,\n  switchMap,\n  tap\n} from \"rxjs\"\n\nimport { getElement } from \"~/browser\"\n\nimport { Component } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Dialog\n */\nexport interface Dialog {\n  message: string                      /* Dialog message */\n  active: boolean                      /* Dialog is active */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  alert$: Subject<string>              /* Alert subject */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  alert$: Subject<string>              /* Alert subject */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch dialog\n *\n * @param _el - Dialog element\n * @param options - Options\n *\n * @returns Dialog observable\n */\nexport function watchDialog(\n  _el: HTMLElement, { alert$ }: WatchOptions\n): Observable<Dialog> {\n  return alert$\n    .pipe(\n      switchMap(message => merge(\n        of(true),\n        of(false).pipe(delay(2000))\n      )\n        .pipe(\n          map(active => ({ message, active }))\n        )\n      )\n    )\n}\n\n/**\n * Mount dialog\n *\n * This function reveals the dialog in the right corner when a new alert is\n * emitted through the subject that is passed as part of the options.\n *\n * @param el - Dialog element\n * @param options - Options\n *\n * @returns Dialog component observable\n */\nexport function mountDialog(\n  el: HTMLElement, options: MountOptions\n): Observable<Component<Dialog>> {\n  const inner = getElement(\".md-typeset\", el)\n  return defer(() => {\n    const push$ = new Subject<Dialog>()\n    push$.subscribe(({ message, active }) => {\n      el.classList.toggle(\"md-dialog--active\", active)\n      inner.textContent = message\n    })\n\n    /* Create and return component */\n    return watchDialog(el, options)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  bufferCount,\n  combineLatest,\n  combineLatestWith,\n  defer,\n  distinctUntilChanged,\n  distinctUntilKeyChanged,\n  endWith,\n  filter,\n  ignoreElements,\n  map,\n  of,\n  shareReplay,\n  startWith,\n  switchMap,\n  takeUntil\n} from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport {\n  Viewport,\n  watchElementSize,\n  watchToggle\n} from \"~/browser\"\n\nimport { Component } from \"../../_\"\nimport { Main } from \"../../main\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Header\n */\nexport interface Header {\n  height: number                       /* Header visible height */\n  hidden: boolean                      /* Header is hidden */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n  main$: Observable<Main>              /* Main area observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Compute whether the header is hidden\n *\n * If the user scrolls past a certain threshold, the header can be hidden when\n * scrolling down, and shown when scrolling up.\n *\n * @param options - Options\n *\n * @returns Toggle observable\n */\nfunction isHidden({ viewport$ }: WatchOptions): Observable<boolean> {\n  if (!feature(\"header.autohide\"))\n    return of(false)\n\n  /* Compute direction and turning point */\n  const direction$ = viewport$\n    .pipe(\n      map(({ offset: { y } }) => y),\n      bufferCount(2, 1),\n      map(([a, b]) => [a < b, b] as const),\n      distinctUntilKeyChanged(0)\n    )\n\n  /* Compute whether header should be hidden */\n  const hidden$ = combineLatest([viewport$, direction$])\n    .pipe(\n      filter(([{ offset }, [, y]]) => Math.abs(y - offset.y) > 100),\n      map(([, [direction]]) => direction),\n      distinctUntilChanged()\n    )\n\n  /* Compute threshold for hiding */\n  const search$ = watchToggle(\"search\")\n  return combineLatest([viewport$, search$])\n    .pipe(\n      map(([{ offset }, search]) => offset.y > 400 && !search),\n      distinctUntilChanged(),\n      switchMap(active => active ? hidden$ : of(false)),\n      startWith(false)\n    )\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch header\n *\n * @param el - Header element\n * @param options - Options\n *\n * @returns Header observable\n */\nexport function watchHeader(\n  el: HTMLElement, options: WatchOptions\n): Observable<Header> {\n  return defer(() => combineLatest([\n    watchElementSize(el),\n    isHidden(options)\n  ]))\n    .pipe(\n      map(([{ height }, hidden]) => ({\n        height,\n        hidden\n      })),\n      distinctUntilChanged((a, b) => (\n        a.height === b.height &&\n        a.hidden === b.hidden\n      )),\n      shareReplay(1)\n    )\n}\n\n/**\n * Mount header\n *\n * This function manages the different states of the header, i.e. whether it's\n * hidden or rendered with a shadow. This depends heavily on the main area.\n *\n * @param el - Header element\n * @param options - Options\n *\n * @returns Header component observable\n */\nexport function mountHeader(\n  el: HTMLElement, { header$, main$ }: MountOptions\n): Observable<Component<Header>> {\n  return defer(() => {\n    const push$ = new Subject<Main>()\n    const done$ = push$.pipe(ignoreElements(), endWith(true))\n    push$\n      .pipe(\n        distinctUntilKeyChanged(\"active\"),\n        combineLatestWith(header$)\n      )\n        .subscribe(([{ active }, { hidden }]) => {\n          el.classList.toggle(\"md-header--shadow\", active && !hidden)\n          el.hidden = hidden\n        })\n\n    /* Link to main area */\n    main$.subscribe(push$)\n\n    /* Create and return component */\n    return header$\n      .pipe(\n        takeUntil(done$),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  defer,\n  distinctUntilKeyChanged,\n  finalize,\n  map,\n  tap\n} from \"rxjs\"\n\nimport {\n  Viewport,\n  getElementSize,\n  getOptionalElement,\n  watchViewportAt\n} from \"~/browser\"\n\nimport { Component } from \"../../_\"\nimport { Header } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Header\n */\nexport interface HeaderTitle {\n  active: boolean                      /* Header title is active */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch header title\n *\n * @param el - Heading element\n * @param options - Options\n *\n * @returns Header title observable\n */\nexport function watchHeaderTitle(\n  el: HTMLElement, { viewport$, header$ }: WatchOptions\n): Observable<HeaderTitle> {\n  return watchViewportAt(el, { viewport$, header$ })\n    .pipe(\n      map(({ offset: { y } }) => {\n        const { height } = getElementSize(el)\n        return {\n          active: y >= height\n        }\n      }),\n      distinctUntilKeyChanged(\"active\")\n    )\n}\n\n/**\n * Mount header title\n *\n * This function swaps the header title from the site title to the title of the\n * current page when the user scrolls past the first headline.\n *\n * @param el - Header title element\n * @param options - Options\n *\n * @returns Header title component observable\n */\nexport function mountHeaderTitle(\n  el: HTMLElement, options: MountOptions\n): Observable<Component<HeaderTitle>> {\n  return defer(() => {\n    const push$ = new Subject<HeaderTitle>()\n    push$.subscribe({\n\n      /* Handle emission */\n      next({ active }) {\n        el.classList.toggle(\"md-header__title--active\", active)\n      },\n\n      /* Handle complete */\n      complete() {\n        el.classList.remove(\"md-header__title--active\")\n      }\n    })\n\n    /* Obtain headline, if any */\n    const heading = getOptionalElement(\".md-content h1\")\n    if (typeof heading === \"undefined\")\n      return EMPTY\n\n    /* Create and return component */\n    return watchHeaderTitle(heading, options)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  combineLatest,\n  distinctUntilChanged,\n  distinctUntilKeyChanged,\n  map,\n  switchMap\n} from \"rxjs\"\n\nimport {\n  Viewport,\n  watchElementSize\n} from \"~/browser\"\n\nimport { Header } from \"../header\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Main area\n */\nexport interface Main {\n  offset: number                       /* Main area top offset */\n  height: number                       /* Main area visible height */\n  active: boolean                      /* Main area is active */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch main area\n *\n * This function returns an observable that computes the visual parameters of\n * the main area which depends on the viewport vertical offset and height, as\n * well as the height of the header element, if the header is fixed.\n *\n * @param el - Main area element\n * @param options - Options\n *\n * @returns Main area observable\n */\nexport function watchMain(\n  el: HTMLElement, { viewport$, header$ }: WatchOptions\n): Observable<Main> {\n\n  /* Compute necessary adjustment for header */\n  const adjust$ = header$\n    .pipe(\n      map(({ height }) => height),\n      distinctUntilChanged()\n    )\n\n  /* Compute the main area's top and bottom borders */\n  const border$ = adjust$\n    .pipe(\n      switchMap(() => watchElementSize(el)\n        .pipe(\n          map(({ height }) => ({\n            top:    el.offsetTop,\n            bottom: el.offsetTop + height\n          })),\n          distinctUntilKeyChanged(\"bottom\")\n        )\n      )\n    )\n\n  /* Compute the main area's offset, visible height and if we scrolled past */\n  return combineLatest([adjust$, border$, viewport$])\n    .pipe(\n      map(([header, { top, bottom }, { offset: { y }, size: { height } }]) => {\n        height = Math.max(0, height\n          - Math.max(0, top    - y,  header)\n          - Math.max(0, height + y - bottom)\n        )\n        return {\n          offset: top - header,\n          height,\n          active: top - header <= y\n        }\n      }),\n      distinctUntilChanged((a, b) => (\n        a.offset === b.offset &&\n        a.height === b.height &&\n        a.active === b.active\n      ))\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  asyncScheduler,\n  defer,\n  finalize,\n  fromEvent,\n  map,\n  mergeMap,\n  observeOn,\n  of,\n  shareReplay,\n  startWith,\n  tap\n} from \"rxjs\"\n\nimport { getElements } from \"~/browser\"\nimport { h } from \"~/utilities\"\n\nimport {\n  Component,\n  getComponentElement\n} from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Palette colors\n */\nexport interface PaletteColor {\n  scheme?: string                      /* Color scheme */\n  primary?: string                     /* Primary color */\n  accent?: string                      /* Accent color */\n}\n\n/**\n * Palette\n */\nexport interface Palette {\n  index: number                        /* Palette index */\n  color: PaletteColor                  /* Palette colors */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch color palette\n *\n * @param inputs - Color palette element\n *\n * @returns Color palette observable\n */\nexport function watchPalette(\n  inputs: HTMLInputElement[]\n): Observable<Palette> {\n  const current = __md_get<Palette>(\"__palette\") || {\n    index: inputs.findIndex(input => matchMedia(\n      input.getAttribute(\"data-md-color-media\")!\n    ).matches)\n  }\n\n  /* Emit changes in color palette */\n  return of(...inputs)\n    .pipe(\n      mergeMap(input => fromEvent(input, \"change\")\n        .pipe(\n          map(() => input)\n        )\n      ),\n      startWith(inputs[Math.max(0, current.index)]),\n      map(input => ({\n        index: inputs.indexOf(input),\n        color: {\n          scheme:  input.getAttribute(\"data-md-color-scheme\"),\n          primary: input.getAttribute(\"data-md-color-primary\"),\n          accent:  input.getAttribute(\"data-md-color-accent\")\n        }\n      } as Palette)),\n      shareReplay(1)\n    )\n}\n\n/**\n * Mount color palette\n *\n * @param el - Color palette element\n *\n * @returns Color palette component observable\n */\nexport function mountPalette(\n  el: HTMLElement\n): Observable<Component<Palette>> {\n  const meta = h(\"meta\", { name: \"theme-color\" })\n  document.head.appendChild(meta)\n\n  // Add color scheme meta tag\n  const scheme = h(\"meta\", { name: \"color-scheme\" })\n  document.head.appendChild(scheme)\n\n  /* Mount component on subscription */\n  return defer(() => {\n    const push$ = new Subject<Palette>()\n    push$.subscribe(palette => {\n      document.body.setAttribute(\"data-md-color-switching\", \"\")\n\n      /* Set color palette */\n      for (const [key, value] of Object.entries(palette.color))\n        document.body.setAttribute(`data-md-color-${key}`, value)\n\n      /* Toggle visibility */\n      for (let index = 0; index < inputs.length; index++) {\n        const label = inputs[index].nextElementSibling\n        if (label instanceof HTMLElement)\n          label.hidden = palette.index !== index\n      }\n\n      /* Persist preference in local storage */\n      __md_set(\"__palette\", palette)\n    })\n\n    /* Update theme-color meta tag */\n    push$\n      .pipe(\n        map(() => {\n          const header = getComponentElement(\"header\")\n          const style  = window.getComputedStyle(header)\n\n          // Set color scheme\n          scheme.content = style.colorScheme\n\n          /* Return color in hexadecimal format */\n          return style.backgroundColor.match(/\\d+/g)!\n            .map(value => (+value).toString(16).padStart(2, \"0\"))\n            .join(\"\")\n        })\n      )\n        .subscribe(color => meta.content = `#${color}`)\n\n    /* Revert transition durations after color switch */\n    push$.pipe(observeOn(asyncScheduler))\n      .subscribe(() => {\n        document.body.removeAttribute(\"data-md-color-switching\")\n      })\n\n    /* Create and return component */\n    const inputs = getElements<HTMLInputElement>(\"input\", el)\n    return watchPalette(inputs)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  defer,\n  finalize,\n  map,\n  tap\n} from \"rxjs\"\n\nimport { Component } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Progress indicator\n */\nexport interface Progress {\n  value: number                        // Progress value\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  progress$: Subject<number>           // Progress subject\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount progress indicator\n *\n * @param el - Progress indicator element\n * @param options - Options\n *\n * @returns Progress indicator component observable\n */\nexport function mountProgress(\n  el: HTMLElement, { progress$ }: MountOptions\n): Observable<Component<Progress>> {\n\n  // Mount component on subscription\n  return defer(() => {\n    const push$ = new Subject<Progress>()\n    push$.subscribe(({ value }) => {\n      el.style.setProperty(\"--md-progress-value\", `${value}`)\n    })\n\n    // Create and return component\n    return progress$\n      .pipe(\n        tap(value => push$.next({ value })),\n        finalize(() => push$.complete()),\n        map(value => ({ ref: el, value }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport ClipboardJS from \"clipboard\"\nimport {\n  Observable,\n  Subject,\n  map,\n  tap\n} from \"rxjs\"\n\nimport { translation } from \"~/_\"\nimport { getElement } from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Setup options\n */\ninterface SetupOptions {\n  alert$: Subject<string>              /* Alert subject */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Extract text to copy\n *\n * @param el - HTML element\n *\n * @returns Extracted text\n */\nfunction extract(el: HTMLElement): string {\n  el.setAttribute(\"data-md-copying\", \"\")\n  const copy = el.closest(\"[data-copy]\")\n  const text = copy\n    ? copy.getAttribute(\"data-copy\")!\n    : el.innerText\n  el.removeAttribute(\"data-md-copying\")\n  return text.trimEnd()\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Set up Clipboard.js integration\n *\n * @param options - Options\n */\nexport function setupClipboardJS(\n  { alert$ }: SetupOptions\n): void {\n  if (ClipboardJS.isSupported()) {\n    new Observable<ClipboardJS.Event>(subscriber => {\n      new ClipboardJS(\"[data-clipboard-target], [data-clipboard-text]\", {\n        text: el => (\n          el.getAttribute(\"data-clipboard-text\")! ||\n          extract(getElement(\n            el.getAttribute(\"data-clipboard-target\")!\n          ))\n        )\n      })\n        .on(\"success\", ev => subscriber.next(ev))\n    })\n      .pipe(\n        tap(ev => {\n          const trigger = ev.trigger as HTMLElement\n          trigger.focus()\n        }),\n        map(() => translation(\"clipboard.copied\"))\n      )\n        .subscribe(alert$)\n  }\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  catchError,\n  defaultIfEmpty,\n  map,\n  of,\n  tap\n} from \"rxjs\"\n\nimport { configuration } from \"~/_\"\nimport { getElements, requestXML } from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Sitemap, i.e. a list of URLs\n */\nexport type Sitemap = string[]\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Preprocess a list of URLs\n *\n * This function replaces the `site_url` in the sitemap with the actual base\n * URL, to allow instant navigation to work in occasions like Netlify previews.\n *\n * @param urls - URLs\n *\n * @returns URL path parts\n */\nfunction preprocess(urls: Sitemap): Sitemap {\n  if (urls.length < 2)\n    return [\"\"]\n\n  /* Take the first two URLs and remove everything after the last slash */\n  const [root, next] = [...urls]\n    .sort((a, b) => a.length - b.length)\n    .map(url => url.replace(/[^/]+$/, \"\"))\n\n  /* Compute common prefix */\n  let index = 0\n  if (root === next)\n    index = root.length\n  else\n    while (root.charCodeAt(index) === next.charCodeAt(index))\n      index++\n\n  /* Remove common prefix and return in original order */\n  return urls.map(url => url.replace(root.slice(0, index), \"\"))\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch the sitemap for the given base URL\n *\n * @param base - Base URL\n *\n * @returns Sitemap observable\n */\nexport function fetchSitemap(base?: URL): Observable<Sitemap> {\n  const cached = __md_get<Sitemap>(\"__sitemap\", sessionStorage, base)\n  if (cached) {\n    return of(cached)\n  } else {\n    const config = configuration()\n    return requestXML(new URL(\"sitemap.xml\", base || config.base))\n      .pipe(\n        map(sitemap => preprocess(getElements(\"loc\", sitemap)\n          .map(node => node.textContent!)\n        )),\n        catchError(() => EMPTY), // @todo refactor instant loading\n        defaultIfEmpty([]),\n        tap(sitemap => __md_set(\"__sitemap\", sitemap, sessionStorage, base))\n      )\n  }\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  bufferCount,\n  catchError,\n  concat,\n  debounceTime,\n  distinctUntilKeyChanged,\n  endWith,\n  filter,\n  fromEvent,\n  ignoreElements,\n  map,\n  of,\n  sample,\n  share,\n  skip,\n  startWith,\n  switchMap,\n  take,\n  withLatestFrom\n} from \"rxjs\"\n\nimport { configuration, feature } from \"~/_\"\nimport {\n  Viewport,\n  getElements,\n  getLocation,\n  getOptionalElement,\n  request,\n  setLocation,\n  setLocationHash\n} from \"~/browser\"\nimport { getComponentElement } from \"~/components\"\n\nimport { fetchSitemap } from \"../sitemap\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Setup options\n */\ninterface SetupOptions {\n  location$: Subject<URL>              // Location subject\n  viewport$: Observable<Viewport>      // Viewport observable\n  progress$: Subject<number>           // Progress suject\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Create a map of head elements for lookup and replacement\n *\n * @param head - Document head\n *\n * @returns Element map\n */\nfunction lookup(head: HTMLHeadElement): Map<string, HTMLElement> {\n\n  // @todo When resolving URLs, we must make sure to use the correct base for\n  // resolution. The next time we refactor instant loading, we should use the\n  // location subject as a source, which is also used for anchor links tracking,\n  // but for now we just rely on canonical.\n  const canonical = getOptionalElement<HTMLLinkElement>(\"[rel=canonical]\", head)\n  if (typeof canonical !== \"undefined\")\n    canonical.href = canonical.href.replace(\"//localhost:\", \"//127.0.0.1:\")\n\n  // Create tag map and index elements in head\n  const tags = new Map<string, HTMLElement>()\n  for (const el of getElements(\":scope > *\", head)) {\n    let html = el.outerHTML\n\n    // If the current element is a style sheet or script, we must resolve the\n    // URL relative to the current location and make it absolute, so it's easy\n    // to deduplicate it later on by comparing the outer HTML of tags. We must\n    // keep identical style sheets and scripts without replacing them.\n    for (const key of [\"href\", \"src\"]) {\n      const value = el.getAttribute(key)!\n      if (value === null)\n        continue\n\n      // Resolve URL relative to current location\n      const url = new URL(value, canonical?.href)\n      const ref = el.cloneNode() as HTMLElement\n\n      // Set resolved URL and retrieve HTML for deduplication\n      ref.setAttribute(key, `${url}`)\n      html = ref.outerHTML\n      break\n    }\n\n    // Index element in tag map\n    tags.set(html, el)\n  }\n\n  // Return tag map\n  return tags\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Set up instant navigation\n *\n * This is a heavily orchestrated operation - see inline comments to learn how\n * this works with Material for MkDocs, and how you can hook into it.\n *\n * @param options - Options\n *\n * @returns Document observable\n */\nexport function setupInstantNavigation(\n  { location$, viewport$, progress$ }: SetupOptions\n): Observable<Document> {\n  const config = configuration()\n  if (location.protocol === \"file:\")\n    return EMPTY\n\n  // Load sitemap immediately, so we have it available when the user initiates\n  // the first instant navigation request, and canonicalize URLs to the current\n  // base URL. The base URL will remain stable in between loads, as it's only\n  // read at the first initialization of the application.\n  const sitemap$ = fetchSitemap()\n    .pipe(\n      map(paths => paths.map(path => `${new URL(path, config.base)}`))\n    )\n\n  // Intercept inter-site navigation - to keep the number of event listeners\n  // low we use the fact that uncaptured events bubble up to the body. This also\n  // has the nice property that we don't need to detach and then again attach\n  // event listeners when instant navigation occurs.\n  const instant$ = fromEvent<MouseEvent>(document.body, \"click\")\n    .pipe(\n      withLatestFrom(sitemap$),\n      switchMap(([ev, sitemap]) => {\n        if (!(ev.target instanceof Element))\n          return EMPTY\n\n        // Skip, as target is not within a link - clicks on non-link elements\n        // are also captured, which we need to exclude from processing\n        const el = ev.target.closest(\"a\")\n        if (el === null)\n          return EMPTY\n\n        // Skip, as link opens in new window - we now know we have captured a\n        // click on a link, but the link either has a `target` property defined,\n        // or the user pressed the `meta` or `ctrl` key to open it in a new\n        // window. Thus, we need to filter those events, too.\n        if (el.target || ev.metaKey || ev.ctrlKey)\n          return EMPTY\n\n        // Next, we must check if the URL is relevant for us, i.e., if it's an\n        // internal link to a page that is managed by MkDocs. Only then we can\n        // be sure that the structure of the page to be loaded adheres to the\n        // current document structure and can subsequently be injected into it\n        // without doing a full reload. For this reason, we must canonicalize\n        // the URL by removing all search parameters and hash fragments.\n        const url = new URL(el.href)\n        url.search = url.hash = \"\"\n\n        // Skip, if URL is not included in the sitemap - this could be the case\n        // when linking between versions or languages, or to another page that\n        // the author included as part of the build, but that is not managed by\n        // MkDocs. In that case we must not continue with instant navigation.\n        if (!sitemap.includes(`${url}`))\n          return EMPTY\n\n        // We now know that we have a link to an internal page, so we prevent\n        // the browser from navigation and emit the URL for instant navigation.\n        // Note that this also includes anchor links, which means we need to\n        // implement anchor positioning ourselves. The reason for this is that\n        // if we wouldn't manage anchor links as well, scroll restoration will\n        // not work correctly (e.g. following an anchor link and scrolling).\n        ev.preventDefault()\n        return of(new URL(el.href))\n      }),\n      share()\n    )\n\n  // Before fetching for the first time, resolve the absolute favicon position,\n  // as the browser will try to fetch the icon immediately\n  instant$.pipe(take(1))\n    .subscribe(() => {\n      const favicon = getOptionalElement<HTMLLinkElement>(\"link[rel=icon]\")\n      if (typeof favicon !== \"undefined\")\n        favicon.href = favicon.href\n    })\n\n  // Enable scroll restoration before window unloads - this is essential to\n  // ensure that full reloads (F5) restore the viewport offset correctly. If\n  // only popstate events wouldn't reset the scroll position prior to their\n  // emission, we could just reset this in popstate. Meh.\n  fromEvent(window, \"beforeunload\")\n    .subscribe(() => {\n      history.scrollRestoration = \"auto\"\n    })\n\n  // When an instant navigation event occurs, disable scroll restoration, since\n  // we must normalize and synchronize the behavior across all browsers. For\n  // instance, when the user clicks the back or forward button, the browser\n  // would immediately jump to the position of the previous document.\n  instant$.pipe(withLatestFrom(viewport$))\n    .subscribe(([url, { offset }]) => {\n      history.scrollRestoration = \"manual\"\n\n      // While it would be better UX to defer the history state change until the\n      // document was fully fetched and parsed, we must schedule it here, since\n      // popstate events are emitted when history state changes happen. Moreover\n      // we need to back up the current viewport offset, so we can restore it\n      // when popstate events occur, e.g., when the browser's back and forward\n      // buttons are used for navigation.\n      history.replaceState(offset, \"\")\n      history.pushState(null, \"\", url)\n    })\n\n  // Emit URL that should be fetched via instant navigation on location subject,\n  // which was passed into this function. Instant navigation can be intercepted\n  // by other parts of the application, which can synchronously back up or\n  // restore state before instant navigation happens.\n  instant$.subscribe(location$)\n\n  // Fetch document - when fetching, we could use `responseType: document`, but\n  // since all MkDocs links are relative, we need to make sure that the current\n  // location matches the document we just loaded. Otherwise any relative links\n  // in the document might use the old location. If the request fails for some\n  // reason, we fall back to regular navigation and set the location explicitly,\n  // which will force-load the page. Furthermore, we must pre-warm the buffer\n  // for the duplicate check, or the first click on an anchor link will also\n  // trigger an instant navigation event, which doesn't make sense.\n  const response$ = location$\n    .pipe(\n      startWith(getLocation()),\n      distinctUntilKeyChanged(\"pathname\"),\n      skip(1),\n      switchMap(url => request(url, { progress$ })\n        .pipe(\n          catchError(() => {\n            setLocation(url, true)\n            return EMPTY\n          })\n        )\n      )\n    )\n\n  // Initialize the DOM parser, parse the returned HTML, and replace selected\n  // components before handing control down to the application\n  const dom = new DOMParser()\n  const document$ = response$\n    .pipe(\n      switchMap(res => res.text()),\n      switchMap(res => {\n        const next = dom.parseFromString(res, \"text/html\")\n        for (const selector of [\n          \"[data-md-component=announce]\",\n          \"[data-md-component=container]\",\n          \"[data-md-component=header-topic]\",\n          \"[data-md-component=outdated]\",\n          \"[data-md-component=logo]\",\n          \"[data-md-component=skip]\",\n          ...feature(\"navigation.tabs.sticky\")\n            ? [\"[data-md-component=tabs]\"]\n            : []\n        ]) {\n          const source = getOptionalElement(selector)\n          const target = getOptionalElement(selector, next)\n          if (\n            typeof source !== \"undefined\" &&\n            typeof target !== \"undefined\"\n          ) {\n            source.replaceWith(target)\n          }\n        }\n\n        // Update meta tags\n        const source = lookup(document.head)\n        const target = lookup(next.head)\n        for (const [html, el] of target) {\n\n          // Hack: skip stylesheets and scripts until we manage to replace them\n          // entirely in order to omit flashes of white content @todo refactor\n          if (\n            el.getAttribute(\"rel\") === \"stylesheet\" ||\n            el.hasAttribute(\"src\")\n          )\n            continue\n\n          if (source.has(html)) {\n            source.delete(html)\n          } else {\n            document.head.appendChild(el)\n          }\n        }\n\n        // Remove meta tags that are not present in the new document\n        for (const el of source.values())\n\n          // Hack: skip stylesheets and scripts until we manage to replace them\n          // entirely in order to omit flashes of white content @todo refactor\n          if (\n            el.getAttribute(\"rel\") === \"stylesheet\" ||\n            el.hasAttribute(\"src\")\n          )\n            continue\n          else\n            el.remove()\n\n        // After components and meta tags were replaced, re-evaluate scripts\n        // that were provided by the author as part of Markdown files\n        const container = getComponentElement(\"container\")\n        return concat(getElements(\"script\", container))\n          .pipe(\n            switchMap(el => {\n              const script = next.createElement(\"script\")\n              if (el.src) {\n                for (const name of el.getAttributeNames())\n                  script.setAttribute(name, el.getAttribute(name)!)\n                el.replaceWith(script)\n\n                // Complete when script is loaded\n                return new Observable(observer => {\n                  script.onload = () => observer.complete()\n                })\n\n              // Complete immediately\n              } else {\n                script.textContent = el.textContent\n                el.replaceWith(script)\n                return EMPTY\n              }\n            }),\n            ignoreElements(),\n            endWith(next)\n          )\n      }),\n      share()\n    )\n\n  // Intercept popstate events, e.g. when using the browser's back and forward\n  // buttons, and emit new location for fetching and parsing\n  const popstate$ = fromEvent<PopStateEvent>(window, \"popstate\")\n  popstate$.pipe(map(getLocation))\n    .subscribe(location$)\n\n  // Intercept clicks on anchor links, and scroll document into position - as\n  // we disabled scroll restoration, we need to do this manually here\n  location$\n    .pipe(\n      startWith(getLocation()),\n      bufferCount(2, 1),\n      filter(([prev, next]) => (\n        prev.pathname === next.pathname &&\n        prev.hash     !== next.hash\n      )),\n      map(([, next]) => next)\n    )\n      .subscribe(url => {\n        if (history.state !== null || !url.hash) {\n          window.scrollTo(0, history.state?.y ?? 0)\n        } else {\n          history.scrollRestoration = \"auto\"\n          setLocationHash(url.hash)\n          history.scrollRestoration = \"manual\"\n        }\n      })\n\n  // Intercept clicks on the same anchor link - we must use a distinct pipeline\n  // for this, or we'd end up in a loop, setting the hash again and again\n  location$\n    .pipe(\n      sample(instant$),\n      startWith(getLocation()),\n      bufferCount(2, 1),\n      filter(([prev, next]) => (\n        prev.pathname === next.pathname &&\n        prev.hash     === next.hash\n      )),\n      map(([, next]) => next)\n    )\n      .subscribe(url => {\n        history.scrollRestoration = \"auto\"\n        setLocationHash(url.hash)\n        history.scrollRestoration = \"manual\"\n\n        // Hack: we need to make sure that we don't end up with multiple history\n        // entries for the same anchor link, so we just remove the last entry\n        history.back()\n      })\n\n  // After parsing the document, check if the current history entry has a state.\n  // This may happen when users press the back or forward button to visit a page\n  // that was already seen. If there's no state, it means a new page was visited\n  // and we should scroll to the top, unless an anchor is given.\n  document$.pipe(withLatestFrom(location$))\n    .subscribe(([, url]) => {\n      if (history.state !== null || !url.hash) {\n        window.scrollTo(0, history.state?.y ?? 0)\n      } else {\n        setLocationHash(url.hash)\n      }\n    })\n\n  // If the current history is not empty, register an event listener updating\n  // the current history state whenever the scroll position changes. This must\n  // be debounced and cannot be done in popstate, as popstate has already\n  // removed the entry from the history.\n  viewport$\n    .pipe(\n      distinctUntilKeyChanged(\"offset\"),\n      debounceTime(100)\n    )\n      .subscribe(({ offset }) => {\n        history.replaceState(offset, \"\")\n      })\n\n  // Return document\n  return document$\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport escapeHTML from \"escape-html\"\n\nimport { SearchConfig } from \"../config\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search highlight function\n *\n * @param value - Value\n *\n * @returns Highlighted value\n */\nexport type SearchHighlightFn = (value: string) => string\n\n/**\n * Search highlight factory function\n *\n * @param query - Query value\n *\n * @returns Search highlight function\n */\nexport type SearchHighlightFactoryFn = (query: string) => SearchHighlightFn\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Create a search highlighter\n *\n * @param config - Search configuration\n *\n * @returns Search highlight factory function\n */\nexport function setupSearchHighlighter(\n  config: SearchConfig\n): SearchHighlightFactoryFn {\n  // Hack: temporarily remove pure lookaheads and lookbehinds\n  const regex = config.separator.split(\"|\").map(term => {\n    const temp = term.replace(/(\\(\\?[!=<][^)]+\\))/g, \"\")\n    return temp.length === 0 ? \"\uFFFD\" : term\n  })\n    .join(\"|\")\n\n  const separator = new RegExp(regex, \"img\")\n  const highlight = (_: unknown, data: string, term: string) => {\n    return `${data}<mark data-md-highlight>${term}</mark>`\n  }\n\n  /* Return factory function */\n  return (query: string) => {\n    query = query\n      .replace(/[\\s*+\\-:~^]+/g, \" \")\n      .trim()\n\n    /* Create search term match expression */\n    const match = new RegExp(`(^|${config.separator}|)(${\n      query\n        .replace(/[|\\\\{}()[\\]^$+*?.-]/g, \"\\\\$&\")\n        .replace(separator, \"|\")\n    })`, \"img\")\n\n    /* Highlight string value */\n    return value => escapeHTML(value)\n      .replace(match, highlight)\n      .replace(/<\\/mark>(\\s+)<mark[^>]*>/img, \"$1\")\n  }\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A RTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { SearchResult } from \"../../_\"\nimport { SearchIndex } from \"../../config\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search message type\n */\nexport const enum SearchMessageType {\n  SETUP,                               /* Search index setup */\n  READY,                               /* Search index ready */\n  QUERY,                               /* Search query */\n  RESULT                               /* Search results */\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Message containing the data necessary to setup the search index\n */\nexport interface SearchSetupMessage {\n  type: SearchMessageType.SETUP        /* Message type */\n  data: SearchIndex                    /* Message data */\n}\n\n/**\n * Message indicating the search index is ready\n */\nexport interface SearchReadyMessage {\n  type: SearchMessageType.READY        /* Message type */\n}\n\n/**\n * Message containing a search query\n */\nexport interface SearchQueryMessage {\n  type: SearchMessageType.QUERY        /* Message type */\n  data: string                         /* Message data */\n}\n\n/**\n * Message containing results for a search query\n */\nexport interface SearchResultMessage {\n  type: SearchMessageType.RESULT       /* Message type */\n  data: SearchResult                   /* Message data */\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Message exchanged with the search worker\n */\nexport type SearchMessage =\n  | SearchSetupMessage\n  | SearchReadyMessage\n  | SearchQueryMessage\n  | SearchResultMessage\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Type guard for search ready messages\n *\n * @param message - Search worker message\n *\n * @returns Test result\n */\nexport function isSearchReadyMessage(\n  message: SearchMessage\n): message is SearchReadyMessage {\n  return message.type === SearchMessageType.READY\n}\n\n/**\n * Type guard for search result messages\n *\n * @param message - Search worker message\n *\n * @returns Test result\n */\nexport function isSearchResultMessage(\n  message: SearchMessage\n): message is SearchResultMessage {\n  return message.type === SearchMessageType.RESULT\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A RTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  ObservableInput,\n  Subject,\n  first,\n  merge,\n  of,\n  switchMap\n} from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport { watchToggle, watchWorker } from \"~/browser\"\n\nimport { SearchIndex } from \"../../config\"\nimport {\n  SearchMessage,\n  SearchMessageType\n} from \"../message\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Set up search worker\n *\n * This function creates and initializes a web worker that is used for search,\n * so that the user interface doesn't freeze. In general, the application does\n * not care how search is implemented, as long as the web worker conforms to\n * the format expected by the application as defined in `SearchMessage`. This\n * allows the author to implement custom search functionality, by providing a\n * custom web worker via configuration.\n *\n * Material for MkDocs' built-in search implementation makes use of Lunr.js, an\n * efficient and fast implementation for client-side search. Leveraging a tiny\n * iframe-based web worker shim, search is even supported for the `file://`\n * protocol, enabling search for local non-hosted builds.\n *\n * If the protocol is `file://`, search initialization is deferred to mitigate\n * freezing, as it's now synchronous by design - see https://bit.ly/3C521EO\n *\n * @see https://bit.ly/3igvtQv - How to implement custom search\n *\n * @param url - Worker URL\n * @param index$ - Search index observable input\n *\n * @returns Search worker\n */\nexport function setupSearchWorker(\n  url: string, index$: ObservableInput<SearchIndex>\n): Subject<SearchMessage> {\n  const worker$ = watchWorker<SearchMessage>(url)\n  merge(\n    of(location.protocol !== \"file:\"),\n    watchToggle(\"search\")\n  )\n    .pipe(\n      first(active => active),\n      switchMap(() => index$)\n    )\n      .subscribe(({ config, docs }) => worker$.next({\n        type: SearchMessageType.SETUP,\n        data: {\n          config,\n          docs,\n          options: {\n            suggest: feature(\"search.suggest\")\n          }\n        }\n      }))\n\n  /* Return search worker */\n  return worker$\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Subject,\n  catchError,\n  combineLatest,\n  filter,\n  fromEvent,\n  map,\n  of,\n  switchMap,\n  withLatestFrom\n} from \"rxjs\"\n\nimport { configuration } from \"~/_\"\nimport {\n  getElement,\n  getLocation,\n  requestJSON,\n  setLocation\n} from \"~/browser\"\nimport { getComponentElements } from \"~/components\"\nimport {\n  Version,\n  renderVersionSelector\n} from \"~/templates\"\n\nimport { fetchSitemap } from \"../sitemap\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Setup options\n */\ninterface SetupOptions {\n  document$: Subject<Document>         /* Document subject */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Set up version selector\n *\n * @param options - Options\n */\nexport function setupVersionSelector(\n  { document$ }: SetupOptions\n): void {\n  const config = configuration()\n  const versions$ = requestJSON<Version[]>(\n    new URL(\"../versions.json\", config.base)\n  )\n    .pipe(\n      catchError(() => EMPTY) // @todo refactor instant loading\n    )\n\n  /* Determine current version */\n  const current$ = versions$\n    .pipe(\n      map(versions => {\n        const [, current] = config.base.match(/([^/]+)\\/?$/)!\n        return versions.find(({ version, aliases }) => (\n          version === current || aliases.includes(current)\n        )) || versions[0]\n      })\n    )\n\n  /* Intercept inter-version navigation */\n  versions$\n    .pipe(\n      map(versions => new Map(versions.map(version => [\n        `${new URL(`../${version.version}/`, config.base)}`,\n        version\n      ]))),\n      switchMap(urls => fromEvent<MouseEvent>(document.body, \"click\")\n        .pipe(\n          filter(ev => !ev.metaKey && !ev.ctrlKey),\n          withLatestFrom(current$),\n          switchMap(([ev, current]) => {\n            if (ev.target instanceof Element) {\n              const el = ev.target.closest(\"a\")\n              if (el && !el.target && urls.has(el.href)) {\n                const url = el.href\n                // This is a temporary hack to detect if a version inside the\n                // version selector or on another part of the site was clicked.\n                // If we're inside the version selector, we definitely want to\n                // find the same page, as we might have different deployments\n                // due to aliases. However, if we're outside the version\n                // selector, we must abort here, because we might otherwise\n                // interfere with instant navigation. We need to refactor this\n                // at some point together with instant navigation.\n                //\n                // See https://github.com/squidfunk/mkdocs-material/issues/4012\n                if (!ev.target.closest(\".md-version\")) {\n                  const version = urls.get(url)!\n                  if (version === current)\n                    return EMPTY\n                }\n                ev.preventDefault()\n                return of(url)\n              }\n            }\n            return EMPTY\n          }),\n          switchMap(url => {\n            const { version } = urls.get(url)!\n            return fetchSitemap(new URL(url))\n              .pipe(\n                map(sitemap => {\n                  const location = getLocation()\n                  const path = location.href.replace(config.base, \"\")\n                  return sitemap.includes(path.split(\"#\")[0])\n                    ? new URL(`../${version}/${path}`, config.base)\n                    : new URL(url)\n                })\n              )\n          })\n        )\n      )\n    )\n      .subscribe(url => setLocation(url, true))\n\n  /* Render version selector and warning */\n  combineLatest([versions$, current$])\n    .subscribe(([versions, current]) => {\n      const topic = getElement(\".md-header__topic\")\n      topic.appendChild(renderVersionSelector(versions, current))\n    })\n\n  /* Integrate outdated version banner with instant navigation */\n  document$.pipe(switchMap(() => current$))\n    .subscribe(current => {\n\n      /* Check if version state was already determined */\n      let outdated = __md_get(\"__outdated\", sessionStorage)\n      if (outdated === null) {\n        outdated = true\n\n        /* Obtain and normalize default versions */\n        let ignored = config.version?.default || \"latest\"\n        if (!Array.isArray(ignored))\n          ignored = [ignored]\n\n        /* Check if version is considered a default */\n        main: for (const ignore of ignored)\n          for (const version of current.aliases.concat(current.version))\n            if (new RegExp(ignore, \"i\").test(version)) {\n              outdated = false\n              break main\n            }\n\n        /* Persist version state in session storage */\n        __md_set(\"__outdated\", outdated, sessionStorage)\n      }\n\n      /* Unhide outdated version banner */\n      if (outdated)\n        for (const warning of getComponentElements(\"outdated\"))\n          warning.hidden = false\n    })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  combineLatest,\n  distinctUntilChanged,\n  distinctUntilKeyChanged,\n  endWith,\n  finalize,\n  first,\n  fromEvent,\n  ignoreElements,\n  map,\n  merge,\n  shareReplay,\n  takeUntil,\n  tap\n} from \"rxjs\"\n\nimport {\n  getElement,\n  getLocation,\n  setToggle,\n  watchElementFocus,\n  watchToggle\n} from \"~/browser\"\nimport {\n  SearchMessage,\n  SearchMessageType,\n  isSearchReadyMessage\n} from \"~/integrations\"\n\nimport { Component } from \"../../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search query\n */\nexport interface SearchQuery {\n  value: string                        /* Query value */\n  focus: boolean                       /* Query focus */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  worker$: Subject<SearchMessage>      /* Search worker */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  worker$: Subject<SearchMessage>      /* Search worker */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch search query\n *\n * Note that the focus event which triggers re-reading the current query value\n * is delayed by `1ms` so the input's empty state is allowed to propagate.\n *\n * @param el - Search query element\n * @param options - Options\n *\n * @returns Search query observable\n */\nexport function watchSearchQuery(\n  el: HTMLInputElement, { worker$ }: WatchOptions\n): Observable<SearchQuery> {\n\n  /* Support search deep linking */\n  const { searchParams } = getLocation()\n  if (searchParams.has(\"q\")) {\n    setToggle(\"search\", true)\n\n    /* Set query from parameter */\n    el.value = searchParams.get(\"q\")!\n    el.focus()\n\n    /* Remove query parameter on close */\n    watchToggle(\"search\")\n      .pipe(\n        first(active => !active)\n      )\n        .subscribe(() => {\n          const url = getLocation()\n          url.searchParams.delete(\"q\")\n          history.replaceState({}, \"\", `${url}`)\n        })\n  }\n\n  /* Intercept focus and input events */\n  const focus$ = watchElementFocus(el)\n  const value$ = merge(\n    worker$.pipe(first(isSearchReadyMessage)),\n    fromEvent(el, \"keyup\"),\n    focus$\n  )\n    .pipe(\n      map(() => el.value),\n      distinctUntilChanged()\n    )\n\n  /* Combine into single observable */\n  return combineLatest([value$, focus$])\n    .pipe(\n      map(([value, focus]) => ({ value, focus })),\n      shareReplay(1)\n    )\n}\n\n/**\n * Mount search query\n *\n * @param el - Search query element\n * @param options - Options\n *\n * @returns Search query component observable\n */\nexport function mountSearchQuery(\n  el: HTMLInputElement, { worker$ }: MountOptions\n): Observable<Component<SearchQuery, HTMLInputElement>> {\n  const push$ = new Subject<SearchQuery>()\n  const done$ = push$.pipe(ignoreElements(), endWith(true))\n\n  /* Handle value change */\n  combineLatest([\n    worker$.pipe(first(isSearchReadyMessage)),\n    push$\n  ], (_, query) => query)\n    .pipe(\n      distinctUntilKeyChanged(\"value\")\n    )\n      .subscribe(({ value }) => worker$.next({\n        type: SearchMessageType.QUERY,\n        data: value\n      }))\n\n  /* Handle focus change */\n  push$\n    .pipe(\n      distinctUntilKeyChanged(\"focus\")\n    )\n      .subscribe(({ focus }) => {\n        if (focus)\n          setToggle(\"search\", focus)\n      })\n\n  /* Handle reset */\n  fromEvent(el.form!, \"reset\")\n    .pipe(\n      takeUntil(done$)\n    )\n      .subscribe(() => el.focus())\n\n  // Focus search query on label click - note that this is necessary to bring\n  // up the keyboard on iOS and other mobile platforms, as the search dialog is\n  // not visible at first, and programatically focusing an input element must\n  // be triggered by a user interaction - see https://t.ly/Cb30n\n  const label = getElement(\"header [for=__search]\")\n  fromEvent(label, \"click\")\n    .subscribe(() => el.focus())\n\n  /* Create and return component */\n  return watchSearchQuery(el, { worker$ })\n    .pipe(\n      tap(state => push$.next(state)),\n      finalize(() => push$.complete()),\n      map(state => ({ ref: el, ...state })),\n      shareReplay(1)\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  bufferCount,\n  filter,\n  finalize,\n  first,\n  fromEvent,\n  map,\n  merge,\n  mergeMap,\n  of,\n  share,\n  skipUntil,\n  switchMap,\n  takeUntil,\n  tap,\n  withLatestFrom,\n  zipWith\n} from \"rxjs\"\n\nimport { translation } from \"~/_\"\nimport {\n  getElement,\n  getOptionalElement,\n  watchElementBoundary,\n  watchToggle\n} from \"~/browser\"\nimport {\n  SearchMessage,\n  SearchResult,\n  isSearchReadyMessage,\n  isSearchResultMessage\n} from \"~/integrations\"\nimport { renderSearchResultItem } from \"~/templates\"\nimport { round } from \"~/utilities\"\n\nimport { Component } from \"../../_\"\nimport { SearchQuery } from \"../query\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  query$: Observable<SearchQuery>      /* Search query observable */\n  worker$: Subject<SearchMessage>      /* Search worker */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount search result list\n *\n * This function performs a lazy rendering of the search results, depending on\n * the vertical offset of the search result container.\n *\n * @param el - Search result list element\n * @param options - Options\n *\n * @returns Search result list component observable\n */\nexport function mountSearchResult(\n  el: HTMLElement, { worker$, query$ }: MountOptions\n): Observable<Component<SearchResult>> {\n  const push$ = new Subject<SearchResult>()\n  const boundary$ = watchElementBoundary(el.parentElement!)\n    .pipe(\n      filter(Boolean)\n    )\n\n  /* Retrieve container */\n  const container = el.parentElement!\n\n  /* Retrieve nested components */\n  const meta = getElement(\":scope > :first-child\", el)\n  const list = getElement(\":scope > :last-child\", el)\n\n  /* Reveal to accessibility tree \u2013 see https://bit.ly/3iAA7t8 */\n  watchToggle(\"search\")\n    .subscribe(active => list.setAttribute(\n      \"role\", active ? \"list\" : \"presentation\"\n    ))\n\n  /* Update search result metadata */\n  push$\n    .pipe(\n      withLatestFrom(query$),\n      skipUntil(worker$.pipe(first(isSearchReadyMessage)))\n    )\n      .subscribe(([{ items }, { value }]) => {\n        switch (items.length) {\n\n          /* No results */\n          case 0:\n            meta.textContent = value.length\n              ? translation(\"search.result.none\")\n              : translation(\"search.result.placeholder\")\n            break\n\n          /* One result */\n          case 1:\n            meta.textContent = translation(\"search.result.one\")\n            break\n\n          /* Multiple result */\n          default:\n            const count = round(items.length)\n            meta.textContent = translation(\"search.result.other\", count)\n        }\n      })\n\n  /* Render search result item */\n  const render$ = push$\n    .pipe(\n      tap(() => list.innerHTML = \"\"),\n      switchMap(({ items }) => merge(\n        of(...items.slice(0, 10)),\n        of(...items.slice(10))\n          .pipe(\n            bufferCount(4),\n            zipWith(boundary$),\n            switchMap(([chunk]) => chunk)\n          )\n      )),\n      map(renderSearchResultItem),\n      share()\n    )\n\n  /* Update search result list */\n  render$.subscribe(item => list.appendChild(item))\n  render$\n    .pipe(\n      mergeMap(item => {\n        const details = getOptionalElement(\"details\", item)\n        if (typeof details === \"undefined\")\n          return EMPTY\n\n        /* Keep position of details element stable */\n        return fromEvent(details, \"toggle\")\n          .pipe(\n            takeUntil(push$),\n            map(() => details)\n          )\n      })\n    )\n      .subscribe(details => {\n        if (\n          details.open === false &&\n          details.offsetTop <= container.scrollTop\n        )\n          container.scrollTo({ top: details.offsetTop })\n      })\n\n  /* Filter search result message */\n  const result$ = worker$\n    .pipe(\n      filter(isSearchResultMessage),\n      map(({ data }) => data)\n    )\n\n  /* Create and return component */\n  return result$\n    .pipe(\n      tap(state => push$.next(state)),\n      finalize(() => push$.complete()),\n      map(state => ({ ref: el, ...state }))\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  endWith,\n  finalize,\n  fromEvent,\n  ignoreElements,\n  map,\n  takeUntil,\n  tap\n} from \"rxjs\"\n\nimport { getLocation } from \"~/browser\"\n\nimport { Component } from \"../../_\"\nimport { SearchQuery } from \"../query\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search sharing\n */\nexport interface SearchShare {\n  url: URL                             /* Deep link for sharing */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  query$: Observable<SearchQuery>      /* Search query observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  query$: Observable<SearchQuery>      /* Search query observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount search sharing\n *\n * @param _el - Search sharing element\n * @param options - Options\n *\n * @returns Search sharing observable\n */\nexport function watchSearchShare(\n  _el: HTMLElement, { query$ }: WatchOptions\n): Observable<SearchShare> {\n  return query$\n    .pipe(\n      map(({ value }) => {\n        const url = getLocation()\n        url.hash = \"\"\n\n        /* Compute readable query strings */\n        value = value\n          .replace(/\\s+/g, \"+\")        /* Collapse whitespace */\n          .replace(/&/g, \"%26\")        /* Escape '&' character */\n          .replace(/=/g, \"%3D\")        /* Escape '=' character */\n\n        /* Replace query string */\n        url.search = `q=${value}`\n        return { url }\n      })\n    )\n}\n\n/**\n * Mount search sharing\n *\n * @param el - Search sharing element\n * @param options - Options\n *\n * @returns Search sharing component observable\n */\nexport function mountSearchShare(\n  el: HTMLAnchorElement, options: MountOptions\n): Observable<Component<SearchShare>> {\n  const push$ = new Subject<SearchShare>()\n  const done$ = push$.pipe(ignoreElements(), endWith(true))\n  push$.subscribe(({ url }) => {\n    el.setAttribute(\"data-clipboard-text\", el.href)\n    el.href = `${url}`\n  })\n\n  /* Prevent following of link */\n  fromEvent(el, \"click\")\n    .pipe(\n      takeUntil(done$)\n    )\n      .subscribe(ev => ev.preventDefault())\n\n  /* Create and return component */\n  return watchSearchShare(el, options)\n    .pipe(\n      tap(state => push$.next(state)),\n      finalize(() => push$.complete()),\n      map(state => ({ ref: el, ...state }))\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  asyncScheduler,\n  combineLatestWith,\n  distinctUntilChanged,\n  filter,\n  finalize,\n  fromEvent,\n  map,\n  merge,\n  observeOn,\n  tap\n} from \"rxjs\"\n\nimport { Keyboard } from \"~/browser\"\nimport {\n  SearchMessage,\n  SearchResult,\n  isSearchResultMessage\n} from \"~/integrations\"\n\nimport { Component, getComponentElement } from \"../../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search suggestions\n */\nexport interface SearchSuggest {}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  keyboard$: Observable<Keyboard>      /* Keyboard observable */\n  worker$: Subject<SearchMessage>      /* Search worker */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount search suggestions\n *\n * This function will perform a lazy rendering of the search results, depending\n * on the vertical offset of the search result container.\n *\n * @param el - Search result list element\n * @param options - Options\n *\n * @returns Search result list component observable\n */\nexport function mountSearchSuggest(\n  el: HTMLElement, { worker$, keyboard$ }: MountOptions\n): Observable<Component<SearchSuggest>> {\n  const push$ = new Subject<SearchResult>()\n\n  /* Retrieve query component and track all changes */\n  const query  = getComponentElement(\"search-query\")\n  const query$ = merge(\n    fromEvent(query, \"keydown\"),\n    fromEvent(query, \"focus\")\n  )\n    .pipe(\n      observeOn(asyncScheduler),\n      map(() => query.value),\n      distinctUntilChanged(),\n    )\n\n  /* Update search suggestions */\n  push$\n    .pipe(\n      combineLatestWith(query$),\n      map(([{ suggest }, value]) => {\n        const words = value.split(/([\\s-]+)/)\n        if (suggest?.length && words[words.length - 1]) {\n          const last = suggest[suggest.length - 1]\n          if (last.startsWith(words[words.length - 1]))\n            words[words.length - 1] = last\n        } else {\n          words.length = 0\n        }\n        return words\n      })\n    )\n      .subscribe(words => el.innerHTML = words\n        .join(\"\")\n        .replace(/\\s/g, \"&nbsp;\")\n      )\n\n  /* Set up search keyboard handlers */\n  keyboard$\n    .pipe(\n      filter(({ mode }) => mode === \"search\")\n    )\n      .subscribe(key => {\n        switch (key.type) {\n\n          /* Right arrow: accept current suggestion */\n          case \"ArrowRight\":\n            if (\n              el.innerText.length &&\n              query.selectionStart === query.value.length\n            )\n              query.value = el.innerText\n            break\n        }\n      })\n\n  /* Filter search result message */\n  const result$ = worker$\n    .pipe(\n      filter(isSearchResultMessage),\n      map(({ data }) => data)\n    )\n\n  /* Create and return component */\n  return result$\n    .pipe(\n      tap(state => push$.next(state)),\n      finalize(() => push$.complete()),\n      map(() => ({ ref: el }))\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  NEVER,\n  Observable,\n  ObservableInput,\n  filter,\n  fromEvent,\n  merge,\n  mergeWith\n} from \"rxjs\"\n\nimport { configuration } from \"~/_\"\nimport {\n  Keyboard,\n  getActiveElement,\n  getElements,\n  setToggle\n} from \"~/browser\"\nimport {\n  SearchIndex,\n  SearchResult,\n  setupSearchWorker\n} from \"~/integrations\"\n\nimport {\n  Component,\n  getComponentElement,\n  getComponentElements\n} from \"../../_\"\nimport {\n  SearchQuery,\n  mountSearchQuery\n} from \"../query\"\nimport { mountSearchResult } from \"../result\"\nimport {\n  SearchShare,\n  mountSearchShare\n} from \"../share\"\nimport {\n  SearchSuggest,\n  mountSearchSuggest\n} from \"../suggest\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search\n */\nexport type Search =\n  | SearchQuery\n  | SearchResult\n  | SearchShare\n  | SearchSuggest\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  index$: ObservableInput<SearchIndex> /* Search index observable */\n  keyboard$: Observable<Keyboard>      /* Keyboard observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount search\n *\n * This function sets up the search functionality, including the underlying\n * web worker and all keyboard bindings.\n *\n * @param el - Search element\n * @param options - Options\n *\n * @returns Search component observable\n */\nexport function mountSearch(\n  el: HTMLElement, { index$, keyboard$ }: MountOptions\n): Observable<Component<Search>> {\n  const config = configuration()\n  try {\n    const worker$ = setupSearchWorker(config.search, index$)\n\n    /* Retrieve query and result components */\n    const query  = getComponentElement(\"search-query\", el)\n    const result = getComponentElement(\"search-result\", el)\n\n    /* Always close search on result selection */\n    fromEvent<PointerEvent>(el, \"click\")\n      .pipe(\n        filter(({ target }) => (\n          target instanceof Element && !!target.closest(\"a\")\n        ))\n      )\n        .subscribe(() => setToggle(\"search\", false))\n\n    /* Set up search keyboard handlers */\n    keyboard$\n      .pipe(\n        filter(({ mode }) => mode === \"search\")\n      )\n        .subscribe(key => {\n          const active = getActiveElement()\n          switch (key.type) {\n\n            /* Enter: go to first (best) result */\n            case \"Enter\":\n              if (active === query) {\n                const anchors = new Map<HTMLAnchorElement, number>()\n                for (const anchor of getElements<HTMLAnchorElement>(\n                  \":first-child [href]\", result\n                )) {\n                  const article = anchor.firstElementChild!\n                  anchors.set(anchor, parseFloat(\n                    article.getAttribute(\"data-md-score\")!\n                  ))\n                }\n\n                /* Go to result with highest score, if any */\n                if (anchors.size) {\n                  const [[best]] = [...anchors].sort(([, a], [, b]) => b - a)\n                  best.click()\n                }\n\n                /* Otherwise omit form submission */\n                key.claim()\n              }\n              break\n\n            /* Escape or Tab: close search */\n            case \"Escape\":\n            case \"Tab\":\n              setToggle(\"search\", false)\n              query.blur()\n              break\n\n            /* Vertical arrows: select previous or next search result */\n            case \"ArrowUp\":\n            case \"ArrowDown\":\n              if (typeof active === \"undefined\") {\n                query.focus()\n              } else {\n                const els = [query, ...getElements(\n                  \":not(details) > [href], summary, details[open] [href]\",\n                  result\n                )]\n                const i = Math.max(0, (\n                  Math.max(0, els.indexOf(active)) + els.length + (\n                    key.type === \"ArrowUp\" ? -1 : +1\n                  )\n                ) % els.length)\n                els[i].focus()\n              }\n\n              /* Prevent scrolling of page */\n              key.claim()\n              break\n\n            /* All other keys: hand to search query */\n            default:\n              if (query !== getActiveElement())\n                query.focus()\n          }\n        })\n\n    /* Set up global keyboard handlers */\n    keyboard$\n      .pipe(\n        filter(({ mode }) => mode === \"global\")\n      )\n        .subscribe(key => {\n          switch (key.type) {\n\n            /* Open search and select query */\n            case \"f\":\n            case \"s\":\n            case \"/\":\n              query.focus()\n              query.select()\n\n              /* Prevent scrolling of page */\n              key.claim()\n              break\n          }\n        })\n\n    /* Create and return component */\n    const query$ = mountSearchQuery(query, { worker$ })\n    return merge(\n      query$,\n      mountSearchResult(result, { worker$, query$ })\n    )\n      .pipe(\n        mergeWith(\n\n          /* Search sharing */\n          ...getComponentElements(\"search-share\", el)\n            .map(child => mountSearchShare(child, { query$ })),\n\n          /* Search suggestions */\n          ...getComponentElements(\"search-suggest\", el)\n            .map(child => mountSearchSuggest(child, { worker$, keyboard$ }))\n        )\n      )\n\n  /* Gracefully handle broken search */\n  } catch (err) {\n    el.hidden = true\n    return NEVER\n  }\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  ObservableInput,\n  combineLatest,\n  filter,\n  map,\n  startWith\n} from \"rxjs\"\n\nimport { getLocation } from \"~/browser\"\nimport {\n  SearchIndex,\n  setupSearchHighlighter\n} from \"~/integrations\"\nimport { h } from \"~/utilities\"\n\nimport { Component } from \"../../_\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search highlighting\n */\nexport interface SearchHighlight {\n  nodes: Map<ChildNode, string>        /* Map of replacements */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  index$: ObservableInput<SearchIndex> /* Search index observable */\n  location$: Observable<URL>           /* Location observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Mount search highlighting\n *\n * @param el - Content element\n * @param options - Options\n *\n * @returns Search highlighting component observable\n */\nexport function mountSearchHiglight(\n  el: HTMLElement, { index$, location$ }: MountOptions\n): Observable<Component<SearchHighlight>> {\n  return combineLatest([\n    index$,\n    location$\n      .pipe(\n        startWith(getLocation()),\n        filter(url => !!url.searchParams.get(\"h\"))\n      )\n  ])\n    .pipe(\n      map(([index, url]) => setupSearchHighlighter(index.config)(\n        url.searchParams.get(\"h\")!\n      )),\n      map(fn => {\n        const nodes = new Map<ChildNode, string>()\n\n        /* Traverse text nodes and collect matches */\n        const it = document.createNodeIterator(el, NodeFilter.SHOW_TEXT)\n        for (let node = it.nextNode(); node; node = it.nextNode()) {\n          if (node.parentElement?.offsetHeight) {\n            const original = node.textContent!\n            const replaced = fn(original)\n            if (replaced.length > original.length)\n              nodes.set(node as ChildNode, replaced)\n          }\n        }\n\n        /* Replace original nodes with matches */\n        for (const [node, text] of nodes) {\n          const { childNodes } = h(\"span\", null, text)\n          node.replaceWith(...Array.from(childNodes))\n        }\n\n        /* Return component */\n        return { ref: el, nodes }\n      })\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  animationFrameScheduler,\n  asyncScheduler,\n  auditTime,\n  combineLatest,\n  defer,\n  distinctUntilChanged,\n  endWith,\n  finalize,\n  first,\n  from,\n  fromEvent,\n  ignoreElements,\n  map,\n  mergeMap,\n  observeOn,\n  takeUntil,\n  tap,\n  withLatestFrom\n} from \"rxjs\"\n\nimport {\n  Viewport,\n  getElement,\n  getElementOffset,\n  getElementSize,\n  getElements\n} from \"~/browser\"\n\nimport { Component } from \"../_\"\nimport { Header } from \"../header\"\nimport { Main } from \"../main\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Sidebar\n */\nexport interface Sidebar {\n  height: number                       /* Sidebar height */\n  locked: boolean                      /* Sidebar is locked */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  main$: Observable<Main>              /* Main area observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n  main$: Observable<Main>              /* Main area observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch sidebar\n *\n * This function returns an observable that computes the visual parameters of\n * the sidebar which depends on the vertical viewport offset, as well as the\n * height of the main area. When the page is scrolled beyond the header, the\n * sidebar is locked and fills the remaining space.\n *\n * @param el - Sidebar element\n * @param options - Options\n *\n * @returns Sidebar observable\n */\nexport function watchSidebar(\n  el: HTMLElement, { viewport$, main$ }: WatchOptions\n): Observable<Sidebar> {\n  const parent = el.closest<HTMLElement>(\".md-grid\")!\n  const adjust =\n    parent.offsetTop -\n    parent.parentElement!.offsetTop\n\n  /* Compute the sidebar's available height and if it should be locked */\n  return combineLatest([main$, viewport$])\n    .pipe(\n      map(([{ offset, height }, { offset: { y } }]) => {\n        height = height\n          + Math.min(adjust, Math.max(0, y - offset))\n          - adjust\n        return {\n          height,\n          locked: y >= offset + adjust\n        }\n      }),\n      distinctUntilChanged((a, b) => (\n        a.height === b.height &&\n        a.locked === b.locked\n      ))\n    )\n}\n\n/**\n * Mount sidebar\n *\n * This function doesn't set the height of the actual sidebar, but of its first\n * child \u2013 the `.md-sidebar__scrollwrap` element in order to mitigiate jittery\n * sidebars when the footer is scrolled into view. At some point we switched\n * from `absolute` / `fixed` positioning to `sticky` positioning, significantly\n * reducing jitter in some browsers (respectively Firefox and Safari) when\n * scrolling from the top. However, top-aligned sticky positioning means that\n * the sidebar snaps to the bottom when the end of the container is reached.\n * This is what leads to the mentioned jitter, as the sidebar's height may be\n * updated too slowly.\n *\n * This behaviour can be mitigiated by setting the height of the sidebar to `0`\n * while preserving the padding, and the height on its first element.\n *\n * @param el - Sidebar element\n * @param options - Options\n *\n * @returns Sidebar component observable\n */\nexport function mountSidebar(\n  el: HTMLElement, { header$, ...options }: MountOptions\n): Observable<Component<Sidebar>> {\n  const inner = getElement(\".md-sidebar__scrollwrap\", el)\n  const { y } = getElementOffset(inner)\n  return defer(() => {\n    const push$ = new Subject<Sidebar>()\n    const done$ = push$.pipe(ignoreElements(), endWith(true))\n    const next$ = push$\n      .pipe(\n        auditTime(0, animationFrameScheduler)\n      )\n\n    /* Update sidebar height and offset */\n    next$.pipe(withLatestFrom(header$))\n      .subscribe({\n\n        /* Handle emission */\n        next([{ height }, { height: offset }]) {\n          inner.style.height = `${height - 2 * y}px`\n          el.style.top       = `${offset}px`\n        },\n\n        /* Handle complete */\n        complete() {\n          inner.style.height = \"\"\n          el.style.top       = \"\"\n        }\n      })\n\n    /* Bring active item into view on initial load */\n    next$.pipe(first())\n      .subscribe(() => {\n        for (const item of getElements(\".md-nav__link--active[href]\", el)) {\n          if (!item.clientHeight) // skip invisible toc in left sidebar\n            continue\n          const container = item.closest<HTMLElement>(\".md-sidebar__scrollwrap\")!\n          if (typeof container !== \"undefined\") {\n            const offset = item.offsetTop - container.offsetTop\n            const { height } = getElementSize(container)\n            container.scrollTo({\n              top: offset - height / 2\n            })\n          }\n        }\n      })\n\n    /* Handle accessibility for expandable items, see https://bit.ly/3jaod9p */\n    from(getElements<HTMLLabelElement>(\"label[tabindex]\", el))\n      .pipe(\n        mergeMap(label => fromEvent(label, \"click\")\n          .pipe(\n            observeOn(asyncScheduler),\n            map(() => label),\n            takeUntil(done$)\n          )\n        )\n      )\n        .subscribe(label => {\n          const input = getElement<HTMLInputElement>(`[id=\"${label.htmlFor}\"]`)\n          const nav = getElement(`[aria-labelledby=\"${label.id}\"]`)\n          nav.setAttribute(\"aria-expanded\", `${input.checked}`)\n        })\n\n    /* Create and return component */\n    return watchSidebar(el, options)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { Repo, User } from \"github-types\"\nimport {\n  EMPTY,\n  Observable,\n  catchError,\n  defaultIfEmpty,\n  map,\n  zip\n} from \"rxjs\"\n\nimport { requestJSON } from \"~/browser\"\n\nimport { SourceFacts } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * GitHub release (partial)\n */\ninterface Release {\n  tag_name: string                     /* Tag name */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch GitHub repository facts\n *\n * @param user - GitHub user or organization\n * @param repo - GitHub repository\n *\n * @returns Repository facts observable\n */\nexport function fetchSourceFactsFromGitHub(\n  user: string, repo?: string\n): Observable<SourceFacts> {\n  if (typeof repo !== \"undefined\") {\n    const url = `https://api.github.com/repos/${user}/${repo}`\n    return zip(\n\n      /* Fetch version */\n      requestJSON<Release>(`${url}/releases/latest`)\n        .pipe(\n          catchError(() => EMPTY), // @todo refactor instant loading\n          map(release => ({\n            version: release.tag_name\n          })),\n          defaultIfEmpty({})\n        ),\n\n      /* Fetch stars and forks */\n      requestJSON<Repo>(url)\n        .pipe(\n          catchError(() => EMPTY), // @todo refactor instant loading\n          map(info => ({\n            stars: info.stargazers_count,\n            forks: info.forks_count\n          })),\n          defaultIfEmpty({})\n        )\n    )\n      .pipe(\n        map(([release, info]) => ({ ...release, ...info }))\n      )\n\n  /* User or organization */\n  } else {\n    const url = `https://api.github.com/users/${user}`\n    return requestJSON<User>(url)\n      .pipe(\n        map(info => ({\n          repositories: info.public_repos\n        })),\n        defaultIfEmpty({})\n      )\n  }\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { ProjectSchema } from \"gitlab\"\nimport {\n  EMPTY,\n  Observable,\n  catchError,\n  defaultIfEmpty,\n  map\n} from \"rxjs\"\n\nimport { requestJSON } from \"~/browser\"\n\nimport { SourceFacts } from \"../_\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch GitLab repository facts\n *\n * @param base - GitLab base\n * @param project - GitLab project\n *\n * @returns Repository facts observable\n */\nexport function fetchSourceFactsFromGitLab(\n  base: string, project: string\n): Observable<SourceFacts> {\n  const url = `https://${base}/api/v4/projects/${encodeURIComponent(project)}`\n  return requestJSON<ProjectSchema>(url)\n    .pipe(\n      catchError(() => EMPTY), // @todo refactor instant loading\n      map(({ star_count, forks_count }) => ({\n        stars: star_count,\n        forks: forks_count\n      })),\n      defaultIfEmpty({})\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { EMPTY, Observable } from \"rxjs\"\n\nimport { fetchSourceFactsFromGitHub } from \"../github\"\nimport { fetchSourceFactsFromGitLab } from \"../gitlab\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Repository facts for repositories\n */\nexport interface RepositoryFacts {\n  stars?: number                       /* Number of stars */\n  forks?: number                       /* Number of forks */\n  version?: string                     /* Latest version */\n}\n\n/**\n * Repository facts for organizations\n */\nexport interface OrganizationFacts {\n  repositories?: number                /* Number of repositories */\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Repository facts\n */\nexport type SourceFacts =\n  | RepositoryFacts\n  | OrganizationFacts\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch repository facts\n *\n * @param url - Repository URL\n *\n * @returns Repository facts observable\n */\nexport function fetchSourceFacts(\n  url: string\n): Observable<SourceFacts> {\n\n  /* Try to match GitHub repository */\n  let match = url.match(/^.+github\\.com\\/([^/]+)\\/?([^/]+)?/i)\n  if (match) {\n    const [, user, repo] = match\n    return fetchSourceFactsFromGitHub(user, repo)\n  }\n\n  /* Try to match GitLab repository */\n  match = url.match(/^.+?([^/]*gitlab[^/]+)\\/(.+?)\\/?$/i)\n  if (match) {\n    const [, base, slug] = match\n    return fetchSourceFactsFromGitLab(base, slug)\n  }\n\n  /* Fallback */\n  return EMPTY\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  EMPTY,\n  Observable,\n  Subject,\n  catchError,\n  defer,\n  filter,\n  finalize,\n  map,\n  of,\n  shareReplay,\n  tap\n} from \"rxjs\"\n\nimport { getElement } from \"~/browser\"\nimport { ConsentDefaults } from \"~/components/consent\"\nimport { renderSourceFacts } from \"~/templates\"\n\nimport {\n  Component,\n  getComponentElements\n} from \"../../_\"\nimport {\n  SourceFacts,\n  fetchSourceFacts\n} from \"../facts\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Repository information\n */\nexport interface Source {\n  facts: SourceFacts                   /* Repository facts */\n}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Repository information observable\n */\nlet fetch$: Observable<Source>\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch repository information\n *\n * This function tries to read the repository facts from session storage, and\n * if unsuccessful, fetches them from the underlying provider.\n *\n * @param el - Repository information element\n *\n * @returns Repository information observable\n */\nexport function watchSource(\n  el: HTMLAnchorElement\n): Observable<Source> {\n  return fetch$ ||= defer(() => {\n    const cached = __md_get<SourceFacts>(\"__source\", sessionStorage)\n    if (cached) {\n      return of(cached)\n    } else {\n\n      /* Check if consent is configured and was given */\n      const els = getComponentElements(\"consent\")\n      if (els.length) {\n        const consent = __md_get<ConsentDefaults>(\"__consent\")\n        if (!(consent && consent.github))\n          return EMPTY\n      }\n\n      /* Fetch repository facts */\n      return fetchSourceFacts(el.href)\n        .pipe(\n          tap(facts => __md_set(\"__source\", facts, sessionStorage))\n        )\n    }\n  })\n    .pipe(\n      catchError(() => EMPTY),\n      filter(facts => Object.keys(facts).length > 0),\n      map(facts => ({ facts })),\n      shareReplay(1)\n    )\n}\n\n/**\n * Mount repository information\n *\n * @param el - Repository information element\n *\n * @returns Repository information component observable\n */\nexport function mountSource(\n  el: HTMLAnchorElement\n): Observable<Component<Source>> {\n  const inner = getElement(\":scope > :last-child\", el)\n  return defer(() => {\n    const push$ = new Subject<Source>()\n    push$.subscribe(({ facts }) => {\n      inner.appendChild(renderSourceFacts(facts))\n      inner.classList.add(\"md-source__repository--active\")\n    })\n\n    /* Create and return component */\n    return watchSource(el)\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  defer,\n  distinctUntilKeyChanged,\n  finalize,\n  map,\n  of,\n  switchMap,\n  tap\n} from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport {\n  Viewport,\n  watchElementSize,\n  watchViewportAt\n} from \"~/browser\"\n\nimport { Component } from \"../_\"\nimport { Header } from \"../header\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Navigation tabs\n */\nexport interface Tabs {\n  hidden: boolean                      /* Navigation tabs are hidden */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch navigation tabs\n *\n * @param el - Navigation tabs element\n * @param options - Options\n *\n * @returns Navigation tabs observable\n */\nexport function watchTabs(\n  el: HTMLElement, { viewport$, header$ }: WatchOptions\n): Observable<Tabs> {\n  return watchElementSize(document.body)\n    .pipe(\n      switchMap(() => watchViewportAt(el, { header$, viewport$ })),\n      map(({ offset: { y } }) => {\n        return {\n          hidden: y >= 10\n        }\n      }),\n      distinctUntilKeyChanged(\"hidden\")\n    )\n}\n\n/**\n * Mount navigation tabs\n *\n * This function hides the navigation tabs when scrolling past the threshold\n * and makes them reappear in a nice CSS animation when scrolling back up.\n *\n * @param el - Navigation tabs element\n * @param options - Options\n *\n * @returns Navigation tabs component observable\n */\nexport function mountTabs(\n  el: HTMLElement, options: MountOptions\n): Observable<Component<Tabs>> {\n  return defer(() => {\n    const push$ = new Subject<Tabs>()\n    push$.subscribe({\n\n      /* Handle emission */\n      next({ hidden }) {\n        el.hidden = hidden\n      },\n\n      /* Handle complete */\n      complete() {\n        el.hidden = false\n      }\n    })\n\n    /* Create and return component */\n    return (\n      feature(\"navigation.tabs.sticky\")\n        ? of({ hidden: false })\n        : watchTabs(el, options)\n    )\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  asyncScheduler,\n  bufferCount,\n  combineLatestWith,\n  debounceTime,\n  defer,\n  distinctUntilChanged,\n  distinctUntilKeyChanged,\n  endWith,\n  filter,\n  finalize,\n  ignoreElements,\n  map,\n  merge,\n  observeOn,\n  of,\n  repeat,\n  scan,\n  share,\n  skip,\n  startWith,\n  switchMap,\n  takeUntil,\n  tap,\n  withLatestFrom\n} from \"rxjs\"\n\nimport { feature } from \"~/_\"\nimport {\n  Viewport,\n  getElement,\n  getElementContainer,\n  getElementSize,\n  getElements,\n  getLocation,\n  getOptionalElement,\n  watchElementSize\n} from \"~/browser\"\n\nimport {\n  Component,\n  getComponentElement\n} from \"../_\"\nimport { Header } from \"../header\"\nimport { Main } from \"../main\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Table of contents\n */\nexport interface TableOfContents {\n  prev: HTMLAnchorElement[][]          /* Anchors (previous) */\n  next: HTMLAnchorElement[][]          /* Anchors (next) */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n  main$: Observable<Main>              /* Main area observable */\n  target$: Observable<HTMLElement>     /* Location target observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch table of contents\n *\n * This is effectively a scroll spy implementation which will account for the\n * fixed header and automatically re-calculate anchor offsets when the viewport\n * is resized. The returned observable will only emit if the table of contents\n * needs to be repainted.\n *\n * This implementation tracks an anchor element's entire path starting from its\n * level up to the top-most anchor element, e.g. `[h3, h2, h1]`. Although the\n * Material theme currently doesn't make use of this information, it enables\n * the styling of the entire hierarchy through customization.\n *\n * Note that the current anchor is the last item of the `prev` anchor list.\n *\n * @param el - Table of contents element\n * @param options - Options\n *\n * @returns Table of contents observable\n */\nexport function watchTableOfContents(\n  el: HTMLElement, { viewport$, header$ }: WatchOptions\n): Observable<TableOfContents> {\n  const table = new Map<HTMLAnchorElement, HTMLElement>()\n\n  /* Compute anchor-to-target mapping */\n  const anchors = getElements<HTMLAnchorElement>(\"[href^=\\\\#]\", el)\n  for (const anchor of anchors) {\n    const id = decodeURIComponent(anchor.hash.substring(1))\n    const target = getOptionalElement(`[id=\"${id}\"]`)\n    if (typeof target !== \"undefined\")\n      table.set(anchor, target)\n  }\n\n  /* Compute necessary adjustment for header */\n  const adjust$ = header$\n    .pipe(\n      distinctUntilKeyChanged(\"height\"),\n      map(({ height }) => {\n        const main = getComponentElement(\"main\")\n        const grid = getElement(\":scope > :first-child\", main)\n        return height + 0.8 * (\n          grid.offsetTop -\n          main.offsetTop\n        )\n      }),\n      share()\n    )\n\n  /* Compute partition of previous and next anchors */\n  const partition$ = watchElementSize(document.body)\n    .pipe(\n      distinctUntilKeyChanged(\"height\"),\n\n      /* Build index to map anchor paths to vertical offsets */\n      switchMap(body => defer(() => {\n        let path: HTMLAnchorElement[] = []\n        return of([...table].reduce((index, [anchor, target]) => {\n          while (path.length) {\n            const last = table.get(path[path.length - 1])!\n            if (last.tagName >= target.tagName) {\n              path.pop()\n            } else {\n              break\n            }\n          }\n\n          /* If the current anchor is hidden, continue with its parent */\n          let offset = target.offsetTop\n          while (!offset && target.parentElement) {\n            target = target.parentElement\n            offset = target.offsetTop\n          }\n\n          /* Fix anchor offsets in tables - see https://bit.ly/3CUFOcn */\n          let parent = target.offsetParent as HTMLElement\n          for (; parent; parent = parent.offsetParent as HTMLElement)\n            offset += parent.offsetTop\n\n          /* Map reversed anchor path to vertical offset */\n          return index.set(\n            [...path = [...path, anchor]].reverse(),\n            offset\n          )\n        }, new Map<HTMLAnchorElement[], number>()))\n      })\n        .pipe(\n\n          /* Sort index by vertical offset (see https://bit.ly/30z6QSO) */\n          map(index => new Map([...index].sort(([, a], [, b]) => a - b))),\n          combineLatestWith(adjust$),\n\n          /* Re-compute partition when viewport offset changes */\n          switchMap(([index, adjust]) => viewport$\n            .pipe(\n              scan(([prev, next], { offset: { y }, size }) => {\n                const last = y + size.height >= Math.floor(body.height)\n\n                /* Look forward */\n                while (next.length) {\n                  const [, offset] = next[0]\n                  if (offset - adjust < y || last) {\n                    prev = [...prev, next.shift()!]\n                  } else {\n                    break\n                  }\n                }\n\n                /* Look backward */\n                while (prev.length) {\n                  const [, offset] = prev[prev.length - 1]\n                  if (offset - adjust >= y && !last) {\n                    next = [prev.pop()!, ...next]\n                  } else {\n                    break\n                  }\n                }\n\n                /* Return partition */\n                return [prev, next]\n              }, [[], [...index]]),\n              distinctUntilChanged((a, b) => (\n                a[0] === b[0] &&\n                a[1] === b[1]\n              ))\n            )\n          )\n        )\n      )\n    )\n\n  /* Compute and return anchor list migrations */\n  return partition$\n    .pipe(\n      map(([prev, next]) => ({\n        prev: prev.map(([path]) => path),\n        next: next.map(([path]) => path)\n      })),\n\n      /* Extract anchor list migrations */\n      startWith({ prev: [], next: [] }),\n      bufferCount(2, 1),\n      map(([a, b]) => {\n\n        /* Moving down */\n        if (a.prev.length < b.prev.length) {\n          return {\n            prev: b.prev.slice(Math.max(0, a.prev.length - 1), b.prev.length),\n            next: []\n          }\n\n        /* Moving up */\n        } else {\n          return {\n            prev: b.prev.slice(-1),\n            next: b.next.slice(0, b.next.length - a.next.length)\n          }\n        }\n      })\n    )\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Mount table of contents\n *\n * @param el - Table of contents element\n * @param options - Options\n *\n * @returns Table of contents component observable\n */\nexport function mountTableOfContents(\n  el: HTMLElement, { viewport$, header$, main$, target$ }: MountOptions\n): Observable<Component<TableOfContents>> {\n  return defer(() => {\n    const push$ = new Subject<TableOfContents>()\n    const done$ = push$.pipe(ignoreElements(), endWith(true))\n    push$.subscribe(({ prev, next }) => {\n\n      /* Look forward */\n      for (const [anchor] of next) {\n        anchor.classList.remove(\"md-nav__link--passed\")\n        anchor.classList.remove(\"md-nav__link--active\")\n      }\n\n      /* Look backward */\n      for (const [index, [anchor]] of prev.entries()) {\n        anchor.classList.add(\"md-nav__link--passed\")\n        anchor.classList.toggle(\n          \"md-nav__link--active\",\n          index === prev.length - 1\n        )\n      }\n    })\n\n    /* Set up following, if enabled */\n    if (feature(\"toc.follow\")) {\n\n      /* Toggle smooth scrolling only for anchor clicks */\n      const smooth$ = merge(\n        viewport$.pipe(debounceTime(1), map(() => undefined)),\n        viewport$.pipe(debounceTime(250), map(() => \"smooth\" as const))\n      )\n\n      /* Bring active anchor into view */ // @todo: refactor\n      push$\n        .pipe(\n          filter(({ prev }) => prev.length > 0),\n          combineLatestWith(main$.pipe(observeOn(asyncScheduler))),\n          withLatestFrom(smooth$)\n        )\n          .subscribe(([[{ prev }], behavior]) => {\n            const [anchor] = prev[prev.length - 1]\n            if (anchor.offsetHeight) {\n\n              /* Retrieve overflowing container and scroll */\n              const container = getElementContainer(anchor)\n              if (typeof container !== \"undefined\") {\n                const offset = anchor.offsetTop - container.offsetTop\n                const { height } = getElementSize(container)\n                container.scrollTo({\n                  top: offset - height / 2,\n                  behavior\n                })\n              }\n            }\n          })\n    }\n\n    /* Set up anchor tracking, if enabled */\n    if (feature(\"navigation.tracking\"))\n      viewport$\n        .pipe(\n          takeUntil(done$),\n          distinctUntilKeyChanged(\"offset\"),\n          debounceTime(250),\n          skip(1),\n          takeUntil(target$.pipe(skip(1))),\n          repeat({ delay: 250 }),\n          withLatestFrom(push$)\n        )\n          .subscribe(([, { prev }]) => {\n            const url = getLocation()\n\n            /* Set hash fragment to active anchor */\n            const anchor = prev[prev.length - 1]\n            if (anchor && anchor.length) {\n              const [active] = anchor\n              const { hash } = new URL(active.href)\n              if (url.hash !== hash) {\n                url.hash = hash\n                history.replaceState({}, \"\", `${url}`)\n              }\n\n            /* Reset anchor when at the top */\n            } else {\n              url.hash = \"\"\n              history.replaceState({}, \"\", `${url}`)\n            }\n          })\n\n    /* Create and return component */\n    return watchTableOfContents(el, { viewport$, header$ })\n      .pipe(\n        tap(state => push$.next(state)),\n        finalize(() => push$.complete()),\n        map(state => ({ ref: el, ...state }))\n      )\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  Subject,\n  bufferCount,\n  combineLatest,\n  distinctUntilChanged,\n  distinctUntilKeyChanged,\n  endWith,\n  finalize,\n  fromEvent,\n  ignoreElements,\n  map,\n  repeat,\n  skip,\n  takeUntil,\n  tap\n} from \"rxjs\"\n\nimport { Viewport } from \"~/browser\"\n\nimport { Component } from \"../_\"\nimport { Header } from \"../header\"\nimport { Main } from \"../main\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Back-to-top button\n */\nexport interface BackToTop {\n  hidden: boolean                      /* Back-to-top button is hidden */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch options\n */\ninterface WatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  main$: Observable<Main>              /* Main area observable */\n  target$: Observable<HTMLElement>     /* Location target observable */\n}\n\n/**\n * Mount options\n */\ninterface MountOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  header$: Observable<Header>          /* Header observable */\n  main$: Observable<Main>              /* Main area observable */\n  target$: Observable<HTMLElement>     /* Location target observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Watch back-to-top\n *\n * @param _el - Back-to-top element\n * @param options - Options\n *\n * @returns Back-to-top observable\n */\nexport function watchBackToTop(\n  _el: HTMLElement, { viewport$, main$, target$ }: WatchOptions\n): Observable<BackToTop> {\n\n  /* Compute direction */\n  const direction$ = viewport$\n    .pipe(\n      map(({ offset: { y } }) => y),\n      bufferCount(2, 1),\n      map(([a, b]) => a > b && b > 0),\n      distinctUntilChanged()\n    )\n\n  /* Compute whether main area is active */\n  const active$ = main$\n    .pipe(\n      map(({ active }) => active)\n    )\n\n  /* Compute threshold for hiding */\n  return combineLatest([active$, direction$])\n    .pipe(\n      map(([active, direction]) => !(active && direction)),\n      distinctUntilChanged(),\n      takeUntil(target$.pipe(skip(1))),\n      endWith(true),\n      repeat({ delay: 250 }),\n      map(hidden => ({ hidden }))\n    )\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Mount back-to-top\n *\n * @param el - Back-to-top element\n * @param options - Options\n *\n * @returns Back-to-top component observable\n */\nexport function mountBackToTop(\n  el: HTMLElement, { viewport$, header$, main$, target$ }: MountOptions\n): Observable<Component<BackToTop>> {\n  const push$ = new Subject<BackToTop>()\n  const done$ = push$.pipe(ignoreElements(), endWith(true))\n  push$.subscribe({\n\n    /* Handle emission */\n    next({ hidden }) {\n      el.hidden = hidden\n      if (hidden) {\n        el.setAttribute(\"tabindex\", \"-1\")\n        el.blur()\n      } else {\n        el.removeAttribute(\"tabindex\")\n      }\n    },\n\n    /* Handle complete */\n    complete() {\n      el.style.top = \"\"\n      el.hidden = true\n      el.removeAttribute(\"tabindex\")\n    }\n  })\n\n  /* Watch header height */\n  header$\n    .pipe(\n      takeUntil(done$),\n      distinctUntilKeyChanged(\"height\")\n    )\n      .subscribe(({ height }) => {\n        el.style.top = `${height + 16}px`\n      })\n\n  /* Go back to top */\n  fromEvent(el, \"click\")\n    .subscribe(ev => {\n      ev.preventDefault()\n      window.scrollTo({ top: 0 })\n    })\n\n  /* Create and return component */\n  return watchBackToTop(el, { viewport$, main$, target$ })\n    .pipe(\n      tap(state => push$.next(state)),\n      finalize(() => push$.complete()),\n      map(state => ({ ref: el, ...state }))\n    )\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  fromEvent,\n  map,\n  mergeMap,\n  switchMap,\n  takeWhile,\n  tap,\n  withLatestFrom\n} from \"rxjs\"\n\nimport { getElements } from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Patch options\n */\ninterface PatchOptions {\n  document$: Observable<Document>      /* Document observable */\n  tablet$: Observable<boolean>         /* Media tablet observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Patch indeterminate checkboxes\n *\n * This function replaces the indeterminate \"pseudo state\" with the actual\n * indeterminate state, which is used to keep navigation always expanded.\n *\n * @param options - Options\n */\nexport function patchIndeterminate(\n  { document$, tablet$ }: PatchOptions\n): void {\n  document$\n    .pipe(\n      switchMap(() => getElements<HTMLInputElement>(\n        \".md-toggle--indeterminate\"\n      )),\n      tap(el => {\n        el.indeterminate = true\n        el.checked = false\n      }),\n      mergeMap(el => fromEvent(el, \"change\")\n        .pipe(\n          takeWhile(() => el.classList.contains(\"md-toggle--indeterminate\")),\n          map(() => el)\n        )\n      ),\n      withLatestFrom(tablet$)\n    )\n      .subscribe(([el, tablet]) => {\n        el.classList.remove(\"md-toggle--indeterminate\")\n        if (tablet)\n          el.checked = false\n      })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  filter,\n  fromEvent,\n  map,\n  mergeMap,\n  switchMap,\n  tap\n} from \"rxjs\"\n\nimport { getElements } from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Patch options\n */\ninterface PatchOptions {\n  document$: Observable<Document>      /* Document observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Check whether the given device is an Apple device\n *\n * @returns Test result\n */\nfunction isAppleDevice(): boolean {\n  return /(iPad|iPhone|iPod)/.test(navigator.userAgent)\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Patch all elements with `data-md-scrollfix` attributes\n *\n * This is a year-old patch which ensures that overflow scrolling works at the\n * top and bottom of containers on iOS by ensuring a `1px` scroll offset upon\n * the start of a touch event.\n *\n * @see https://bit.ly/2SCtAOO - Original source\n *\n * @param options - Options\n */\nexport function patchScrollfix(\n  { document$ }: PatchOptions\n): void {\n  document$\n    .pipe(\n      switchMap(() => getElements(\"[data-md-scrollfix]\")),\n      tap(el => el.removeAttribute(\"data-md-scrollfix\")),\n      filter(isAppleDevice),\n      mergeMap(el => fromEvent(el, \"touchstart\")\n        .pipe(\n          map(() => el)\n        )\n      )\n    )\n      .subscribe(el => {\n        const top = el.scrollTop\n\n        /* We're at the top of the container */\n        if (top === 0) {\n          el.scrollTop = 1\n\n        /* We're at the bottom of the container */\n        } else if (top + el.offsetHeight === el.scrollHeight) {\n          el.scrollTop = top - 1\n        }\n      })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  Observable,\n  combineLatest,\n  delay,\n  map,\n  of,\n  switchMap,\n  withLatestFrom\n} from \"rxjs\"\n\nimport {\n  Viewport,\n  watchToggle\n} from \"~/browser\"\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Patch options\n */\ninterface PatchOptions {\n  viewport$: Observable<Viewport>      /* Viewport observable */\n  tablet$: Observable<boolean>         /* Media tablet observable */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Patch the document body to lock when search is open\n *\n * For mobile and tablet viewports, the search is rendered full screen, which\n * leads to scroll leaking when at the top or bottom of the search result. This\n * function locks the body when the search is in full screen mode, and restores\n * the scroll position when leaving.\n *\n * @param options - Options\n */\nexport function patchScrolllock(\n  { viewport$, tablet$ }: PatchOptions\n): void {\n  combineLatest([watchToggle(\"search\"), tablet$])\n    .pipe(\n      map(([active, tablet]) => active && !tablet),\n      switchMap(active => of(active)\n        .pipe(\n          delay(active ? 400 : 100)\n        )\n      ),\n      withLatestFrom(viewport$)\n    )\n      .subscribe(([active, { offset: { y }}]) => {\n        if (active) {\n          document.body.setAttribute(\"data-md-scrolllock\", \"\")\n          document.body.style.top = `-${y}px`\n        } else {\n          const value = -1 * parseInt(document.body.style.top, 10)\n          document.body.removeAttribute(\"data-md-scrolllock\")\n          document.body.style.top = \"\"\n          if (value)\n            window.scrollTo(0, value)\n        }\n      })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Polyfills\n * ------------------------------------------------------------------------- */\n\n/* Polyfill `Object.entries` */\nif (!Object.entries)\n  Object.entries = function (obj: object) {\n    const data: [string, string][] = []\n    for (const key of Object.keys(obj))\n      // @ts-expect-error - ignore property access warning\n      data.push([key, obj[key]])\n\n    /* Return entries */\n    return data\n  }\n\n/* Polyfill `Object.values` */\nif (!Object.values)\n  Object.values = function (obj: object) {\n    const data: string[] = []\n    for (const key of Object.keys(obj))\n      // @ts-expect-error - ignore property access warning\n      data.push(obj[key])\n\n    /* Return values */\n    return data\n  }\n\n/* ------------------------------------------------------------------------- */\n\n/* Polyfills for `Element` */\nif (typeof Element !== \"undefined\") {\n\n  /* Polyfill `Element.scrollTo` */\n  if (!Element.prototype.scrollTo)\n    Element.prototype.scrollTo = function (\n      x?: ScrollToOptions | number, y?: number\n    ): void {\n      if (typeof x === \"object\") {\n        this.scrollLeft = x.left!\n        this.scrollTop = x.top!\n      } else {\n        this.scrollLeft = x!\n        this.scrollTop = y!\n      }\n    }\n\n  /* Polyfill `Element.replaceWith` */\n  if (!Element.prototype.replaceWith)\n    Element.prototype.replaceWith = function (\n      ...nodes: Array<string | Node>\n    ): void {\n      const parent = this.parentNode\n      if (parent) {\n        if (nodes.length === 0)\n          parent.removeChild(this)\n\n        /* Replace children and create text nodes */\n        for (let i = nodes.length - 1; i >= 0; i--) {\n          let node = nodes[i]\n          if (typeof node === \"string\")\n            node = document.createTextNode(node)\n          else if (node.parentNode)\n            node.parentNode.removeChild(node)\n\n          /* Replace child or insert before previous sibling */\n          if (!i)\n            parent.replaceChild(node, this)\n          else\n            parent.insertBefore(this.previousSibling!, node)\n        }\n      }\n    }\n}\n"],
-  "mappings": "6+BAAA,IAAAA,GAAAC,GAAA,CAAAC,GAAAC,KAAA,EAAC,SAAUC,EAAQC,EAAS,CAC1B,OAAOH,IAAY,UAAY,OAAOC,IAAW,YAAcE,EAAQ,EACvE,OAAO,QAAW,YAAc,OAAO,IAAM,OAAOA,CAAO,EAC1DA,EAAQ,CACX,GAAEH,GAAO,UAAY,CAAE,aASrB,SAASI,EAA0BC,EAAO,CACxC,IAAIC,EAAmB,GACnBC,EAA0B,GAC1BC,EAAiC,KAEjCC,EAAsB,CACxB,KAAM,GACN,OAAQ,GACR,IAAK,GACL,IAAK,GACL,MAAO,GACP,SAAU,GACV,OAAQ,GACR,KAAM,GACN,MAAO,GACP,KAAM,GACN,KAAM,GACN,SAAU,GACV,iBAAkB,EACpB,EAOA,SAASC,EAAmBC,EAAI,CAC9B,MACE,GAAAA,GACAA,IAAO,UACPA,EAAG,WAAa,QAChBA,EAAG,WAAa,QAChB,cAAeA,GACf,aAAcA,EAAG,UAKrB,CASA,SAASC,EAA8BD,EAAI,CACzC,IAAIE,GAAOF,EAAG,KACVG,GAAUH,EAAG,QAUjB,MARI,GAAAG,KAAY,SAAWL,EAAoBI,EAAI,GAAK,CAACF,EAAG,UAIxDG,KAAY,YAAc,CAACH,EAAG,UAI9BA,EAAG,kBAKT,CAOA,SAASI,EAAqBJ,EAAI,CAC5BA,EAAG,UAAU,SAAS,eAAe,IAGzCA,EAAG,UAAU,IAAI,eAAe,EAChCA,EAAG,aAAa,2BAA4B,EAAE,EAChD,CAOA,SAASK,EAAwBL,EAAI,CAC9BA,EAAG,aAAa,0BAA0B,IAG/CA,EAAG,UAAU,OAAO,eAAe,EACnCA,EAAG,gBAAgB,0BAA0B,EAC/C,CAUA,SAASM,EAAUC,EAAG,CAChBA,EAAE,SAAWA,EAAE,QAAUA,EAAE,UAI3BR,EAAmBL,EAAM,aAAa,GACxCU,EAAqBV,EAAM,aAAa,EAG1CC,EAAmB,GACrB,CAUA,SAASa,EAAcD,EAAG,CACxBZ,EAAmB,EACrB,CASA,SAASc,EAAQF,EAAG,CAEbR,EAAmBQ,EAAE,MAAM,IAI5BZ,GAAoBM,EAA8BM,EAAE,MAAM,IAC5DH,EAAqBG,EAAE,MAAM,CAEjC,CAMA,SAASG,EAAOH,EAAG,CACZR,EAAmBQ,EAAE,MAAM,IAK9BA,EAAE,OAAO,UAAU,SAAS,eAAe,GAC3CA,EAAE,OAAO,aAAa,0BAA0B,KAMhDX,EAA0B,GAC1B,OAAO,aAAaC,CAA8B,EAClDA,EAAiC,OAAO,WAAW,UAAW,CAC5DD,EAA0B,EAC5B,EAAG,GAAG,EACNS,EAAwBE,EAAE,MAAM,EAEpC,CAOA,SAASI,EAAmBJ,EAAG,CACzB,SAAS,kBAAoB,WAK3BX,IACFD,EAAmB,IAErBiB,EAA+B,EAEnC,CAQA,SAASA,GAAiC,CACxC,SAAS,iBAAiB,YAAaC,CAAoB,EAC3D,SAAS,iBAAiB,YAAaA,CAAoB,EAC3D,SAAS,iBAAiB,UAAWA,CAAoB,EACzD,SAAS,iBAAiB,cAAeA,CAAoB,EAC7D,SAAS,iBAAiB,cAAeA,CAAoB,EAC7D,SAAS,iBAAiB,YAAaA,CAAoB,EAC3D,SAAS,iBAAiB,YAAaA,CAAoB,EAC3D,SAAS,iBAAiB,aAAcA,CAAoB,EAC5D,SAAS,iBAAiB,WAAYA,CAAoB,CAC5D,CAEA,SAASC,GAAoC,CAC3C,SAAS,oBAAoB,YAAaD,CAAoB,EAC9D,SAAS,oBAAoB,YAAaA,CAAoB,EAC9D,SAAS,oBAAoB,UAAWA,CAAoB,EAC5D,SAAS,oBAAoB,cAAeA,CAAoB,EAChE,SAAS,oBAAoB,cAAeA,CAAoB,EAChE,SAAS,oBAAoB,YAAaA,CAAoB,EAC9D,SAAS,oBAAoB,YAAaA,CAAoB,EAC9D,SAAS,oBAAoB,aAAcA,CAAoB,EAC/D,SAAS,oBAAoB,WAAYA,CAAoB,CAC/D,CASA,SAASA,EAAqBN,EAAG,CAG3BA,EAAE,OAAO,UAAYA,EAAE,OAAO,SAAS,YAAY,IAAM,SAI7DZ,EAAmB,GACnBmB,EAAkC,EACpC,CAKA,SAAS,iBAAiB,UAAWR,EAAW,EAAI,EACpD,SAAS,iBAAiB,YAAaE,EAAe,EAAI,EAC1D,SAAS,iBAAiB,cAAeA,EAAe,EAAI,EAC5D,SAAS,iBAAiB,aAAcA,EAAe,EAAI,EAC3D,SAAS,iBAAiB,mBAAoBG,EAAoB,EAAI,EAEtEC,EAA+B,EAM/BlB,EAAM,iBAAiB,QAASe,EAAS,EAAI,EAC7Cf,EAAM,iBAAiB,OAAQgB,EAAQ,EAAI,EAOvChB,EAAM,WAAa,KAAK,wBAA0BA,EAAM,KAI1DA,EAAM,KAAK,aAAa,wBAAyB,EAAE,EAC1CA,EAAM,WAAa,KAAK,gBACjC,SAAS,gBAAgB,UAAU,IAAI,kBAAkB,EACzD,SAAS,gBAAgB,aAAa,wBAAyB,EAAE,EAErE,CAKA,GAAI,OAAO,QAAW,aAAe,OAAO,UAAa,YAAa,CAIpE,OAAO,0BAA4BD,EAInC,IAAIsB,EAEJ,GAAI,CACFA,EAAQ,IAAI,YAAY,8BAA8B,CACxD,OAASC,EAAO,CAEdD,EAAQ,SAAS,YAAY,aAAa,EAC1CA,EAAM,gBAAgB,+BAAgC,GAAO,GAAO,CAAC,CAAC,CACxE,CAEA,OAAO,cAAcA,CAAK,CAC5B,CAEI,OAAO,UAAa,aAGtBtB,EAA0B,QAAQ,CAGtC,CAAE,ICvTF,IAAAwB,GAAAC,GAAA,CAAAC,GAAAC,KAAA;AAAA;AAAA;AAAA;AAAA;AAAA,IAMC,SAA0CC,EAAMC,EAAS,CACtD,OAAOH,IAAY,UAAY,OAAOC,IAAW,SACnDA,GAAO,QAAUE,EAAQ,EAClB,OAAO,QAAW,YAAc,OAAO,IAC9C,OAAO,CAAC,EAAGA,CAAO,EACX,OAAOH,IAAY,SAC1BA,GAAQ,YAAiBG,EAAQ,EAEjCD,EAAK,YAAiBC,EAAQ,CAChC,GAAGH,GAAM,UAAW,CACpB,OAAiB,UAAW,CAClB,IAAII,EAAuB,CAE/B,IACC,SAASC,EAAyBC,EAAqBC,EAAqB,CAEnF,aAGAA,EAAoB,EAAED,EAAqB,CACzC,QAAW,UAAW,CAAE,OAAqBE,EAAW,CAC1D,CAAC,EAGD,IAAIC,EAAeF,EAAoB,GAAG,EACtCG,EAAoCH,EAAoB,EAAEE,CAAY,EAEtEE,EAASJ,EAAoB,GAAG,EAChCK,EAA8BL,EAAoB,EAAEI,CAAM,EAE1DE,EAAaN,EAAoB,GAAG,EACpCO,EAA8BP,EAAoB,EAAEM,CAAU,EAOlE,SAASE,EAAQC,EAAM,CACrB,GAAI,CACF,OAAO,SAAS,YAAYA,CAAI,CAClC,OAASC,EAAK,CACZ,MAAO,EACT,CACF,CAUA,IAAIC,EAAqB,SAA4BC,EAAQ,CAC3D,IAAIC,EAAeN,EAAe,EAAEK,CAAM,EAC1C,OAAAJ,EAAQ,KAAK,EACNK,CACT,EAEiCC,EAAeH,EAOhD,SAASI,EAAkBC,EAAO,CAChC,IAAIC,EAAQ,SAAS,gBAAgB,aAAa,KAAK,IAAM,MACzDC,EAAc,SAAS,cAAc,UAAU,EAEnDA,EAAY,MAAM,SAAW,OAE7BA,EAAY,MAAM,OAAS,IAC3BA,EAAY,MAAM,QAAU,IAC5BA,EAAY,MAAM,OAAS,IAE3BA,EAAY,MAAM,SAAW,WAC7BA,EAAY,MAAMD,EAAQ,QAAU,MAAM,EAAI,UAE9C,IAAIE,EAAY,OAAO,aAAe,SAAS,gBAAgB,UAC/D,OAAAD,EAAY,MAAM,IAAM,GAAG,OAAOC,EAAW,IAAI,EACjDD,EAAY,aAAa,WAAY,EAAE,EACvCA,EAAY,MAAQF,EACbE,CACT,CAYA,IAAIE,EAAiB,SAAwBJ,EAAOK,EAAS,CAC3D,IAAIH,EAAcH,EAAkBC,CAAK,EACzCK,EAAQ,UAAU,YAAYH,CAAW,EACzC,IAAIL,EAAeN,EAAe,EAAEW,CAAW,EAC/C,OAAAV,EAAQ,MAAM,EACdU,EAAY,OAAO,EACZL,CACT,EASIS,EAAsB,SAA6BV,EAAQ,CAC7D,IAAIS,EAAU,UAAU,OAAS,GAAK,UAAU,CAAC,IAAM,OAAY,UAAU,CAAC,EAAI,CAChF,UAAW,SAAS,IACtB,EACIR,EAAe,GAEnB,OAAI,OAAOD,GAAW,SACpBC,EAAeO,EAAeR,EAAQS,CAAO,EACpCT,aAAkB,kBAAoB,CAAC,CAAC,OAAQ,SAAU,MAAO,MAAO,UAAU,EAAE,SAASA,GAAW,KAA4B,OAASA,EAAO,IAAI,EAEjKC,EAAeO,EAAeR,EAAO,MAAOS,CAAO,GAEnDR,EAAeN,EAAe,EAAEK,CAAM,EACtCJ,EAAQ,MAAM,GAGTK,CACT,EAEiCU,EAAgBD,EAEjD,SAASE,EAAQC,EAAK,CAAE,0BAA2B,OAAI,OAAO,QAAW,YAAc,OAAO,OAAO,UAAa,SAAYD,EAAU,SAAiBC,EAAK,CAAE,OAAO,OAAOA,CAAK,EAAYD,EAAU,SAAiBC,EAAK,CAAE,OAAOA,GAAO,OAAO,QAAW,YAAcA,EAAI,cAAgB,QAAUA,IAAQ,OAAO,UAAY,SAAW,OAAOA,CAAK,EAAYD,EAAQC,CAAG,CAAG,CAUzX,IAAIC,GAAyB,UAAkC,CAC7D,IAAIL,EAAU,UAAU,OAAS,GAAK,UAAU,CAAC,IAAM,OAAY,UAAU,CAAC,EAAI,CAAC,EAE/EM,EAAkBN,EAAQ,OAC1BO,EAASD,IAAoB,OAAS,OAASA,EAC/CE,EAAYR,EAAQ,UACpBT,EAASS,EAAQ,OACjBS,GAAOT,EAAQ,KAEnB,GAAIO,IAAW,QAAUA,IAAW,MAClC,MAAM,IAAI,MAAM,oDAAoD,EAItE,GAAIhB,IAAW,OACb,GAAIA,GAAUY,EAAQZ,CAAM,IAAM,UAAYA,EAAO,WAAa,EAAG,CACnE,GAAIgB,IAAW,QAAUhB,EAAO,aAAa,UAAU,EACrD,MAAM,IAAI,MAAM,mFAAmF,EAGrG,GAAIgB,IAAW,QAAUhB,EAAO,aAAa,UAAU,GAAKA,EAAO,aAAa,UAAU,GACxF,MAAM,IAAI,MAAM,uGAAwG,CAE5H,KACE,OAAM,IAAI,MAAM,6CAA6C,EAKjE,GAAIkB,GACF,OAAOP,EAAaO,GAAM,CACxB,UAAWD,CACb,CAAC,EAIH,GAAIjB,EACF,OAAOgB,IAAW,MAAQd,EAAYF,CAAM,EAAIW,EAAaX,EAAQ,CACnE,UAAWiB,CACb,CAAC,CAEL,EAEiCE,GAAmBL,GAEpD,SAASM,GAAiBP,EAAK,CAAE,0BAA2B,OAAI,OAAO,QAAW,YAAc,OAAO,OAAO,UAAa,SAAYO,GAAmB,SAAiBP,EAAK,CAAE,OAAO,OAAOA,CAAK,EAAYO,GAAmB,SAAiBP,EAAK,CAAE,OAAOA,GAAO,OAAO,QAAW,YAAcA,EAAI,cAAgB,QAAUA,IAAQ,OAAO,UAAY,SAAW,OAAOA,CAAK,EAAYO,GAAiBP,CAAG,CAAG,CAE7Z,SAASQ,GAAgBC,EAAUC,EAAa,CAAE,GAAI,EAAED,aAAoBC,GAAgB,MAAM,IAAI,UAAU,mCAAmC,CAAK,CAExJ,SAASC,GAAkBxB,EAAQyB,EAAO,CAAE,QAASC,EAAI,EAAGA,EAAID,EAAM,OAAQC,IAAK,CAAE,IAAIC,EAAaF,EAAMC,CAAC,EAAGC,EAAW,WAAaA,EAAW,YAAc,GAAOA,EAAW,aAAe,GAAU,UAAWA,IAAYA,EAAW,SAAW,IAAM,OAAO,eAAe3B,EAAQ2B,EAAW,IAAKA,CAAU,CAAG,CAAE,CAE5T,SAASC,GAAaL,EAAaM,EAAYC,EAAa,CAAE,OAAID,GAAYL,GAAkBD,EAAY,UAAWM,CAAU,EAAOC,GAAaN,GAAkBD,EAAaO,CAAW,EAAUP,CAAa,CAEtN,SAASQ,GAAUC,EAAUC,EAAY,CAAE,GAAI,OAAOA,GAAe,YAAcA,IAAe,KAAQ,MAAM,IAAI,UAAU,oDAAoD,EAAKD,EAAS,UAAY,OAAO,OAAOC,GAAcA,EAAW,UAAW,CAAE,YAAa,CAAE,MAAOD,EAAU,SAAU,GAAM,aAAc,EAAK,CAAE,CAAC,EAAOC,GAAYC,GAAgBF,EAAUC,CAAU,CAAG,CAEhY,SAASC,GAAgBC,EAAGC,EAAG,CAAE,OAAAF,GAAkB,OAAO,gBAAkB,SAAyBC,EAAGC,EAAG,CAAE,OAAAD,EAAE,UAAYC,EAAUD,CAAG,EAAUD,GAAgBC,EAAGC,CAAC,CAAG,CAEzK,SAASC,GAAaC,EAAS,CAAE,IAAIC,EAA4BC,GAA0B,EAAG,OAAO,UAAgC,CAAE,IAAIC,EAAQC,GAAgBJ,CAAO,EAAGK,EAAQ,GAAIJ,EAA2B,CAAE,IAAIK,EAAYF,GAAgB,IAAI,EAAE,YAAaC,EAAS,QAAQ,UAAUF,EAAO,UAAWG,CAAS,CAAG,MAASD,EAASF,EAAM,MAAM,KAAM,SAAS,EAAK,OAAOI,GAA2B,KAAMF,CAAM,CAAG,CAAG,CAExa,SAASE,GAA2BC,EAAMC,EAAM,CAAE,OAAIA,IAAS3B,GAAiB2B,CAAI,IAAM,UAAY,OAAOA,GAAS,YAAsBA,EAAeC,GAAuBF,CAAI,CAAG,CAEzL,SAASE,GAAuBF,EAAM,CAAE,GAAIA,IAAS,OAAU,MAAM,IAAI,eAAe,2DAA2D,EAAK,OAAOA,CAAM,CAErK,SAASN,IAA4B,CAA0E,GAApE,OAAO,SAAY,aAAe,CAAC,QAAQ,WAA6B,QAAQ,UAAU,KAAM,MAAO,GAAO,GAAI,OAAO,OAAU,WAAY,MAAO,GAAM,GAAI,CAAE,YAAK,UAAU,SAAS,KAAK,QAAQ,UAAU,KAAM,CAAC,EAAG,UAAY,CAAC,CAAC,CAAC,EAAU,EAAM,OAASS,EAAG,CAAE,MAAO,EAAO,CAAE,CAEnU,SAASP,GAAgBP,EAAG,CAAE,OAAAO,GAAkB,OAAO,eAAiB,OAAO,eAAiB,SAAyBP,EAAG,CAAE,OAAOA,EAAE,WAAa,OAAO,eAAeA,CAAC,CAAG,EAAUO,GAAgBP,CAAC,CAAG,CAa5M,SAASe,GAAkBC,EAAQC,EAAS,CAC1C,IAAIC,EAAY,kBAAkB,OAAOF,CAAM,EAE/C,GAAKC,EAAQ,aAAaC,CAAS,EAInC,OAAOD,EAAQ,aAAaC,CAAS,CACvC,CAOA,IAAIC,GAAyB,SAAUC,EAAU,CAC/CxB,GAAUuB,EAAWC,CAAQ,EAE7B,IAAIC,EAASnB,GAAaiB,CAAS,EAMnC,SAASA,EAAUG,EAAShD,EAAS,CACnC,IAAIiD,EAEJ,OAAArC,GAAgB,KAAMiC,CAAS,EAE/BI,EAAQF,EAAO,KAAK,IAAI,EAExBE,EAAM,eAAejD,CAAO,EAE5BiD,EAAM,YAAYD,CAAO,EAElBC,CACT,CAQA,OAAA9B,GAAa0B,EAAW,CAAC,CACvB,IAAK,iBACL,MAAO,UAA0B,CAC/B,IAAI7C,EAAU,UAAU,OAAS,GAAK,UAAU,CAAC,IAAM,OAAY,UAAU,CAAC,EAAI,CAAC,EACnF,KAAK,OAAS,OAAOA,EAAQ,QAAW,WAAaA,EAAQ,OAAS,KAAK,cAC3E,KAAK,OAAS,OAAOA,EAAQ,QAAW,WAAaA,EAAQ,OAAS,KAAK,cAC3E,KAAK,KAAO,OAAOA,EAAQ,MAAS,WAAaA,EAAQ,KAAO,KAAK,YACrE,KAAK,UAAYW,GAAiBX,EAAQ,SAAS,IAAM,SAAWA,EAAQ,UAAY,SAAS,IACnG,CAMF,EAAG,CACD,IAAK,cACL,MAAO,SAAqBgD,EAAS,CACnC,IAAIE,EAAS,KAEb,KAAK,SAAWlE,EAAe,EAAEgE,EAAS,QAAS,SAAUR,GAAG,CAC9D,OAAOU,EAAO,QAAQV,EAAC,CACzB,CAAC,CACH,CAMF,EAAG,CACD,IAAK,UACL,MAAO,SAAiBA,EAAG,CACzB,IAAIQ,EAAUR,EAAE,gBAAkBA,EAAE,cAChCjC,GAAS,KAAK,OAAOyC,CAAO,GAAK,OACjCvC,GAAOC,GAAgB,CACzB,OAAQH,GACR,UAAW,KAAK,UAChB,OAAQ,KAAK,OAAOyC,CAAO,EAC3B,KAAM,KAAK,KAAKA,CAAO,CACzB,CAAC,EAED,KAAK,KAAKvC,GAAO,UAAY,QAAS,CACpC,OAAQF,GACR,KAAME,GACN,QAASuC,EACT,eAAgB,UAA0B,CACpCA,GACFA,EAAQ,MAAM,EAGhB,OAAO,aAAa,EAAE,gBAAgB,CACxC,CACF,CAAC,CACH,CAMF,EAAG,CACD,IAAK,gBACL,MAAO,SAAuBA,EAAS,CACrC,OAAOP,GAAkB,SAAUO,CAAO,CAC5C,CAMF,EAAG,CACD,IAAK,gBACL,MAAO,SAAuBA,EAAS,CACrC,IAAIG,EAAWV,GAAkB,SAAUO,CAAO,EAElD,GAAIG,EACF,OAAO,SAAS,cAAcA,CAAQ,CAE1C,CAQF,EAAG,CACD,IAAK,cAML,MAAO,SAAqBH,EAAS,CACnC,OAAOP,GAAkB,OAAQO,CAAO,CAC1C,CAKF,EAAG,CACD,IAAK,UACL,MAAO,UAAmB,CACxB,KAAK,SAAS,QAAQ,CACxB,CACF,CAAC,EAAG,CAAC,CACH,IAAK,OACL,MAAO,SAAczD,EAAQ,CAC3B,IAAIS,EAAU,UAAU,OAAS,GAAK,UAAU,CAAC,IAAM,OAAY,UAAU,CAAC,EAAI,CAChF,UAAW,SAAS,IACtB,EACA,OAAOE,EAAaX,EAAQS,CAAO,CACrC,CAOF,EAAG,CACD,IAAK,MACL,MAAO,SAAaT,EAAQ,CAC1B,OAAOE,EAAYF,CAAM,CAC3B,CAOF,EAAG,CACD,IAAK,cACL,MAAO,UAAuB,CAC5B,IAAIgB,EAAS,UAAU,OAAS,GAAK,UAAU,CAAC,IAAM,OAAY,UAAU,CAAC,EAAI,CAAC,OAAQ,KAAK,EAC3F6C,EAAU,OAAO7C,GAAW,SAAW,CAACA,CAAM,EAAIA,EAClD8C,GAAU,CAAC,CAAC,SAAS,sBACzB,OAAAD,EAAQ,QAAQ,SAAU7C,GAAQ,CAChC8C,GAAUA,IAAW,CAAC,CAAC,SAAS,sBAAsB9C,EAAM,CAC9D,CAAC,EACM8C,EACT,CACF,CAAC,CAAC,EAEKR,CACT,EAAG/D,EAAqB,CAAE,EAEOF,GAAaiE,EAExC,EAEA,IACC,SAASxE,EAAQ,CAExB,IAAIiF,EAAqB,EAKzB,GAAI,OAAO,SAAY,aAAe,CAAC,QAAQ,UAAU,QAAS,CAC9D,IAAIC,EAAQ,QAAQ,UAEpBA,EAAM,QAAUA,EAAM,iBACNA,EAAM,oBACNA,EAAM,mBACNA,EAAM,kBACNA,EAAM,qBAC1B,CASA,SAASC,EAASb,EAASQ,EAAU,CACjC,KAAOR,GAAWA,EAAQ,WAAaW,GAAoB,CACvD,GAAI,OAAOX,EAAQ,SAAY,YAC3BA,EAAQ,QAAQQ,CAAQ,EAC1B,OAAOR,EAETA,EAAUA,EAAQ,UACtB,CACJ,CAEAtE,EAAO,QAAUmF,CAGX,EAEA,IACC,SAASnF,EAAQoF,EAA0B9E,EAAqB,CAEvE,IAAI6E,EAAU7E,EAAoB,GAAG,EAYrC,SAAS+E,EAAUf,EAASQ,EAAU/D,EAAMuE,EAAUC,EAAY,CAC9D,IAAIC,EAAaC,EAAS,MAAM,KAAM,SAAS,EAE/C,OAAAnB,EAAQ,iBAAiBvD,EAAMyE,EAAYD,CAAU,EAE9C,CACH,QAAS,UAAW,CAChBjB,EAAQ,oBAAoBvD,EAAMyE,EAAYD,CAAU,CAC5D,CACJ,CACJ,CAYA,SAASG,EAASC,EAAUb,EAAU/D,EAAMuE,EAAUC,EAAY,CAE9D,OAAI,OAAOI,EAAS,kBAAqB,WAC9BN,EAAU,MAAM,KAAM,SAAS,EAItC,OAAOtE,GAAS,WAGTsE,EAAU,KAAK,KAAM,QAAQ,EAAE,MAAM,KAAM,SAAS,GAI3D,OAAOM,GAAa,WACpBA,EAAW,SAAS,iBAAiBA,CAAQ,GAI1C,MAAM,UAAU,IAAI,KAAKA,EAAU,SAAUrB,EAAS,CACzD,OAAOe,EAAUf,EAASQ,EAAU/D,EAAMuE,EAAUC,CAAU,CAClE,CAAC,EACL,CAWA,SAASE,EAASnB,EAASQ,EAAU/D,EAAMuE,EAAU,CACjD,OAAO,SAASnB,EAAG,CACfA,EAAE,eAAiBgB,EAAQhB,EAAE,OAAQW,CAAQ,EAEzCX,EAAE,gBACFmB,EAAS,KAAKhB,EAASH,CAAC,CAEhC,CACJ,CAEAnE,EAAO,QAAU0F,CAGX,EAEA,IACC,SAAStF,EAAyBL,EAAS,CAQlDA,EAAQ,KAAO,SAASuB,EAAO,CAC3B,OAAOA,IAAU,QACVA,aAAiB,aACjBA,EAAM,WAAa,CAC9B,EAQAvB,EAAQ,SAAW,SAASuB,EAAO,CAC/B,IAAIP,EAAO,OAAO,UAAU,SAAS,KAAKO,CAAK,EAE/C,OAAOA,IAAU,SACTP,IAAS,qBAAuBA,IAAS,4BACzC,WAAYO,IACZA,EAAM,SAAW,GAAKvB,EAAQ,KAAKuB,EAAM,CAAC,CAAC,EACvD,EAQAvB,EAAQ,OAAS,SAASuB,EAAO,CAC7B,OAAO,OAAOA,GAAU,UACjBA,aAAiB,MAC5B,EAQAvB,EAAQ,GAAK,SAASuB,EAAO,CACzB,IAAIP,EAAO,OAAO,UAAU,SAAS,KAAKO,CAAK,EAE/C,OAAOP,IAAS,mBACpB,CAGM,EAEA,IACC,SAASf,EAAQoF,EAA0B9E,EAAqB,CAEvE,IAAIsF,EAAKtF,EAAoB,GAAG,EAC5BoF,EAAWpF,EAAoB,GAAG,EAWtC,SAASI,EAAOQ,EAAQH,EAAMuE,EAAU,CACpC,GAAI,CAACpE,GAAU,CAACH,GAAQ,CAACuE,EACrB,MAAM,IAAI,MAAM,4BAA4B,EAGhD,GAAI,CAACM,EAAG,OAAO7E,CAAI,EACf,MAAM,IAAI,UAAU,kCAAkC,EAG1D,GAAI,CAAC6E,EAAG,GAAGN,CAAQ,EACf,MAAM,IAAI,UAAU,mCAAmC,EAG3D,GAAIM,EAAG,KAAK1E,CAAM,EACd,OAAO2E,EAAW3E,EAAQH,EAAMuE,CAAQ,EAEvC,GAAIM,EAAG,SAAS1E,CAAM,EACvB,OAAO4E,EAAe5E,EAAQH,EAAMuE,CAAQ,EAE3C,GAAIM,EAAG,OAAO1E,CAAM,EACrB,OAAO6E,EAAe7E,EAAQH,EAAMuE,CAAQ,EAG5C,MAAM,IAAI,UAAU,2EAA2E,CAEvG,CAWA,SAASO,EAAWG,EAAMjF,EAAMuE,EAAU,CACtC,OAAAU,EAAK,iBAAiBjF,EAAMuE,CAAQ,EAE7B,CACH,QAAS,UAAW,CAChBU,EAAK,oBAAoBjF,EAAMuE,CAAQ,CAC3C,CACJ,CACJ,CAWA,SAASQ,EAAeG,EAAUlF,EAAMuE,EAAU,CAC9C,aAAM,UAAU,QAAQ,KAAKW,EAAU,SAASD,EAAM,CAClDA,EAAK,iBAAiBjF,EAAMuE,CAAQ,CACxC,CAAC,EAEM,CACH,QAAS,UAAW,CAChB,MAAM,UAAU,QAAQ,KAAKW,EAAU,SAASD,EAAM,CAClDA,EAAK,oBAAoBjF,EAAMuE,CAAQ,CAC3C,CAAC,CACL,CACJ,CACJ,CAWA,SAASS,EAAejB,EAAU/D,EAAMuE,EAAU,CAC9C,OAAOI,EAAS,SAAS,KAAMZ,EAAU/D,EAAMuE,CAAQ,CAC3D,CAEAtF,EAAO,QAAUU,CAGX,EAEA,IACC,SAASV,EAAQ,CAExB,SAASkG,EAAO5B,EAAS,CACrB,IAAInD,EAEJ,GAAImD,EAAQ,WAAa,SACrBA,EAAQ,MAAM,EAEdnD,EAAemD,EAAQ,cAElBA,EAAQ,WAAa,SAAWA,EAAQ,WAAa,WAAY,CACtE,IAAI6B,EAAa7B,EAAQ,aAAa,UAAU,EAE3C6B,GACD7B,EAAQ,aAAa,WAAY,EAAE,EAGvCA,EAAQ,OAAO,EACfA,EAAQ,kBAAkB,EAAGA,EAAQ,MAAM,MAAM,EAE5C6B,GACD7B,EAAQ,gBAAgB,UAAU,EAGtCnD,EAAemD,EAAQ,KAC3B,KACK,CACGA,EAAQ,aAAa,iBAAiB,GACtCA,EAAQ,MAAM,EAGlB,IAAI8B,EAAY,OAAO,aAAa,EAChCC,EAAQ,SAAS,YAAY,EAEjCA,EAAM,mBAAmB/B,CAAO,EAChC8B,EAAU,gBAAgB,EAC1BA,EAAU,SAASC,CAAK,EAExBlF,EAAeiF,EAAU,SAAS,CACtC,CAEA,OAAOjF,CACX,CAEAnB,EAAO,QAAUkG,CAGX,EAEA,IACC,SAASlG,EAAQ,CAExB,SAASsG,GAAK,CAGd,CAEAA,EAAE,UAAY,CACZ,GAAI,SAAUC,EAAMjB,EAAUkB,EAAK,CACjC,IAAIrC,EAAI,KAAK,IAAM,KAAK,EAAI,CAAC,GAE7B,OAACA,EAAEoC,CAAI,IAAMpC,EAAEoC,CAAI,EAAI,CAAC,IAAI,KAAK,CAC/B,GAAIjB,EACJ,IAAKkB,CACP,CAAC,EAEM,IACT,EAEA,KAAM,SAAUD,EAAMjB,EAAUkB,EAAK,CACnC,IAAIxC,EAAO,KACX,SAASyB,GAAY,CACnBzB,EAAK,IAAIuC,EAAMd,CAAQ,EACvBH,EAAS,MAAMkB,EAAK,SAAS,CAC/B,CAEA,OAAAf,EAAS,EAAIH,EACN,KAAK,GAAGiB,EAAMd,EAAUe,CAAG,CACpC,EAEA,KAAM,SAAUD,EAAM,CACpB,IAAIE,EAAO,CAAC,EAAE,MAAM,KAAK,UAAW,CAAC,EACjCC,IAAW,KAAK,IAAM,KAAK,EAAI,CAAC,IAAIH,CAAI,GAAK,CAAC,GAAG,MAAM,EACvD3D,EAAI,EACJ+D,EAAMD,EAAO,OAEjB,IAAK9D,EAAGA,EAAI+D,EAAK/D,IACf8D,EAAO9D,CAAC,EAAE,GAAG,MAAM8D,EAAO9D,CAAC,EAAE,IAAK6D,CAAI,EAGxC,OAAO,IACT,EAEA,IAAK,SAAUF,EAAMjB,EAAU,CAC7B,IAAInB,EAAI,KAAK,IAAM,KAAK,EAAI,CAAC,GACzByC,EAAOzC,EAAEoC,CAAI,EACbM,EAAa,CAAC,EAElB,GAAID,GAAQtB,EACV,QAAS1C,EAAI,EAAG+D,EAAMC,EAAK,OAAQhE,EAAI+D,EAAK/D,IACtCgE,EAAKhE,CAAC,EAAE,KAAO0C,GAAYsB,EAAKhE,CAAC,EAAE,GAAG,IAAM0C,GAC9CuB,EAAW,KAAKD,EAAKhE,CAAC,CAAC,EAQ7B,OAACiE,EAAW,OACR1C,EAAEoC,CAAI,EAAIM,EACV,OAAO1C,EAAEoC,CAAI,EAEV,IACT,CACF,EAEAvG,EAAO,QAAUsG,EACjBtG,EAAO,QAAQ,YAAcsG,CAGvB,CAEI,EAGIQ,EAA2B,CAAC,EAGhC,SAASxG,EAAoByG,EAAU,CAEtC,GAAGD,EAAyBC,CAAQ,EACnC,OAAOD,EAAyBC,CAAQ,EAAE,QAG3C,IAAI/G,EAAS8G,EAAyBC,CAAQ,EAAI,CAGjD,QAAS,CAAC,CACX,EAGA,OAAA5G,EAAoB4G,CAAQ,EAAE/G,EAAQA,EAAO,QAASM,CAAmB,EAGlEN,EAAO,OACf,CAIA,OAAC,UAAW,CAEXM,EAAoB,EAAI,SAASN,EAAQ,CACxC,IAAIgH,EAAShH,GAAUA,EAAO,WAC7B,UAAW,CAAE,OAAOA,EAAO,OAAY,EACvC,UAAW,CAAE,OAAOA,CAAQ,EAC7B,OAAAM,EAAoB,EAAE0G,EAAQ,CAAE,EAAGA,CAAO,CAAC,EACpCA,CACR,CACD,EAAE,EAGD,UAAW,CAEX1G,EAAoB,EAAI,SAASP,EAASkH,EAAY,CACrD,QAAQC,KAAOD,EACX3G,EAAoB,EAAE2G,EAAYC,CAAG,GAAK,CAAC5G,EAAoB,EAAEP,EAASmH,CAAG,GAC/E,OAAO,eAAenH,EAASmH,EAAK,CAAE,WAAY,GAAM,IAAKD,EAAWC,CAAG,CAAE,CAAC,CAGjF,CACD,EAAE,EAGD,UAAW,CACX5G,EAAoB,EAAI,SAASyB,EAAKoF,EAAM,CAAE,OAAO,OAAO,UAAU,eAAe,KAAKpF,EAAKoF,CAAI,CAAG,CACvG,EAAE,EAMK7G,EAAoB,GAAG,CAC/B,EAAG,EACX,OACD,CAAC,ICz3BD,IAAA8G,GAAAC,GAAA,CAAAC,GAAAC,KAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,GAeA,IAAIC,GAAkB,UAOtBD,GAAO,QAAUE,GAUjB,SAASA,GAAWC,EAAQ,CAC1B,IAAIC,EAAM,GAAKD,EACXE,EAAQJ,GAAgB,KAAKG,CAAG,EAEpC,GAAI,CAACC,EACH,OAAOD,EAGT,IAAIE,EACAC,EAAO,GACPC,EAAQ,EACRC,EAAY,EAEhB,IAAKD,EAAQH,EAAM,MAAOG,EAAQJ,EAAI,OAAQI,IAAS,CACrD,OAAQJ,EAAI,WAAWI,CAAK,EAAG,CAC7B,IAAK,IACHF,EAAS,SACT,MACF,IAAK,IACHA,EAAS,QACT,MACF,IAAK,IACHA,EAAS,QACT,MACF,IAAK,IACHA,EAAS,OACT,MACF,IAAK,IACHA,EAAS,OACT,MACF,QACE,QACJ,CAEIG,IAAcD,IAChBD,GAAQH,EAAI,UAAUK,EAAWD,CAAK,GAGxCC,EAAYD,EAAQ,EACpBD,GAAQD,CACV,CAEA,OAAOG,IAAcD,EACjBD,EAAOH,EAAI,UAAUK,EAAWD,CAAK,EACrCD,CACN,ICvDA,IAAAG,GAAO,SCtBP;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,gFAgBA,IAAIC,GAAgB,SAASC,EAAGC,EAAG,CAC/B,OAAAF,GAAgB,OAAO,gBAClB,CAAE,UAAW,CAAC,CAAE,YAAa,OAAS,SAAUC,EAAGC,EAAG,CAAED,EAAE,UAAYC,CAAG,GAC1E,SAAUD,EAAGC,EAAG,CAAE,QAASC,KAAKD,EAAO,OAAO,UAAU,eAAe,KAAKA,EAAGC,CAAC,IAAGF,EAAEE,CAAC,EAAID,EAAEC,CAAC,EAAG,EAC7FH,GAAcC,EAAGC,CAAC,CAC7B,EAEO,SAASE,GAAUH,EAAGC,EAAG,CAC5B,GAAI,OAAOA,GAAM,YAAcA,IAAM,KACjC,MAAM,IAAI,UAAU,uBAAyB,OAAOA,CAAC,EAAI,+BAA+B,EAC5FF,GAAcC,EAAGC,CAAC,EAClB,SAASG,GAAK,CAAE,KAAK,YAAcJ,CAAG,CACtCA,EAAE,UAAYC,IAAM,KAAO,OAAO,OAAOA,CAAC,GAAKG,EAAG,UAAYH,EAAE,UAAW,IAAIG,EACnF,CAwCO,SAASC,GAAUC,EAASC,EAAYC,EAAGC,EAAW,CACzD,SAASC,EAAMC,EAAO,CAAE,OAAOA,aAAiBH,EAAIG,EAAQ,IAAIH,EAAE,SAAUI,EAAS,CAAEA,EAAQD,CAAK,CAAG,CAAC,CAAG,CAC3G,OAAO,IAAKH,IAAMA,EAAI,UAAU,SAAUI,EAASC,EAAQ,CACvD,SAASC,EAAUH,EAAO,CAAE,GAAI,CAAEI,EAAKN,EAAU,KAAKE,CAAK,CAAC,CAAG,OAASK,EAAG,CAAEH,EAAOG,CAAC,CAAG,CAAE,CAC1F,SAASC,EAASN,EAAO,CAAE,GAAI,CAAEI,EAAKN,EAAU,MAASE,CAAK,CAAC,CAAG,OAASK,EAAG,CAAEH,EAAOG,CAAC,CAAG,CAAE,CAC7F,SAASD,EAAKG,EAAQ,CAAEA,EAAO,KAAON,EAAQM,EAAO,KAAK,EAAIR,EAAMQ,EAAO,KAAK,EAAE,KAAKJ,EAAWG,CAAQ,CAAG,CAC7GF,GAAMN,EAAYA,EAAU,MAAMH,EAASC,GAAc,CAAC,CAAC,GAAG,KAAK,CAAC,CACxE,CAAC,CACL,CAEO,SAASY,GAAYb,EAASc,EAAM,CACvC,IAAIC,EAAI,CAAE,MAAO,EAAG,KAAM,UAAW,CAAE,GAAIC,EAAE,CAAC,EAAI,EAAG,MAAMA,EAAE,CAAC,EAAG,OAAOA,EAAE,CAAC,CAAG,EAAG,KAAM,CAAC,EAAG,IAAK,CAAC,CAAE,EAAGC,EAAGC,EAAGF,EAAGG,EAC/G,OAAOA,EAAI,CAAE,KAAMC,EAAK,CAAC,EAAG,MAASA,EAAK,CAAC,EAAG,OAAUA,EAAK,CAAC,CAAE,EAAG,OAAO,QAAW,aAAeD,EAAE,OAAO,QAAQ,EAAI,UAAW,CAAE,OAAO,IAAM,GAAIA,EACvJ,SAASC,EAAKC,EAAG,CAAE,OAAO,SAAUC,EAAG,CAAE,OAAOb,EAAK,CAACY,EAAGC,CAAC,CAAC,CAAG,CAAG,CACjE,SAASb,EAAKc,EAAI,CACd,GAAIN,EAAG,MAAM,IAAI,UAAU,iCAAiC,EAC5D,KAAOF,GAAG,GAAI,CACV,GAAIE,EAAI,EAAGC,IAAMF,EAAIO,EAAG,CAAC,EAAI,EAAIL,EAAE,OAAYK,EAAG,CAAC,EAAIL,EAAE,SAAcF,EAAIE,EAAE,SAAcF,EAAE,KAAKE,CAAC,EAAG,GAAKA,EAAE,OAAS,EAAEF,EAAIA,EAAE,KAAKE,EAAGK,EAAG,CAAC,CAAC,GAAG,KAAM,OAAOP,EAE3J,OADIE,EAAI,EAAGF,IAAGO,EAAK,CAACA,EAAG,CAAC,EAAI,EAAGP,EAAE,KAAK,GAC9BO,EAAG,CAAC,EAAG,CACX,IAAK,GAAG,IAAK,GAAGP,EAAIO,EAAI,MACxB,IAAK,GAAG,OAAAR,EAAE,QAAgB,CAAE,MAAOQ,EAAG,CAAC,EAAG,KAAM,EAAM,EACtD,IAAK,GAAGR,EAAE,QAASG,EAAIK,EAAG,CAAC,EAAGA,EAAK,CAAC,CAAC,EAAG,SACxC,IAAK,GAAGA,EAAKR,EAAE,IAAI,IAAI,EAAGA,EAAE,KAAK,IAAI,EAAG,SACxC,QACI,GAAMC,EAAID,EAAE,KAAM,EAAAC,EAAIA,EAAE,OAAS,GAAKA,EAAEA,EAAE,OAAS,CAAC,KAAOO,EAAG,CAAC,IAAM,GAAKA,EAAG,CAAC,IAAM,GAAI,CAAER,EAAI,EAAG,QAAU,CAC3G,GAAIQ,EAAG,CAAC,IAAM,IAAM,CAACP,GAAMO,EAAG,CAAC,EAAIP,EAAE,CAAC,GAAKO,EAAG,CAAC,EAAIP,EAAE,CAAC,GAAK,CAAED,EAAE,MAAQQ,EAAG,CAAC,EAAG,KAAO,CACrF,GAAIA,EAAG,CAAC,IAAM,GAAKR,EAAE,MAAQC,EAAE,CAAC,EAAG,CAAED,EAAE,MAAQC,EAAE,CAAC,EAAGA,EAAIO,EAAI,KAAO,CACpE,GAAIP,GAAKD,EAAE,MAAQC,EAAE,CAAC,EAAG,CAAED,EAAE,MAAQC,EAAE,CAAC,EAAGD,EAAE,IAAI,KAAKQ,CAAE,EAAG,KAAO,CAC9DP,EAAE,CAAC,GAAGD,EAAE,IAAI,IAAI,EACpBA,EAAE,KAAK,IAAI,EAAG,QACtB,CACAQ,EAAKT,EAAK,KAAKd,EAASe,CAAC,CAC7B,OAASL,EAAG,CAAEa,EAAK,CAAC,EAAGb,CAAC,EAAGQ,EAAI,CAAG,QAAE,CAAUD,EAAID,EAAI,CAAG,CACzD,GAAIO,EAAG,CAAC,EAAI,EAAG,MAAMA,EAAG,CAAC,EAAG,MAAO,CAAE,MAAOA,EAAG,CAAC,EAAIA,EAAG,CAAC,EAAI,OAAQ,KAAM,EAAK,CACnF,CACJ,CAcO,SAASC,GAASC,EAAG,CACxB,IAAIC,EAAI,OAAO,QAAW,YAAc,OAAO,SAAUC,EAAID,GAAKD,EAAEC,CAAC,EAAGE,EAAI,EAC5E,GAAID,EAAG,OAAOA,EAAE,KAAKF,CAAC,EACtB,GAAIA,GAAK,OAAOA,EAAE,QAAW,SAAU,MAAO,CAC1C,KAAM,UAAY,CACd,OAAIA,GAAKG,GAAKH,EAAE,SAAQA,EAAI,QACrB,CAAE,MAAOA,GAAKA,EAAEG,GAAG,EAAG,KAAM,CAACH,CAAE,CAC1C,CACJ,EACA,MAAM,IAAI,UAAUC,EAAI,0BAA4B,iCAAiC,CACzF,CAEO,SAASG,EAAOJ,EAAGK,EAAG,CACzB,IAAIH,EAAI,OAAO,QAAW,YAAcF,EAAE,OAAO,QAAQ,EACzD,GAAI,CAACE,EAAG,OAAOF,EACf,IAAIG,EAAID,EAAE,KAAKF,CAAC,EAAGM,EAAGC,EAAK,CAAC,EAAGC,EAC/B,GAAI,CACA,MAAQH,IAAM,QAAUA,KAAM,IAAM,EAAEC,EAAIH,EAAE,KAAK,GAAG,MAAMI,EAAG,KAAKD,EAAE,KAAK,CAC7E,OACOG,EAAO,CAAED,EAAI,CAAE,MAAOC,CAAM,CAAG,QACtC,CACI,GAAI,CACIH,GAAK,CAACA,EAAE,OAASJ,EAAIC,EAAE,SAAYD,EAAE,KAAKC,CAAC,CACnD,QACA,CAAU,GAAIK,EAAG,MAAMA,EAAE,KAAO,CACpC,CACA,OAAOD,CACX,CAkBO,SAASG,EAAcC,EAAIC,EAAMC,EAAM,CAC1C,GAAIA,GAAQ,UAAU,SAAW,EAAG,QAASC,EAAI,EAAGC,EAAIH,EAAK,OAAQI,EAAIF,EAAIC,EAAGD,KACxEE,GAAM,EAAEF,KAAKF,MACRI,IAAIA,EAAK,MAAM,UAAU,MAAM,KAAKJ,EAAM,EAAGE,CAAC,GACnDE,EAAGF,CAAC,EAAIF,EAAKE,CAAC,GAGtB,OAAOH,EAAG,OAAOK,GAAM,MAAM,UAAU,MAAM,KAAKJ,CAAI,CAAC,CAC3D,CAEO,SAASK,GAAQC,EAAG,CACvB,OAAO,gBAAgBD,IAAW,KAAK,EAAIC,EAAG,MAAQ,IAAID,GAAQC,CAAC,CACvE,CAEO,SAASC,GAAiBC,EAASC,EAAYC,EAAW,CAC7D,GAAI,CAAC,OAAO,cAAe,MAAM,IAAI,UAAU,sCAAsC,EACrF,IAAIC,EAAID,EAAU,MAAMF,EAASC,GAAc,CAAC,CAAC,EAAGP,EAAGU,EAAI,CAAC,EAC5D,OAAOV,EAAI,CAAC,EAAGW,EAAK,MAAM,EAAGA,EAAK,OAAO,EAAGA,EAAK,QAAQ,EAAGX,EAAE,OAAO,aAAa,EAAI,UAAY,CAAE,OAAO,IAAM,EAAGA,EACpH,SAASW,EAAKC,EAAG,CAAMH,EAAEG,CAAC,IAAGZ,EAAEY,CAAC,EAAI,SAAUR,EAAG,CAAE,OAAO,IAAI,QAAQ,SAAUS,EAAG,EAAG,CAAEH,EAAE,KAAK,CAACE,EAAGR,EAAGS,EAAG,CAAC,CAAC,EAAI,GAAKC,EAAOF,EAAGR,CAAC,CAAG,CAAC,CAAG,EAAG,CACzI,SAASU,EAAOF,EAAGR,EAAG,CAAE,GAAI,CAAEW,EAAKN,EAAEG,CAAC,EAAER,CAAC,CAAC,CAAG,OAASY,EAAG,CAAEC,EAAOP,EAAE,CAAC,EAAE,CAAC,EAAGM,CAAC,CAAG,CAAE,CACjF,SAASD,EAAKG,EAAG,CAAEA,EAAE,iBAAiBf,GAAU,QAAQ,QAAQe,EAAE,MAAM,CAAC,EAAE,KAAKC,EAASC,CAAM,EAAIH,EAAOP,EAAE,CAAC,EAAE,CAAC,EAAGQ,CAAC,CAAG,CACvH,SAASC,EAAQE,EAAO,CAAEP,EAAO,OAAQO,CAAK,CAAG,CACjD,SAASD,EAAOC,EAAO,CAAEP,EAAO,QAASO,CAAK,CAAG,CACjD,SAASJ,EAAOK,EAAGlB,EAAG,CAAMkB,EAAElB,CAAC,EAAGM,EAAE,MAAM,EAAGA,EAAE,QAAQI,EAAOJ,EAAE,CAAC,EAAE,CAAC,EAAGA,EAAE,CAAC,EAAE,CAAC,CAAC,CAAG,CACrF,CAQO,SAASa,GAAcC,EAAG,CAC7B,GAAI,CAAC,OAAO,cAAe,MAAM,IAAI,UAAU,sCAAsC,EACrF,IAAIC,EAAID,EAAE,OAAO,aAAa,EAAGE,EACjC,OAAOD,EAAIA,EAAE,KAAKD,CAAC,GAAKA,EAAI,OAAOG,IAAa,WAAaA,GAASH,CAAC,EAAIA,EAAE,OAAO,QAAQ,EAAE,EAAGE,EAAI,CAAC,EAAGE,EAAK,MAAM,EAAGA,EAAK,OAAO,EAAGA,EAAK,QAAQ,EAAGF,EAAE,OAAO,aAAa,EAAI,UAAY,CAAE,OAAO,IAAM,EAAGA,GAC9M,SAASE,EAAKC,EAAG,CAAEH,EAAEG,CAAC,EAAIL,EAAEK,CAAC,GAAK,SAAUC,EAAG,CAAE,OAAO,IAAI,QAAQ,SAAUC,EAASC,EAAQ,CAAEF,EAAIN,EAAEK,CAAC,EAAEC,CAAC,EAAGG,EAAOF,EAASC,EAAQF,EAAE,KAAMA,EAAE,KAAK,CAAG,CAAC,CAAG,CAAG,CAC/J,SAASG,EAAOF,EAASC,EAAQE,EAAGJ,EAAG,CAAE,QAAQ,QAAQA,CAAC,EAAE,KAAK,SAASA,EAAG,CAAEC,EAAQ,CAAE,MAAOD,EAAG,KAAMI,CAAE,CAAC,CAAG,EAAGF,CAAM,CAAG,CAC/H,CCtMM,SAAUG,EAAWC,EAAU,CACnC,OAAO,OAAOA,GAAU,UAC1B,CCGM,SAAUC,GAAoBC,EAAgC,CAClE,IAAMC,EAAS,SAACC,EAAa,CAC3B,MAAM,KAAKA,CAAQ,EACnBA,EAAS,MAAQ,IAAI,MAAK,EAAG,KAC/B,EAEMC,EAAWH,EAAWC,CAAM,EAClC,OAAAE,EAAS,UAAY,OAAO,OAAO,MAAM,SAAS,EAClDA,EAAS,UAAU,YAAcA,EAC1BA,CACT,CCDO,IAAMC,GAA+CC,GAC1D,SAACC,EAAM,CACL,OAAA,SAA4CC,EAA0B,CACpED,EAAO,IAAI,EACX,KAAK,QAAUC,EACRA,EAAO,OAAM;EACxBA,EAAO,IAAI,SAACC,EAAKC,EAAC,CAAK,OAAGA,EAAI,EAAC,KAAKD,EAAI,SAAQ,CAAzB,CAA6B,EAAE,KAAK;GAAM,EACzD,GACJ,KAAK,KAAO,sBACZ,KAAK,OAASD,CAChB,CARA,CAQC,ECvBC,SAAUG,GAAaC,EAA6BC,EAAO,CAC/D,GAAID,EAAK,CACP,IAAME,EAAQF,EAAI,QAAQC,CAAI,EAC9B,GAAKC,GAASF,EAAI,OAAOE,EAAO,CAAC,EAErC,CCOA,IAAAC,GAAA,UAAA,CAyBE,SAAAA,EAAoBC,EAA4B,CAA5B,KAAA,gBAAAA,EAdb,KAAA,OAAS,GAER,KAAA,WAAmD,KAMnD,KAAA,YAAqD,IAMV,CAQnD,OAAAD,EAAA,UAAA,YAAA,UAAA,aACME,EAEJ,GAAI,CAAC,KAAK,OAAQ,CAChB,KAAK,OAAS,GAGN,IAAAC,EAAe,KAAI,WAC3B,GAAIA,EAEF,GADA,KAAK,WAAa,KACd,MAAM,QAAQA,CAAU,MAC1B,QAAqBC,EAAAC,GAAAF,CAAU,EAAAG,EAAAF,EAAA,KAAA,EAAA,CAAAE,EAAA,KAAAA,EAAAF,EAAA,KAAA,EAAE,CAA5B,IAAMG,EAAMD,EAAA,MACfC,EAAO,OAAO,IAAI,yGAGpBJ,EAAW,OAAO,IAAI,EAIlB,IAAiBK,EAAqB,KAAI,gBAClD,GAAIC,EAAWD,CAAgB,EAC7B,GAAI,CACFA,EAAgB,QACTE,EAAG,CACVR,EAASQ,aAAaC,GAAsBD,EAAE,OAAS,CAACA,CAAC,EAIrD,IAAAE,EAAgB,KAAI,YAC5B,GAAIA,EAAa,CACf,KAAK,YAAc,SACnB,QAAwBC,EAAAR,GAAAO,CAAW,EAAAE,EAAAD,EAAA,KAAA,EAAA,CAAAC,EAAA,KAAAA,EAAAD,EAAA,KAAA,EAAE,CAAhC,IAAME,EAASD,EAAA,MAClB,GAAI,CACFE,GAAcD,CAAS,QAChBE,EAAK,CACZf,EAASA,GAAM,KAANA,EAAU,CAAA,EACfe,aAAeN,GACjBT,EAAMgB,EAAAA,EAAA,CAAA,EAAAC,EAAOjB,CAAM,CAAA,EAAAiB,EAAKF,EAAI,MAAM,CAAA,EAElCf,EAAO,KAAKe,CAAG,sGAMvB,GAAIf,EACF,MAAM,IAAIS,GAAoBT,CAAM,EAG1C,EAoBAF,EAAA,UAAA,IAAA,SAAIoB,EAAuB,OAGzB,GAAIA,GAAYA,IAAa,KAC3B,GAAI,KAAK,OAGPJ,GAAcI,CAAQ,MACjB,CACL,GAAIA,aAAoBpB,EAAc,CAGpC,GAAIoB,EAAS,QAAUA,EAAS,WAAW,IAAI,EAC7C,OAEFA,EAAS,WAAW,IAAI,GAEzB,KAAK,aAAcC,EAAA,KAAK,eAAW,MAAAA,IAAA,OAAAA,EAAI,CAAA,GAAI,KAAKD,CAAQ,EAG/D,EAOQpB,EAAA,UAAA,WAAR,SAAmBsB,EAAoB,CAC7B,IAAAnB,EAAe,KAAI,WAC3B,OAAOA,IAAemB,GAAW,MAAM,QAAQnB,CAAU,GAAKA,EAAW,SAASmB,CAAM,CAC1F,EASQtB,EAAA,UAAA,WAAR,SAAmBsB,EAAoB,CAC7B,IAAAnB,EAAe,KAAI,WAC3B,KAAK,WAAa,MAAM,QAAQA,CAAU,GAAKA,EAAW,KAAKmB,CAAM,EAAGnB,GAAcA,EAAa,CAACA,EAAYmB,CAAM,EAAIA,CAC5H,EAMQtB,EAAA,UAAA,cAAR,SAAsBsB,EAAoB,CAChC,IAAAnB,EAAe,KAAI,WACvBA,IAAemB,EACjB,KAAK,WAAa,KACT,MAAM,QAAQnB,CAAU,GACjCoB,GAAUpB,EAAYmB,CAAM,CAEhC,EAgBAtB,EAAA,UAAA,OAAA,SAAOoB,EAAsC,CACnC,IAAAR,EAAgB,KAAI,YAC5BA,GAAeW,GAAUX,EAAaQ,CAAQ,EAE1CA,aAAoBpB,GACtBoB,EAAS,cAAc,IAAI,CAE/B,EAlLcpB,EAAA,MAAS,UAAA,CACrB,IAAMwB,EAAQ,IAAIxB,EAClB,OAAAwB,EAAM,OAAS,GACRA,CACT,EAAE,EA+KJxB,GArLA,EAuLO,IAAMyB,GAAqBC,GAAa,MAEzC,SAAUC,GAAeC,EAAU,CACvC,OACEA,aAAiBF,IAChBE,GAAS,WAAYA,GAASC,EAAWD,EAAM,MAAM,GAAKC,EAAWD,EAAM,GAAG,GAAKC,EAAWD,EAAM,WAAW,CAEpH,CAEA,SAASE,GAAcC,EAAwC,CACzDF,EAAWE,CAAS,EACtBA,EAAS,EAETA,EAAU,YAAW,CAEzB,CChNO,IAAMC,GAAuB,CAClC,iBAAkB,KAClB,sBAAuB,KACvB,QAAS,OACT,sCAAuC,GACvC,yBAA0B,ICGrB,IAAMC,GAAmC,CAG9C,WAAA,SAAWC,EAAqBC,EAAgB,SAAEC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,EAAA,CAAA,EAAA,UAAAA,CAAA,EACxC,IAAAC,EAAaL,GAAe,SACpC,OAAIK,GAAQ,MAARA,EAAU,WACLA,EAAS,WAAU,MAAnBA,EAAQC,EAAA,CAAYL,EAASC,CAAO,EAAAK,EAAKJ,CAAI,CAAA,CAAA,EAE/C,WAAU,MAAA,OAAAG,EAAA,CAACL,EAASC,CAAO,EAAAK,EAAKJ,CAAI,CAAA,CAAA,CAC7C,EACA,aAAA,SAAaK,EAAM,CACT,IAAAH,EAAaL,GAAe,SACpC,QAAQK,GAAQ,KAAA,OAARA,EAAU,eAAgB,cAAcG,CAAa,CAC/D,EACA,SAAU,QCjBN,SAAUC,GAAqBC,EAAQ,CAC3CC,GAAgB,WAAW,UAAA,CACjB,IAAAC,EAAqBC,GAAM,iBACnC,GAAID,EAEFA,EAAiBF,CAAG,MAGpB,OAAMA,CAEV,CAAC,CACH,CCtBM,SAAUI,IAAI,CAAK,CCMlB,IAAMC,GAAyB,UAAA,CAAM,OAAAC,GAAmB,IAAK,OAAW,MAAS,CAA5C,EAAsE,EAO5G,SAAUC,GAAkBC,EAAU,CAC1C,OAAOF,GAAmB,IAAK,OAAWE,CAAK,CACjD,CAOM,SAAUC,GAAoBC,EAAQ,CAC1C,OAAOJ,GAAmB,IAAKI,EAAO,MAAS,CACjD,CAQM,SAAUJ,GAAmBK,EAAuBD,EAAYF,EAAU,CAC9E,MAAO,CACL,KAAIG,EACJ,MAAKD,EACL,MAAKF,EAET,CCrCA,IAAII,GAAuD,KASrD,SAAUC,GAAaC,EAAc,CACzC,GAAIC,GAAO,sCAAuC,CAChD,IAAMC,EAAS,CAACJ,GAKhB,GAJII,IACFJ,GAAU,CAAE,YAAa,GAAO,MAAO,IAAI,GAE7CE,EAAE,EACEE,EAAQ,CACJ,IAAAC,EAAyBL,GAAvBM,EAAWD,EAAA,YAAEE,EAAKF,EAAA,MAE1B,GADAL,GAAU,KACNM,EACF,MAAMC,QAMVL,EAAE,CAEN,CAMM,SAAUM,GAAaC,EAAQ,CAC/BN,GAAO,uCAAyCH,KAClDA,GAAQ,YAAc,GACtBA,GAAQ,MAAQS,EAEpB,CCrBA,IAAAC,GAAA,SAAAC,EAAA,CAAmCC,GAAAF,EAAAC,CAAA,EA6BjC,SAAAD,EAAYG,EAA6C,CAAzD,IAAAC,EACEH,EAAA,KAAA,IAAA,GAAO,KATC,OAAAG,EAAA,UAAqB,GAUzBD,GACFC,EAAK,YAAcD,EAGfE,GAAeF,CAAW,GAC5BA,EAAY,IAAIC,CAAI,GAGtBA,EAAK,YAAcE,IAEvB,CAzBO,OAAAN,EAAA,OAAP,SAAiBO,EAAwBC,EAA2BC,EAAqB,CACvF,OAAO,IAAIC,GAAeH,EAAMC,EAAOC,CAAQ,CACjD,EAgCAT,EAAA,UAAA,KAAA,SAAKW,EAAS,CACR,KAAK,UACPC,GAA0BC,GAAiBF,CAAK,EAAG,IAAI,EAEvD,KAAK,MAAMA,CAAM,CAErB,EASAX,EAAA,UAAA,MAAA,SAAMc,EAAS,CACT,KAAK,UACPF,GAA0BG,GAAkBD,CAAG,EAAG,IAAI,GAEtD,KAAK,UAAY,GACjB,KAAK,OAAOA,CAAG,EAEnB,EAQAd,EAAA,UAAA,SAAA,UAAA,CACM,KAAK,UACPY,GAA0BI,GAAuB,IAAI,GAErD,KAAK,UAAY,GACjB,KAAK,UAAS,EAElB,EAEAhB,EAAA,UAAA,YAAA,UAAA,CACO,KAAK,SACR,KAAK,UAAY,GACjBC,EAAA,UAAM,YAAW,KAAA,IAAA,EACjB,KAAK,YAAc,KAEvB,EAEUD,EAAA,UAAA,MAAV,SAAgBW,EAAQ,CACtB,KAAK,YAAY,KAAKA,CAAK,CAC7B,EAEUX,EAAA,UAAA,OAAV,SAAiBc,EAAQ,CACvB,GAAI,CACF,KAAK,YAAY,MAAMA,CAAG,UAE1B,KAAK,YAAW,EAEpB,EAEUd,EAAA,UAAA,UAAV,UAAA,CACE,GAAI,CACF,KAAK,YAAY,SAAQ,UAEzB,KAAK,YAAW,EAEpB,EACFA,CAAA,EApHmCiB,EAAY,EA2H/C,IAAMC,GAAQ,SAAS,UAAU,KAEjC,SAASC,GAAyCC,EAAQC,EAAY,CACpE,OAAOH,GAAM,KAAKE,EAAIC,CAAO,CAC/B,CAMA,IAAAC,GAAA,UAAA,CACE,SAAAA,EAAoBC,EAAqC,CAArC,KAAA,gBAAAA,CAAwC,CAE5D,OAAAD,EAAA,UAAA,KAAA,SAAKE,EAAQ,CACH,IAAAD,EAAoB,KAAI,gBAChC,GAAIA,EAAgB,KAClB,GAAI,CACFA,EAAgB,KAAKC,CAAK,QACnBC,EAAO,CACdC,GAAqBD,CAAK,EAGhC,EAEAH,EAAA,UAAA,MAAA,SAAMK,EAAQ,CACJ,IAAAJ,EAAoB,KAAI,gBAChC,GAAIA,EAAgB,MAClB,GAAI,CACFA,EAAgB,MAAMI,CAAG,QAClBF,EAAO,CACdC,GAAqBD,CAAK,OAG5BC,GAAqBC,CAAG,CAE5B,EAEAL,EAAA,UAAA,SAAA,UAAA,CACU,IAAAC,EAAoB,KAAI,gBAChC,GAAIA,EAAgB,SAClB,GAAI,CACFA,EAAgB,SAAQ,QACjBE,EAAO,CACdC,GAAqBD,CAAK,EAGhC,EACFH,CAAA,EArCA,EAuCAM,GAAA,SAAAC,EAAA,CAAuCC,GAAAF,EAAAC,CAAA,EACrC,SAAAD,EACEG,EACAN,EACAO,EAA8B,CAHhC,IAAAC,EAKEJ,EAAA,KAAA,IAAA,GAAO,KAEHN,EACJ,GAAIW,EAAWH,CAAc,GAAK,CAACA,EAGjCR,EAAkB,CAChB,KAAOQ,GAAc,KAAdA,EAAkB,OACzB,MAAON,GAAK,KAALA,EAAS,OAChB,SAAUO,GAAQ,KAARA,EAAY,YAEnB,CAEL,IAAIG,EACAF,GAAQG,GAAO,0BAIjBD,EAAU,OAAO,OAAOJ,CAAc,EACtCI,EAAQ,YAAc,UAAA,CAAM,OAAAF,EAAK,YAAW,CAAhB,EAC5BV,EAAkB,CAChB,KAAMQ,EAAe,MAAQZ,GAAKY,EAAe,KAAMI,CAAO,EAC9D,MAAOJ,EAAe,OAASZ,GAAKY,EAAe,MAAOI,CAAO,EACjE,SAAUJ,EAAe,UAAYZ,GAAKY,EAAe,SAAUI,CAAO,IAI5EZ,EAAkBQ,EAMtB,OAAAE,EAAK,YAAc,IAAIX,GAAiBC,CAAe,GACzD,CACF,OAAAK,CAAA,EAzCuCS,EAAU,EA2CjD,SAASC,GAAqBC,EAAU,CAClCC,GAAO,sCACTC,GAAaF,CAAK,EAIlBG,GAAqBH,CAAK,CAE9B,CAQA,SAASI,GAAoBC,EAAQ,CACnC,MAAMA,CACR,CAOA,SAASC,GAA0BC,EAA2CC,EAA2B,CAC/F,IAAAC,EAA0BR,GAAM,sBACxCQ,GAAyBC,GAAgB,WAAW,UAAA,CAAM,OAAAD,EAAsBF,EAAcC,CAAU,CAA9C,CAA+C,CAC3G,CAOO,IAAMG,GAA6D,CACxE,OAAQ,GACR,KAAMC,GACN,MAAOR,GACP,SAAUQ,IC5QL,IAAMC,GAA+B,UAAA,CAAM,OAAC,OAAO,QAAW,YAAc,OAAO,YAAe,cAAvD,EAAsE,ECoClH,SAAUC,GAAYC,EAAI,CAC9B,OAAOA,CACT,CCiCM,SAAUC,IAAI,SAACC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACnB,OAAOC,GAAcF,CAAG,CAC1B,CAGM,SAAUE,GAAoBF,EAA+B,CACjE,OAAIA,EAAI,SAAW,EACVG,GAGLH,EAAI,SAAW,EACVA,EAAI,CAAC,EAGP,SAAeI,EAAQ,CAC5B,OAAOJ,EAAI,OAAO,SAACK,EAAWC,EAAuB,CAAK,OAAAA,EAAGD,CAAI,CAAP,EAAUD,CAAY,CAClF,CACF,CC9EA,IAAAG,EAAA,UAAA,CAkBE,SAAAA,EAAYC,EAA6E,CACnFA,IACF,KAAK,WAAaA,EAEtB,CA4BA,OAAAD,EAAA,UAAA,KAAA,SAAQE,EAAyB,CAC/B,IAAMC,EAAa,IAAIH,EACvB,OAAAG,EAAW,OAAS,KACpBA,EAAW,SAAWD,EACfC,CACT,EA6IAH,EAAA,UAAA,UAAA,SACEI,EACAC,EACAC,EAA8B,CAHhC,IAAAC,EAAA,KAKQC,EAAaC,GAAaL,CAAc,EAAIA,EAAiB,IAAIM,GAAeN,EAAgBC,EAAOC,CAAQ,EAErH,OAAAK,GAAa,UAAA,CACL,IAAAC,EAAuBL,EAArBL,EAAQU,EAAA,SAAEC,EAAMD,EAAA,OACxBJ,EAAW,IACTN,EAGIA,EAAS,KAAKM,EAAYK,CAAM,EAChCA,EAIAN,EAAK,WAAWC,CAAU,EAG1BD,EAAK,cAAcC,CAAU,CAAC,CAEtC,CAAC,EAEMA,CACT,EAGUR,EAAA,UAAA,cAAV,SAAwBc,EAAmB,CACzC,GAAI,CACF,OAAO,KAAK,WAAWA,CAAI,QACpBC,EAAK,CAIZD,EAAK,MAAMC,CAAG,EAElB,EA6DAf,EAAA,UAAA,QAAA,SAAQgB,EAA0BC,EAAoC,CAAtE,IAAAV,EAAA,KACE,OAAAU,EAAcC,GAAeD,CAAW,EAEjC,IAAIA,EAAkB,SAACE,EAASC,EAAM,CAC3C,IAAMZ,EAAa,IAAIE,GAAkB,CACvC,KAAM,SAACW,EAAK,CACV,GAAI,CACFL,EAAKK,CAAK,QACHN,EAAK,CACZK,EAAOL,CAAG,EACVP,EAAW,YAAW,EAE1B,EACA,MAAOY,EACP,SAAUD,EACX,EACDZ,EAAK,UAAUC,CAAU,CAC3B,CAAC,CACH,EAGUR,EAAA,UAAA,WAAV,SAAqBQ,EAA2B,OAC9C,OAAOI,EAAA,KAAK,UAAM,MAAAA,IAAA,OAAA,OAAAA,EAAE,UAAUJ,CAAU,CAC1C,EAOAR,EAAA,UAACG,EAAiB,EAAlB,UAAA,CACE,OAAO,IACT,EA4FAH,EAAA,UAAA,KAAA,UAAA,SAAKsB,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACH,OAAOC,GAAcF,CAAU,EAAE,IAAI,CACvC,EA6BAtB,EAAA,UAAA,UAAA,SAAUiB,EAAoC,CAA9C,IAAAV,EAAA,KACE,OAAAU,EAAcC,GAAeD,CAAW,EAEjC,IAAIA,EAAY,SAACE,EAASC,EAAM,CACrC,IAAIC,EACJd,EAAK,UACH,SAACkB,EAAI,CAAK,OAACJ,EAAQI,CAAT,EACV,SAACV,EAAQ,CAAK,OAAAK,EAAOL,CAAG,CAAV,EACd,UAAA,CAAM,OAAAI,EAAQE,CAAK,CAAb,CAAc,CAExB,CAAC,CACH,EA1aOrB,EAAA,OAAkC,SAAIC,EAAwD,CACnG,OAAO,IAAID,EAAcC,CAAS,CACpC,EAyaFD,GA9cA,EAudA,SAAS0B,GAAeC,EAA+C,OACrE,OAAOC,EAAAD,GAAW,KAAXA,EAAeE,GAAO,WAAO,MAAAD,IAAA,OAAAA,EAAI,OAC1C,CAEA,SAASE,GAAcC,EAAU,CAC/B,OAAOA,GAASC,EAAWD,EAAM,IAAI,GAAKC,EAAWD,EAAM,KAAK,GAAKC,EAAWD,EAAM,QAAQ,CAChG,CAEA,SAASE,GAAgBF,EAAU,CACjC,OAAQA,GAASA,aAAiBG,IAAgBJ,GAAWC,CAAK,GAAKI,GAAeJ,CAAK,CAC7F,CCzeM,SAAUK,GAAQC,EAAW,CACjC,OAAOC,EAAWD,GAAM,KAAA,OAANA,EAAQ,IAAI,CAChC,CAMM,SAAUE,EACdC,EAAqF,CAErF,OAAO,SAACH,EAAqB,CAC3B,GAAID,GAAQC,CAAM,EAChB,OAAOA,EAAO,KAAK,SAA+BI,EAA2B,CAC3E,GAAI,CACF,OAAOD,EAAKC,EAAc,IAAI,QACvBC,EAAK,CACZ,KAAK,MAAMA,CAAG,EAElB,CAAC,EAEH,MAAM,IAAI,UAAU,wCAAwC,CAC9D,CACF,CCjBM,SAAUC,EACdC,EACAC,EACAC,EACAC,EACAC,EAAuB,CAEvB,OAAO,IAAIC,GAAmBL,EAAaC,EAAQC,EAAYC,EAASC,CAAU,CACpF,CAMA,IAAAC,GAAA,SAAAC,EAAA,CAA2CC,GAAAF,EAAAC,CAAA,EAiBzC,SAAAD,EACEL,EACAC,EACAC,EACAC,EACQC,EACAI,EAAiC,CAN3C,IAAAC,EAoBEH,EAAA,KAAA,KAAMN,CAAW,GAAC,KAfV,OAAAS,EAAA,WAAAL,EACAK,EAAA,kBAAAD,EAeRC,EAAK,MAAQR,EACT,SAAuCS,EAAQ,CAC7C,GAAI,CACFT,EAAOS,CAAK,QACLC,EAAK,CACZX,EAAY,MAAMW,CAAG,EAEzB,EACAL,EAAA,UAAM,MACVG,EAAK,OAASN,EACV,SAAuCQ,EAAQ,CAC7C,GAAI,CACFR,EAAQQ,CAAG,QACJA,EAAK,CAEZX,EAAY,MAAMW,CAAG,UAGrB,KAAK,YAAW,EAEpB,EACAL,EAAA,UAAM,OACVG,EAAK,UAAYP,EACb,UAAA,CACE,GAAI,CACFA,EAAU,QACHS,EAAK,CAEZX,EAAY,MAAMW,CAAG,UAGrB,KAAK,YAAW,EAEpB,EACAL,EAAA,UAAM,WACZ,CAEA,OAAAD,EAAA,UAAA,YAAA,UAAA,OACE,GAAI,CAAC,KAAK,mBAAqB,KAAK,kBAAiB,EAAI,CAC/C,IAAAO,EAAW,KAAI,OACvBN,EAAA,UAAM,YAAW,KAAA,IAAA,EAEjB,CAACM,KAAUC,EAAA,KAAK,cAAU,MAAAA,IAAA,QAAAA,EAAA,KAAf,IAAI,GAEnB,EACFR,CAAA,EAnF2CS,EAAU,ECd9C,IAAMC,GAAiD,CAG5D,SAAA,SAASC,EAAQ,CACf,IAAIC,EAAU,sBACVC,EAAkD,qBAC9CC,EAAaJ,GAAsB,SACvCI,IACFF,EAAUE,EAAS,sBACnBD,EAASC,EAAS,sBAEpB,IAAMC,EAASH,EAAQ,SAACI,EAAS,CAI/BH,EAAS,OACTF,EAASK,CAAS,CACpB,CAAC,EACD,OAAO,IAAIC,GAAa,UAAA,CAAM,OAAAJ,GAAM,KAAA,OAANA,EAASE,CAAM,CAAf,CAAgB,CAChD,EACA,sBAAqB,UAAA,SAACG,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACZ,IAAAL,EAAaJ,GAAsB,SAC3C,QAAQI,GAAQ,KAAA,OAARA,EAAU,wBAAyB,uBAAsB,MAAA,OAAAM,EAAA,CAAA,EAAAC,EAAIH,CAAI,CAAA,CAAA,CAC3E,EACA,qBAAoB,UAAA,SAACA,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACX,IAAAL,EAAaJ,GAAsB,SAC3C,QAAQI,GAAQ,KAAA,OAARA,EAAU,uBAAwB,sBAAqB,MAAA,OAAAM,EAAA,CAAA,EAAAC,EAAIH,CAAI,CAAA,CAAA,CACzE,EACA,SAAU,QCrBL,IAAMI,GAAuDC,GAClE,SAACC,EAAM,CACL,OAAA,UAAoC,CAClCA,EAAO,IAAI,EACX,KAAK,KAAO,0BACZ,KAAK,QAAU,qBACjB,CAJA,CAIC,ECXL,IAAAC,EAAA,SAAAC,EAAA,CAAgCC,GAAAF,EAAAC,CAAA,EAwB9B,SAAAD,GAAA,CAAA,IAAAG,EAEEF,EAAA,KAAA,IAAA,GAAO,KAzBT,OAAAE,EAAA,OAAS,GAEDA,EAAA,iBAAyC,KAGjDA,EAAA,UAA2B,CAAA,EAE3BA,EAAA,UAAY,GAEZA,EAAA,SAAW,GAEXA,EAAA,YAAmB,MAenB,CAGA,OAAAH,EAAA,UAAA,KAAA,SAAQI,EAAwB,CAC9B,IAAMC,EAAU,IAAIC,GAAiB,KAAM,IAAI,EAC/C,OAAAD,EAAQ,SAAWD,EACZC,CACT,EAGUL,EAAA,UAAA,eAAV,UAAA,CACE,GAAI,KAAK,OACP,MAAM,IAAIO,EAEd,EAEAP,EAAA,UAAA,KAAA,SAAKQ,EAAQ,CAAb,IAAAL,EAAA,KACEM,GAAa,UAAA,SAEX,GADAN,EAAK,eAAc,EACf,CAACA,EAAK,UAAW,CACdA,EAAK,mBACRA,EAAK,iBAAmB,MAAM,KAAKA,EAAK,SAAS,OAEnD,QAAuBO,EAAAC,GAAAR,EAAK,gBAAgB,EAAAS,EAAAF,EAAA,KAAA,EAAA,CAAAE,EAAA,KAAAA,EAAAF,EAAA,KAAA,EAAE,CAAzC,IAAMG,EAAQD,EAAA,MACjBC,EAAS,KAAKL,CAAK,qGAGzB,CAAC,CACH,EAEAR,EAAA,UAAA,MAAA,SAAMc,EAAQ,CAAd,IAAAX,EAAA,KACEM,GAAa,UAAA,CAEX,GADAN,EAAK,eAAc,EACf,CAACA,EAAK,UAAW,CACnBA,EAAK,SAAWA,EAAK,UAAY,GACjCA,EAAK,YAAcW,EAEnB,QADQC,EAAcZ,EAAI,UACnBY,EAAU,QACfA,EAAU,MAAK,EAAI,MAAMD,CAAG,EAGlC,CAAC,CACH,EAEAd,EAAA,UAAA,SAAA,UAAA,CAAA,IAAAG,EAAA,KACEM,GAAa,UAAA,CAEX,GADAN,EAAK,eAAc,EACf,CAACA,EAAK,UAAW,CACnBA,EAAK,UAAY,GAEjB,QADQY,EAAcZ,EAAI,UACnBY,EAAU,QACfA,EAAU,MAAK,EAAI,SAAQ,EAGjC,CAAC,CACH,EAEAf,EAAA,UAAA,YAAA,UAAA,CACE,KAAK,UAAY,KAAK,OAAS,GAC/B,KAAK,UAAY,KAAK,iBAAmB,IAC3C,EAEA,OAAA,eAAIA,EAAA,UAAA,WAAQ,KAAZ,UAAA,OACE,QAAOgB,EAAA,KAAK,aAAS,MAAAA,IAAA,OAAA,OAAAA,EAAE,QAAS,CAClC,kCAGUhB,EAAA,UAAA,cAAV,SAAwBiB,EAAyB,CAC/C,YAAK,eAAc,EACZhB,EAAA,UAAM,cAAa,KAAA,KAACgB,CAAU,CACvC,EAGUjB,EAAA,UAAA,WAAV,SAAqBiB,EAAyB,CAC5C,YAAK,eAAc,EACnB,KAAK,wBAAwBA,CAAU,EAChC,KAAK,gBAAgBA,CAAU,CACxC,EAGUjB,EAAA,UAAA,gBAAV,SAA0BiB,EAA2B,CAArD,IAAAd,EAAA,KACQa,EAAqC,KAAnCE,EAAQF,EAAA,SAAEG,EAASH,EAAA,UAAED,EAASC,EAAA,UACtC,OAAIE,GAAYC,EACPC,IAET,KAAK,iBAAmB,KACxBL,EAAU,KAAKE,CAAU,EAClB,IAAII,GAAa,UAAA,CACtBlB,EAAK,iBAAmB,KACxBmB,GAAUP,EAAWE,CAAU,CACjC,CAAC,EACH,EAGUjB,EAAA,UAAA,wBAAV,SAAkCiB,EAA2B,CACrD,IAAAD,EAAuC,KAArCE,EAAQF,EAAA,SAAEO,EAAWP,EAAA,YAAEG,EAASH,EAAA,UACpCE,EACFD,EAAW,MAAMM,CAAW,EACnBJ,GACTF,EAAW,SAAQ,CAEvB,EAQAjB,EAAA,UAAA,aAAA,UAAA,CACE,IAAMwB,EAAkB,IAAIC,EAC5B,OAAAD,EAAW,OAAS,KACbA,CACT,EAxHOxB,EAAA,OAAkC,SAAI0B,EAA0BC,EAAqB,CAC1F,OAAO,IAAIrB,GAAoBoB,EAAaC,CAAM,CACpD,EAuHF3B,GA7IgCyB,CAAU,EAkJ1C,IAAAG,GAAA,SAAAC,EAAA,CAAyCC,GAAAF,EAAAC,CAAA,EACvC,SAAAD,EAESG,EACPC,EAAsB,CAHxB,IAAAC,EAKEJ,EAAA,KAAA,IAAA,GAAO,KAHA,OAAAI,EAAA,YAAAF,EAIPE,EAAK,OAASD,GAChB,CAEA,OAAAJ,EAAA,UAAA,KAAA,SAAKM,EAAQ,UACXC,GAAAC,EAAA,KAAK,eAAW,MAAAA,IAAA,OAAA,OAAAA,EAAE,QAAI,MAAAD,IAAA,QAAAA,EAAA,KAAAC,EAAGF,CAAK,CAChC,EAEAN,EAAA,UAAA,MAAA,SAAMS,EAAQ,UACZF,GAAAC,EAAA,KAAK,eAAW,MAAAA,IAAA,OAAA,OAAAA,EAAE,SAAK,MAAAD,IAAA,QAAAA,EAAA,KAAAC,EAAGC,CAAG,CAC/B,EAEAT,EAAA,UAAA,SAAA,UAAA,UACEO,GAAAC,EAAA,KAAK,eAAW,MAAAA,IAAA,OAAA,OAAAA,EAAE,YAAQ,MAAAD,IAAA,QAAAA,EAAA,KAAAC,CAAA,CAC5B,EAGUR,EAAA,UAAA,WAAV,SAAqBU,EAAyB,SAC5C,OAAOH,GAAAC,EAAA,KAAK,UAAM,MAAAA,IAAA,OAAA,OAAAA,EAAE,UAAUE,CAAU,KAAC,MAAAH,IAAA,OAAAA,EAAII,EAC/C,EACFX,CAAA,EA1ByCY,CAAO,EC5JzC,IAAMC,GAA+C,CAC1D,IAAG,UAAA,CAGD,OAAQA,GAAsB,UAAY,MAAM,IAAG,CACrD,EACA,SAAU,QCwBZ,IAAAC,GAAA,SAAAC,EAAA,CAAsCC,GAAAF,EAAAC,CAAA,EAUpC,SAAAD,EACUG,EACAC,EACAC,EAA6D,CAF7DF,IAAA,SAAAA,EAAA,KACAC,IAAA,SAAAA,EAAA,KACAC,IAAA,SAAAA,EAAAC,IAHV,IAAAC,EAKEN,EAAA,KAAA,IAAA,GAAO,KAJC,OAAAM,EAAA,YAAAJ,EACAI,EAAA,YAAAH,EACAG,EAAA,mBAAAF,EAZFE,EAAA,QAA0B,CAAA,EAC1BA,EAAA,oBAAsB,GAc5BA,EAAK,oBAAsBH,IAAgB,IAC3CG,EAAK,YAAc,KAAK,IAAI,EAAGJ,CAAW,EAC1CI,EAAK,YAAc,KAAK,IAAI,EAAGH,CAAW,GAC5C,CAEA,OAAAJ,EAAA,UAAA,KAAA,SAAKQ,EAAQ,CACL,IAAAC,EAA+E,KAA7EC,EAASD,EAAA,UAAEE,EAAOF,EAAA,QAAEG,EAAmBH,EAAA,oBAAEJ,EAAkBI,EAAA,mBAAEL,EAAWK,EAAA,YAC3EC,IACHC,EAAQ,KAAKH,CAAK,EAClB,CAACI,GAAuBD,EAAQ,KAAKN,EAAmB,IAAG,EAAKD,CAAW,GAE7E,KAAK,YAAW,EAChBH,EAAA,UAAM,KAAI,KAAA,KAACO,CAAK,CAClB,EAGUR,EAAA,UAAA,WAAV,SAAqBa,EAAyB,CAC5C,KAAK,eAAc,EACnB,KAAK,YAAW,EAQhB,QANMC,EAAe,KAAK,gBAAgBD,CAAU,EAE9CJ,EAAmC,KAAjCG,EAAmBH,EAAA,oBAAEE,EAAOF,EAAA,QAG9BM,EAAOJ,EAAQ,MAAK,EACjBK,EAAI,EAAGA,EAAID,EAAK,QAAU,CAACF,EAAW,OAAQG,GAAKJ,EAAsB,EAAI,EACpFC,EAAW,KAAKE,EAAKC,CAAC,CAAM,EAG9B,YAAK,wBAAwBH,CAAU,EAEhCC,CACT,EAEQd,EAAA,UAAA,YAAR,UAAA,CACQ,IAAAS,EAAoE,KAAlEN,EAAWM,EAAA,YAAEJ,EAAkBI,EAAA,mBAAEE,EAAOF,EAAA,QAAEG,EAAmBH,EAAA,oBAK/DQ,GAAsBL,EAAsB,EAAI,GAAKT,EAK3D,GAJAA,EAAc,KAAYc,EAAqBN,EAAQ,QAAUA,EAAQ,OAAO,EAAGA,EAAQ,OAASM,CAAkB,EAIlH,CAACL,EAAqB,CAKxB,QAJMM,EAAMb,EAAmB,IAAG,EAC9Bc,EAAO,EAGFH,EAAI,EAAGA,EAAIL,EAAQ,QAAWA,EAAQK,CAAC,GAAgBE,EAAKF,GAAK,EACxEG,EAAOH,EAETG,GAAQR,EAAQ,OAAO,EAAGQ,EAAO,CAAC,EAEtC,EACFnB,CAAA,EAzEsCoB,CAAO,EClB7C,IAAAC,GAAA,SAAAC,EAAA,CAA+BC,GAAAF,EAAAC,CAAA,EAC7B,SAAAD,EAAYG,EAAsBC,EAAmD,QACnFH,EAAA,KAAA,IAAA,GAAO,IACT,CAWO,OAAAD,EAAA,UAAA,SAAP,SAAgBK,EAAWC,EAAiB,CAAjB,OAAAA,IAAA,SAAAA,EAAA,GAClB,IACT,EACFN,CAAA,EAjB+BO,EAAY,ECHpC,IAAMC,GAAqC,CAGhD,YAAA,SAAYC,EAAqBC,EAAgB,SAAEC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,EAAA,CAAA,EAAA,UAAAA,CAAA,EACzC,IAAAC,EAAaL,GAAgB,SACrC,OAAIK,GAAQ,MAARA,EAAU,YACLA,EAAS,YAAW,MAApBA,EAAQC,EAAA,CAAaL,EAASC,CAAO,EAAAK,EAAKJ,CAAI,CAAA,CAAA,EAEhD,YAAW,MAAA,OAAAG,EAAA,CAACL,EAASC,CAAO,EAAAK,EAAKJ,CAAI,CAAA,CAAA,CAC9C,EACA,cAAA,SAAcK,EAAM,CACV,IAAAH,EAAaL,GAAgB,SACrC,QAAQK,GAAQ,KAAA,OAARA,EAAU,gBAAiB,eAAeG,CAAa,CACjE,EACA,SAAU,QCrBZ,IAAAC,GAAA,SAAAC,EAAA,CAAoCC,GAAAF,EAAAC,CAAA,EAOlC,SAAAD,EAAsBG,EAAqCC,EAAmD,CAA9G,IAAAC,EACEJ,EAAA,KAAA,KAAME,EAAWC,CAAI,GAAC,KADF,OAAAC,EAAA,UAAAF,EAAqCE,EAAA,KAAAD,EAFjDC,EAAA,QAAmB,IAI7B,CAEO,OAAAL,EAAA,UAAA,SAAP,SAAgBM,EAAWC,EAAiB,OAC1C,GADyBA,IAAA,SAAAA,EAAA,GACrB,KAAK,OACP,OAAO,KAIT,KAAK,MAAQD,EAEb,IAAME,EAAK,KAAK,GACVL,EAAY,KAAK,UAuBvB,OAAIK,GAAM,OACR,KAAK,GAAK,KAAK,eAAeL,EAAWK,EAAID,CAAK,GAKpD,KAAK,QAAU,GAEf,KAAK,MAAQA,EAEb,KAAK,IAAKE,EAAA,KAAK,MAAE,MAAAA,IAAA,OAAAA,EAAI,KAAK,eAAeN,EAAW,KAAK,GAAII,CAAK,EAE3D,IACT,EAEUP,EAAA,UAAA,eAAV,SAAyBG,EAA2BO,EAAmBH,EAAiB,CAAjB,OAAAA,IAAA,SAAAA,EAAA,GAC9DI,GAAiB,YAAYR,EAAU,MAAM,KAAKA,EAAW,IAAI,EAAGI,CAAK,CAClF,EAEUP,EAAA,UAAA,eAAV,SAAyBY,EAA4BJ,EAAkBD,EAAwB,CAE7F,GAFqEA,IAAA,SAAAA,EAAA,GAEjEA,GAAS,MAAQ,KAAK,QAAUA,GAAS,KAAK,UAAY,GAC5D,OAAOC,EAILA,GAAM,MACRG,GAAiB,cAAcH,CAAE,CAIrC,EAMOR,EAAA,UAAA,QAAP,SAAeM,EAAUC,EAAa,CACpC,GAAI,KAAK,OACP,OAAO,IAAI,MAAM,8BAA8B,EAGjD,KAAK,QAAU,GACf,IAAMM,EAAQ,KAAK,SAASP,EAAOC,CAAK,EACxC,GAAIM,EACF,OAAOA,EACE,KAAK,UAAY,IAAS,KAAK,IAAM,OAc9C,KAAK,GAAK,KAAK,eAAe,KAAK,UAAW,KAAK,GAAI,IAAI,EAE/D,EAEUb,EAAA,UAAA,SAAV,SAAmBM,EAAUQ,EAAc,CACzC,IAAIC,EAAmB,GACnBC,EACJ,GAAI,CACF,KAAK,KAAKV,CAAK,QACRW,EAAG,CACVF,EAAU,GAIVC,EAAaC,GAAQ,IAAI,MAAM,oCAAoC,EAErE,GAAIF,EACF,YAAK,YAAW,EACTC,CAEX,EAEAhB,EAAA,UAAA,YAAA,UAAA,CACE,GAAI,CAAC,KAAK,OAAQ,CACV,IAAAS,EAAoB,KAAlBD,EAAEC,EAAA,GAAEN,EAASM,EAAA,UACbS,EAAYf,EAAS,QAE7B,KAAK,KAAO,KAAK,MAAQ,KAAK,UAAY,KAC1C,KAAK,QAAU,GAEfgB,GAAUD,EAAS,IAAI,EACnBV,GAAM,OACR,KAAK,GAAK,KAAK,eAAeL,EAAWK,EAAI,IAAI,GAGnD,KAAK,MAAQ,KACbP,EAAA,UAAM,YAAW,KAAA,IAAA,EAErB,EACFD,CAAA,EA9IoCoB,EAAM,ECgB1C,IAAAC,GAAA,UAAA,CAGE,SAAAA,EAAoBC,EAAoCC,EAAiC,CAAjCA,IAAA,SAAAA,EAAoBF,EAAU,KAAlE,KAAA,oBAAAC,EAClB,KAAK,IAAMC,CACb,CA6BO,OAAAF,EAAA,UAAA,SAAP,SAAmBG,EAAqDC,EAAmBC,EAAS,CAA5B,OAAAD,IAAA,SAAAA,EAAA,GAC/D,IAAI,KAAK,oBAAuB,KAAMD,CAAI,EAAE,SAASE,EAAOD,CAAK,CAC1E,EAnCcJ,EAAA,IAAoBM,GAAsB,IAoC1DN,GArCA,ECnBA,IAAAO,GAAA,SAAAC,EAAA,CAAoCC,GAAAF,EAAAC,CAAA,EAkBlC,SAAAD,EAAYG,EAAgCC,EAAiC,CAAjCA,IAAA,SAAAA,EAAoBC,GAAU,KAA1E,IAAAC,EACEL,EAAA,KAAA,KAAME,EAAiBC,CAAG,GAAC,KAlBtB,OAAAE,EAAA,QAAmC,CAAA,EAOnCA,EAAA,QAAmB,IAY1B,CAEO,OAAAN,EAAA,UAAA,MAAP,SAAaO,EAAwB,CAC3B,IAAAC,EAAY,KAAI,QAExB,GAAI,KAAK,QAAS,CAChBA,EAAQ,KAAKD,CAAM,EACnB,OAGF,IAAIE,EACJ,KAAK,QAAU,GAEf,EACE,IAAKA,EAAQF,EAAO,QAAQA,EAAO,MAAOA,EAAO,KAAK,EACpD,YAEMA,EAASC,EAAQ,MAAK,GAIhC,GAFA,KAAK,QAAU,GAEXC,EAAO,CACT,KAAQF,EAASC,EAAQ,MAAK,GAC5BD,EAAO,YAAW,EAEpB,MAAME,EAEV,EACFT,CAAA,EAhDoCK,EAAS,EC6CtC,IAAMK,GAAiB,IAAIC,GAAeC,EAAW,EAK/CC,GAAQH,GCjDrB,IAAAI,GAAA,SAAAC,EAAA,CAA6CC,GAAAF,EAAAC,CAAA,EAC3C,SAAAD,EAAsBG,EAA8CC,EAAmD,CAAvH,IAAAC,EACEJ,EAAA,KAAA,KAAME,EAAWC,CAAI,GAAC,KADF,OAAAC,EAAA,UAAAF,EAA8CE,EAAA,KAAAD,GAEpE,CAEU,OAAAJ,EAAA,UAAA,eAAV,SAAyBG,EAAoCG,EAAkBC,EAAiB,CAE9F,OAF6EA,IAAA,SAAAA,EAAA,GAEzEA,IAAU,MAAQA,EAAQ,EACrBN,EAAA,UAAM,eAAc,KAAA,KAACE,EAAWG,EAAIC,CAAK,GAGlDJ,EAAU,QAAQ,KAAK,IAAI,EAIpBA,EAAU,aAAeA,EAAU,WAAaK,GAAuB,sBAAsB,UAAA,CAAM,OAAAL,EAAU,MAAM,MAAS,CAAzB,CAA0B,GACtI,EAEUH,EAAA,UAAA,eAAV,SAAyBG,EAAoCG,EAAkBC,EAAiB,OAI9F,GAJ6EA,IAAA,SAAAA,EAAA,GAIzEA,GAAS,KAAOA,EAAQ,EAAI,KAAK,MAAQ,EAC3C,OAAON,EAAA,UAAM,eAAc,KAAA,KAACE,EAAWG,EAAIC,CAAK,EAK1C,IAAAE,EAAYN,EAAS,QACzBG,GAAM,QAAQI,EAAAD,EAAQA,EAAQ,OAAS,CAAC,KAAC,MAAAC,IAAA,OAAA,OAAAA,EAAE,MAAOJ,IACpDE,GAAuB,qBAAqBF,CAAY,EACxDH,EAAU,WAAa,OAI3B,EACFH,CAAA,EApC6CW,EAAW,ECHxD,IAAAC,GAAA,SAAAC,EAAA,CAA6CC,GAAAF,EAAAC,CAAA,EAA7C,SAAAD,GAAA,+CAkCA,CAjCS,OAAAA,EAAA,UAAA,MAAP,SAAaG,EAAyB,CACpC,KAAK,QAAU,GAUf,IAAMC,EAAU,KAAK,WACrB,KAAK,WAAa,OAEV,IAAAC,EAAY,KAAI,QACpBC,EACJH,EAASA,GAAUE,EAAQ,MAAK,EAEhC,EACE,IAAKC,EAAQH,EAAO,QAAQA,EAAO,MAAOA,EAAO,KAAK,EACpD,aAEMA,EAASE,EAAQ,CAAC,IAAMF,EAAO,KAAOC,GAAWC,EAAQ,MAAK,GAIxE,GAFA,KAAK,QAAU,GAEXC,EAAO,CACT,MAAQH,EAASE,EAAQ,CAAC,IAAMF,EAAO,KAAOC,GAAWC,EAAQ,MAAK,GACpEF,EAAO,YAAW,EAEpB,MAAMG,EAEV,EACFN,CAAA,EAlC6CO,EAAc,ECgCpD,IAAMC,GAA0B,IAAIC,GAAwBC,EAAoB,EC8BhF,IAAMC,EAAQ,IAAIC,EAAkB,SAACC,EAAU,CAAK,OAAAA,EAAW,SAAQ,CAAnB,CAAqB,EC9D1E,SAAUC,GAAYC,EAAU,CACpC,OAAOA,GAASC,EAAWD,EAAM,QAAQ,CAC3C,CCDA,SAASE,GAAQC,EAAQ,CACvB,OAAOA,EAAIA,EAAI,OAAS,CAAC,CAC3B,CAEM,SAAUC,GAAkBC,EAAW,CAC3C,OAAOC,EAAWJ,GAAKG,CAAI,CAAC,EAAIA,EAAK,IAAG,EAAK,MAC/C,CAEM,SAAUE,GAAaF,EAAW,CACtC,OAAOG,GAAYN,GAAKG,CAAI,CAAC,EAAIA,EAAK,IAAG,EAAK,MAChD,CAEM,SAAUI,GAAUJ,EAAaK,EAAoB,CACzD,OAAO,OAAOR,GAAKG,CAAI,GAAM,SAAWA,EAAK,IAAG,EAAMK,CACxD,CClBO,IAAMC,GAAe,SAAIC,EAAM,CAAwB,OAAAA,GAAK,OAAOA,EAAE,QAAW,UAAY,OAAOA,GAAM,UAAlD,ECMxD,SAAUC,GAAUC,EAAU,CAClC,OAAOC,EAAWD,GAAK,KAAA,OAALA,EAAO,IAAI,CAC/B,CCHM,SAAUE,GAAoBC,EAAU,CAC5C,OAAOC,EAAWD,EAAME,EAAiB,CAAC,CAC5C,CCLM,SAAUC,GAAmBC,EAAQ,CACzC,OAAO,OAAO,eAAiBC,EAAWD,GAAG,KAAA,OAAHA,EAAM,OAAO,aAAa,CAAC,CACvE,CCAM,SAAUE,GAAiCC,EAAU,CAEzD,OAAO,IAAI,UACT,iBACEA,IAAU,MAAQ,OAAOA,GAAU,SAAW,oBAAsB,IAAIA,EAAK,KAAG,0HACwC,CAE9H,CCXM,SAAUC,IAAiB,CAC/B,OAAI,OAAO,QAAW,YAAc,CAAC,OAAO,SACnC,aAGF,OAAO,QAChB,CAEO,IAAMC,GAAWD,GAAiB,ECJnC,SAAUE,GAAWC,EAAU,CACnC,OAAOC,EAAWD,GAAK,KAAA,OAALA,EAAQE,EAAe,CAAC,CAC5C,CCHM,SAAiBC,GAAsCC,EAAqC,mGAC1FC,EAASD,EAAe,UAAS,2DAGX,MAAA,CAAA,EAAAE,GAAMD,EAAO,KAAI,CAAE,CAAA,gBAArCE,EAAkBC,EAAA,KAAA,EAAhBC,EAAKF,EAAA,MAAEG,EAAIH,EAAA,KACfG,iBAAA,CAAA,EAAA,CAAA,SACF,MAAA,CAAA,EAAAF,EAAA,KAAA,CAAA,qBAEIC,CAAM,CAAA,SAAZ,MAAA,CAAA,EAAAD,EAAA,KAAA,CAAA,SAAA,OAAAA,EAAA,KAAA,mCAGF,OAAAH,EAAO,YAAW,6BAIhB,SAAUM,GAAwBC,EAAQ,CAG9C,OAAOC,EAAWD,GAAG,KAAA,OAAHA,EAAK,SAAS,CAClC,CCPM,SAAUE,EAAaC,EAAyB,CACpD,GAAIA,aAAiBC,EACnB,OAAOD,EAET,GAAIA,GAAS,KAAM,CACjB,GAAIE,GAAoBF,CAAK,EAC3B,OAAOG,GAAsBH,CAAK,EAEpC,GAAII,GAAYJ,CAAK,EACnB,OAAOK,GAAcL,CAAK,EAE5B,GAAIM,GAAUN,CAAK,EACjB,OAAOO,GAAYP,CAAK,EAE1B,GAAIQ,GAAgBR,CAAK,EACvB,OAAOS,GAAkBT,CAAK,EAEhC,GAAIU,GAAWV,CAAK,EAClB,OAAOW,GAAaX,CAAK,EAE3B,GAAIY,GAAqBZ,CAAK,EAC5B,OAAOa,GAAuBb,CAAK,EAIvC,MAAMc,GAAiCd,CAAK,CAC9C,CAMM,SAAUG,GAAyBY,EAAQ,CAC/C,OAAO,IAAId,EAAW,SAACe,EAAyB,CAC9C,IAAMC,EAAMF,EAAIG,EAAiB,EAAC,EAClC,GAAIC,EAAWF,EAAI,SAAS,EAC1B,OAAOA,EAAI,UAAUD,CAAU,EAGjC,MAAM,IAAI,UAAU,gEAAgE,CACtF,CAAC,CACH,CASM,SAAUX,GAAiBe,EAAmB,CAClD,OAAO,IAAInB,EAAW,SAACe,EAAyB,CAU9C,QAASK,EAAI,EAAGA,EAAID,EAAM,QAAU,CAACJ,EAAW,OAAQK,IACtDL,EAAW,KAAKI,EAAMC,CAAC,CAAC,EAE1BL,EAAW,SAAQ,CACrB,CAAC,CACH,CAEM,SAAUT,GAAee,EAAuB,CACpD,OAAO,IAAIrB,EAAW,SAACe,EAAyB,CAC9CM,EACG,KACC,SAACC,EAAK,CACCP,EAAW,SACdA,EAAW,KAAKO,CAAK,EACrBP,EAAW,SAAQ,EAEvB,EACA,SAACQ,EAAQ,CAAK,OAAAR,EAAW,MAAMQ,CAAG,CAApB,CAAqB,EAEpC,KAAK,KAAMC,EAAoB,CACpC,CAAC,CACH,CAEM,SAAUd,GAAgBe,EAAqB,CACnD,OAAO,IAAIzB,EAAW,SAACe,EAAyB,aAC9C,QAAoBW,EAAAC,GAAAF,CAAQ,EAAAG,EAAAF,EAAA,KAAA,EAAA,CAAAE,EAAA,KAAAA,EAAAF,EAAA,KAAA,EAAE,CAAzB,IAAMJ,EAAKM,EAAA,MAEd,GADAb,EAAW,KAAKO,CAAK,EACjBP,EAAW,OACb,yGAGJA,EAAW,SAAQ,CACrB,CAAC,CACH,CAEM,SAAUP,GAAqBqB,EAA+B,CAClE,OAAO,IAAI7B,EAAW,SAACe,EAAyB,CAC9Ce,GAAQD,EAAed,CAAU,EAAE,MAAM,SAACQ,EAAG,CAAK,OAAAR,EAAW,MAAMQ,CAAG,CAApB,CAAqB,CACzE,CAAC,CACH,CAEM,SAAUX,GAA0BmB,EAAqC,CAC7E,OAAOvB,GAAkBwB,GAAmCD,CAAc,CAAC,CAC7E,CAEA,SAAeD,GAAWD,EAAiCd,EAAyB,uIACxDkB,EAAAC,GAAAL,CAAa,gFAIrC,GAJeP,EAAKa,EAAA,MACpBpB,EAAW,KAAKO,CAAK,EAGjBP,EAAW,OACb,MAAA,CAAA,CAAA,6RAGJ,OAAAA,EAAW,SAAQ,WChHf,SAAUqB,GACdC,EACAC,EACAC,EACAC,EACAC,EAAc,CADdD,IAAA,SAAAA,EAAA,GACAC,IAAA,SAAAA,EAAA,IAEA,IAAMC,EAAuBJ,EAAU,SAAS,UAAA,CAC9CC,EAAI,EACAE,EACFJ,EAAmB,IAAI,KAAK,SAAS,KAAMG,CAAK,CAAC,EAEjD,KAAK,YAAW,CAEpB,EAAGA,CAAK,EAIR,GAFAH,EAAmB,IAAIK,CAAoB,EAEvC,CAACD,EAKH,OAAOC,CAEX,CCeM,SAAUC,GAAaC,EAA0BC,EAAS,CAAT,OAAAA,IAAA,SAAAA,EAAA,GAC9CC,EAAQ,SAACC,EAAQC,EAAU,CAChCD,EAAO,UACLE,EACED,EACA,SAACE,EAAK,CAAK,OAAAC,GAAgBH,EAAYJ,EAAW,UAAA,CAAM,OAAAI,EAAW,KAAKE,CAAK,CAArB,EAAwBL,CAAK,CAA1E,EACX,UAAA,CAAM,OAAAM,GAAgBH,EAAYJ,EAAW,UAAA,CAAM,OAAAI,EAAW,SAAQ,CAAnB,EAAuBH,CAAK,CAAzE,EACN,SAACO,EAAG,CAAK,OAAAD,GAAgBH,EAAYJ,EAAW,UAAA,CAAM,OAAAI,EAAW,MAAMI,CAAG,CAApB,EAAuBP,CAAK,CAAzE,CAA0E,CACpF,CAEL,CAAC,CACH,CCPM,SAAUQ,GAAeC,EAA0BC,EAAiB,CAAjB,OAAAA,IAAA,SAAAA,EAAA,GAChDC,EAAQ,SAACC,EAAQC,EAAU,CAChCA,EAAW,IAAIJ,EAAU,SAAS,UAAA,CAAM,OAAAG,EAAO,UAAUC,CAAU,CAA3B,EAA8BH,CAAK,CAAC,CAC9E,CAAC,CACH,CC7DM,SAAUI,GAAsBC,EAA6BC,EAAwB,CACzF,OAAOC,EAAUF,CAAK,EAAE,KAAKG,GAAYF,CAAS,EAAGG,GAAUH,CAAS,CAAC,CAC3E,CCFM,SAAUI,GAAmBC,EAAuBC,EAAwB,CAChF,OAAOC,EAAUF,CAAK,EAAE,KAAKG,GAAYF,CAAS,EAAGG,GAAUH,CAAS,CAAC,CAC3E,CCJM,SAAUI,GAAiBC,EAAqBC,EAAwB,CAC5E,OAAO,IAAIC,EAAc,SAACC,EAAU,CAElC,IAAIC,EAAI,EAER,OAAOH,EAAU,SAAS,UAAA,CACpBG,IAAMJ,EAAM,OAGdG,EAAW,SAAQ,GAInBA,EAAW,KAAKH,EAAMI,GAAG,CAAC,EAIrBD,EAAW,QACd,KAAK,SAAQ,EAGnB,CAAC,CACH,CAAC,CACH,CCfM,SAAUE,GAAoBC,EAAoBC,EAAwB,CAC9E,OAAO,IAAIC,EAAc,SAACC,EAAU,CAClC,IAAIC,EAKJ,OAAAC,GAAgBF,EAAYF,EAAW,UAAA,CAErCG,EAAYJ,EAAcI,EAAe,EAAC,EAE1CC,GACEF,EACAF,EACA,UAAA,OACMK,EACAC,EACJ,GAAI,CAEDC,EAAkBJ,EAAS,KAAI,EAA7BE,EAAKE,EAAA,MAAED,EAAIC,EAAA,WACPC,EAAK,CAEZN,EAAW,MAAMM,CAAG,EACpB,OAGEF,EAKFJ,EAAW,SAAQ,EAGnBA,EAAW,KAAKG,CAAK,CAEzB,EACA,EACA,EAAI,CAER,CAAC,EAMM,UAAA,CAAM,OAAAI,EAAWN,GAAQ,KAAA,OAARA,EAAU,MAAM,GAAKA,EAAS,OAAM,CAA/C,CACf,CAAC,CACH,CCvDM,SAAUO,GAAyBC,EAAyBC,EAAwB,CACxF,GAAI,CAACD,EACH,MAAM,IAAI,MAAM,yBAAyB,EAE3C,OAAO,IAAIE,EAAc,SAACC,EAAU,CAClCC,GAAgBD,EAAYF,EAAW,UAAA,CACrC,IAAMI,EAAWL,EAAM,OAAO,aAAa,EAAC,EAC5CI,GACED,EACAF,EACA,UAAA,CACEI,EAAS,KAAI,EAAG,KAAK,SAACC,EAAM,CACtBA,EAAO,KAGTH,EAAW,SAAQ,EAEnBA,EAAW,KAAKG,EAAO,KAAK,CAEhC,CAAC,CACH,EACA,EACA,EAAI,CAER,CAAC,CACH,CAAC,CACH,CCzBM,SAAUC,GAA8BC,EAA8BC,EAAwB,CAClG,OAAOC,GAAsBC,GAAmCH,CAAK,EAAGC,CAAS,CACnF,CCoBM,SAAUG,GAAaC,EAA2BC,EAAwB,CAC9E,GAAID,GAAS,KAAM,CACjB,GAAIE,GAAoBF,CAAK,EAC3B,OAAOG,GAAmBH,EAAOC,CAAS,EAE5C,GAAIG,GAAYJ,CAAK,EACnB,OAAOK,GAAcL,EAAOC,CAAS,EAEvC,GAAIK,GAAUN,CAAK,EACjB,OAAOO,GAAgBP,EAAOC,CAAS,EAEzC,GAAIO,GAAgBR,CAAK,EACvB,OAAOS,GAAsBT,EAAOC,CAAS,EAE/C,GAAIS,GAAWV,CAAK,EAClB,OAAOW,GAAiBX,EAAOC,CAAS,EAE1C,GAAIW,GAAqBZ,CAAK,EAC5B,OAAOa,GAA2Bb,EAAOC,CAAS,EAGtD,MAAMa,GAAiCd,CAAK,CAC9C,CCoDM,SAAUe,GAAQC,EAA2BC,EAAyB,CAC1E,OAAOA,EAAYC,GAAUF,EAAOC,CAAS,EAAIE,EAAUH,CAAK,CAClE,CCxBM,SAAUI,GAAE,SAAIC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACpB,IAAMC,EAAYC,GAAaH,CAAI,EACnC,OAAOI,GAAKJ,EAAaE,CAAS,CACpC,CCsCM,SAAUG,GAAWC,EAA0BC,EAAyB,CAC5E,IAAMC,EAAeC,EAAWH,CAAmB,EAAIA,EAAsB,UAAA,CAAM,OAAAA,CAAA,EAC7EI,EAAO,SAACC,EAA6B,CAAK,OAAAA,EAAW,MAAMH,EAAY,CAAE,CAA/B,EAChD,OAAO,IAAII,EAAWL,EAAY,SAACI,EAAU,CAAK,OAAAJ,EAAU,SAASG,EAAa,EAAGC,CAAU,CAA7C,EAAiDD,CAAI,CACzG,CCpGO,IAAMG,GAA6BC,GAAiB,SAACC,EAAM,CAAK,OAAA,UAAuB,CAC5FA,EAAO,IAAI,EACX,KAAK,KAAO,aACZ,KAAK,QAAU,yBACjB,CAJuE,CAItE,ECrBK,SAAUC,GAAYC,EAAU,CACpC,OAAOA,aAAiB,MAAQ,CAAC,MAAMA,CAAY,CACrD,CCsCM,SAAUC,EAAUC,EAAyCC,EAAa,CAC9E,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAEhC,IAAIC,EAAQ,EAGZF,EAAO,UACLG,EAAyBF,EAAY,SAACG,EAAQ,CAG5CH,EAAW,KAAKJ,EAAQ,KAAKC,EAASM,EAAOF,GAAO,CAAC,CACvD,CAAC,CAAC,CAEN,CAAC,CACH,CC1DQ,IAAAG,GAAY,MAAK,QAEzB,SAASC,GAAkBC,EAA6BC,EAAW,CAC/D,OAAOH,GAAQG,CAAI,EAAID,EAAE,MAAA,OAAAE,EAAA,CAAA,EAAAC,EAAIF,CAAI,CAAA,CAAA,EAAID,EAAGC,CAAI,CAChD,CAMM,SAAUG,GAAuBJ,EAA2B,CAC9D,OAAOK,EAAI,SAAAJ,EAAI,CAAI,OAAAF,GAAYC,EAAIC,CAAI,CAApB,CAAqB,CAC5C,CCfQ,IAAAK,GAAY,MAAK,QACjBC,GAA0D,OAAM,eAArCC,GAA+B,OAAM,UAAlBC,GAAY,OAAM,KAQlE,SAAUC,GAAqDC,EAAuB,CAC1F,GAAIA,EAAK,SAAW,EAAG,CACrB,IAAMC,EAAQD,EAAK,CAAC,EACpB,GAAIL,GAAQM,CAAK,EACf,MAAO,CAAE,KAAMA,EAAO,KAAM,IAAI,EAElC,GAAIC,GAAOD,CAAK,EAAG,CACjB,IAAME,EAAOL,GAAQG,CAAK,EAC1B,MAAO,CACL,KAAME,EAAK,IAAI,SAACC,EAAG,CAAK,OAAAH,EAAMG,CAAG,CAAT,CAAU,EAClC,KAAID,IAKV,MAAO,CAAE,KAAMH,EAAa,KAAM,IAAI,CACxC,CAEA,SAASE,GAAOG,EAAQ,CACtB,OAAOA,GAAO,OAAOA,GAAQ,UAAYT,GAAeS,CAAG,IAAMR,EACnE,CC7BM,SAAUS,GAAaC,EAAgBC,EAAa,CACxD,OAAOD,EAAK,OAAO,SAACE,EAAQC,EAAKC,EAAC,CAAK,OAAEF,EAAOC,CAAG,EAAIF,EAAOG,CAAC,EAAIF,CAA5B,EAAqC,CAAA,CAAS,CACvF,CCsMM,SAAUG,GAAa,SAAoCC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAC/D,IAAMC,EAAYC,GAAaH,CAAI,EAC7BI,EAAiBC,GAAkBL,CAAI,EAEvCM,EAA8BC,GAAqBP,CAAI,EAA/CQ,EAAWF,EAAA,KAAEG,EAAIH,EAAA,KAE/B,GAAIE,EAAY,SAAW,EAIzB,OAAOE,GAAK,CAAA,EAAIR,CAAgB,EAGlC,IAAMS,EAAS,IAAIC,EACjBC,GACEL,EACAN,EACAO,EAEI,SAACK,EAAM,CAAK,OAAAC,GAAaN,EAAMK,CAAM,CAAzB,EAEZE,EAAQ,CACb,EAGH,OAAOZ,EAAkBO,EAAO,KAAKM,GAAiBb,CAAc,CAAC,EAAsBO,CAC7F,CAEM,SAAUE,GACdL,EACAN,EACAgB,EAAiD,CAAjD,OAAAA,IAAA,SAAAA,EAAAF,IAEO,SAACG,EAA2B,CAGjCC,GACElB,EACA,UAAA,CAaE,QAZQmB,EAAWb,EAAW,OAExBM,EAAS,IAAI,MAAMO,CAAM,EAG3BC,EAASD,EAITE,EAAuBF,aAGlBG,EAAC,CACRJ,GACElB,EACA,UAAA,CACE,IAAMuB,EAASf,GAAKF,EAAYgB,CAAC,EAAGtB,CAAgB,EAChDwB,EAAgB,GACpBD,EAAO,UACLE,EACER,EACA,SAACS,EAAK,CAEJd,EAAOU,CAAC,EAAII,EACPF,IAEHA,EAAgB,GAChBH,KAEGA,GAGHJ,EAAW,KAAKD,EAAeJ,EAAO,MAAK,CAAE,CAAC,CAElD,EACA,UAAA,CACO,EAAEQ,GAGLH,EAAW,SAAQ,CAEvB,CAAC,CACF,CAEL,EACAA,CAAU,GAjCLK,EAAI,EAAGA,EAAIH,EAAQG,MAAnBA,CAAC,CAoCZ,EACAL,CAAU,CAEd,CACF,CAMA,SAASC,GAAclB,EAAsC2B,EAAqBC,EAA0B,CACtG5B,EACF6B,GAAgBD,EAAc5B,EAAW2B,CAAO,EAEhDA,EAAO,CAEX,CC3RM,SAAUG,GACdC,EACAC,EACAC,EACAC,EACAC,EACAC,EACAC,EACAC,EAAgC,CAGhC,IAAMC,EAAc,CAAA,EAEhBC,EAAS,EAETC,EAAQ,EAERC,EAAa,GAKXC,EAAgB,UAAA,CAIhBD,GAAc,CAACH,EAAO,QAAU,CAACC,GACnCR,EAAW,SAAQ,CAEvB,EAGMY,EAAY,SAACC,EAAQ,CAAK,OAACL,EAASN,EAAaY,EAAWD,CAAK,EAAIN,EAAO,KAAKM,CAAK,CAA5D,EAE1BC,EAAa,SAACD,EAAQ,CAI1BT,GAAUJ,EAAW,KAAKa,CAAY,EAItCL,IAKA,IAAIO,EAAgB,GAGpBC,EAAUf,EAAQY,EAAOJ,GAAO,CAAC,EAAE,UACjCQ,EACEjB,EACA,SAACkB,EAAU,CAGTf,GAAY,MAAZA,EAAee,CAAU,EAErBd,EAGFQ,EAAUM,CAAiB,EAG3BlB,EAAW,KAAKkB,CAAU,CAE9B,EACA,UAAA,CAGEH,EAAgB,EAClB,EAEA,OACA,UAAA,CAIE,GAAIA,EAKF,GAAI,CAIFP,IAKA,qBACE,IAAMW,EAAgBZ,EAAO,MAAK,EAI9BF,EACFe,GAAgBpB,EAAYK,EAAmB,UAAA,CAAM,OAAAS,EAAWK,CAAa,CAAxB,CAAyB,EAE9EL,EAAWK,CAAa,GARrBZ,EAAO,QAAUC,EAASN,OAYjCS,EAAa,QACNU,EAAK,CACZrB,EAAW,MAAMqB,CAAG,EAG1B,CAAC,CACF,CAEL,EAGA,OAAAtB,EAAO,UACLkB,EAAyBjB,EAAYY,EAAW,UAAA,CAE9CF,EAAa,GACbC,EAAa,CACf,CAAC,CAAC,EAKG,UAAA,CACLL,GAAmB,MAAnBA,EAAmB,CACrB,CACF,CClEM,SAAUgB,GACdC,EACAC,EACAC,EAA6B,CAE7B,OAFAA,IAAA,SAAAA,EAAA,KAEIC,EAAWF,CAAc,EAEpBF,GAAS,SAACK,EAAGC,EAAC,CAAK,OAAAC,EAAI,SAACC,EAAQC,EAAU,CAAK,OAAAP,EAAeG,EAAGG,EAAGF,EAAGG,CAAE,CAA1B,CAA2B,EAAEC,EAAUT,EAAQI,EAAGC,CAAC,CAAC,CAAC,CAAjF,EAAoFH,CAAU,GAC/G,OAAOD,GAAmB,WACnCC,EAAaD,GAGRS,EAAQ,SAACC,EAAQC,EAAU,CAAK,OAAAC,GAAeF,EAAQC,EAAYZ,EAASE,CAAU,CAAtD,CAAuD,EAChG,CChCM,SAAUY,GAAyCC,EAA6B,CAA7B,OAAAA,IAAA,SAAAA,EAAA,KAChDC,GAASC,GAAUF,CAAU,CACtC,CCNM,SAAUG,IAAS,CACvB,OAAOC,GAAS,CAAC,CACnB,CCmDM,SAAUC,IAAM,SAACC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACrB,OAAOC,GAAS,EAAGC,GAAKH,EAAMI,GAAaJ,CAAI,CAAC,CAAC,CACnD,CC9DM,SAAUK,EAAsCC,EAA0B,CAC9E,OAAO,IAAIC,EAA+B,SAACC,EAAU,CACnDC,EAAUH,EAAiB,CAAE,EAAE,UAAUE,CAAU,CACrD,CAAC,CACH,CChDA,IAAME,GAA0B,CAAC,cAAe,gBAAgB,EAC1DC,GAAqB,CAAC,mBAAoB,qBAAqB,EAC/DC,GAAgB,CAAC,KAAM,KAAK,EAkO5B,SAAUC,EACdC,EACAC,EACAC,EACAC,EAAsC,CAMtC,GAJIC,EAAWF,CAAO,IACpBC,EAAiBD,EACjBA,EAAU,QAERC,EACF,OAAOJ,EAAaC,EAAQC,EAAWC,CAA+B,EAAE,KAAKG,GAAiBF,CAAc,CAAC,EAUzG,IAAAG,EAAAC,EAEJC,GAAcR,CAAM,EAChBH,GAAmB,IAAI,SAACY,EAAU,CAAK,OAAA,SAACC,EAAY,CAAK,OAAAV,EAAOS,CAAU,EAAER,EAAWS,EAASR,CAA+B,CAAtE,CAAlB,CAAyF,EAElIS,GAAwBX,CAAM,EAC5BJ,GAAwB,IAAIgB,GAAwBZ,EAAQC,CAAS,CAAC,EACtEY,GAA0Bb,CAAM,EAChCF,GAAc,IAAIc,GAAwBZ,EAAQC,CAAS,CAAC,EAC5D,CAAA,EAAE,CAAA,EATDa,EAAGR,EAAA,CAAA,EAAES,EAAMT,EAAA,CAAA,EAgBlB,GAAI,CAACQ,GACCE,GAAYhB,CAAM,EACpB,OAAOiB,GAAS,SAACC,EAAc,CAAK,OAAAnB,EAAUmB,EAAWjB,EAAWC,CAA+B,CAA/D,CAAgE,EAClGiB,EAAUnB,CAAM,CAAC,EAOvB,GAAI,CAACc,EACH,MAAM,IAAI,UAAU,sBAAsB,EAG5C,OAAO,IAAIM,EAAc,SAACC,EAAU,CAIlC,IAAMX,EAAU,UAAA,SAACY,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAAmB,OAAAF,EAAW,KAAK,EAAIC,EAAK,OAASA,EAAOA,EAAK,CAAC,CAAC,CAAhD,EAEpC,OAAAR,EAAIJ,CAAO,EAEJ,UAAA,CAAM,OAAAK,EAAQL,CAAO,CAAf,CACf,CAAC,CACH,CASA,SAASE,GAAwBZ,EAAaC,EAAiB,CAC7D,OAAO,SAACQ,EAAkB,CAAK,OAAA,SAACC,EAAY,CAAK,OAAAV,EAAOS,CAAU,EAAER,EAAWS,CAAO,CAArC,CAAlB,CACjC,CAOA,SAASC,GAAwBX,EAAW,CAC1C,OAAOI,EAAWJ,EAAO,WAAW,GAAKI,EAAWJ,EAAO,cAAc,CAC3E,CAOA,SAASa,GAA0Bb,EAAW,CAC5C,OAAOI,EAAWJ,EAAO,EAAE,GAAKI,EAAWJ,EAAO,GAAG,CACvD,CAOA,SAASQ,GAAcR,EAAW,CAChC,OAAOI,EAAWJ,EAAO,gBAAgB,GAAKI,EAAWJ,EAAO,mBAAmB,CACrF,CCnMM,SAAUwB,GACdC,EACAC,EACAC,EAAsC,CAEtC,OAAIA,EACKH,GAAoBC,EAAYC,CAAa,EAAE,KAAKE,GAAiBD,CAAc,CAAC,EAGtF,IAAIE,EAAoB,SAACC,EAAU,CACxC,IAAMC,EAAU,UAAA,SAACC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAAc,OAAAH,EAAW,KAAKE,EAAE,SAAW,EAAIA,EAAE,CAAC,EAAIA,CAAC,CAAzC,EACzBE,EAAWT,EAAWM,CAAO,EACnC,OAAOI,EAAWT,CAAa,EAAI,UAAA,CAAM,OAAAA,EAAcK,EAASG,CAAQ,CAA/B,EAAmC,MAC9E,CAAC,CACH,CCtBM,SAAUE,GACdC,EACAC,EACAC,EAAyC,CAFzCF,IAAA,SAAAA,EAAA,GAEAE,IAAA,SAAAA,EAAAC,IAIA,IAAIC,EAAmB,GAEvB,OAAIH,GAAuB,OAIrBI,GAAYJ,CAAmB,EACjCC,EAAYD,EAIZG,EAAmBH,GAIhB,IAAIK,EAAW,SAACC,EAAU,CAI/B,IAAIC,EAAMC,GAAYT,CAAO,EAAI,CAACA,EAAUE,EAAW,IAAG,EAAKF,EAE3DQ,EAAM,IAERA,EAAM,GAIR,IAAIE,EAAI,EAGR,OAAOR,EAAU,SAAS,UAAA,CACnBK,EAAW,SAEdA,EAAW,KAAKG,GAAG,EAEf,GAAKN,EAGP,KAAK,SAAS,OAAWA,CAAgB,EAGzCG,EAAW,SAAQ,EAGzB,EAAGC,CAAG,CACR,CAAC,CACH,CChGM,SAAUG,GAAK,SAACC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACpB,IAAMC,EAAYC,GAAaH,CAAI,EAC7BI,EAAaC,GAAUL,EAAM,GAAQ,EACrCM,EAAUN,EAChB,OAAQM,EAAQ,OAGZA,EAAQ,SAAW,EAEnBC,EAAUD,EAAQ,CAAC,CAAC,EAEpBE,GAASJ,CAAU,EAAEK,GAAKH,EAASJ,CAAS,CAAC,EAL7CQ,CAMN,CCjEO,IAAMC,GAAQ,IAAIC,EAAkBC,EAAI,ECpCvC,IAAAC,GAAY,MAAK,QAMnB,SAAUC,GAAkBC,EAAiB,CACjD,OAAOA,EAAK,SAAW,GAAKF,GAAQE,EAAK,CAAC,CAAC,EAAIA,EAAK,CAAC,EAAKA,CAC5D,CCoDM,SAAUC,EAAUC,EAAiDC,EAAa,CACtF,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAEhC,IAAIC,EAAQ,EAIZF,EAAO,UAILG,EAAyBF,EAAY,SAACG,EAAK,CAAK,OAAAP,EAAU,KAAKC,EAASM,EAAOF,GAAO,GAAKD,EAAW,KAAKG,CAAK,CAAhE,CAAiE,CAAC,CAEtH,CAAC,CACH,CCxBM,SAAUC,IAAG,SAACC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAClB,IAAMC,EAAiBC,GAAkBH,CAAI,EAEvCI,EAAUC,GAAeL,CAAI,EAEnC,OAAOI,EAAQ,OACX,IAAIE,EAAsB,SAACC,EAAU,CAGnC,IAAIC,EAAuBJ,EAAQ,IAAI,UAAA,CAAM,MAAA,CAAA,CAAA,CAAE,EAK3CK,EAAYL,EAAQ,IAAI,UAAA,CAAM,MAAA,EAAA,CAAK,EAGvCG,EAAW,IAAI,UAAA,CACbC,EAAUC,EAAY,IACxB,CAAC,EAKD,mBAASC,EAAW,CAClBC,EAAUP,EAAQM,CAAW,CAAC,EAAE,UAC9BE,EACEL,EACA,SAACM,EAAK,CAKJ,GAJAL,EAAQE,CAAW,EAAE,KAAKG,CAAK,EAI3BL,EAAQ,MAAM,SAACM,EAAM,CAAK,OAAAA,EAAO,MAAP,CAAa,EAAG,CAC5C,IAAMC,EAAcP,EAAQ,IAAI,SAACM,EAAM,CAAK,OAAAA,EAAO,MAAK,CAAZ,CAAe,EAE3DP,EAAW,KAAKL,EAAiBA,EAAc,MAAA,OAAAc,EAAA,CAAA,EAAAC,EAAIF,CAAM,CAAA,CAAA,EAAIA,CAAM,EAI/DP,EAAQ,KAAK,SAACM,EAAQI,EAAC,CAAK,MAAA,CAACJ,EAAO,QAAUL,EAAUS,CAAC,CAA7B,CAA8B,GAC5DX,EAAW,SAAQ,EAGzB,EACA,UAAA,CAGEE,EAAUC,CAAW,EAAI,GAIzB,CAACF,EAAQE,CAAW,EAAE,QAAUH,EAAW,SAAQ,CACrD,CAAC,CACF,GA9BIG,EAAc,EAAG,CAACH,EAAW,QAAUG,EAAcN,EAAQ,OAAQM,MAArEA,CAAW,EAmCpB,OAAO,UAAA,CACLF,EAAUC,EAAY,IACxB,CACF,CAAC,EACDU,CACN,CC9DM,SAAUC,GAASC,EAAoD,CAC3E,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAW,GACXC,EAAsB,KACtBC,EAA6C,KAC7CC,EAAa,GAEXC,EAAc,UAAA,CAGlB,GAFAF,GAAkB,MAAlBA,EAAoB,YAAW,EAC/BA,EAAqB,KACjBF,EAAU,CACZA,EAAW,GACX,IAAMK,EAAQJ,EACdA,EAAY,KACZF,EAAW,KAAKM,CAAK,EAEvBF,GAAcJ,EAAW,SAAQ,CACnC,EAEMO,EAAkB,UAAA,CACtBJ,EAAqB,KACrBC,GAAcJ,EAAW,SAAQ,CACnC,EAEAD,EAAO,UACLS,EACER,EACA,SAACM,EAAK,CACJL,EAAW,GACXC,EAAYI,EACPH,GACHM,EAAUZ,EAAiBS,CAAK,CAAC,EAAE,UAChCH,EAAqBK,EAAyBR,EAAYK,EAAaE,CAAe,CAAE,CAG/F,EACA,UAAA,CACEH,EAAa,IACZ,CAACH,GAAY,CAACE,GAAsBA,EAAmB,SAAWH,EAAW,SAAQ,CACxF,CAAC,CACF,CAEL,CAAC,CACH,CC3CM,SAAUU,GAAaC,EAAkBC,EAAyC,CAAzC,OAAAA,IAAA,SAAAA,EAAAC,IACtCC,GAAM,UAAA,CAAM,OAAAC,GAAMJ,EAAUC,CAAS,CAAzB,CAA0B,CAC/C,CCEM,SAAUI,GAAeC,EAAoBC,EAAsC,CAAtC,OAAAA,IAAA,SAAAA,EAAA,MAGjDA,EAAmBA,GAAgB,KAAhBA,EAAoBD,EAEhCE,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAiB,CAAA,EACjBC,EAAQ,EAEZH,EAAO,UACLI,EACEH,EACA,SAACI,EAAK,aACAC,EAAuB,KAKvBH,IAAUL,IAAsB,GAClCI,EAAQ,KAAK,CAAA,CAAE,MAIjB,QAAqBK,EAAAC,GAAAN,CAAO,EAAAO,EAAAF,EAAA,KAAA,EAAA,CAAAE,EAAA,KAAAA,EAAAF,EAAA,KAAA,EAAE,CAAzB,IAAMG,EAAMD,EAAA,MACfC,EAAO,KAAKL,CAAK,EAMbR,GAAca,EAAO,SACvBJ,EAASA,GAAM,KAANA,EAAU,CAAA,EACnBA,EAAO,KAAKI,CAAM,qGAItB,GAAIJ,MAIF,QAAqBK,EAAAH,GAAAF,CAAM,EAAAM,EAAAD,EAAA,KAAA,EAAA,CAAAC,EAAA,KAAAA,EAAAD,EAAA,KAAA,EAAE,CAAxB,IAAMD,EAAME,EAAA,MACfC,GAAUX,EAASQ,CAAM,EACzBT,EAAW,KAAKS,CAAM,oGAG5B,EACA,UAAA,aAGE,QAAqBI,EAAAN,GAAAN,CAAO,EAAAa,EAAAD,EAAA,KAAA,EAAA,CAAAC,EAAA,KAAAA,EAAAD,EAAA,KAAA,EAAE,CAAzB,IAAMJ,EAAMK,EAAA,MACfd,EAAW,KAAKS,CAAM,oGAExBT,EAAW,SAAQ,CACrB,EAEA,OACA,UAAA,CAEEC,EAAU,IACZ,CAAC,CACF,CAEL,CAAC,CACH,CCbM,SAAUc,GACdC,EAAgD,CAEhD,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAgC,KAChCC,EAAY,GACZC,EAEJF,EAAWF,EAAO,UAChBK,EAAyBJ,EAAY,OAAW,OAAW,SAACK,EAAG,CAC7DF,EAAgBG,EAAUT,EAASQ,EAAKT,GAAWC,CAAQ,EAAEE,CAAM,CAAC,CAAC,EACjEE,GACFA,EAAS,YAAW,EACpBA,EAAW,KACXE,EAAc,UAAUH,CAAU,GAIlCE,EAAY,EAEhB,CAAC,CAAC,EAGAA,IAMFD,EAAS,YAAW,EACpBA,EAAW,KACXE,EAAe,UAAUH,CAAU,EAEvC,CAAC,CACH,CC/HM,SAAUO,GACdC,EACAC,EACAC,EACAC,EACAC,EAAqC,CAErC,OAAO,SAACC,EAAuBC,EAA2B,CAIxD,IAAIC,EAAWL,EAIXM,EAAaP,EAEbQ,EAAQ,EAGZJ,EAAO,UACLK,EACEJ,EACA,SAACK,EAAK,CAEJ,IAAMC,EAAIH,IAEVD,EAAQD,EAEJP,EAAYQ,EAAOG,EAAOC,CAAC,GAIzBL,EAAW,GAAOI,GAGxBR,GAAcG,EAAW,KAAKE,CAAK,CACrC,EAGAJ,GACG,UAAA,CACCG,GAAYD,EAAW,KAAKE,CAAK,EACjCF,EAAW,SAAQ,CACrB,CAAE,CACL,CAEL,CACF,CCnCM,SAAUO,IAAa,SAAOC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAClC,IAAMC,EAAiBC,GAAkBH,CAAI,EAC7C,OAAOE,EACHE,GAAKL,GAAa,MAAA,OAAAM,EAAA,CAAA,EAAAC,EAAKN,CAAoC,CAAA,CAAA,EAAGO,GAAiBL,CAAc,CAAC,EAC9FM,EAAQ,SAACC,EAAQC,EAAU,CACzBC,GAAiBN,EAAA,CAAEI,CAAM,EAAAH,EAAKM,GAAeZ,CAAI,CAAC,CAAA,CAAA,EAAGU,CAAU,CACjE,CAAC,CACP,CCUM,SAAUG,IAAiB,SAC/BC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAEA,OAAOC,GAAa,MAAA,OAAAC,EAAA,CAAA,EAAAC,EAAIJ,CAAY,CAAA,CAAA,CACtC,CCgBM,SAAUK,GAAgBC,EAAiBC,EAAyC,CAAzC,OAAAA,IAAA,SAAAA,EAAAC,IACxCC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAkC,KAClCC,EAAsB,KACtBC,EAA0B,KAExBC,EAAO,UAAA,CACX,GAAIH,EAAY,CAEdA,EAAW,YAAW,EACtBA,EAAa,KACb,IAAMI,EAAQH,EACdA,EAAY,KACZF,EAAW,KAAKK,CAAK,EAEzB,EACA,SAASC,GAAY,CAInB,IAAMC,EAAaJ,EAAYR,EACzBa,EAAMZ,EAAU,IAAG,EACzB,GAAIY,EAAMD,EAAY,CAEpBN,EAAa,KAAK,SAAS,OAAWM,EAAaC,CAAG,EACtDR,EAAW,IAAIC,CAAU,EACzB,OAGFG,EAAI,CACN,CAEAL,EAAO,UACLU,EACET,EACA,SAACK,EAAQ,CACPH,EAAYG,EACZF,EAAWP,EAAU,IAAG,EAGnBK,IACHA,EAAaL,EAAU,SAASU,EAAcX,CAAO,EACrDK,EAAW,IAAIC,CAAU,EAE7B,EACA,UAAA,CAGEG,EAAI,EACJJ,EAAW,SAAQ,CACrB,EAEA,OACA,UAAA,CAEEE,EAAYD,EAAa,IAC3B,CAAC,CACF,CAEL,CAAC,CACH,CCpFM,SAAUS,GAAqBC,EAAe,CAClD,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAW,GACfF,EAAO,UACLG,EACEF,EACA,SAACG,EAAK,CACJF,EAAW,GACXD,EAAW,KAAKG,CAAK,CACvB,EACA,UAAA,CACOF,GACHD,EAAW,KAAKH,CAAa,EAE/BG,EAAW,SAAQ,CACrB,CAAC,CACF,CAEL,CAAC,CACH,CCXM,SAAUI,GAAQC,EAAa,CACnC,OAAOA,GAAS,EAEZ,UAAA,CAAM,OAAAC,CAAA,EACNC,EAAQ,SAACC,EAAQC,EAAU,CACzB,IAAIC,EAAO,EACXF,EAAO,UACLG,EAAyBF,EAAY,SAACG,EAAK,CAIrC,EAAEF,GAAQL,IACZI,EAAW,KAAKG,CAAK,EAIjBP,GAASK,GACXD,EAAW,SAAQ,EAGzB,CAAC,CAAC,CAEN,CAAC,CACP,CC9BM,SAAUI,GAAc,CAC5B,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChCD,EAAO,UAAUE,EAAyBD,EAAYE,EAAI,CAAC,CAC7D,CAAC,CACH,CCCM,SAAUC,GAASC,EAAQ,CAC/B,OAAOC,EAAI,UAAA,CAAM,OAAAD,CAAA,CAAK,CACxB,CC4CM,SAAUE,GACdC,EACAC,EAAmC,CAEnC,OAAIA,EAEK,SAACC,EAAqB,CAC3B,OAAAC,GAAOF,EAAkB,KAAKG,GAAK,CAAC,EAAGC,EAAc,CAAE,EAAGH,EAAO,KAAKH,GAAUC,CAAqB,CAAC,CAAC,CAAvG,EAGGM,GAAS,SAACC,EAAOC,EAAK,CAAK,OAAAC,EAAUT,EAAsBO,EAAOC,CAAK,CAAC,EAAE,KAAKJ,GAAK,CAAC,EAAGM,GAAMH,CAAK,CAAC,CAAzE,CAA0E,CAC9G,CCzCM,SAAUI,GAASC,EAAoBC,EAAyC,CAAzCA,IAAA,SAAAA,EAAAC,IAC3C,IAAMC,EAAWC,GAAMJ,EAAKC,CAAS,EACrC,OAAOI,GAAU,UAAA,CAAM,OAAAF,CAAA,CAAQ,CACjC,CC0EM,SAAUG,EACdC,EACAC,EAA0D,CAA1D,OAAAA,IAAA,SAAAA,EAA+BC,IAK/BF,EAAaA,GAAU,KAAVA,EAAcG,GAEpBC,EAAQ,SAACC,EAAQC,EAAU,CAGhC,IAAIC,EAEAC,EAAQ,GAEZH,EAAO,UACLI,EAAyBH,EAAY,SAACI,EAAK,CAEzC,IAAMC,EAAaV,EAAYS,CAAK,GAKhCF,GAAS,CAACR,EAAYO,EAAaI,CAAU,KAM/CH,EAAQ,GACRD,EAAcI,EAGdL,EAAW,KAAKI,CAAK,EAEzB,CAAC,CAAC,CAEN,CAAC,CACH,CAEA,SAASP,GAAeS,EAAQC,EAAM,CACpC,OAAOD,IAAMC,CACf,CCjHM,SAAUC,GAA8CC,EAAQC,EAAuC,CAC3G,OAAOC,EAAqB,SAACC,EAAMC,EAAI,CAAK,OAAAH,EAAUA,EAAQE,EAAEH,CAAG,EAAGI,EAAEJ,CAAG,CAAC,EAAIG,EAAEH,CAAG,IAAMI,EAAEJ,CAAG,CAApD,CAAqD,CACnG,CC7BM,SAAUK,GAAgBC,EAA6C,CAA7C,OAAAA,IAAA,SAAAA,EAAAC,IACvBC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAW,GACfF,EAAO,UACLG,EACEF,EACA,SAACG,EAAK,CACJF,EAAW,GACXD,EAAW,KAAKG,CAAK,CACvB,EACA,UAAA,CAAM,OAACF,EAAWD,EAAW,SAAQ,EAAKA,EAAW,MAAMJ,EAAY,CAAE,CAAnE,CAAqE,CAC5E,CAEL,CAAC,CACH,CAEA,SAASC,IAAmB,CAC1B,OAAO,IAAIO,EACb,CCMM,SAAUC,IAAO,SAAIC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACzB,OAAO,SAACC,EAAqB,CAAK,OAAAC,GAAOD,EAAQE,EAAE,MAAA,OAAAC,EAAA,CAAA,EAAAC,EAAIN,CAAM,CAAA,CAAA,CAAA,CAA3B,CACpC,CCHM,SAAUO,EAAYC,EAAoB,CAC9C,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAGhC,GAAI,CACFD,EAAO,UAAUC,CAAU,UAE3BA,EAAW,IAAIH,CAAQ,EAE3B,CAAC,CACH,CCMM,SAAUI,GACdC,EACAC,EAAgB,CAEhB,IAAMC,EAAkB,UAAU,QAAU,EAC5C,OAAO,SAACC,EAAqB,CAC3B,OAAAA,EAAO,KACLH,EAAYI,EAAO,SAACC,EAAG,EAAC,CAAK,OAAAL,EAAUK,EAAG,EAAGF,CAAM,CAAtB,CAAuB,EAAIG,GACxDC,GAAK,CAAC,EACNL,EAAkBM,GAAeP,CAAa,EAAIQ,GAAa,UAAA,CAAM,OAAA,IAAIC,EAAJ,CAAgB,CAAC,CAHxF,CAKJ,CCrEM,SAAUC,IAAK,SAAIC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACvB,IAAMC,EAAYC,GAAaH,CAAI,EAC7BI,EAAaC,GAAUL,EAAM,GAAQ,EAC3C,OAAAA,EAAOM,GAAeN,CAAI,EAEnBO,EAAQ,SAACC,EAAQC,EAAU,CAChCC,GAASN,CAAU,EAAEO,GAAIC,EAAA,CAAEJ,CAAM,EAAAK,EAAMb,CAA6B,CAAA,EAAGE,CAAS,CAAC,EAAE,UAAUO,CAAU,CACzG,CAAC,CACH,CCcM,SAAUK,IAAS,SACvBC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAEA,OAAOC,GAAK,MAAA,OAAAC,EAAA,CAAA,EAAAC,EAAIJ,CAAY,CAAA,CAAA,CAC9B,CCmEM,SAAUK,GAAUC,EAAqC,OACzDC,EAAQ,IACRC,EAEJ,OAAIF,GAAiB,OACf,OAAOA,GAAkB,UACxBG,EAA4BH,EAAa,MAAzCC,EAAKE,IAAA,OAAG,IAAQA,EAAED,EAAUF,EAAa,OAE5CC,EAAQD,GAILC,GAAS,EACZ,UAAA,CAAM,OAAAG,CAAA,EACNC,EAAQ,SAACC,EAAQC,EAAU,CACzB,IAAIC,EAAQ,EACRC,EAEEC,EAAc,UAAA,CAGlB,GAFAD,GAAS,MAATA,EAAW,YAAW,EACtBA,EAAY,KACRP,GAAS,KAAM,CACjB,IAAMS,EAAW,OAAOT,GAAU,SAAWU,GAAMV,CAAK,EAAIW,EAAUX,EAAMM,CAAK,CAAC,EAC5EM,EAAqBC,EAAyBR,EAAY,UAAA,CAC9DO,EAAmB,YAAW,EAC9BE,EAAiB,CACnB,CAAC,EACDL,EAAS,UAAUG,CAAkB,OAErCE,EAAiB,CAErB,EAEMA,EAAoB,UAAA,CACxB,IAAIC,EAAY,GAChBR,EAAYH,EAAO,UACjBS,EAAyBR,EAAY,OAAW,UAAA,CAC1C,EAAEC,EAAQP,EACRQ,EACFC,EAAW,EAEXO,EAAY,GAGdV,EAAW,SAAQ,CAEvB,CAAC,CAAC,EAGAU,GACFP,EAAW,CAEf,EAEAM,EAAiB,CACnB,CAAC,CACP,CC7HM,SAAUE,GAAUC,EAA8B,CACtD,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAW,GACXC,EAAsB,KAC1BH,EAAO,UACLI,EAAyBH,EAAY,SAACI,EAAK,CACzCH,EAAW,GACXC,EAAYE,CACd,CAAC,CAAC,EAEJC,EAAUR,CAAQ,EAAE,UAClBM,EACEH,EACA,UAAA,CACE,GAAIC,EAAU,CACZA,EAAW,GACX,IAAMG,EAAQF,EACdA,EAAY,KACZF,EAAW,KAAKI,CAAK,EAEzB,EACAE,EAAI,CACL,CAEL,CAAC,CACH,CCgBM,SAAUC,GAAcC,EAA6DC,EAAQ,CAMjG,OAAOC,EAAQC,GAAcH,EAAaC,EAAW,UAAU,QAAU,EAAG,EAAI,CAAC,CACnF,CC+CM,SAAUG,GAASC,EAA4B,CAA5BA,IAAA,SAAAA,EAAA,CAAA,GACf,IAAAC,EAAgHD,EAAO,UAAvHE,EAASD,IAAA,OAAG,UAAA,CAAM,OAAA,IAAIE,CAAJ,EAAgBF,EAAEG,EAA4EJ,EAAO,aAAnFK,EAAYD,IAAA,OAAG,GAAIA,EAAEE,EAAuDN,EAAO,gBAA9DO,EAAeD,IAAA,OAAG,GAAIA,EAAEE,EAA+BR,EAAO,oBAAtCS,EAAmBD,IAAA,OAAG,GAAIA,EAUnH,OAAO,SAACE,EAAa,CACnB,IAAIC,EACAC,EACAC,EACAC,EAAW,EACXC,EAAe,GACfC,EAAa,GAEXC,EAAc,UAAA,CAClBL,GAAe,MAAfA,EAAiB,YAAW,EAC5BA,EAAkB,MACpB,EAGMM,EAAQ,UAAA,CACZD,EAAW,EACXN,EAAaE,EAAU,OACvBE,EAAeC,EAAa,EAC9B,EACMG,EAAsB,UAAA,CAG1B,IAAMC,EAAOT,EACbO,EAAK,EACLE,GAAI,MAAJA,EAAM,YAAW,CACnB,EAEA,OAAOC,EAAc,SAACC,EAAQC,GAAU,CACtCT,IACI,CAACE,GAAc,CAACD,GAClBE,EAAW,EAOb,IAAMO,GAAQX,EAAUA,GAAO,KAAPA,EAAWX,EAAS,EAO5CqB,GAAW,IAAI,UAAA,CACbT,IAKIA,IAAa,GAAK,CAACE,GAAc,CAACD,IACpCH,EAAkBa,GAAYN,EAAqBV,CAAmB,EAE1E,CAAC,EAIDe,GAAK,UAAUD,EAAU,EAGvB,CAACZ,GAIDG,EAAW,IAOXH,EAAa,IAAIe,GAAe,CAC9B,KAAM,SAACC,GAAK,CAAK,OAAAH,GAAK,KAAKG,EAAK,CAAf,EACjB,MAAO,SAACC,GAAG,CACTZ,EAAa,GACbC,EAAW,EACXL,EAAkBa,GAAYP,EAAOb,EAAcuB,EAAG,EACtDJ,GAAK,MAAMI,EAAG,CAChB,EACA,SAAU,UAAA,CACRb,EAAe,GACfE,EAAW,EACXL,EAAkBa,GAAYP,EAAOX,CAAe,EACpDiB,GAAK,SAAQ,CACf,EACD,EACDK,EAAUP,CAAM,EAAE,UAAUX,CAAU,EAE1C,CAAC,EAAED,CAAa,CAClB,CACF,CAEA,SAASe,GACPP,EACAY,EAAoD,SACpDC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,EAAA,CAAA,EAAA,UAAAA,CAAA,EAEA,GAAIF,IAAO,GAAM,CACfZ,EAAK,EACL,OAGF,GAAIY,IAAO,GAIX,KAAMG,EAAe,IAAIP,GAAe,CACtC,KAAM,UAAA,CACJO,EAAa,YAAW,EACxBf,EAAK,CACP,EACD,EAED,OAAOW,EAAUC,EAAE,MAAA,OAAAI,EAAA,CAAA,EAAAC,EAAIJ,CAAI,CAAA,CAAA,CAAA,EAAG,UAAUE,CAAY,EACtD,CChHM,SAAUG,EACdC,EACAC,EACAC,EAAyB,WAErBC,EACAC,EAAW,GACf,OAAIJ,GAAsB,OAAOA,GAAuB,UACnDK,EAA8EL,EAAkB,WAAhGG,EAAUE,IAAA,OAAG,IAAQA,EAAEC,EAAuDN,EAAkB,WAAzEC,EAAUK,IAAA,OAAG,IAAQA,EAAEC,EAAgCP,EAAkB,SAAlDI,EAAQG,IAAA,OAAG,GAAKA,EAAEL,EAAcF,EAAkB,WAEnGG,EAAcH,GAAkB,KAAlBA,EAAsB,IAE/BQ,GAAS,CACd,UAAW,UAAA,CAAM,OAAA,IAAIC,GAAcN,EAAYF,EAAYC,CAAS,CAAnD,EACjB,aAAc,GACd,gBAAiB,GACjB,oBAAqBE,EACtB,CACH,CCxIM,SAAUM,GAAQC,EAAa,CACnC,OAAOC,EAAO,SAACC,EAAGC,EAAK,CAAK,OAAAH,GAASG,CAAT,CAAc,CAC5C,CCaM,SAAUC,GAAaC,EAA8B,CACzD,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAS,GAEPC,EAAiBC,EACrBH,EACA,UAAA,CACEE,GAAc,MAAdA,EAAgB,YAAW,EAC3BD,EAAS,EACX,EACAG,EAAI,EAGNC,EAAUR,CAAQ,EAAE,UAAUK,CAAc,EAE5CH,EAAO,UAAUI,EAAyBH,EAAY,SAACM,EAAK,CAAK,OAAAL,GAAUD,EAAW,KAAKM,CAAK,CAA/B,CAAgC,CAAC,CACpG,CAAC,CACH,CCVM,SAAUC,GAAS,SAAOC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EAC9B,IAAMC,EAAYC,GAAaH,CAAM,EACrC,OAAOI,EAAQ,SAACC,EAAQC,EAAU,EAI/BJ,EAAYK,GAAOP,EAAQK,EAAQH,CAAS,EAAIK,GAAOP,EAAQK,CAAM,GAAG,UAAUC,CAAU,CAC/F,CAAC,CACH,CCmBM,SAAUE,EACdC,EACAC,EAA6G,CAE7G,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAyD,KACzDC,EAAQ,EAERC,EAAa,GAIXC,EAAgB,UAAA,CAAM,OAAAD,GAAc,CAACF,GAAmBD,EAAW,SAAQ,CAArD,EAE5BD,EAAO,UACLM,EACEL,EACA,SAACM,EAAK,CAEJL,GAAe,MAAfA,EAAiB,YAAW,EAC5B,IAAIM,EAAa,EACXC,EAAaN,IAEnBO,EAAUb,EAAQU,EAAOE,CAAU,CAAC,EAAE,UACnCP,EAAkBI,EACjBL,EAIA,SAACU,EAAU,CAAK,OAAAV,EAAW,KAAKH,EAAiBA,EAAeS,EAAOI,EAAYF,EAAYD,GAAY,EAAIG,CAAU,CAAzG,EAChB,UAAA,CAIET,EAAkB,KAClBG,EAAa,CACf,CAAC,CACD,CAEN,EACA,UAAA,CACED,EAAa,GACbC,EAAa,CACf,CAAC,CACF,CAEL,CAAC,CACH,CCvFM,SAAUO,EAAaC,EAA8B,CACzD,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChCC,EAAUJ,CAAQ,EAAE,UAAUK,EAAyBF,EAAY,UAAA,CAAM,OAAAA,EAAW,SAAQ,CAAnB,EAAuBG,EAAI,CAAC,EACrG,CAACH,EAAW,QAAUD,EAAO,UAAUC,CAAU,CACnD,CAAC,CACH,CCIM,SAAUI,GAAaC,EAAiDC,EAAiB,CAAjB,OAAAA,IAAA,SAAAA,EAAA,IACrEC,EAAQ,SAACC,EAAQC,EAAU,CAChC,IAAIC,EAAQ,EACZF,EAAO,UACLG,EAAyBF,EAAY,SAACG,EAAK,CACzC,IAAMC,EAASR,EAAUO,EAAOF,GAAO,GACtCG,GAAUP,IAAcG,EAAW,KAAKG,CAAK,EAC9C,CAACC,GAAUJ,EAAW,SAAQ,CAChC,CAAC,CAAC,CAEN,CAAC,CACH,CCqGM,SAAUK,EACdC,EACAC,EACAC,EAA8B,CAK9B,IAAMC,EACJC,EAAWJ,CAAc,GAAKC,GAASC,EAElC,CAAE,KAAMF,EAA2E,MAAKC,EAAE,SAAQC,CAAA,EACnGF,EAEN,OAAOG,EACHE,EAAQ,SAACC,EAAQC,EAAU,QACzBC,EAAAL,EAAY,aAAS,MAAAK,IAAA,QAAAA,EAAA,KAArBL,CAAW,EACX,IAAIM,EAAU,GACdH,EAAO,UACLI,EACEH,EACA,SAACI,EAAK,QACJH,EAAAL,EAAY,QAAI,MAAAK,IAAA,QAAAA,EAAA,KAAhBL,EAAmBQ,CAAK,EACxBJ,EAAW,KAAKI,CAAK,CACvB,EACA,UAAA,OACEF,EAAU,IACVD,EAAAL,EAAY,YAAQ,MAAAK,IAAA,QAAAA,EAAA,KAApBL,CAAW,EACXI,EAAW,SAAQ,CACrB,EACA,SAACK,EAAG,OACFH,EAAU,IACVD,EAAAL,EAAY,SAAK,MAAAK,IAAA,QAAAA,EAAA,KAAjBL,EAAoBS,CAAG,EACvBL,EAAW,MAAMK,CAAG,CACtB,EACA,UAAA,SACMH,KACFD,EAAAL,EAAY,eAAW,MAAAK,IAAA,QAAAA,EAAA,KAAvBL,CAAW,IAEbU,EAAAV,EAAY,YAAQ,MAAAU,IAAA,QAAAA,EAAA,KAApBV,CAAW,CACb,CAAC,CACF,CAEL,CAAC,EAIDW,EACN,CCnIM,SAAUC,GAAYC,EAAsDC,EAAuB,CACvG,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAC1B,IAAAC,EAAuCJ,GAAM,KAANA,EAAU,CAAA,EAA/CK,EAAAD,EAAA,QAAAE,EAAOD,IAAA,OAAG,GAAIA,EAAEE,EAAAH,EAAA,SAAAI,EAAQD,IAAA,OAAG,GAAKA,EACpCE,EAAW,GACXC,EAAsB,KACtBC,EAAiC,KACjCC,EAAa,GAEXC,EAAgB,UAAA,CACpBF,GAAS,MAATA,EAAW,YAAW,EACtBA,EAAY,KACRH,IACFM,EAAI,EACJF,GAAcT,EAAW,SAAQ,EAErC,EAEMY,EAAoB,UAAA,CACxBJ,EAAY,KACZC,GAAcT,EAAW,SAAQ,CACnC,EAEMa,EAAgB,SAACC,EAAQ,CAC7B,OAACN,EAAYO,EAAUnB,EAAiBkB,CAAK,CAAC,EAAE,UAAUE,EAAyBhB,EAAYU,EAAeE,CAAiB,CAAC,CAAhI,EAEID,EAAO,UAAA,CACX,GAAIL,EAAU,CAIZA,EAAW,GACX,IAAMQ,EAAQP,EACdA,EAAY,KAEZP,EAAW,KAAKc,CAAK,EACrB,CAACL,GAAcI,EAAcC,CAAK,EAEtC,EAEAf,EAAO,UACLiB,EACEhB,EAMA,SAACc,EAAK,CACJR,EAAW,GACXC,EAAYO,EACZ,EAAEN,GAAa,CAACA,EAAU,UAAYL,EAAUQ,EAAI,EAAKE,EAAcC,CAAK,EAC9E,EACA,UAAA,CACEL,EAAa,GACb,EAAEJ,GAAYC,GAAYE,GAAa,CAACA,EAAU,SAAWR,EAAW,SAAQ,CAClF,CAAC,CACF,CAEL,CAAC,CACH,CCxFM,SAAUiB,GACdC,EACAC,EACAC,EAAuB,CADvBD,IAAA,SAAAA,EAAAE,IAGA,IAAMC,EAAYC,GAAML,EAAUC,CAAS,EAC3C,OAAOK,GAAS,UAAA,CAAM,OAAAF,CAAA,EAAWF,CAAM,CACzC,CCJM,SAAUK,IAAc,SAAOC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACnC,IAAMC,EAAUC,GAAkBH,CAAM,EAExC,OAAOI,EAAQ,SAACC,EAAQC,EAAU,CAehC,QAdMC,EAAMP,EAAO,OACbQ,EAAc,IAAI,MAAMD,CAAG,EAI7BE,EAAWT,EAAO,IAAI,UAAA,CAAM,MAAA,EAAA,CAAK,EAGjCU,EAAQ,cAMHC,EAAC,CACRC,EAAUZ,EAAOW,CAAC,CAAC,EAAE,UACnBE,EACEP,EACA,SAACQ,EAAK,CACJN,EAAYG,CAAC,EAAIG,EACb,CAACJ,GAAS,CAACD,EAASE,CAAC,IAEvBF,EAASE,CAAC,EAAI,IAKbD,EAAQD,EAAS,MAAMM,EAAQ,KAAON,EAAW,MAEtD,EAGAO,EAAI,CACL,GAnBIL,EAAI,EAAGA,EAAIJ,EAAKI,MAAhBA,CAAC,EAwBVN,EAAO,UACLQ,EAAyBP,EAAY,SAACQ,EAAK,CACzC,GAAIJ,EAAO,CAET,IAAMO,EAAMC,EAAA,CAAIJ,CAAK,EAAAK,EAAKX,CAAW,CAAA,EACrCF,EAAW,KAAKJ,EAAUA,EAAO,MAAA,OAAAgB,EAAA,CAAA,EAAAC,EAAIF,CAAM,CAAA,CAAA,EAAIA,CAAM,EAEzD,CAAC,CAAC,CAEN,CAAC,CACH,CCxFM,SAAUG,IAAG,SAAOC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACxB,OAAOC,EAAQ,SAACC,EAAQC,EAAU,CAChCL,GAAS,MAAA,OAAAM,EAAA,CAACF,CAA8B,EAAAG,EAAMN,CAAuC,CAAA,CAAA,EAAE,UAAUI,CAAU,CAC7G,CAAC,CACH,CCCM,SAAUG,IAAO,SAAkCC,EAAA,CAAA,EAAAC,EAAA,EAAAA,EAAA,UAAA,OAAAA,IAAAD,EAAAC,CAAA,EAAA,UAAAA,CAAA,EACvD,OAAOC,GAAG,MAAA,OAAAC,EAAA,CAAA,EAAAC,EAAIJ,CAAW,CAAA,CAAA,CAC3B,CCYO,SAASK,IAAmC,CACjD,IAAMC,EAAY,IAAIC,GAAwB,CAAC,EAC/C,OAAAC,EAAU,SAAU,mBAAoB,CAAE,KAAM,EAAK,CAAC,EACnD,UAAU,IAAMF,EAAU,KAAK,QAAQ,CAAC,EAGpCA,CACT,CCHO,SAASG,EACdC,EAAkBC,EAAmB,SAChC,CACL,OAAO,MAAM,KAAKA,EAAK,iBAAoBD,CAAQ,CAAC,CACtD,CAuBO,SAASE,EACdF,EAAkBC,EAAmB,SAClC,CACH,IAAME,EAAKC,GAAsBJ,EAAUC,CAAI,EAC/C,GAAI,OAAOE,GAAO,YAChB,MAAM,IAAI,eACR,8BAA8BH,CAAQ,iBACxC,EAGF,OAAOG,CACT,CAsBO,SAASC,GACdJ,EAAkBC,EAAmB,SACtB,CACf,OAAOA,EAAK,cAAiBD,CAAQ,GAAK,MAC5C,CAOO,SAASK,IAA4C,CAC1D,OAAO,SAAS,yBAAyB,aACrC,SAAS,eAAiB,MAEhC,CCrEA,IAAMC,GAAYC,EAChBC,EAAU,SAAS,KAAM,SAAS,EAClCA,EAAU,SAAS,KAAM,UAAU,CACrC,EACG,KACCC,GAAa,CAAC,EACdC,EAAU,MAAS,EACnBC,EAAI,IAAMC,GAAiB,GAAK,SAAS,IAAI,EAC7CC,EAAY,CAAC,CACf,EAaK,SAASC,GACdC,EACqB,CACrB,OAAOT,GACJ,KACCK,EAAIK,GAAUD,EAAG,SAASC,CAAM,CAAC,EACjCC,EAAqB,CACvB,CACJ,CCzBO,SAASC,GACdC,EACe,CACf,MAAO,CACL,EAAGA,EAAG,WACN,EAAGA,EAAG,SACR,CACF,CAWO,SAASC,GACdD,EAC2B,CAC3B,OAAOE,EACLC,EAAU,OAAQ,MAAM,EACxBA,EAAU,OAAQ,QAAQ,CAC5B,EACG,KACCC,GAAU,EAAGC,EAAuB,EACpCC,EAAI,IAAMP,GAAiBC,CAAE,CAAC,EAC9BO,EAAUR,GAAiBC,CAAE,CAAC,CAChC,CACJ,CCxCO,SAASQ,GACdC,EACe,CACf,MAAO,CACL,EAAGA,EAAG,WACN,EAAGA,EAAG,SACR,CACF,CAWO,SAASC,GACdD,EAC2B,CAC3B,OAAOE,EACLC,EAAUH,EAAI,QAAQ,EACtBG,EAAU,OAAQ,QAAQ,CAC5B,EACG,KACCC,GAAU,EAAGC,EAAuB,EACpCC,EAAI,IAAMP,GAAwBC,CAAE,CAAC,EACrCO,EAAUR,GAAwBC,CAAE,CAAC,CACvC,CACJ,CCnBA,SAASQ,GAAYC,EAAiBC,EAA8B,CAGlE,GAAI,OAAOA,GAAU,UAAY,OAAOA,GAAU,SAChDD,EAAG,WAAaC,EAAM,SAAS,UAGtBA,aAAiB,KAC1BD,EAAG,YAAYC,CAAK,UAGX,MAAM,QAAQA,CAAK,EAC5B,QAAWC,KAAQD,EACjBF,GAAYC,EAAIE,CAAI,CAE1B,CAyBO,SAASC,EACdC,EAAaC,KAAmCC,EAC7C,CACH,IAAMN,EAAK,SAAS,cAAcI,CAAG,EAGrC,GAAIC,EACF,QAAWE,KAAQ,OAAO,KAAKF,CAAU,EACnC,OAAOA,EAAWE,CAAI,GAAM,cAI5B,OAAOF,EAAWE,CAAI,GAAM,UAC9BP,EAAG,aAAaO,EAAMF,EAAWE,CAAI,CAAC,EAEtCP,EAAG,aAAaO,EAAM,EAAE,GAI9B,QAAWN,KAASK,EAClBP,GAAYC,EAAIC,CAAK,EAGvB,OAAOD,CACT,CC9EO,SAASQ,GAAMC,EAAuB,CAC3C,GAAIA,EAAQ,IAAK,CACf,IAAMC,EAAS,GAAGD,EAAQ,KAAO,IAAO,IACxC,MAAO,KAAKA,EAAQ,MAAY,KAAM,QAAQC,CAAM,CAAC,GACvD,KACE,QAAOD,EAAM,SAAS,CAE1B,CCCO,SAASE,GAAYC,EAA+B,CACzD,IAAMC,EAASC,EAAE,SAAU,CAAE,IAAAF,CAAI,CAAC,EAClC,OAAOG,EAAM,KACX,SAAS,KAAK,YAAYF,CAAM,EACzBG,EACLC,EAAUJ,EAAQ,MAAM,EACxBI,EAAUJ,EAAQ,OAAO,EACtB,KACCK,EAAU,IACRC,GAAW,IAAM,IAAI,eAAe,mBAAmBP,CAAG,EAAE,CAAC,CAC9D,CACH,CACJ,EACG,KACCQ,EAAI,IAAG,EAAY,EACnBC,EAAS,IAAM,SAAS,KAAK,YAAYR,CAAM,CAAC,EAChDS,GAAK,CAAC,CACR,EACH,CACH,CCVA,IAAMC,GAAS,IAAIC,EAiBbC,GAAYC,EAAM,IACtB,OAAO,gBAAmB,YACtBC,GAAY,4CAA4C,EACxDC,EAAG,MAAS,CACjB,EACE,KACCC,EAAI,IAAM,IAAI,eAAeC,GAAW,CACtC,QAAWC,KAASD,EAClBP,GAAO,KAAKQ,CAAK,CACrB,CAAC,CAAC,EACFC,EAAUC,GAAYC,EAAMC,GAAOP,EAAGK,CAAQ,CAAC,EAC5C,KACCG,EAAS,IAAMH,EAAS,WAAW,CAAC,CACtC,CACF,EACAI,EAAY,CAAC,CACf,EAaK,SAASC,GACdC,EACa,CACb,MAAO,CACL,MAAQA,EAAG,YACX,OAAQA,EAAG,YACb,CACF,CAuBO,SAASC,GACdD,EACyB,CACzB,OAAOd,GACJ,KACCgB,EAAIR,GAAYA,EAAS,QAAQM,CAAE,CAAC,EACpCP,EAAUC,GAAYV,GACnB,KACCmB,EAAO,CAAC,CAAE,OAAAC,CAAO,IAAMA,IAAWJ,CAAE,EACpCH,EAAS,IAAMH,EAAS,UAAUM,CAAE,CAAC,EACrCV,EAAI,IAAMS,GAAeC,CAAE,CAAC,CAC9B,CACF,EACAK,EAAUN,GAAeC,CAAE,CAAC,CAC9B,CACJ,CCnHO,SAASM,GACdC,EACa,CACb,MAAO,CACL,MAAQA,EAAG,YACX,OAAQA,EAAG,YACb,CACF,CASO,SAASC,GACdD,EACyB,CACzB,IAAIE,EAASF,EAAG,cAChB,KAAOE,IAEHF,EAAG,aAAgBE,EAAO,aAC1BF,EAAG,cAAgBE,EAAO,eAE1BA,GAAUF,EAAKE,GAAQ,cAK3B,OAAOA,EAASF,EAAK,MACvB,CCfA,IAAMG,GAAS,IAAIC,EAUbC,GAAYC,EAAM,IAAMC,EAC5B,IAAI,qBAAqBC,GAAW,CAClC,QAAWC,KAASD,EAClBL,GAAO,KAAKM,CAAK,CACrB,EAAG,CACD,UAAW,CACb,CAAC,CACH,CAAC,EACE,KACCC,EAAUC,GAAYC,EAAMC,GAAON,EAAGI,CAAQ,CAAC,EAC5C,KACCG,EAAS,IAAMH,EAAS,WAAW,CAAC,CACtC,CACF,EACAI,EAAY,CAAC,CACf,EAaK,SAASC,GACdC,EACqB,CACrB,OAAOZ,GACJ,KACCa,EAAIP,GAAYA,EAAS,QAAQM,CAAE,CAAC,EACpCP,EAAUC,GAAYR,GACnB,KACCgB,EAAO,CAAC,CAAE,OAAAC,CAAO,IAAMA,IAAWH,CAAE,EACpCH,EAAS,IAAMH,EAAS,UAAUM,CAAE,CAAC,EACrCI,EAAI,CAAC,CAAE,eAAAC,CAAe,IAAMA,CAAc,CAC5C,CACF,CACF,CACJ,CAaO,SAASC,GACdN,EAAiBO,EAAY,GACR,CACrB,OAAOC,GAA0BR,CAAE,EAChC,KACCI,EAAI,CAAC,CAAE,EAAAK,CAAE,IAAM,CACb,IAAMC,EAAUC,GAAeX,CAAE,EAC3BY,EAAUC,GAAsBb,CAAE,EACxC,OAAOS,GACLG,EAAQ,OAASF,EAAQ,OAASH,CAEtC,CAAC,EACDO,EAAqB,CACvB,CACJ,CCjFA,IAAMC,GAA4C,CAChD,OAAQC,EAAW,yBAAyB,EAC5C,OAAQA,EAAW,yBAAyB,CAC9C,EAaO,SAASC,GAAUC,EAAuB,CAC/C,OAAOH,GAAQG,CAAI,EAAE,OACvB,CAaO,SAASC,GAAUD,EAAcE,EAAsB,CACxDL,GAAQG,CAAI,EAAE,UAAYE,GAC5BL,GAAQG,CAAI,EAAE,MAAM,CACxB,CAWO,SAASG,GAAYH,EAAmC,CAC7D,IAAMI,EAAKP,GAAQG,CAAI,EACvB,OAAOK,EAAUD,EAAI,QAAQ,EAC1B,KACCE,EAAI,IAAMF,EAAG,OAAO,EACpBG,EAAUH,EAAG,OAAO,CACtB,CACJ,CC9BA,SAASI,GACPC,EAAiBC,EACR,CACT,OAAQD,EAAG,YAAa,CAGtB,KAAK,iBAEH,OAAIA,EAAG,OAAS,QACP,SAAS,KAAKC,CAAI,EAElB,GAGX,KAAK,kBACL,KAAK,oBACH,MAAO,GAGT,QACE,OAAOD,EAAG,iBACd,CACF,CAWO,SAASE,IAAwC,CACtD,OAAOC,EACLC,EAAU,OAAQ,kBAAkB,EAAE,KAAKC,EAAI,IAAM,EAAI,CAAC,EAC1DD,EAAU,OAAQ,gBAAgB,EAAE,KAAKC,EAAI,IAAM,EAAK,CAAC,CAC3D,EACG,KACCC,EAAU,EAAK,CACjB,CACJ,CAOO,SAASC,IAAsC,CACpD,IAAMC,EAAYJ,EAAyB,OAAQ,SAAS,EACzD,KACCK,EAAOC,GAAM,EAAEA,EAAG,SAAWA,EAAG,QAAQ,EACxCL,EAAIK,IAAO,CACT,KAAMC,GAAU,QAAQ,EAAI,SAAW,SACvC,KAAMD,EAAG,IACT,OAAQ,CACNA,EAAG,eAAe,EAClBA,EAAG,gBAAgB,CACrB,CACF,EAAc,EACdD,EAAO,CAAC,CAAE,KAAAG,EAAM,KAAAX,CAAK,IAAM,CACzB,GAAIW,IAAS,SAAU,CACrB,IAAMC,EAASC,GAAiB,EAChC,GAAI,OAAOD,GAAW,YACpB,MAAO,CAACd,GAAwBc,EAAQZ,CAAI,CAChD,CACA,MAAO,EACT,CAAC,EACDc,GAAM,CACR,EAGF,OAAOb,GAAiB,EACrB,KACCc,EAAUH,GAAWA,EAAqBI,EAAZT,CAAiB,CACjD,CACJ,CC1GO,SAASU,IAAmB,CACjC,OAAO,IAAI,IAAI,SAAS,IAAI,CAC9B,CAgBO,SAASC,GACdC,EAA4BC,EAAW,GACjC,CACN,GAAIC,GAAQ,oBAAoB,GAAK,CAACD,EAAU,CAC9C,IAAME,EAAKC,EAAE,IAAK,CAAE,KAAMJ,EAAI,IAAK,CAAC,EACpC,SAAS,KAAK,YAAYG,CAAE,EAC5BA,EAAG,MAAM,EACTA,EAAG,OAAO,CAIZ,MACE,SAAS,KAAOH,EAAI,IAExB,CASO,SAASK,IAA8B,CAC5C,OAAO,IAAIC,CACb,CCxCO,SAASC,IAA0B,CACxC,OAAO,SAAS,KAAK,MAAM,CAAC,CAC9B,CAYO,SAASC,GAAgBC,EAAoB,CAClD,IAAMC,EAAKC,EAAE,IAAK,CAAE,KAAMF,CAAK,CAAC,EAChCC,EAAG,iBAAiB,QAASE,GAAMA,EAAG,gBAAgB,CAAC,EACvDF,EAAG,MAAM,CACX,CAWO,SAASG,GACdC,EACoB,CACpB,OAAOC,EACLC,EAA2B,OAAQ,YAAY,EAC/CF,CACF,EACG,KACCG,EAAIV,EAAe,EACnBW,EAAUX,GAAgB,CAAC,EAC3BY,EAAOV,GAAQA,EAAK,OAAS,CAAC,EAC9BW,EAAY,CAAC,CACf,CACJ,CASO,SAASC,GACdP,EACyB,CACzB,OAAOD,GAAkBC,CAAS,EAC/B,KACCG,EAAIK,GAAMC,GAAmB,QAAQD,CAAE,IAAI,CAAE,EAC7CH,EAAOT,GAAM,OAAOA,GAAO,WAAW,CACxC,CACJ,CCtDO,SAASc,GAAWC,EAAoC,CAC7D,IAAMC,EAAQ,WAAWD,CAAK,EAC9B,OAAOE,GAA0BC,GAC/BF,EAAM,YAAY,IAAME,EAAKF,EAAM,OAAO,CAAC,CAC5C,EACE,KACCG,EAAUH,EAAM,OAAO,CACzB,CACJ,CAOO,SAASI,IAAkC,CAChD,IAAMJ,EAAQ,WAAW,OAAO,EAChC,OAAOK,EACLC,EAAU,OAAQ,aAAa,EAAE,KAAKC,EAAI,IAAM,EAAI,CAAC,EACrDD,EAAU,OAAQ,YAAY,EAAE,KAAKC,EAAI,IAAM,EAAK,CAAC,CACvD,EACG,KACCJ,EAAUH,EAAM,OAAO,CACzB,CACJ,CAcO,SAASQ,GACdC,EAA6BC,EACd,CACf,OAAOD,EACJ,KACCE,EAAUC,GAAUA,EAASF,EAAQ,EAAIG,CAAK,CAChD,CACJ,CCtCO,SAASC,GACdC,EAAmBC,EACD,CAClB,OAAO,IAAIC,EAAiBC,GAAY,CACtC,IAAMC,EAAM,IAAI,eAChBA,EAAI,KAAK,MAAQ,GAAGJ,CAAG,EAAE,EACzBI,EAAI,aAAe,OAGnBA,EAAI,iBAAiB,OAAQ,IAAM,CAC7BA,EAAI,QAAU,KAAOA,EAAI,OAAS,KACpCD,EAAS,KAAKC,EAAI,QAAQ,EAC1BD,EAAS,SAAS,GAElBA,EAAS,MAAM,IAAI,MAAMC,EAAI,UAAU,CAAC,CAE5C,CAAC,EAGDA,EAAI,iBAAiB,QAAS,IAAM,CAClCD,EAAS,MAAM,IAAI,MAAM,eAAe,CAAC,CAC3C,CAAC,EAGDC,EAAI,iBAAiB,QAAS,IAAM,CAClCD,EAAS,MAAM,IAAI,MAAM,iBAAiB,CAAC,CAC7C,CAAC,EAGG,OAAOF,GAAA,YAAAA,EAAS,YAAc,cAChCG,EAAI,iBAAiB,WAAYC,GAAS,CACxC,GAAIA,EAAM,iBACRJ,EAAQ,UAAW,KAAMI,EAAM,OAASA,EAAM,MAAS,GAAG,MACrD,CACL,IAAMC,EAAkB,OAAOF,EAAI,kBAAkB,gBAAgB,CAAC,GAAK,EAC3EH,EAAQ,UAAW,KAAMI,EAAM,OAASC,EAAmB,GAAG,CAChE,CACF,CAAC,EAGDL,EAAQ,UAAU,KAAK,CAAC,GAI1BG,EAAI,KAAK,CACX,CAAC,CACH,CAcO,SAASG,GACdP,EAAmBC,EACJ,CACf,OAAOF,GAAQC,EAAKC,CAAO,EACxB,KACCO,EAAUC,GAAOA,EAAI,KAAK,CAAC,EAC3BC,EAAIC,GAAQ,KAAK,MAAMA,CAAI,CAAM,EACjCC,EAAY,CAAC,CACf,CACJ,CAUO,SAASC,GACdb,EAAmBC,EACG,CACtB,IAAMa,EAAM,IAAI,UAChB,OAAOf,GAAQC,EAAKC,CAAO,EACxB,KACCO,EAAUC,GAAOA,EAAI,KAAK,CAAC,EAC3BC,EAAID,GAAOK,EAAI,gBAAgBL,EAAK,UAAU,CAAC,EAC/CG,EAAY,CAAC,CACf,CACJ,CC3FO,SAASG,IAAoC,CAClD,MAAO,CACL,EAAG,KAAK,IAAI,EAAG,OAAO,EACtB,EAAG,KAAK,IAAI,EAAG,OAAO,CACxB,CACF,CASO,SAASC,IAAkD,CAChE,OAAOC,EACLC,EAAU,OAAQ,SAAU,CAAE,QAAS,EAAK,CAAC,EAC7CA,EAAU,OAAQ,SAAU,CAAE,QAAS,EAAK,CAAC,CAC/C,EACG,KACCC,EAAIJ,EAAiB,EACrBK,EAAUL,GAAkB,CAAC,CAC/B,CACJ,CC3BO,SAASM,IAAgC,CAC9C,MAAO,CACL,MAAQ,WACR,OAAQ,WACV,CACF,CASO,SAASC,IAA8C,CAC5D,OAAOC,EAAU,OAAQ,SAAU,CAAE,QAAS,EAAK,CAAC,EACjD,KACCC,EAAIH,EAAe,EACnBI,EAAUJ,GAAgB,CAAC,CAC7B,CACJ,CCXO,SAASK,IAAsC,CACpD,OAAOC,EAAc,CACnBC,GAAoB,EACpBC,GAAkB,CACpB,CAAC,EACE,KACCC,EAAI,CAAC,CAACC,EAAQC,CAAI,KAAO,CAAE,OAAAD,EAAQ,KAAAC,CAAK,EAAE,EAC1CC,EAAY,CAAC,CACf,CACJ,CCVO,SAASC,GACdC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EAChB,CACtB,IAAMC,EAAQF,EACX,KACCG,GAAwB,MAAM,CAChC,EAGIC,EAAUC,EAAc,CAACH,EAAOD,CAAO,CAAC,EAC3C,KACCK,EAAI,IAAMC,GAAiBR,CAAE,CAAC,CAChC,EAGF,OAAOM,EAAc,CAACJ,EAASD,EAAWI,CAAO,CAAC,EAC/C,KACCE,EAAI,CAAC,CAAC,CAAE,OAAAE,CAAO,EAAG,CAAE,OAAAC,EAAQ,KAAAC,CAAK,EAAG,CAAE,EAAAC,EAAG,EAAAC,CAAE,CAAC,KAAO,CACjD,OAAQ,CACN,EAAGH,EAAO,EAAIE,EACd,EAAGF,EAAO,EAAIG,EAAIJ,CACpB,EACA,KAAAE,CACF,EAAE,CACJ,CACJ,CCzBA,SAASG,GAAQC,EAA+B,CAC9C,OAAOC,EAA8BD,EAAQ,UAAWE,GAAMA,EAAG,IAAI,CACvE,CAWA,SAASC,GAAQH,EAA4B,CAC3C,IAAMI,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAUE,GAAQN,EAAO,YAAYM,CAAI,CAAC,EAGzCF,CACT,CAgBO,SAASG,GACdC,EAAaR,EAAS,IAAI,OAAOQ,CAAG,EACxB,CACZ,IAAMC,EAAQV,GAAQC,CAAM,EACtBI,EAAQD,GAAQH,CAAM,EAGtBU,EAAU,IAAIL,EACpBK,EAAQ,UAAUN,CAAK,EAGvB,IAAMO,EAAQP,EAAM,KAAKQ,EAAe,EAAGC,GAAQ,EAAI,CAAC,EACxD,OAAOH,EACJ,KACCE,EAAe,EACfE,GAAUL,EAAM,KAAKM,EAAUJ,CAAK,CAAC,CAAC,EACtCK,GAAM,CACR,CACJ,CCNA,IAAMC,GAASC,EAAW,WAAW,EAC/BC,GAAiB,KAAK,MAAMF,GAAO,WAAY,EACrDE,GAAO,KAAO,GAAG,IAAI,IAAIA,GAAO,KAAMC,GAAY,CAAC,CAAC,GAW7C,SAASC,IAAwB,CACtC,OAAOF,EACT,CASO,SAASG,GAAQC,EAAqB,CAC3C,OAAOJ,GAAO,SAAS,SAASI,CAAI,CACtC,CAUO,SAASC,GACdC,EAAkBC,EACV,CACR,OAAO,OAAOA,GAAU,YACpBP,GAAO,aAAaM,CAAG,EAAE,QAAQ,IAAKC,EAAM,SAAS,CAAC,EACtDP,GAAO,aAAaM,CAAG,CAC7B,CC9BO,SAASE,GACdC,EAASC,EAAmB,SACP,CACrB,OAAOC,EAAW,sBAAsBF,CAAI,IAAKC,CAAI,CACvD,CAYO,SAASE,GACdH,EAASC,EAAmB,SACL,CACvB,OAAOG,EAAY,sBAAsBJ,CAAI,IAAKC,CAAI,CACxD,CC7EO,SAASI,GACdC,EACsB,CACtB,IAAMC,EAASC,EAAW,6BAA8BF,CAAE,EAC1D,OAAOG,EAAUF,EAAQ,QAAS,CAAE,KAAM,EAAK,CAAC,EAC7C,KACCG,EAAI,IAAMF,EAAW,cAAeF,CAAE,CAAC,EACvCI,EAAIC,IAAY,CAAE,KAAM,UAAUA,EAAQ,SAAS,CAAE,EAAE,CACzD,CACJ,CASO,SAASC,GACdN,EACiC,CACjC,GAAI,CAACO,GAAQ,kBAAkB,GAAK,CAACP,EAAG,kBACtC,OAAOQ,EAGT,GAAI,CAACR,EAAG,OAAQ,CACd,IAAMK,EAAUH,EAAW,cAAeF,CAAE,EACxC,UAAUK,EAAQ,SAAS,IAAM,SAAS,YAAY,IACxDL,EAAG,OAAS,GAChB,CAGA,OAAOS,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAU,CAAC,CAAE,KAAAE,CAAK,IAAM,CAC5BZ,EAAG,OAAS,GAGZ,SAAiB,aAAcY,CAAI,CACrC,CAAC,EAGMb,GAAcC,CAAE,EACpB,KACCa,EAAIC,GAASJ,EAAM,KAAKI,CAAK,CAAC,EAC9BC,EAAS,IAAML,EAAM,SAAS,CAAC,EAC/BN,EAAIU,GAAUE,EAAA,CAAE,IAAKhB,GAAOc,EAAQ,CACtC,CACJ,CAAC,CACH,CC5BO,SAASG,GACdC,EAAiB,CAAE,QAAAC,CAAQ,EACN,CACrB,OAAOA,EACJ,KACCC,EAAIC,IAAW,CAAE,OAAQA,IAAWH,CAAG,EAAE,CAC3C,CACJ,CAYO,SAASI,GACdJ,EAAiBK,EACe,CAChC,IAAMC,EAAY,IAAIC,EACtB,OAAAD,EAAU,UAAU,CAAC,CAAE,OAAAE,CAAO,IAAM,CAClCR,EAAG,OAASQ,CACd,CAAC,EAGMT,GAAaC,EAAIK,CAAO,EAC5B,KACCI,EAAIC,GAASJ,EAAU,KAAKI,CAAK,CAAC,EAClCC,EAAS,IAAML,EAAU,SAAS,CAAC,EACnCJ,EAAIQ,GAAUE,EAAA,CAAE,IAAKZ,GAAOU,EAAQ,CACtC,CACJ,CCtBO,SAASG,GACdC,EAAiBC,EACO,CACxB,IAAMC,EAAUC,EAAM,IAAMC,EAAc,CACxCC,GAAmBL,CAAE,EACrBM,GAA0BL,CAAS,CACrC,CAAC,CAAC,EACC,KACCM,EAAI,CAAC,CAAC,CAAE,EAAAC,EAAG,EAAAC,CAAE,EAAGC,CAAM,IAAqB,CACzC,GAAM,CAAE,MAAAC,EAAO,OAAAC,CAAO,EAAIC,GAAeb,CAAE,EAC3C,MAAQ,CACN,EAAGQ,EAAIE,EAAO,EAAIC,EAAS,EAC3B,EAAGF,EAAIC,EAAO,EAAIE,EAAS,CAC7B,CACF,CAAC,CACH,EAGF,OAAOE,GAAkBd,CAAE,EACxB,KACCe,EAAUC,GAAUd,EACjB,KACCK,EAAIU,IAAW,CAAE,OAAAD,EAAQ,OAAAC,CAAO,EAAE,EAClCC,GAAK,CAAC,CAACF,GAAU,GAAQ,CAC3B,CACF,CACF,CACJ,CAWO,SAASG,GACdnB,EAAiBC,EAAwB,CAAE,QAAAmB,CAAQ,EAChB,CACnC,GAAM,CAACC,EAASC,CAAK,EAAI,MAAM,KAAKtB,EAAG,QAAQ,EAG/C,OAAOG,EAAM,IAAM,CACjB,IAAMoB,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,EAAe,EAAGC,GAAQ,EAAI,CAAC,EACxD,OAAAJ,EAAM,UAAU,CAGd,KAAK,CAAE,OAAAN,CAAO,EAAG,CACfjB,EAAG,MAAM,YAAY,iBAAkB,GAAGiB,EAAO,CAAC,IAAI,EACtDjB,EAAG,MAAM,YAAY,iBAAkB,GAAGiB,EAAO,CAAC,IAAI,CACxD,EAGA,UAAW,CACTjB,EAAG,MAAM,eAAe,gBAAgB,EACxCA,EAAG,MAAM,eAAe,gBAAgB,CAC1C,CACF,CAAC,EAGD4B,GAAuB5B,CAAE,EACtB,KACC6B,EAAUJ,CAAK,CACjB,EACG,UAAUK,GAAW,CACpB9B,EAAG,gBAAgB,kBAAmB8B,CAAO,CAC/C,CAAC,EAGLC,EACER,EAAM,KAAKS,EAAO,CAAC,CAAE,OAAAhB,CAAO,IAAMA,CAAM,CAAC,EACzCO,EAAM,KAAKU,GAAa,GAAG,EAAGD,EAAO,CAAC,CAAE,OAAAhB,CAAO,IAAM,CAACA,CAAM,CAAC,CAC/D,EACG,UAAU,CAGT,KAAK,CAAE,OAAAA,CAAO,EAAG,CACXA,EACFhB,EAAG,QAAQqB,CAAO,EAElBA,EAAQ,OAAO,CACnB,EAGA,UAAW,CACTrB,EAAG,QAAQqB,CAAO,CACpB,CACF,CAAC,EAGHE,EACG,KACCW,GAAU,GAAIC,EAAuB,CACvC,EACG,UAAU,CAAC,CAAE,OAAAnB,CAAO,IAAM,CACzBK,EAAQ,UAAU,OAAO,qBAAsBL,CAAM,CACvD,CAAC,EAGLO,EACG,KACCa,GAAa,IAAKD,EAAuB,EACzCH,EAAO,IAAM,CAAC,CAAChC,EAAG,YAAY,EAC9BO,EAAI,IAAMP,EAAG,aAAc,sBAAsB,CAAC,EAClDO,EAAI,CAAC,CAAE,EAAAC,CAAE,IAAMA,CAAC,CAClB,EACG,UAAU,CAGT,KAAK6B,EAAQ,CACPA,EACFrC,EAAG,MAAM,YAAY,iBAAkB,GAAG,CAACqC,CAAM,IAAI,EAErDrC,EAAG,MAAM,eAAe,gBAAgB,CAC5C,EAGA,UAAW,CACTA,EAAG,MAAM,eAAe,gBAAgB,CAC1C,CACF,CAAC,EAGLsC,EAAsBhB,EAAO,OAAO,EACjC,KACCO,EAAUJ,CAAK,EACfO,EAAOO,GAAM,EAAEA,EAAG,SAAWA,EAAG,QAAQ,CAC1C,EACG,UAAUA,GAAM,CACfA,EAAG,gBAAgB,EACnBA,EAAG,eAAe,CACpB,CAAC,EAGLD,EAAsBhB,EAAO,WAAW,EACrC,KACCO,EAAUJ,CAAK,EACfe,GAAejB,CAAK,CACtB,EACG,UAAU,CAAC,CAACgB,EAAI,CAAE,OAAAvB,CAAO,CAAC,IAAM,CA3OzC,IAAAyB,EA8OU,GAAIF,EAAG,SAAW,GAAKA,EAAG,SAAWA,EAAG,QACtCA,EAAG,eAAe,UAGTvB,EAAQ,CACjBuB,EAAG,eAAe,EAGlB,IAAMG,EAAS1C,EAAG,cAAe,QAAQ,gBAAgB,EACrD0C,aAAkB,YACpBA,EAAO,MAAM,GAEbD,EAAAE,GAAiB,IAAjB,MAAAF,EAAoB,MACxB,CACF,CAAC,EAGLrB,EACG,KACCS,EAAUJ,CAAK,EACfO,EAAOY,GAAUA,IAAWvB,CAAO,EACnCwB,GAAM,GAAG,CACX,EACG,UAAU,IAAM7C,EAAG,MAAM,CAAC,EAGxBD,GAAgBC,EAAIC,CAAS,EACjC,KACC6C,EAAIC,GAASxB,EAAM,KAAKwB,CAAK,CAAC,EAC9BC,EAAS,IAAMzB,EAAM,SAAS,CAAC,EAC/BhB,EAAIwC,GAAUE,EAAA,CAAE,IAAKjD,GAAO+C,EAAQ,CACtC,CACJ,CAAC,CACH,CC5OO,SAASG,GAAcC,EAA0B,CACtD,OACEC,EAAC,OAAI,MAAM,aAAa,GAAID,GAC1BC,EAAC,OAAI,MAAM,+BAA+B,CAC5C,CAEJ,CCHO,SAASC,GACdC,EAAqBC,EACR,CAIb,GAHAA,EAASA,EAAS,GAAGA,CAAM,eAAeD,CAAE,GAAK,OAG7CC,EAAQ,CACV,IAAMC,EAASD,EAAS,IAAIA,CAAM,GAAK,OACvC,OACEE,EAAC,SAAM,MAAM,gBAAgB,SAAU,GACpCC,GAAcH,CAAM,EACrBE,EAAC,KAAE,KAAMD,EAAQ,MAAM,uBAAuB,SAAU,IACtDC,EAAC,QAAK,wBAAuBH,EAAI,CACnC,CACF,CAEJ,KACE,QACEG,EAAC,SAAM,MAAM,gBAAgB,SAAU,GACpCC,GAAcH,CAAM,EACrBE,EAAC,QAAK,MAAM,uBAAuB,SAAU,IAC3CA,EAAC,QAAK,wBAAuBH,EAAI,CACnC,CACF,CAGN,CC5BO,SAASK,GAAsBC,EAAyB,CAC7D,OACEC,EAAC,UACC,MAAM,uBACN,MAAOC,GAAY,gBAAgB,EACnC,wBAAuB,IAAIF,CAAE,UAC9B,CAEL,CCQA,SAASG,GACPC,EAAsBC,EACT,CACb,IAAMC,EAASD,EAAO,EAChBE,EAASF,EAAO,EAGhBG,EAAU,OAAO,KAAKJ,EAAS,KAAK,EACvC,OAAOK,GAAO,CAACL,EAAS,MAAMK,CAAG,CAAC,EAClC,OAAyB,CAACC,EAAMD,IAAQ,CACvC,GAAGC,EAAMC,EAAC,WAAKF,CAAI,EAAQ,GAC7B,EAAG,CAAC,CAAC,EACJ,MAAM,EAAG,EAAE,EAGRG,EAASC,GAAc,EACvBC,EAAM,IAAI,IAAIV,EAAS,SAAUQ,EAAO,IAAI,EAC9CG,GAAQ,kBAAkB,GAC5BD,EAAI,aAAa,IAAI,IAAK,OAAO,QAAQV,EAAS,KAAK,EACpD,OAAO,CAAC,CAAC,CAAEY,CAAK,IAAMA,CAAK,EAC3B,OAAO,CAACC,EAAW,CAACC,CAAK,IAAM,GAAGD,CAAS,IAAIC,CAAK,GAAG,KAAK,EAAG,EAAE,CACpE,EAGF,GAAM,CAAE,KAAAC,CAAK,EAAIN,GAAc,EAC/B,OACEF,EAAC,KAAE,KAAM,GAAGG,CAAG,GAAI,MAAM,yBAAyB,SAAU,IAC1DH,EAAC,WACC,MAAM,uCACN,gBAAeP,EAAS,MAAM,QAAQ,CAAC,GAEtCE,EAAS,GAAKK,EAAC,OAAI,MAAM,iCAAiC,EAC1DL,EAAS,GAAKK,EAAC,UAAIP,EAAS,KAAM,EAClCE,GAAU,GAAKK,EAAC,UAAIP,EAAS,KAAM,EACnCG,EAAS,GAAKH,EAAS,KAAK,OAAS,GACpCA,EAAS,KAEVA,EAAS,MAAQA,EAAS,KAAK,IAAIgB,GAAO,CACzC,IAAMC,EAAOF,EACTC,KAAOD,EACL,uBAAuBA,EAAKC,CAAG,CAAC,GAChC,cACF,GACJ,OACET,EAAC,QAAK,MAAO,UAAUU,CAAI,IAAKD,CAAI,CAExC,CAAC,EACAb,EAAS,GAAKC,EAAQ,OAAS,GAC9BG,EAAC,KAAE,MAAM,2BACNW,GAAY,4BAA4B,EAAE,KAAG,GAAGd,CACnD,CAEJ,CACF,CAEJ,CAaO,SAASe,GACdC,EACa,CACb,IAAMC,EAAYD,EAAO,CAAC,EAAE,MACtBE,EAAO,CAAC,GAAGF,CAAM,EAEjBZ,EAASC,GAAc,EAGvBP,EAASoB,EAAK,UAAUC,GAErB,CADG,GAAG,IAAI,IAAIA,EAAI,SAAUf,EAAO,IAAI,CAAC,GACrC,SAAS,GAAG,CACvB,EACK,CAACgB,CAAO,EAAIF,EAAK,OAAOpB,EAAQ,CAAC,EAGnCuB,EAAQH,EAAK,UAAUC,GAAOA,EAAI,MAAQF,CAAS,EACnDI,IAAU,KACZA,EAAQH,EAAK,QAGf,IAAMI,EAAOJ,EAAK,MAAM,EAAGG,CAAK,EAC1BE,EAAOL,EAAK,MAAMG,CAAK,EAGvBG,EAAW,CACf7B,GAAqByB,EAAS,EAAc,EAAE,CAACtB,GAAUuB,IAAU,EAAE,EACrE,GAAGC,EAAK,IAAIG,GAAW9B,GAAqB8B,EAAS,CAAW,CAAC,EACjE,GAAGF,EAAK,OAAS,CACfpB,EAAC,WAAQ,MAAM,0BACbA,EAAC,WAAQ,SAAU,IACjBA,EAAC,WACEoB,EAAK,OAAS,GAAKA,EAAK,SAAW,EAChCT,GAAY,wBAAwB,EACpCA,GAAY,2BAA4BS,EAAK,MAAM,CAEzD,CACF,EACC,GAAGA,EAAK,IAAIE,GAAW9B,GAAqB8B,EAAS,CAAW,CAAC,CACpE,CACF,EAAI,CAAC,CACP,EAGA,OACEtB,EAAC,MAAG,MAAM,0BACPqB,CACH,CAEJ,CCrIO,SAASE,GAAkBC,EAAiC,CACjE,OACEC,EAAC,MAAG,MAAM,oBACP,OAAO,QAAQD,CAAK,EAAE,IAAI,CAAC,CAACE,EAAKC,CAAK,IACrCF,EAAC,MAAG,MAAO,oCAAoCC,CAAG,IAC/C,OAAOC,GAAU,SAAWC,GAAMD,CAAK,EAAIA,CAC9C,CACD,CACH,CAEJ,CCAO,SAASE,GACdC,EACa,CACb,IAAMC,EAAU,kCAAkCD,CAAI,GACtD,OACEE,EAAC,OAAI,MAAOD,EAAS,OAAM,IACzBC,EAAC,UAAO,MAAM,gBAAgB,SAAU,GAAI,cAAY,OAAO,CACjE,CAEJ,CCpBO,SAASC,GAAYC,EAAiC,CAC3D,OACEC,EAAC,OAAI,MAAM,0BACTA,EAAC,OAAI,MAAM,qBACRD,CACH,CACF,CAEJ,CCMA,SAASE,GAAcC,EAA+B,CACpD,IAAMC,EAASC,GAAc,EAGvBC,EAAM,IAAI,IAAI,MAAMH,EAAQ,OAAO,IAAKC,EAAO,IAAI,EACzD,OACEG,EAAC,MAAG,MAAM,oBACRA,EAAC,KAAE,KAAM,GAAGD,CAAG,GAAI,MAAM,oBACtBH,EAAQ,KACX,CACF,CAEJ,CAcO,SAASK,GACdC,EAAqBC,EACR,CACb,OACEH,EAAC,OAAI,MAAM,cACTA,EAAC,UACC,MAAM,sBACN,aAAYI,GAAY,gBAAgB,GAEvCD,EAAO,KACV,EACAH,EAAC,MAAG,MAAM,oBACPE,EAAS,IAAIP,EAAa,CAC7B,CACF,CAEJ,CCpBA,SAASU,GAAUC,EAAuC,CACxD,OAAOA,EAAU,UAAY,OACzBC,EAAY,eAAgBD,CAAS,EACrC,CAACA,CAAS,CAChB,CASA,SAASE,GAAYF,EAAgC,CACnD,IAAMG,EAAkB,CAAC,EACzB,QAAWC,KAAML,GAAUC,CAAS,EAAG,CACrC,IAAMK,EAAgB,CAAC,EAGjBC,EAAK,SAAS,mBAAmBF,EAAI,WAAW,SAAS,EAC/D,QAASG,EAAOD,EAAG,SAAS,EAAGC,EAAMA,EAAOD,EAAG,SAAS,EACtDD,EAAM,KAAKE,CAAY,EAGzB,QAASC,KAAQH,EAAO,CACtB,IAAII,EAGJ,KAAQA,EAAQ,gBAAgB,KAAKD,EAAK,WAAY,GAAI,CACxD,GAAM,CAAC,CAAEE,EAAIC,CAAK,EAAIF,EACtB,GAAI,OAAOE,GAAU,YAAa,CAChC,IAAMC,EAASJ,EAAK,UAAUC,EAAM,KAAK,EACzCD,EAAOI,EAAO,UAAUF,EAAG,MAAM,EACjCP,EAAQ,KAAKS,CAAM,CAGrB,KAAO,CACLJ,EAAK,YAAcE,EACnBP,EAAQ,KAAKK,CAAI,EACjB,KACF,CACF,CACF,CACF,CACA,OAAOL,CACT,CAQA,SAASU,GAAKC,EAAqBC,EAA2B,CAC5DA,EAAO,OAAO,GAAG,MAAM,KAAKD,EAAO,UAAU,CAAC,CAChD,CAoBO,SAASE,GACdZ,EAAiBJ,EAAwB,CAAE,QAAAiB,EAAS,OAAAC,CAAO,EACxB,CAGnC,IAAMC,EAASnB,EAAU,QAAQ,MAAM,EACjCoB,EAASD,GAAA,YAAAA,EAAQ,GAGjBE,EAAc,IAAI,IACxB,QAAWT,KAAUV,GAAYF,CAAS,EAAG,CAC3C,GAAM,CAAC,CAAEU,CAAE,EAAIE,EAAO,YAAa,MAAM,WAAW,EAChDU,GAAmB,yBAAyBZ,CAAE,IAAKN,CAAE,IACvDiB,EAAY,IAAIX,EAAIa,GAAiBb,EAAIU,CAAM,CAAC,EAChDR,EAAO,YAAYS,EAAY,IAAIX,CAAE,CAAE,EAE3C,CAGA,OAAIW,EAAY,OAAS,EAChBG,EAGFC,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,EAAe,EAAGC,GAAQ,EAAI,CAAC,EAGlDC,EAAsC,CAAC,EAC7C,OAAW,CAACrB,EAAIsB,CAAU,IAAKX,EAC7BU,EAAM,KAAK,CACTE,EAAW,cAAeD,CAAU,EACpCC,EAAW,yBAAyBvB,CAAE,IAAKN,CAAE,CAC/C,CAAC,EAGH,OAAAc,EAAO,KAAKgB,EAAUN,CAAK,CAAC,EACzB,UAAUO,GAAU,CACnB/B,EAAG,OAAS,CAAC+B,EAGb/B,EAAG,UAAU,OAAO,qBAAsB+B,CAAM,EAGhD,OAAW,CAACC,EAAOC,CAAK,IAAKN,EACtBI,EAGHtB,GAAKuB,EAAOC,CAAK,EAFjBxB,GAAKwB,EAAOD,CAAK,CAGvB,CAAC,EAGIE,EAAM,GAAG,CAAC,GAAGjB,CAAW,EAC5B,IAAI,CAAC,CAAC,CAAEW,CAAU,IACjBO,GAAgBP,EAAYhC,EAAW,CAAE,QAAAiB,CAAQ,CAAC,CACnD,CACH,EACG,KACCuB,EAAS,IAAMd,EAAM,SAAS,CAAC,EAC/Be,GAAM,CACR,CACJ,CAAC,CACH,CC7JA,SAASC,GAASC,EAA0C,CAC1D,GAAIA,EAAG,mBAAoB,CACzB,IAAMC,EAAUD,EAAG,mBACnB,GAAIC,EAAQ,UAAY,KACtB,OAAOA,EAGJ,GAAIA,EAAQ,UAAY,KAAO,CAACA,EAAQ,SAAS,OACpD,OAAOF,GAASE,CAAO,CAC3B,CAIF,CAcO,SAASC,GACdF,EAAiBG,EACkB,CACnC,OAAOC,EAAM,IAAM,CACjB,IAAMC,EAAON,GAASC,CAAE,EACxB,OAAO,OAAOK,GAAS,YACnBC,GAAoBD,EAAML,EAAIG,CAAO,EACrCI,CACN,CAAC,CACH,CCjEA,IAAAC,GAAwB,SA6DxB,IAAIC,GAAW,EAaf,SAASC,GAAkBC,EAA0C,CACnE,GAAIA,EAAG,mBAAoB,CACzB,IAAMC,EAAUD,EAAG,mBACnB,GAAIC,EAAQ,UAAY,KACtB,OAAOA,EAGJ,GAAIA,EAAQ,UAAY,KAAO,CAACA,EAAQ,SAAS,OACpD,OAAOF,GAAkBE,CAAO,CACpC,CAIF,CAgBO,SAASC,GACdF,EACuB,CACvB,OAAOG,GAAiBH,CAAE,EACvB,KACCI,EAAI,CAAC,CAAE,MAAAC,CAAM,KAEJ,CACL,WAFcC,GAAsBN,CAAE,EAElB,MAAQK,CAC9B,EACD,EACDE,GAAwB,YAAY,CACtC,CACJ,CAoBO,SAASC,GACdR,EAAiBS,EAC8B,CAC/C,GAAM,CAAE,QAASC,CAAM,EAAI,WAAW,SAAS,EAGzCC,EAAWC,EAAM,IAAM,CAC3B,IAAMC,EAAQ,IAAIC,EASlB,GARAD,EAAM,UAAU,CAAC,CAAE,WAAAE,CAAW,IAAM,CAC9BA,GAAcL,EAChBV,EAAG,aAAa,WAAY,GAAG,EAE/BA,EAAG,gBAAgB,UAAU,CACjC,CAAC,EAGG,GAAAgB,QAAY,YAAY,IACtBhB,EAAG,QAAQ,OAAO,GACpBiB,GAAQ,mBAAmB,GAAK,CAACjB,EAAG,QAAQ,UAAU,GACrD,CACD,IAAMkB,EAASlB,EAAG,QAAQ,KAAK,EAC/BkB,EAAO,GAAK,UAAUpB,IAAU,GAChCoB,EAAO,aACLC,GAAsBD,EAAO,EAAE,EAC/BlB,CACF,CACF,CAIF,IAAMoB,EAAYpB,EAAG,QAAQ,YAAY,EACzC,GAAIoB,aAAqB,YAAa,CACpC,IAAMC,EAAOtB,GAAkBqB,CAAS,EAGxC,GAAI,OAAOC,GAAS,cAClBD,EAAU,UAAU,SAAS,UAAU,GACvCH,GAAQ,uBAAuB,GAC9B,CACD,IAAMK,EAAeC,GAAoBF,EAAMrB,EAAIS,CAAO,EAG1D,OAAOP,GAAeF,CAAE,EACrB,KACCwB,EAAIC,GAASZ,EAAM,KAAKY,CAAK,CAAC,EAC9BC,EAAS,IAAMb,EAAM,SAAS,CAAC,EAC/BT,EAAIqB,GAAUE,EAAA,CAAE,IAAK3B,GAAOyB,EAAQ,EACpCG,GACEzB,GAAiBiB,CAAS,EACvB,KACChB,EAAI,CAAC,CAAE,MAAAC,EAAO,OAAAwB,CAAO,IAAMxB,GAASwB,CAAM,EAC1CC,EAAqB,EACrBC,EAAUC,GAAUA,EAASV,EAAeW,CAAK,CACnD,CACJ,CACF,CACJ,CACF,CAGA,OAAO/B,GAAeF,CAAE,EACrB,KACCwB,EAAIC,GAASZ,EAAM,KAAKY,CAAK,CAAC,EAC9BC,EAAS,IAAMb,EAAM,SAAS,CAAC,EAC/BT,EAAIqB,GAAUE,EAAA,CAAE,IAAK3B,GAAOyB,EAAQ,CACtC,CACJ,CAAC,EAGD,OAAIR,GAAQ,cAAc,EACjBiB,GAAuBlC,CAAE,EAC7B,KACCmC,EAAOC,GAAWA,CAAO,EACzBC,GAAK,CAAC,EACNN,EAAU,IAAMpB,CAAQ,CAC1B,EAGGA,CACT,CC9JO,SAAS2B,GACdC,EAAwB,CAAE,QAAAC,EAAS,OAAAC,CAAO,EACrB,CACrB,IAAIC,EAAO,GACX,OAAOC,EAGLH,EACG,KACCI,EAAIC,GAAUA,EAAO,QAAQ,qBAAqB,CAAE,EACpDC,EAAOC,GAAWR,IAAOQ,CAAO,EAChCH,EAAI,KAAO,CACT,OAAQ,OAAQ,OAAQ,EAC1B,EAAa,CACf,EAGFH,EACG,KACCK,EAAOE,GAAUA,GAAU,CAACN,CAAI,EAChCO,EAAI,IAAMP,EAAOH,EAAG,IAAI,EACxBK,EAAII,IAAW,CACb,OAAQA,EAAS,OAAS,OAC5B,EAAa,CACf,CACJ,CACF,CAaO,SAASE,GACdX,EAAwBY,EACQ,CAChC,OAAOC,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAU,CAAC,CAAE,OAAAE,EAAQ,OAAAC,CAAO,IAAM,CACtCjB,EAAG,gBAAgB,OAAQgB,IAAW,MAAM,EACxCC,GACFjB,EAAG,eAAe,CACtB,CAAC,EAGMD,GAAaC,EAAIY,CAAO,EAC5B,KACCF,EAAIQ,GAASJ,EAAM,KAAKI,CAAK,CAAC,EAC9BC,EAAS,IAAML,EAAM,SAAS,CAAC,EAC/BT,EAAIa,GAAUE,EAAA,CAAE,IAAKpB,GAAOkB,EAAQ,CACtC,CACJ,CAAC,CACH,CCzIA,IAAAG,GAAA,qsLCqDA,IAAIC,GAKAC,GAAW,EAWf,SAASC,IAAiC,CACxC,OAAO,OAAO,SAAY,aAAe,mBAAmB,QACxDC,GAAY,qDAAqD,EACjEC,EAAG,MAAS,CAClB,CAaO,SAASC,GACdC,EACgC,CAChC,OAAAA,EAAG,UAAU,OAAO,SAAS,EAC7BN,QAAaE,GAAa,EACvB,KACCK,EAAI,IAAM,QAAQ,WAAW,CAC3B,YAAa,GACb,SAAAC,GACA,SAAU,CACR,cAAe,OACf,gBAAiB,OACjB,aAAc,MAChB,CACF,CAAC,CAAC,EACFC,EAAI,IAAG,EAAY,EACnBC,EAAY,CAAC,CACf,GAGFV,GAAS,UAAU,IAAM,CACvBM,EAAG,UAAU,IAAI,SAAS,EAC1B,IAAMK,EAAK,aAAaV,IAAU,GAG5BW,EAAOC,EAAE,MAAO,CAAE,MAAO,SAAU,CAAC,EACpCC,EAAOR,EAAG,YAGhB,QAAQ,WAAW,OAAOK,EAAIG,EAAM,CAACC,EAAaC,IAAiB,CAGjE,IAAMC,EAASL,EAAK,aAAa,CAAE,KAAM,QAAS,CAAC,EACnDK,EAAO,UAAYF,EAGnBT,EAAG,YAAYM,CAAI,EACnBI,GAAA,MAAAA,EAAKC,EACP,CAAC,CACH,CAAC,EAGMjB,GACJ,KACCS,EAAI,KAAO,CAAE,IAAKH,CAAG,EAAE,CACzB,CACJ,CCvFA,IAAMY,GAAWC,EAAE,OAAO,EAgBnB,SAASC,GACdC,EACkC,CAClC,OAAAA,EAAG,YAAYH,EAAQ,EACvBA,GAAS,YAAYI,GAAYD,CAAE,CAAC,EAG7BE,EAAG,CAAE,IAAKF,CAAG,CAAC,CACvB,CCwBO,SAASG,GACdC,EACyB,CACzB,IAAMC,EAASC,EAA8B,iBAAkBF,CAAE,EAC3DG,EAAUF,EAAO,KAAKG,GAASA,EAAM,OAAO,GAAKH,EAAO,CAAC,EAC/D,OAAOI,EAAM,GAAGJ,EAAO,IAAIG,GAASE,EAAUF,EAAO,QAAQ,EAC1D,KACCG,EAAI,IAAMC,EAA6B,cAAcJ,EAAM,EAAE,IAAI,CAAC,CACpE,CACF,CAAC,EACE,KACCK,EAAUD,EAA6B,cAAcL,EAAQ,EAAE,IAAI,CAAC,EACpEI,EAAIG,IAAW,CAAE,OAAAA,CAAO,EAAE,CAC5B,CACJ,CAeO,SAASC,GACdX,EAAiB,CAAE,UAAAY,CAAU,EACO,CAGpC,IAAMC,EAAOC,GAAoB,MAAM,EACvCd,EAAG,OAAOa,CAAI,EAGd,IAAME,EAAOD,GAAoB,MAAM,EACvCd,EAAG,OAAOe,CAAI,EAGd,IAAMC,EAAYR,EAAW,iBAAkBR,CAAE,EACjD,OAAOiB,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,EAAe,EAAGC,GAAQ,EAAI,CAAC,EACxD,OAAAC,EAAc,CAACL,EAAOM,GAAiBxB,CAAE,CAAC,CAAC,EACxC,KACCyB,GAAU,EAAGC,EAAuB,EACpCC,EAAUP,CAAK,CACjB,EACG,UAAU,CAGT,KAAK,CAAC,CAAE,OAAAV,CAAO,EAAGkB,CAAI,EAAG,CACvB,IAAMC,EAASC,GAAiBpB,CAAM,EAChC,CAAE,MAAAqB,CAAM,EAAIC,GAAetB,CAAM,EAGvCV,EAAG,MAAM,YAAY,mBAAoB,GAAG6B,EAAO,CAAC,IAAI,EACxD7B,EAAG,MAAM,YAAY,uBAAwB,GAAG+B,CAAK,IAAI,EAGzD,IAAME,EAAUC,GAAwBlB,CAAS,GAE/Ca,EAAO,EAAYI,EAAQ,GAC3BJ,EAAO,EAAIE,EAAQE,EAAQ,EAAIL,EAAK,QAEpCZ,EAAU,SAAS,CACjB,KAAM,KAAK,IAAI,EAAGa,EAAO,EAAI,EAAE,EAC/B,SAAU,QACZ,CAAC,CACL,EAGA,UAAW,CACT7B,EAAG,MAAM,eAAe,kBAAkB,EAC1CA,EAAG,MAAM,eAAe,sBAAsB,CAChD,CACF,CAAC,EAGLuB,EAAc,CACZY,GAA0BnB,CAAS,EACnCQ,GAAiBR,CAAS,CAC5B,CAAC,EACE,KACCW,EAAUP,CAAK,CACjB,EACG,UAAU,CAAC,CAACS,EAAQD,CAAI,IAAM,CAC7B,IAAMK,EAAUG,GAAsBpB,CAAS,EAC/CH,EAAK,OAASgB,EAAO,EAAI,GACzBd,EAAK,OAASc,EAAO,EAAII,EAAQ,MAAQL,EAAK,MAAQ,EACxD,CAAC,EAGLvB,EACEC,EAAUO,EAAM,OAAO,EAAE,KAAKN,EAAI,IAAM,EAAE,CAAC,EAC3CD,EAAUS,EAAM,OAAO,EAAE,KAAKR,EAAI,IAAM,CAAE,CAAC,CAC7C,EACG,KACCoB,EAAUP,CAAK,CACjB,EACG,UAAUiB,GAAa,CACtB,GAAM,CAAE,MAAAN,CAAM,EAAIC,GAAehB,CAAS,EAC1CA,EAAU,SAAS,CACjB,KAAMe,EAAQM,EACd,SAAU,QACZ,CAAC,CACH,CAAC,EAGDC,GAAQ,mBAAmB,GAC7BpB,EAAM,KACJqB,GAAK,CAAC,EACNC,GAAe5B,CAAS,CAC1B,EACG,UAAU,CAAC,CAAC,CAAE,OAAAF,CAAO,EAAG,CAAE,OAAAmB,CAAO,CAAC,IAAM,CACvC,IAAMY,EAAM/B,EAAO,UAAU,KAAK,EAClC,GAAIA,EAAO,aAAa,mBAAmB,EACzCA,EAAO,gBAAgB,mBAAmB,MAGrC,CACL,IAAMgC,EAAI1C,EAAG,UAAY6B,EAAO,EAGhC,QAAWc,KAAOzC,EAAY,aAAa,EACzC,QAAWE,KAASF,EAClB,iBAAkByC,CACpB,EAAG,CACD,IAAMC,EAAQpC,EAAW,cAAcJ,EAAM,EAAE,IAAI,EACnD,GACEwC,IAAUlC,GACVkC,EAAM,UAAU,KAAK,IAAMH,EAC3B,CACAG,EAAM,aAAa,oBAAqB,EAAE,EAC1CxC,EAAM,MAAM,EACZ,KACF,CACF,CAGF,OAAO,SAAS,CACd,IAAKJ,EAAG,UAAY0C,CACtB,CAAC,EAGD,IAAMG,EAAO,SAAmB,QAAQ,GAAK,CAAC,EAC9C,SAAS,SAAU,CAAC,GAAG,IAAI,IAAI,CAACJ,EAAK,GAAGI,CAAI,CAAC,CAAC,CAAC,CACjD,CACF,CAAC,EAGL3B,EAAM,KAAKS,EAAUP,CAAK,CAAC,EACxB,UAAU,IAAM,CACf,QAAW0B,KAAS5C,EAA8B,eAAgBF,CAAE,EAClE8C,EAAM,MAAM,CAChB,CAAC,EAGI/C,GAAiBC,CAAE,EACvB,KACC+C,EAAIC,GAAS9B,EAAM,KAAK8B,CAAK,CAAC,EAC9BC,EAAS,IAAM/B,EAAM,SAAS,CAAC,EAC/BX,EAAIyC,GAAUE,EAAA,CAAE,IAAKlD,GAAOgD,EAAQ,CACtC,CACJ,CAAC,EACE,KACCG,GAAYC,EAAc,CAC5B,CACJ,CCzKO,SAASC,GACdC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,EAAS,OAAAC,CAAO,EACd,CAChC,OAAOC,EAGL,GAAGC,EAAY,4BAA6BL,CAAE,EAC3C,IAAIM,GAASC,GAAqBD,EAAO,CAAE,QAAAJ,EAAS,OAAAC,CAAO,CAAC,CAAC,EAGhE,GAAGE,EAAY,2BAA4BL,CAAE,EAC1C,IAAIM,GAASE,GAAeF,EAAO,CAAE,QAAAJ,EAAS,OAAAC,CAAO,CAAC,CAAC,EAG1D,GAAGE,EAAY,cAAeL,CAAE,EAC7B,IAAIM,GAASG,GAAaH,CAAK,CAAC,EAGnC,GAAGD,EAAY,qBAAsBL,CAAE,EACpC,IAAIM,GAASI,GAAeJ,CAAK,CAAC,EAGrC,GAAGD,EAAY,UAAWL,CAAE,EACzB,IAAIM,GAASK,GAAaL,EAAO,CAAE,QAAAJ,EAAS,OAAAC,CAAO,CAAC,CAAC,EAGxD,GAAGE,EAAY,cAAeL,CAAE,EAC7B,IAAIM,GAASM,GAAiBN,EAAO,CAAE,UAAAL,CAAU,CAAC,CAAC,CACxD,CACF,CC3CO,SAASY,GACdC,EAAkB,CAAE,OAAAC,CAAO,EACP,CACpB,OAAOA,EACJ,KACCC,EAAUC,GAAWC,EACnBC,EAAG,EAAI,EACPA,EAAG,EAAK,EAAE,KAAKC,GAAM,GAAI,CAAC,CAC5B,EACG,KACCC,EAAIC,IAAW,CAAE,QAAAL,EAAS,OAAAK,CAAO,EAAE,CACrC,CACF,CACF,CACJ,CAaO,SAASC,GACdC,EAAiBC,EACc,CAC/B,IAAMC,EAAQC,EAAW,cAAeH,CAAE,EAC1C,OAAOI,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAU,CAAC,CAAE,QAAAZ,EAAS,OAAAK,CAAO,IAAM,CACvCE,EAAG,UAAU,OAAO,oBAAqBF,CAAM,EAC/CI,EAAM,YAAcT,CACtB,CAAC,EAGMJ,GAAYW,EAAIC,CAAO,EAC3B,KACCM,EAAIC,GAASH,EAAM,KAAKG,CAAK,CAAC,EAC9BC,EAAS,IAAMJ,EAAM,SAAS,CAAC,EAC/BR,EAAIW,GAAUE,EAAA,CAAE,IAAKV,GAAOQ,EAAQ,CACtC,CACJ,CAAC,CACH,CC7BA,SAASG,GAAS,CAAE,UAAAC,CAAU,EAAsC,CAClE,GAAI,CAACC,GAAQ,iBAAiB,EAC5B,OAAOC,EAAG,EAAK,EAGjB,IAAMC,EAAaH,EAChB,KACCI,EAAI,CAAC,CAAE,OAAQ,CAAE,EAAAC,CAAE,CAAE,IAAMA,CAAC,EAC5BC,GAAY,EAAG,CAAC,EAChBF,EAAI,CAAC,CAACG,EAAGC,CAAC,IAAM,CAACD,EAAIC,EAAGA,CAAC,CAAU,EACnCC,GAAwB,CAAC,CAC3B,EAGIC,EAAUC,EAAc,CAACX,EAAWG,CAAU,CAAC,EAClD,KACCS,EAAO,CAAC,CAAC,CAAE,OAAAC,CAAO,EAAG,CAAC,CAAER,CAAC,CAAC,IAAM,KAAK,IAAIA,EAAIQ,EAAO,CAAC,EAAI,GAAG,EAC5DT,EAAI,CAAC,CAAC,CAAE,CAACU,CAAS,CAAC,IAAMA,CAAS,EAClCC,EAAqB,CACvB,EAGIC,EAAUC,GAAY,QAAQ,EACpC,OAAON,EAAc,CAACX,EAAWgB,CAAO,CAAC,EACtC,KACCZ,EAAI,CAAC,CAAC,CAAE,OAAAS,CAAO,EAAGK,CAAM,IAAML,EAAO,EAAI,KAAO,CAACK,CAAM,EACvDH,EAAqB,EACrBI,EAAUC,GAAUA,EAASV,EAAUR,EAAG,EAAK,CAAC,EAChDmB,EAAU,EAAK,CACjB,CACJ,CAcO,SAASC,GACdC,EAAiBC,EACG,CACpB,OAAOC,EAAM,IAAMd,EAAc,CAC/Be,GAAiBH,CAAE,EACnBxB,GAASyB,CAAO,CAClB,CAAC,CAAC,EACC,KACCpB,EAAI,CAAC,CAAC,CAAE,OAAAuB,CAAO,EAAGC,CAAM,KAAO,CAC7B,OAAAD,EACA,OAAAC,CACF,EAAE,EACFb,EAAqB,CAACR,EAAGC,IACvBD,EAAE,SAAWC,EAAE,QACfD,EAAE,SAAWC,EAAE,MAChB,EACDqB,EAAY,CAAC,CACf,CACJ,CAaO,SAASC,GACdP,EAAiB,CAAE,QAAAQ,EAAS,MAAAC,CAAM,EACH,CAC/B,OAAOP,EAAM,IAAM,CACjB,IAAMQ,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,EAAe,EAAGC,GAAQ,EAAI,CAAC,EACxD,OAAAJ,EACG,KACCxB,GAAwB,QAAQ,EAChC6B,GAAkBP,CAAO,CAC3B,EACG,UAAU,CAAC,CAAC,CAAE,OAAAX,CAAO,EAAG,CAAE,OAAAQ,CAAO,CAAC,IAAM,CACvCL,EAAG,UAAU,OAAO,oBAAqBH,GAAU,CAACQ,CAAM,EAC1DL,EAAG,OAASK,CACd,CAAC,EAGLI,EAAM,UAAUC,CAAK,EAGdF,EACJ,KACCQ,EAAUJ,CAAK,EACf/B,EAAIoC,GAAUC,EAAA,CAAE,IAAKlB,GAAOiB,EAAQ,CACtC,CACJ,CAAC,CACH,CCjHO,SAASE,GACdC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EACb,CACzB,OAAOC,GAAgBH,EAAI,CAAE,UAAAC,EAAW,QAAAC,CAAQ,CAAC,EAC9C,KACCE,EAAI,CAAC,CAAE,OAAQ,CAAE,EAAAC,CAAE,CAAE,IAAM,CACzB,GAAM,CAAE,OAAAC,CAAO,EAAIC,GAAeP,CAAE,EACpC,MAAO,CACL,OAAQK,GAAKC,CACf,CACF,CAAC,EACDE,GAAwB,QAAQ,CAClC,CACJ,CAaO,SAASC,GACdT,EAAiBU,EACmB,CACpC,OAAOC,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClBD,EAAM,UAAU,CAGd,KAAK,CAAE,OAAAE,CAAO,EAAG,CACfd,EAAG,UAAU,OAAO,2BAA4Bc,CAAM,CACxD,EAGA,UAAW,CACTd,EAAG,UAAU,OAAO,0BAA0B,CAChD,CACF,CAAC,EAGD,IAAMe,EAAUC,GAAmB,gBAAgB,EACnD,OAAI,OAAOD,GAAY,YACdE,EAGFlB,GAAiBgB,EAASL,CAAO,EACrC,KACCQ,EAAIC,GAASP,EAAM,KAAKO,CAAK,CAAC,EAC9BC,EAAS,IAAMR,EAAM,SAAS,CAAC,EAC/BR,EAAIe,GAAUE,EAAA,CAAE,IAAKrB,GAAOmB,EAAQ,CACtC,CACJ,CAAC,CACH,CChEO,SAASG,GACdC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EACpB,CAGlB,IAAMC,EAAUD,EACb,KACCE,EAAI,CAAC,CAAE,OAAAC,CAAO,IAAMA,CAAM,EAC1BC,EAAqB,CACvB,EAGIC,EAAUJ,EACb,KACCK,EAAU,IAAMC,GAAiBT,CAAE,EAChC,KACCI,EAAI,CAAC,CAAE,OAAAC,CAAO,KAAO,CACnB,IAAQL,EAAG,UACX,OAAQA,EAAG,UAAYK,CACzB,EAAE,EACFK,GAAwB,QAAQ,CAClC,CACF,CACF,EAGF,OAAOC,EAAc,CAACR,EAASI,EAASN,CAAS,CAAC,EAC/C,KACCG,EAAI,CAAC,CAACQ,EAAQ,CAAE,IAAAC,EAAK,OAAAC,CAAO,EAAG,CAAE,OAAQ,CAAE,EAAAC,CAAE,EAAG,KAAM,CAAE,OAAAV,CAAO,CAAE,CAAC,KAChEA,EAAS,KAAK,IAAI,EAAGA,EACjB,KAAK,IAAI,EAAGQ,EAASE,EAAIH,CAAM,EAC/B,KAAK,IAAI,EAAGP,EAASU,EAAID,CAAM,CACnC,EACO,CACL,OAAQD,EAAMD,EACd,OAAAP,EACA,OAAQQ,EAAMD,GAAUG,CAC1B,EACD,EACDT,EAAqB,CAACU,EAAGC,IACvBD,EAAE,SAAWC,EAAE,QACfD,EAAE,SAAWC,EAAE,QACfD,EAAE,SAAWC,EAAE,MAChB,CACH,CACJ,CC9CO,SAASC,GACdC,EACqB,CACrB,IAAMC,EAAU,SAAkB,WAAW,GAAK,CAChD,MAAOD,EAAO,UAAUE,GAAS,WAC/BA,EAAM,aAAa,qBAAqB,CAC1C,EAAE,OAAO,CACX,EAGA,OAAOC,EAAG,GAAGH,CAAM,EAChB,KACCI,GAASF,GAASG,EAAUH,EAAO,QAAQ,EACxC,KACCI,EAAI,IAAMJ,CAAK,CACjB,CACF,EACAK,EAAUP,EAAO,KAAK,IAAI,EAAGC,EAAQ,KAAK,CAAC,CAAC,EAC5CK,EAAIJ,IAAU,CACZ,MAAOF,EAAO,QAAQE,CAAK,EAC3B,MAAO,CACL,OAASA,EAAM,aAAa,sBAAsB,EAClD,QAASA,EAAM,aAAa,uBAAuB,EACnD,OAASA,EAAM,aAAa,sBAAsB,CACpD,CACF,EAAa,EACbM,EAAY,CAAC,CACf,CACJ,CASO,SAASC,GACdC,EACgC,CAChC,IAAMC,EAAOC,EAAE,OAAQ,CAAE,KAAM,aAAc,CAAC,EAC9C,SAAS,KAAK,YAAYD,CAAI,EAG9B,IAAME,EAASD,EAAE,OAAQ,CAAE,KAAM,cAAe,CAAC,EACjD,gBAAS,KAAK,YAAYC,CAAM,EAGzBC,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClBD,EAAM,UAAUE,GAAW,CACzB,SAAS,KAAK,aAAa,0BAA2B,EAAE,EAGxD,OAAW,CAACC,EAAKC,CAAK,IAAK,OAAO,QAAQF,EAAQ,KAAK,EACrD,SAAS,KAAK,aAAa,iBAAiBC,CAAG,GAAIC,CAAK,EAG1D,QAASC,EAAQ,EAAGA,EAAQpB,EAAO,OAAQoB,IAAS,CAClD,IAAMC,EAAQrB,EAAOoB,CAAK,EAAE,mBACxBC,aAAiB,cACnBA,EAAM,OAASJ,EAAQ,QAAUG,EACrC,CAGA,SAAS,YAAaH,CAAO,CAC/B,CAAC,EAGDF,EACG,KACCT,EAAI,IAAM,CACR,IAAMgB,EAASC,GAAoB,QAAQ,EACrCC,EAAS,OAAO,iBAAiBF,CAAM,EAG7C,OAAAT,EAAO,QAAUW,EAAM,YAGhBA,EAAM,gBAAgB,MAAM,MAAM,EACtC,IAAIL,IAAU,CAACA,GAAO,SAAS,EAAE,EAAE,SAAS,EAAG,GAAG,CAAC,EACnD,KAAK,EAAE,CACZ,CAAC,CACH,EACG,UAAUM,GAASd,EAAK,QAAU,IAAIc,CAAK,EAAE,EAGlDV,EAAM,KAAKW,GAAUC,EAAc,CAAC,EACjC,UAAU,IAAM,CACf,SAAS,KAAK,gBAAgB,yBAAyB,CACzD,CAAC,EAGH,IAAM3B,EAAS4B,EAA8B,QAASlB,CAAE,EACxD,OAAOX,GAAaC,CAAM,EACvB,KACC6B,EAAIC,GAASf,EAAM,KAAKe,CAAK,CAAC,EAC9BC,EAAS,IAAMhB,EAAM,SAAS,CAAC,EAC/BT,EAAIwB,GAAUE,EAAA,CAAE,IAAKtB,GAAOoB,EAAQ,CACtC,CACJ,CAAC,CACH,CChHO,SAASG,GACdC,EAAiB,CAAE,UAAAC,CAAU,EACI,CAGjC,OAAOC,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAU,CAAC,CAAE,MAAAE,CAAM,IAAM,CAC7BL,EAAG,MAAM,YAAY,sBAAuB,GAAGK,CAAK,EAAE,CACxD,CAAC,EAGMJ,EACJ,KACCK,EAAID,GAASF,EAAM,KAAK,CAAE,MAAAE,CAAM,CAAC,CAAC,EAClCE,EAAS,IAAMJ,EAAM,SAAS,CAAC,EAC/BK,EAAIH,IAAU,CAAE,IAAKL,EAAI,MAAAK,CAAM,EAAE,CACnC,CACJ,CAAC,CACH,CChEA,IAAAI,GAAwB,SAiCxB,SAASC,GAAQC,EAAyB,CACxCA,EAAG,aAAa,kBAAmB,EAAE,EACrC,IAAMC,EAAOD,EAAG,QAAQ,aAAa,EAC/BE,EAAOD,EACTA,EAAK,aAAa,WAAW,EAC7BD,EAAG,UACP,OAAAA,EAAG,gBAAgB,iBAAiB,EAC7BE,EAAK,QAAQ,CACtB,CAWO,SAASC,GACd,CAAE,OAAAC,CAAO,EACH,CACF,GAAAC,QAAY,YAAY,GAC1B,IAAIC,EAA8BC,GAAc,CAC9C,IAAI,GAAAF,QAAY,iDAAkD,CAChE,KAAML,GACJA,EAAG,aAAa,qBAAqB,GACrCD,GAAQS,EACNR,EAAG,aAAa,uBAAuB,CACzC,CAAC,CAEL,CAAC,EACE,GAAG,UAAWS,GAAMF,EAAW,KAAKE,CAAE,CAAC,CAC5C,CAAC,EACE,KACCC,EAAID,GAAM,CACQA,EAAG,QACX,MAAM,CAChB,CAAC,EACDE,EAAI,IAAMC,GAAY,kBAAkB,CAAC,CAC3C,EACG,UAAUR,CAAM,CAEzB,CCxCA,SAASS,GAAWC,EAAwB,CAC1C,GAAIA,EAAK,OAAS,EAChB,MAAO,CAAC,EAAE,EAGZ,GAAM,CAACC,EAAMC,CAAI,EAAI,CAAC,GAAGF,CAAI,EAC1B,KAAK,CAACG,EAAGC,IAAMD,EAAE,OAASC,EAAE,MAAM,EAClC,IAAIC,GAAOA,EAAI,QAAQ,SAAU,EAAE,CAAC,EAGnCC,EAAQ,EACZ,GAAIL,IAASC,EACXI,EAAQL,EAAK,WAEb,MAAOA,EAAK,WAAWK,CAAK,IAAMJ,EAAK,WAAWI,CAAK,GACrDA,IAGJ,OAAON,EAAK,IAAIK,GAAOA,EAAI,QAAQJ,EAAK,MAAM,EAAGK,CAAK,EAAG,EAAE,CAAC,CAC9D,CAaO,SAASC,GAAaC,EAAiC,CAC5D,IAAMC,EAAS,SAAkB,YAAa,eAAgBD,CAAI,EAClE,GAAIC,EACF,OAAOC,EAAGD,CAAM,EACX,CACL,IAAME,EAASC,GAAc,EAC7B,OAAOC,GAAW,IAAI,IAAI,cAAeL,GAAQG,EAAO,IAAI,CAAC,EAC1D,KACCG,EAAIC,GAAWhB,GAAWiB,EAAY,MAAOD,CAAO,EACjD,IAAIE,GAAQA,EAAK,WAAY,CAChC,CAAC,EACDC,GAAW,IAAMC,CAAK,EACtBC,GAAe,CAAC,CAAC,EACjBC,EAAIN,GAAW,SAAS,YAAaA,EAAS,eAAgBP,CAAI,CAAC,CACrE,CACJ,CACF,CCtBA,SAASc,GAAOC,EAAiD,CAM/D,IAAMC,EAAYC,GAAoC,kBAAmBF,CAAI,EACzE,OAAOC,GAAc,cACvBA,EAAU,KAAOA,EAAU,KAAK,QAAQ,eAAgB,cAAc,GAGxE,IAAME,EAAO,IAAI,IACjB,QAAWC,KAAMC,EAAY,aAAcL,CAAI,EAAG,CAChD,IAAIM,EAAOF,EAAG,UAMd,QAAWG,IAAO,CAAC,OAAQ,KAAK,EAAG,CACjC,IAAMC,EAAQJ,EAAG,aAAaG,CAAG,EACjC,GAAIC,IAAU,KACZ,SAGF,IAAMC,EAAM,IAAI,IAAID,EAAOP,GAAA,YAAAA,EAAW,IAAI,EACpCS,EAAMN,EAAG,UAAU,EAGzBM,EAAI,aAAaH,EAAK,GAAGE,CAAG,EAAE,EAC9BH,EAAOI,EAAI,UACX,KACF,CAGAP,EAAK,IAAIG,EAAMF,CAAE,CACnB,CAGA,OAAOD,CACT,CAgBO,SAASQ,GACd,CAAE,UAAAC,EAAW,UAAAC,EAAW,UAAAC,CAAU,EACZ,CACtB,IAAMC,EAASC,GAAc,EAC7B,GAAI,SAAS,WAAa,QACxB,OAAOC,EAMT,IAAMC,EAAWC,GAAa,EAC3B,KACCC,EAAIC,GAASA,EAAM,IAAIC,GAAQ,GAAG,IAAI,IAAIA,EAAMP,EAAO,IAAI,CAAC,EAAE,CAAC,CACjE,EAMIQ,EAAWC,EAAsB,SAAS,KAAM,OAAO,EAC1D,KACCC,GAAeP,CAAQ,EACvBQ,EAAU,CAAC,CAACC,EAAIC,CAAO,IAAM,CAC3B,GAAI,EAAED,EAAG,kBAAkB,SACzB,OAAOV,EAIT,IAAMb,EAAKuB,EAAG,OAAO,QAAQ,GAAG,EAChC,GAAIvB,IAAO,KACT,OAAOa,EAMT,GAAIb,EAAG,QAAUuB,EAAG,SAAWA,EAAG,QAChC,OAAOV,EAQT,IAAMR,EAAM,IAAI,IAAIL,EAAG,IAAI,EAO3B,OANAK,EAAI,OAASA,EAAI,KAAO,GAMnBmB,EAAQ,SAAS,GAAGnB,CAAG,EAAE,GAS9BkB,EAAG,eAAe,EACXE,EAAG,IAAI,IAAIzB,EAAG,IAAI,CAAC,GATjBa,CAUX,CAAC,EACDa,GAAM,CACR,EAIFP,EAAS,KAAKQ,GAAK,CAAC,CAAC,EAClB,UAAU,IAAM,CACf,IAAMC,EAAU9B,GAAoC,gBAAgB,EAChE,OAAO8B,GAAY,cACrBA,EAAQ,KAAOA,EAAQ,KAC3B,CAAC,EAMHR,EAAU,OAAQ,cAAc,EAC7B,UAAU,IAAM,CACf,QAAQ,kBAAoB,MAC9B,CAAC,EAMHD,EAAS,KAAKE,GAAeZ,CAAS,CAAC,EACpC,UAAU,CAAC,CAACJ,EAAK,CAAE,OAAAwB,CAAO,CAAC,IAAM,CAChC,QAAQ,kBAAoB,SAQ5B,QAAQ,aAAaA,EAAQ,EAAE,EAC/B,QAAQ,UAAU,KAAM,GAAIxB,CAAG,CACjC,CAAC,EAMHc,EAAS,UAAUX,CAAS,EAU5B,IAAMsB,EAAYtB,EACf,KACCuB,EAAUC,GAAY,CAAC,EACvBC,GAAwB,UAAU,EAClCC,GAAK,CAAC,EACNZ,EAAUjB,GAAO8B,GAAQ9B,EAAK,CAAE,UAAAK,CAAU,CAAC,EACxC,KACC0B,GAAW,KACTC,GAAYhC,EAAK,EAAI,EACdQ,EACR,CACH,CACF,CACF,EAIIyB,EAAM,IAAI,UACVC,EAAYT,EACf,KACCR,EAAUkB,GAAOA,EAAI,KAAK,CAAC,EAC3BlB,EAAUkB,GAAO,CACf,IAAMC,EAAOH,EAAI,gBAAgBE,EAAK,WAAW,EACjD,QAAWE,IAAY,CACrB,+BACA,gCACA,mCACA,+BACA,2BACA,2BACA,GAAGC,GAAQ,wBAAwB,EAC/B,CAAC,0BAA0B,EAC3B,CAAC,CACP,EAAG,CACD,IAAMC,EAAS9C,GAAmB4C,CAAQ,EACpCG,EAAS/C,GAAmB4C,EAAUD,CAAI,EAE9C,OAAOG,GAAW,aAClB,OAAOC,GAAW,aAElBD,EAAO,YAAYC,CAAM,CAE7B,CAGA,IAAMD,EAASjD,GAAO,SAAS,IAAI,EAC7BkD,EAASlD,GAAO8C,EAAK,IAAI,EAC/B,OAAW,CAACvC,EAAMF,CAAE,IAAK6C,EAKrB7C,EAAG,aAAa,KAAK,IAAM,cAC3BA,EAAG,aAAa,KAAK,IAInB4C,EAAO,IAAI1C,CAAI,EACjB0C,EAAO,OAAO1C,CAAI,EAElB,SAAS,KAAK,YAAYF,CAAE,GAKhC,QAAWA,KAAM4C,EAAO,OAAO,EAK3B5C,EAAG,aAAa,KAAK,IAAM,cAC3BA,EAAG,aAAa,KAAK,GAIrBA,EAAG,OAAO,EAId,IAAM8C,EAAYC,GAAoB,WAAW,EACjD,OAAOC,GAAO/C,EAAY,SAAU6C,CAAS,CAAC,EAC3C,KACCxB,EAAUtB,GAAM,CACd,IAAMiD,EAASR,EAAK,cAAc,QAAQ,EAC1C,GAAIzC,EAAG,IAAK,CACV,QAAWkD,KAAQlD,EAAG,kBAAkB,EACtCiD,EAAO,aAAaC,EAAMlD,EAAG,aAAakD,CAAI,CAAE,EAClD,OAAAlD,EAAG,YAAYiD,CAAM,EAGd,IAAIE,EAAWC,GAAY,CAChCH,EAAO,OAAS,IAAMG,EAAS,SAAS,CAC1C,CAAC,CAGH,KACE,QAAAH,EAAO,YAAcjD,EAAG,YACxBA,EAAG,YAAYiD,CAAM,EACdpC,CAEX,CAAC,EACDwC,EAAe,EACfC,GAAQb,CAAI,CACd,CACJ,CAAC,EACDf,GAAM,CACR,EAKF,OADkBN,EAAyB,OAAQ,UAAU,EACnD,KAAKJ,EAAIgB,EAAW,CAAC,EAC5B,UAAUxB,CAAS,EAItBA,EACG,KACCuB,EAAUC,GAAY,CAAC,EACvBuB,GAAY,EAAG,CAAC,EAChBC,EAAO,CAAC,CAACC,EAAMhB,CAAI,IACjBgB,EAAK,WAAahB,EAAK,UACvBgB,EAAK,OAAahB,EAAK,IACxB,EACDzB,EAAI,CAAC,CAAC,CAAEyB,CAAI,IAAMA,CAAI,CACxB,EACG,UAAUpC,GAAO,CAhYxB,IAAAqD,EAAAC,EAiYY,QAAQ,QAAU,MAAQ,CAACtD,EAAI,KACjC,OAAO,SAAS,GAAGsD,GAAAD,EAAA,QAAQ,QAAR,YAAAA,EAAe,IAAf,KAAAC,EAAoB,CAAC,GAExC,QAAQ,kBAAoB,OAC5BC,GAAgBvD,EAAI,IAAI,EACxB,QAAQ,kBAAoB,SAEhC,CAAC,EAILG,EACG,KACCqD,GAAO1C,CAAQ,EACfY,EAAUC,GAAY,CAAC,EACvBuB,GAAY,EAAG,CAAC,EAChBC,EAAO,CAAC,CAACC,EAAMhB,CAAI,IACjBgB,EAAK,WAAahB,EAAK,UACvBgB,EAAK,OAAahB,EAAK,IACxB,EACDzB,EAAI,CAAC,CAAC,CAAEyB,CAAI,IAAMA,CAAI,CACxB,EACG,UAAUpC,GAAO,CAChB,QAAQ,kBAAoB,OAC5BuD,GAAgBvD,EAAI,IAAI,EACxB,QAAQ,kBAAoB,SAI5B,QAAQ,KAAK,CACf,CAAC,EAMLkC,EAAU,KAAKlB,GAAeb,CAAS,CAAC,EACrC,UAAU,CAAC,CAAC,CAAEH,CAAG,IAAM,CAta5B,IAAAqD,EAAAC,EAuaU,QAAQ,QAAU,MAAQ,CAACtD,EAAI,KACjC,OAAO,SAAS,GAAGsD,GAAAD,EAAA,QAAQ,QAAR,YAAAA,EAAe,IAAf,KAAAC,EAAoB,CAAC,EAExCC,GAAgBvD,EAAI,IAAI,CAE5B,CAAC,EAMHI,EACG,KACCwB,GAAwB,QAAQ,EAChC6B,GAAa,GAAG,CAClB,EACG,UAAU,CAAC,CAAE,OAAAjC,CAAO,IAAM,CACzB,QAAQ,aAAaA,EAAQ,EAAE,CACjC,CAAC,EAGEU,CACT,CCvaA,IAAAwB,GAAuB,SAqChB,SAASC,GACdC,EAC0B,CAE1B,IAAMC,EAAQD,EAAO,UAAU,MAAM,GAAG,EAAE,IAAIE,GAC/BA,EAAK,QAAQ,sBAAuB,EAAE,EACvC,SAAW,EAAI,SAAMA,CAClC,EACE,KAAK,GAAG,EAELC,EAAY,IAAI,OAAOF,EAAO,KAAK,EACnCG,EAAY,CAACC,EAAYC,EAAcJ,IACpC,GAAGI,CAAI,2BAA2BJ,CAAI,UAI/C,OAAQK,GAAkB,CACxBA,EAAQA,EACL,QAAQ,gBAAiB,GAAG,EAC5B,KAAK,EAGR,IAAMC,EAAQ,IAAI,OAAO,MAAMR,EAAO,SAAS,MAC7CO,EACG,QAAQ,uBAAwB,MAAM,EACtC,QAAQJ,EAAW,GAAG,CAC3B,IAAK,KAAK,EAGV,OAAOM,MAAS,GAAAC,SAAWD,CAAK,EAC7B,QAAQD,EAAOJ,CAAS,EACxB,QAAQ,8BAA+B,IAAI,CAChD,CACF,CCEO,SAASO,GACdC,EAC+B,CAC/B,OAAOA,EAAQ,OAAS,CAC1B,CASO,SAASC,GACdD,EACgC,CAChC,OAAOA,EAAQ,OAAS,CAC1B,CC1CO,SAASE,GACdC,EAAaC,EACW,CACxB,IAAMC,EAAUC,GAA2BH,CAAG,EAC9C,OAAAI,EACEC,EAAG,SAAS,WAAa,OAAO,EAChCC,GAAY,QAAQ,CACtB,EACG,KACCC,GAAMC,GAAUA,CAAM,EACtBC,EAAU,IAAMR,CAAM,CACxB,EACG,UAAU,CAAC,CAAE,OAAAS,EAAQ,KAAAC,CAAK,IAAMT,EAAQ,KAAK,CAC5C,OACA,KAAM,CACJ,OAAAQ,EACA,KAAAC,EACA,QAAS,CACP,QAASC,GAAQ,gBAAgB,CACnC,CACF,CACF,CAAC,CAAC,EAGCV,CACT,CCxBO,SAASW,GACd,CAAE,UAAAC,CAAU,EACN,CACN,IAAMC,EAASC,GAAc,EACvBC,EAAYC,GAChB,IAAI,IAAI,mBAAoBH,EAAO,IAAI,CACzC,EACG,KACCI,GAAW,IAAMC,CAAK,CACxB,EAGIC,EAAWJ,EACd,KACCK,EAAIC,GAAY,CACd,GAAM,CAAC,CAAEC,CAAO,EAAIT,EAAO,KAAK,MAAM,aAAa,EACnD,OAAOQ,EAAS,KAAK,CAAC,CAAE,QAAAE,EAAS,QAAAC,CAAQ,IACvCD,IAAYD,GAAWE,EAAQ,SAASF,CAAO,CAChD,GAAKD,EAAS,CAAC,CAClB,CAAC,CACH,EAGFN,EACG,KACCK,EAAIC,GAAY,IAAI,IAAIA,EAAS,IAAIE,GAAW,CAC9C,GAAG,IAAI,IAAI,MAAMA,EAAQ,OAAO,IAAKV,EAAO,IAAI,CAAC,GACjDU,CACF,CAAC,CAAC,CAAC,EACHE,EAAUC,GAAQC,EAAsB,SAAS,KAAM,OAAO,EAC3D,KACCC,EAAOC,GAAM,CAACA,EAAG,SAAW,CAACA,EAAG,OAAO,EACvCC,GAAeX,CAAQ,EACvBM,EAAU,CAAC,CAACI,EAAIP,CAAO,IAAM,CAC3B,GAAIO,EAAG,kBAAkB,QAAS,CAChC,IAAME,EAAKF,EAAG,OAAO,QAAQ,GAAG,EAChC,GAAIE,GAAM,CAACA,EAAG,QAAUL,EAAK,IAAIK,EAAG,IAAI,EAAG,CACzC,IAAMC,EAAMD,EAAG,KAWf,MAAI,CAACF,EAAG,OAAO,QAAQ,aAAa,GAClBH,EAAK,IAAIM,CAAG,IACZV,EACPJ,GAEXW,EAAG,eAAe,EACXI,EAAGD,CAAG,EACf,CACF,CACA,OAAOd,CACT,CAAC,EACDO,EAAUO,GAAO,CACf,GAAM,CAAE,QAAAT,CAAQ,EAAIG,EAAK,IAAIM,CAAG,EAChC,OAAOE,GAAa,IAAI,IAAIF,CAAG,CAAC,EAC7B,KACCZ,EAAIe,GAAW,CAEb,IAAMC,EADWC,GAAY,EACP,KAAK,QAAQxB,EAAO,KAAM,EAAE,EAClD,OAAOsB,EAAQ,SAASC,EAAK,MAAM,GAAG,EAAE,CAAC,CAAC,EACtC,IAAI,IAAI,MAAMb,CAAO,IAAIa,CAAI,GAAIvB,EAAO,IAAI,EAC5C,IAAI,IAAImB,CAAG,CACjB,CAAC,CACH,CACJ,CAAC,CACH,CACF,CACF,EACG,UAAUA,GAAOM,GAAYN,EAAK,EAAI,CAAC,EAG5CO,EAAc,CAACxB,EAAWI,CAAQ,CAAC,EAChC,UAAU,CAAC,CAACE,EAAUC,CAAO,IAAM,CACpBkB,EAAW,mBAAmB,EACtC,YAAYC,GAAsBpB,EAAUC,CAAO,CAAC,CAC5D,CAAC,EAGHV,EAAU,KAAKa,EAAU,IAAMN,CAAQ,CAAC,EACrC,UAAUG,GAAW,CA5J1B,IAAAoB,EA+JM,IAAIC,EAAW,SAAS,aAAc,cAAc,EACpD,GAAIA,IAAa,KAAM,CACrBA,EAAW,GAGX,IAAIC,IAAUF,EAAA7B,EAAO,UAAP,YAAA6B,EAAgB,UAAW,SACpC,MAAM,QAAQE,CAAO,IACxBA,EAAU,CAACA,CAAO,GAGpBC,EAAM,QAAWC,KAAUF,EACzB,QAAWrB,KAAWD,EAAQ,QAAQ,OAAOA,EAAQ,OAAO,EAC1D,GAAI,IAAI,OAAOwB,EAAQ,GAAG,EAAE,KAAKvB,CAAO,EAAG,CACzCoB,EAAW,GACX,MAAME,CACR,CAGJ,SAAS,aAAcF,EAAU,cAAc,CACjD,CAGA,GAAIA,EACF,QAAWI,KAAWC,GAAqB,UAAU,EACnDD,EAAQ,OAAS,EACvB,CAAC,CACL,CCrFO,SAASE,GACdC,EAAsB,CAAE,QAAAC,CAAQ,EACP,CAGzB,GAAM,CAAE,aAAAC,CAAa,EAAIC,GAAY,EACjCD,EAAa,IAAI,GAAG,IACtBE,GAAU,SAAU,EAAI,EAGxBJ,EAAG,MAAQE,EAAa,IAAI,GAAG,EAC/BF,EAAG,MAAM,EAGTK,GAAY,QAAQ,EACjB,KACCC,GAAMC,GAAU,CAACA,CAAM,CACzB,EACG,UAAU,IAAM,CACf,IAAMC,EAAML,GAAY,EACxBK,EAAI,aAAa,OAAO,GAAG,EAC3B,QAAQ,aAAa,CAAC,EAAG,GAAI,GAAGA,CAAG,EAAE,CACvC,CAAC,GAIP,IAAMC,EAASC,GAAkBV,CAAE,EAC7BW,EAASC,EACbX,EAAQ,KAAKK,GAAMO,EAAoB,CAAC,EACxCC,EAAUd,EAAI,OAAO,EACrBS,CACF,EACG,KACCM,EAAI,IAAMf,EAAG,KAAK,EAClBgB,EAAqB,CACvB,EAGF,OAAOC,EAAc,CAACN,EAAQF,CAAM,CAAC,EAClC,KACCM,EAAI,CAAC,CAACG,EAAOC,CAAK,KAAO,CAAE,MAAAD,EAAO,MAAAC,CAAM,EAAE,EAC1CC,EAAY,CAAC,CACf,CACJ,CAUO,SAASC,GACdrB,EAAsB,CAAE,QAAAC,CAAQ,EACsB,CACtD,IAAMqB,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,EAAe,EAAGC,GAAQ,EAAI,CAAC,EAGxDT,EAAc,CACZhB,EAAQ,KAAKK,GAAMO,EAAoB,CAAC,EACxCS,CACF,EAAG,CAACK,EAAGC,IAAUA,CAAK,EACnB,KACCC,GAAwB,OAAO,CACjC,EACG,UAAU,CAAC,CAAE,MAAAX,CAAM,IAAMjB,EAAQ,KAAK,CACrC,OACA,KAAMiB,CACR,CAAC,CAAC,EAGNI,EACG,KACCO,GAAwB,OAAO,CACjC,EACG,UAAU,CAAC,CAAE,MAAAV,CAAM,IAAM,CACpBA,GACFf,GAAU,SAAUe,CAAK,CAC7B,CAAC,EAGLL,EAAUd,EAAG,KAAO,OAAO,EACxB,KACC8B,EAAUN,CAAK,CACjB,EACG,UAAU,IAAMxB,EAAG,MAAM,CAAC,EAM/B,IAAM+B,EAAQC,EAAW,uBAAuB,EAChD,OAAAlB,EAAUiB,EAAO,OAAO,EACrB,UAAU,IAAM/B,EAAG,MAAM,CAAC,EAGtBD,GAAiBC,EAAI,CAAE,QAAAC,CAAQ,CAAC,EACpC,KACCgC,EAAIC,GAASZ,EAAM,KAAKY,CAAK,CAAC,EAC9BC,EAAS,IAAMb,EAAM,SAAS,CAAC,EAC/BP,EAAImB,GAAUE,EAAA,CAAE,IAAKpC,GAAOkC,EAAQ,EACpCd,EAAY,CAAC,CACf,CACJ,CCnHO,SAASiB,GACdC,EAAiB,CAAE,QAAAC,EAAS,OAAAC,CAAO,EACE,CACrC,IAAMC,EAAQ,IAAIC,EACZC,EAAYC,GAAqBN,EAAG,aAAc,EACrD,KACCO,EAAO,OAAO,CAChB,EAGIC,EAAYR,EAAG,cAGfS,EAAOC,EAAW,wBAAyBV,CAAE,EAC7CW,EAAOD,EAAW,uBAAwBV,CAAE,EAGlDY,GAAY,QAAQ,EACjB,UAAUC,GAAUF,EAAK,aACxB,OAAQE,EAAS,OAAS,cAC5B,CAAC,EAGHV,EACG,KACCW,GAAeZ,CAAM,EACrBa,GAAUd,EAAQ,KAAKe,GAAMC,EAAoB,CAAC,CAAC,CACrD,EACG,UAAU,CAAC,CAAC,CAAE,MAAAC,CAAM,EAAG,CAAE,MAAAC,CAAM,CAAC,IAAM,CACrC,OAAQD,EAAM,OAAQ,CAGpB,IAAK,GACHT,EAAK,YAAcU,EAAM,OACrBC,GAAY,oBAAoB,EAChCA,GAAY,2BAA2B,EAC3C,MAGF,IAAK,GACHX,EAAK,YAAcW,GAAY,mBAAmB,EAClD,MAGF,QACE,IAAMC,EAAQC,GAAMJ,EAAM,MAAM,EAChCT,EAAK,YAAcW,GAAY,sBAAuBC,CAAK,CAC/D,CACF,CAAC,EAGL,IAAME,EAAUpB,EACb,KACCqB,EAAI,IAAMb,EAAK,UAAY,EAAE,EAC7Bc,EAAU,CAAC,CAAE,MAAAP,CAAM,IAAMQ,EACvBC,EAAG,GAAGT,EAAM,MAAM,EAAG,EAAE,CAAC,EACxBS,EAAG,GAAGT,EAAM,MAAM,EAAE,CAAC,EAClB,KACCU,GAAY,CAAC,EACbC,GAAQxB,CAAS,EACjBoB,EAAU,CAAC,CAACK,CAAK,IAAMA,CAAK,CAC9B,CACJ,CAAC,EACDC,EAAIC,EAAsB,EAC1BC,GAAM,CACR,EAGF,OAAAV,EAAQ,UAAUW,GAAQvB,EAAK,YAAYuB,CAAI,CAAC,EAChDX,EACG,KACCY,GAASD,GAAQ,CACf,IAAME,EAAUC,GAAmB,UAAWH,CAAI,EAClD,OAAI,OAAOE,GAAY,YACdE,EAGFC,EAAUH,EAAS,QAAQ,EAC/B,KACCI,EAAUrC,CAAK,EACf4B,EAAI,IAAMK,CAAO,CACnB,CACJ,CAAC,CACH,EACG,UAAUA,GAAW,CAElBA,EAAQ,OAAS,IACjBA,EAAQ,WAAa5B,EAAU,WAE/BA,EAAU,SAAS,CAAE,IAAK4B,EAAQ,SAAU,CAAC,CACjD,CAAC,EAGWnC,EACb,KACCM,EAAOkC,EAAqB,EAC5BV,EAAI,CAAC,CAAE,KAAAW,CAAK,IAAMA,CAAI,CACxB,EAIC,KACClB,EAAImB,GAASxC,EAAM,KAAKwC,CAAK,CAAC,EAC9BC,EAAS,IAAMzC,EAAM,SAAS,CAAC,EAC/B4B,EAAIY,GAAUE,EAAA,CAAE,IAAK7C,GAAO2C,EAAQ,CACtC,CACJ,CCpHO,SAASG,GACdC,EAAkB,CAAE,OAAAC,CAAO,EACF,CACzB,OAAOA,EACJ,KACCC,EAAI,CAAC,CAAE,MAAAC,CAAM,IAAM,CACjB,IAAMC,EAAMC,GAAY,EACxB,OAAAD,EAAI,KAAO,GAGXD,EAAQA,EACL,QAAQ,OAAQ,GAAG,EACnB,QAAQ,KAAM,KAAK,EACnB,QAAQ,KAAM,KAAK,EAGtBC,EAAI,OAAS,KAAKD,CAAK,GAChB,CAAE,IAAAC,CAAI,CACf,CAAC,CACH,CACJ,CAUO,SAASE,GACdC,EAAuBC,EACa,CACpC,IAAMC,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,EAAe,EAAGC,GAAQ,EAAI,CAAC,EACxD,OAAAJ,EAAM,UAAU,CAAC,CAAE,IAAAL,CAAI,IAAM,CAC3BG,EAAG,aAAa,sBAAuBA,EAAG,IAAI,EAC9CA,EAAG,KAAO,GAAGH,CAAG,EAClB,CAAC,EAGDU,EAAUP,EAAI,OAAO,EAClB,KACCQ,EAAUJ,CAAK,CACjB,EACG,UAAUK,GAAMA,EAAG,eAAe,CAAC,EAGjCjB,GAAiBQ,EAAIC,CAAO,EAChC,KACCS,EAAIC,GAAST,EAAM,KAAKS,CAAK,CAAC,EAC9BC,EAAS,IAAMV,EAAM,SAAS,CAAC,EAC/BP,EAAIgB,GAAUE,EAAA,CAAE,IAAKb,GAAOW,EAAQ,CACtC,CACJ,CCpDO,SAASG,GACdC,EAAiB,CAAE,QAAAC,EAAS,UAAAC,CAAU,EACA,CACtC,IAAMC,EAAQ,IAAIC,EAGZC,EAASC,GAAoB,cAAc,EAC3CC,EAASC,EACbC,EAAUJ,EAAO,SAAS,EAC1BI,EAAUJ,EAAO,OAAO,CAC1B,EACG,KACCK,GAAUC,EAAc,EACxBC,EAAI,IAAMP,EAAM,KAAK,EACrBQ,EAAqB,CACvB,EAGF,OAAAV,EACG,KACCW,GAAkBP,CAAM,EACxBK,EAAI,CAAC,CAAC,CAAE,QAAAG,CAAQ,EAAGC,CAAK,IAAM,CAC5B,IAAMC,EAAQD,EAAM,MAAM,UAAU,EACpC,GAAID,GAAA,MAAAA,EAAS,QAAUE,EAAMA,EAAM,OAAS,CAAC,EAAG,CAC9C,IAAMC,EAAOH,EAAQA,EAAQ,OAAS,CAAC,EACnCG,EAAK,WAAWD,EAAMA,EAAM,OAAS,CAAC,CAAC,IACzCA,EAAMA,EAAM,OAAS,CAAC,EAAIC,EAC9B,MACED,EAAM,OAAS,EAEjB,OAAOA,CACT,CAAC,CACH,EACG,UAAUA,GAASjB,EAAG,UAAYiB,EAChC,KAAK,EAAE,EACP,QAAQ,MAAO,QAAQ,CAC1B,EAGJf,EACG,KACCiB,EAAO,CAAC,CAAE,KAAAC,CAAK,IAAMA,IAAS,QAAQ,CACxC,EACG,UAAUC,GAAO,CAChB,OAAQA,EAAI,KAAM,CAGhB,IAAK,aAEDrB,EAAG,UAAU,QACbK,EAAM,iBAAmBA,EAAM,MAAM,SAErCA,EAAM,MAAQL,EAAG,WACnB,KACJ,CACF,CAAC,EAGWC,EACb,KACCkB,EAAOG,EAAqB,EAC5BV,EAAI,CAAC,CAAE,KAAAW,CAAK,IAAMA,CAAI,CACxB,EAIC,KACCC,EAAIC,GAAStB,EAAM,KAAKsB,CAAK,CAAC,EAC9BC,EAAS,IAAMvB,EAAM,SAAS,CAAC,EAC/BS,EAAI,KAAO,CAAE,IAAKZ,CAAG,EAAE,CACzB,CACJ,CCjDO,SAAS2B,GACdC,EAAiB,CAAE,OAAAC,EAAQ,UAAAC,CAAU,EACN,CAC/B,IAAMC,EAASC,GAAc,EAC7B,GAAI,CACF,IAAMC,EAAUC,GAAkBH,EAAO,OAAQF,CAAM,EAGjDM,EAASC,GAAoB,eAAgBR,CAAE,EAC/CS,EAASD,GAAoB,gBAAiBR,CAAE,EAGtDU,EAAwBV,EAAI,OAAO,EAChC,KACCW,EAAO,CAAC,CAAE,OAAAC,CAAO,IACfA,aAAkB,SAAW,CAAC,CAACA,EAAO,QAAQ,GAAG,CAClD,CACH,EACG,UAAU,IAAMC,GAAU,SAAU,EAAK,CAAC,EAG/CX,EACG,KACCS,EAAO,CAAC,CAAE,KAAAG,CAAK,IAAMA,IAAS,QAAQ,CACxC,EACG,UAAUC,GAAO,CAChB,IAAMC,EAASC,GAAiB,EAChC,OAAQF,EAAI,KAAM,CAGhB,IAAK,QACH,GAAIC,IAAWT,EAAO,CACpB,IAAMW,EAAU,IAAI,IACpB,QAAWC,KAAUC,EACnB,sBAAuBX,CACzB,EAAG,CACD,IAAMY,EAAUF,EAAO,kBACvBD,EAAQ,IAAIC,EAAQ,WAClBE,EAAQ,aAAa,eAAe,CACtC,CAAC,CACH,CAGA,GAAIH,EAAQ,KAAM,CAChB,GAAM,CAAC,CAACI,CAAI,CAAC,EAAI,CAAC,GAAGJ,CAAO,EAAE,KAAK,CAAC,CAAC,CAAEK,CAAC,EAAG,CAAC,CAAEC,CAAC,IAAMA,EAAID,CAAC,EAC1DD,EAAK,MAAM,CACb,CAGAP,EAAI,MAAM,CACZ,CACA,MAGF,IAAK,SACL,IAAK,MACHF,GAAU,SAAU,EAAK,EACzBN,EAAM,KAAK,EACX,MAGF,IAAK,UACL,IAAK,YACH,GAAI,OAAOS,GAAW,YACpBT,EAAM,MAAM,MACP,CACL,IAAMkB,EAAM,CAAClB,EAAO,GAAGa,EACrB,wDACAX,CACF,CAAC,EACKiB,EAAI,KAAK,IAAI,GACjB,KAAK,IAAI,EAAGD,EAAI,QAAQT,CAAM,CAAC,EAAIS,EAAI,QACrCV,EAAI,OAAS,UAAY,GAAK,IAE9BU,EAAI,MAAM,EACdA,EAAIC,CAAC,EAAE,MAAM,CACf,CAGAX,EAAI,MAAM,EACV,MAGF,QACMR,IAAUU,GAAiB,GAC7BV,EAAM,MAAM,CAClB,CACF,CAAC,EAGLL,EACG,KACCS,EAAO,CAAC,CAAE,KAAAG,CAAK,IAAMA,IAAS,QAAQ,CACxC,EACG,UAAUC,GAAO,CAChB,OAAQA,EAAI,KAAM,CAGhB,IAAK,IACL,IAAK,IACL,IAAK,IACHR,EAAM,MAAM,EACZA,EAAM,OAAO,EAGbQ,EAAI,MAAM,EACV,KACJ,CACF,CAAC,EAGL,IAAMY,EAASC,GAAiBrB,EAAO,CAAE,QAAAF,CAAQ,CAAC,EAClD,OAAOwB,EACLF,EACAG,GAAkBrB,EAAQ,CAAE,QAAAJ,EAAS,OAAAsB,CAAO,CAAC,CAC/C,EACG,KACCI,GAGE,GAAGC,GAAqB,eAAgBhC,CAAE,EACvC,IAAIiC,GAASC,GAAiBD,EAAO,CAAE,OAAAN,CAAO,CAAC,CAAC,EAGnD,GAAGK,GAAqB,iBAAkBhC,CAAE,EACzC,IAAIiC,GAASE,GAAmBF,EAAO,CAAE,QAAA5B,EAAS,UAAAH,CAAU,CAAC,CAAC,CACnE,CACF,CAGJ,OAASkC,EAAK,CACZ,OAAApC,EAAG,OAAS,GACLqC,EACT,CACF,CCnKO,SAASC,GACdC,EAAiB,CAAE,OAAAC,EAAQ,UAAAC,CAAU,EACG,CACxC,OAAOC,EAAc,CACnBF,EACAC,EACG,KACCE,EAAUC,GAAY,CAAC,EACvBC,EAAOC,GAAO,CAAC,CAACA,EAAI,aAAa,IAAI,GAAG,CAAC,CAC3C,CACJ,CAAC,EACE,KACCC,EAAI,CAAC,CAACC,EAAOF,CAAG,IAAMG,GAAuBD,EAAM,MAAM,EACvDF,EAAI,aAAa,IAAI,GAAG,CAC1B,CAAC,EACDC,EAAIG,GAAM,CA1FhB,IAAAC,EA2FQ,IAAMC,EAAQ,IAAI,IAGZC,EAAK,SAAS,mBAAmBd,EAAI,WAAW,SAAS,EAC/D,QAASe,EAAOD,EAAG,SAAS,EAAGC,EAAMA,EAAOD,EAAG,SAAS,EACtD,IAAIF,EAAAG,EAAK,gBAAL,MAAAH,EAAoB,aAAc,CACpC,IAAMI,EAAWD,EAAK,YAChBE,EAAWN,EAAGK,CAAQ,EACxBC,EAAS,OAASD,EAAS,QAC7BH,EAAM,IAAIE,EAAmBE,CAAQ,CACzC,CAIF,OAAW,CAACF,EAAMG,CAAI,IAAKL,EAAO,CAChC,GAAM,CAAE,WAAAM,CAAW,EAAIC,EAAE,OAAQ,KAAMF,CAAI,EAC3CH,EAAK,YAAY,GAAG,MAAM,KAAKI,CAAU,CAAC,CAC5C,CAGA,MAAO,CAAE,IAAKnB,EAAI,MAAAa,CAAM,CAC1B,CAAC,CACH,CACJ,CCPO,SAASQ,GACdC,EAAiB,CAAE,UAAAC,EAAW,MAAAC,CAAM,EACf,CACrB,IAAMC,EAASH,EAAG,QAAqB,UAAU,EAC3CI,EACJD,EAAO,UACPA,EAAO,cAAe,UAGxB,OAAOE,EAAc,CAACH,EAAOD,CAAS,CAAC,EACpC,KACCK,EAAI,CAAC,CAAC,CAAE,OAAAC,EAAQ,OAAAC,CAAO,EAAG,CAAE,OAAQ,CAAE,EAAAC,CAAE,CAAE,CAAC,KACzCD,EAASA,EACL,KAAK,IAAIJ,EAAQ,KAAK,IAAI,EAAGK,EAAIF,CAAM,CAAC,EACxCH,EACG,CACL,OAAAI,EACA,OAAQC,GAAKF,EAASH,CACxB,EACD,EACDM,EAAqB,CAACC,EAAGC,IACvBD,EAAE,SAAWC,EAAE,QACfD,EAAE,SAAWC,EAAE,MAChB,CACH,CACJ,CAuBO,SAASC,GACdb,EAAiBc,EACe,CADf,IAAAC,EAAAD,EAAE,SAAAE,CA5JrB,EA4JmBD,EAAcE,EAAAC,GAAdH,EAAc,CAAZ,YAEnB,IAAMI,EAAQC,EAAW,0BAA2BpB,CAAE,EAChD,CAAE,EAAAS,CAAE,EAAIY,GAAiBF,CAAK,EACpC,OAAOG,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,EAAe,EAAGC,GAAQ,EAAI,CAAC,EAClDC,EAAQL,EACX,KACCM,GAAU,EAAGC,EAAuB,CACtC,EAGF,OAAAF,EAAM,KAAKG,GAAef,CAAO,CAAC,EAC/B,UAAU,CAGT,KAAK,CAAC,CAAE,OAAAR,CAAO,EAAG,CAAE,OAAQD,CAAO,CAAC,EAAG,CACrCY,EAAM,MAAM,OAAS,GAAGX,EAAS,EAAIC,CAAC,KACtCT,EAAG,MAAM,IAAY,GAAGO,CAAM,IAChC,EAGA,UAAW,CACTY,EAAM,MAAM,OAAS,GACrBnB,EAAG,MAAM,IAAY,EACvB,CACF,CAAC,EAGH4B,EAAM,KAAKI,GAAM,CAAC,EACf,UAAU,IAAM,CACf,QAAWC,KAAQC,EAAY,8BAA+BlC,CAAE,EAAG,CACjE,GAAI,CAACiC,EAAK,aACR,SACF,IAAME,EAAYF,EAAK,QAAqB,yBAAyB,EACrE,GAAI,OAAOE,GAAc,YAAa,CACpC,IAAM5B,EAAS0B,EAAK,UAAYE,EAAU,UACpC,CAAE,OAAA3B,CAAO,EAAI4B,GAAeD,CAAS,EAC3CA,EAAU,SAAS,CACjB,IAAK5B,EAASC,EAAS,CACzB,CAAC,CACH,CACF,CACF,CAAC,EAGH6B,GAAKH,EAA8B,kBAAmBlC,CAAE,CAAC,EACtD,KACCsC,GAASC,GAASC,EAAUD,EAAO,OAAO,EACvC,KACCE,GAAUC,EAAc,EACxBpC,EAAI,IAAMiC,CAAK,EACfI,EAAUlB,CAAK,CACjB,CACF,CACF,EACG,UAAUc,GAAS,CAClB,IAAMK,EAAQxB,EAA6B,QAAQmB,EAAM,OAAO,IAAI,EACxDnB,EAAW,qBAAqBmB,EAAM,EAAE,IAAI,EACpD,aAAa,gBAAiB,GAAGK,EAAM,OAAO,EAAE,CACtD,CAAC,EAGE7C,GAAaC,EAAIiB,CAAO,EAC5B,KACC4B,EAAIC,GAASvB,EAAM,KAAKuB,CAAK,CAAC,EAC9BC,EAAS,IAAMxB,EAAM,SAAS,CAAC,EAC/BjB,EAAIwC,GAAUE,EAAA,CAAE,IAAKhD,GAAO8C,EAAQ,CACtC,CACJ,CAAC,CACH,CCxKO,SAASG,GACdC,EAAcC,EACW,CACzB,GAAI,OAAOA,GAAS,YAAa,CAC/B,IAAMC,EAAM,gCAAgCF,CAAI,IAAIC,CAAI,GACxD,OAAOE,GAGLC,GAAqB,GAAGF,CAAG,kBAAkB,EAC1C,KACCG,GAAW,IAAMC,CAAK,EACtBC,EAAIC,IAAY,CACd,QAASA,EAAQ,QACnB,EAAE,EACFC,GAAe,CAAC,CAAC,CACnB,EAGFL,GAAkBF,CAAG,EAClB,KACCG,GAAW,IAAMC,CAAK,EACtBC,EAAIG,IAAS,CACX,MAAOA,EAAK,iBACZ,MAAOA,EAAK,WACd,EAAE,EACFD,GAAe,CAAC,CAAC,CACnB,CACJ,EACG,KACCF,EAAI,CAAC,CAACC,EAASE,CAAI,IAAOC,IAAA,GAAKH,GAAYE,EAAO,CACpD,CAGJ,KAAO,CACL,IAAMR,EAAM,gCAAgCF,CAAI,GAChD,OAAOI,GAAkBF,CAAG,EACzB,KACCK,EAAIG,IAAS,CACX,aAAcA,EAAK,YACrB,EAAE,EACFD,GAAe,CAAC,CAAC,CACnB,CACJ,CACF,CCvDO,SAASG,GACdC,EAAcC,EACW,CACzB,IAAMC,EAAM,WAAWF,CAAI,oBAAoB,mBAAmBC,CAAO,CAAC,GAC1E,OAAOE,GAA2BD,CAAG,EAClC,KACCE,GAAW,IAAMC,CAAK,EACtBC,EAAI,CAAC,CAAE,WAAAC,EAAY,YAAAC,CAAY,KAAO,CACpC,MAAOD,EACP,MAAOC,CACT,EAAE,EACFC,GAAe,CAAC,CAAC,CACnB,CACJ,CCOO,SAASC,GACdC,EACyB,CAGzB,IAAIC,EAAQD,EAAI,MAAM,qCAAqC,EAC3D,GAAIC,EAAO,CACT,GAAM,CAAC,CAAEC,EAAMC,CAAI,EAAIF,EACvB,OAAOG,GAA2BF,EAAMC,CAAI,CAC9C,CAIA,GADAF,EAAQD,EAAI,MAAM,oCAAoC,EAClDC,EAAO,CACT,GAAM,CAAC,CAAEI,EAAMC,CAAI,EAAIL,EACvB,OAAOM,GAA2BF,EAAMC,CAAI,CAC9C,CAGA,OAAOE,CACT,CCpBA,IAAIC,GAgBG,SAASC,GACdC,EACoB,CACpB,OAAOF,QAAWG,EAAM,IAAM,CAC5B,IAAMC,EAAS,SAAsB,WAAY,cAAc,EAC/D,GAAIA,EACF,OAAOC,EAAGD,CAAM,EAKhB,GADYE,GAAqB,SAAS,EAClC,OAAQ,CACd,IAAMC,EAAU,SAA0B,WAAW,EACrD,GAAI,EAAEA,GAAWA,EAAQ,QACvB,OAAOC,CACX,CAGA,OAAOC,GAAiBP,EAAG,IAAI,EAC5B,KACCQ,EAAIC,GAAS,SAAS,WAAYA,EAAO,cAAc,CAAC,CAC1D,CAEN,CAAC,EACE,KACCC,GAAW,IAAMJ,CAAK,EACtBK,EAAOF,GAAS,OAAO,KAAKA,CAAK,EAAE,OAAS,CAAC,EAC7CG,EAAIH,IAAU,CAAE,MAAAA,CAAM,EAAE,EACxBI,EAAY,CAAC,CACf,EACJ,CASO,SAASC,GACdd,EAC+B,CAC/B,IAAMe,EAAQC,EAAW,uBAAwBhB,CAAE,EACnD,OAAOC,EAAM,IAAM,CACjB,IAAMgB,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAU,CAAC,CAAE,MAAAR,CAAM,IAAM,CAC7BM,EAAM,YAAYI,GAAkBV,CAAK,CAAC,EAC1CM,EAAM,UAAU,IAAI,+BAA+B,CACrD,CAAC,EAGMhB,GAAYC,CAAE,EAClB,KACCQ,EAAIY,GAASH,EAAM,KAAKG,CAAK,CAAC,EAC9BC,EAAS,IAAMJ,EAAM,SAAS,CAAC,EAC/BL,EAAIQ,GAAUE,EAAA,CAAE,IAAKtB,GAAOoB,EAAQ,CACtC,CACJ,CAAC,CACH,CCtDO,SAASG,GACdC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EACpB,CAClB,OAAOC,GAAiB,SAAS,IAAI,EAClC,KACCC,EAAU,IAAMC,GAAgBL,EAAI,CAAE,QAAAE,EAAS,UAAAD,CAAU,CAAC,CAAC,EAC3DK,EAAI,CAAC,CAAE,OAAQ,CAAE,EAAAC,CAAE,CAAE,KACZ,CACL,OAAQA,GAAK,EACf,EACD,EACDC,GAAwB,QAAQ,CAClC,CACJ,CAaO,SAASC,GACdT,EAAiBU,EACY,CAC7B,OAAOC,EAAM,IAAM,CACjB,IAAMC,EAAQ,IAAIC,EAClB,OAAAD,EAAM,UAAU,CAGd,KAAK,CAAE,OAAAE,CAAO,EAAG,CACfd,EAAG,OAASc,CACd,EAGA,UAAW,CACTd,EAAG,OAAS,EACd,CACF,CAAC,GAICe,GAAQ,wBAAwB,EAC5BC,EAAG,CAAE,OAAQ,EAAM,CAAC,EACpBjB,GAAUC,EAAIU,CAAO,GAExB,KACCO,EAAIC,GAASN,EAAM,KAAKM,CAAK,CAAC,EAC9BC,EAAS,IAAMP,EAAM,SAAS,CAAC,EAC/BN,EAAIY,GAAUE,EAAA,CAAE,IAAKpB,GAAOkB,EAAQ,CACtC,CACJ,CAAC,CACH,CCfO,SAASG,GACdC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EACT,CAC7B,IAAMC,EAAQ,IAAI,IAGZC,EAAUC,EAA+B,cAAeL,CAAE,EAChE,QAAWM,KAAUF,EAAS,CAC5B,IAAMG,EAAK,mBAAmBD,EAAO,KAAK,UAAU,CAAC,CAAC,EAChDE,EAASC,GAAmB,QAAQF,CAAE,IAAI,EAC5C,OAAOC,GAAW,aACpBL,EAAM,IAAIG,EAAQE,CAAM,CAC5B,CAGA,IAAME,EAAUR,EACb,KACCS,GAAwB,QAAQ,EAChCC,EAAI,CAAC,CAAE,OAAAC,CAAO,IAAM,CAClB,IAAMC,EAAOC,GAAoB,MAAM,EACjCC,EAAOC,EAAW,wBAAyBH,CAAI,EACrD,OAAOD,EAAS,IACdG,EAAK,UACLF,EAAK,UAET,CAAC,EACDI,GAAM,CACR,EAqFF,OAlFmBC,GAAiB,SAAS,IAAI,EAC9C,KACCR,GAAwB,QAAQ,EAGhCS,EAAUC,GAAQC,EAAM,IAAM,CAC5B,IAAIC,EAA4B,CAAC,EACjC,OAAOC,EAAG,CAAC,GAAGrB,CAAK,EAAE,OAAO,CAACsB,EAAO,CAACnB,EAAQE,CAAM,IAAM,CACvD,KAAOe,EAAK,QACGpB,EAAM,IAAIoB,EAAKA,EAAK,OAAS,CAAC,CAAC,EACnC,SAAWf,EAAO,SACzBe,EAAK,IAAI,EAOb,IAAIG,EAASlB,EAAO,UACpB,KAAO,CAACkB,GAAUlB,EAAO,eACvBA,EAASA,EAAO,cAChBkB,EAASlB,EAAO,UAIlB,IAAImB,EAASnB,EAAO,aACpB,KAAOmB,EAAQA,EAASA,EAAO,aAC7BD,GAAUC,EAAO,UAGnB,OAAOF,EAAM,IACX,CAAC,GAAGF,EAAO,CAAC,GAAGA,EAAMjB,CAAM,CAAC,EAAE,QAAQ,EACtCoB,CACF,CACF,EAAG,IAAI,GAAkC,CAAC,CAC5C,CAAC,EACE,KAGCd,EAAIa,GAAS,IAAI,IAAI,CAAC,GAAGA,CAAK,EAAE,KAAK,CAAC,CAAC,CAAEG,CAAC,EAAG,CAAC,CAAEC,CAAC,IAAMD,EAAIC,CAAC,CAAC,CAAC,EAC9DC,GAAkBpB,CAAO,EAGzBU,EAAU,CAAC,CAACK,EAAOM,CAAM,IAAM9B,EAC5B,KACC+B,GAAK,CAAC,CAACC,EAAMC,CAAI,EAAG,CAAE,OAAQ,CAAE,EAAAC,CAAE,EAAG,KAAAC,CAAK,IAAM,CAC9C,IAAMC,EAAOF,EAAIC,EAAK,QAAU,KAAK,MAAMf,EAAK,MAAM,EAGtD,KAAOa,EAAK,QAAQ,CAClB,GAAM,CAAC,CAAER,CAAM,EAAIQ,EAAK,CAAC,EACzB,GAAIR,EAASK,EAASI,GAAKE,EACzBJ,EAAO,CAAC,GAAGA,EAAMC,EAAK,MAAM,CAAE,MAE9B,MAEJ,CAGA,KAAOD,EAAK,QAAQ,CAClB,GAAM,CAAC,CAAEP,CAAM,EAAIO,EAAKA,EAAK,OAAS,CAAC,EACvC,GAAIP,EAASK,GAAUI,GAAK,CAACE,EAC3BH,EAAO,CAACD,EAAK,IAAI,EAAI,GAAGC,CAAI,MAE5B,MAEJ,CAGA,MAAO,CAACD,EAAMC,CAAI,CACpB,EAAG,CAAC,CAAC,EAAG,CAAC,GAAGT,CAAK,CAAC,CAAC,EACnBa,EAAqB,CAACV,EAAGC,IACvBD,EAAE,CAAC,IAAMC,EAAE,CAAC,GACZD,EAAE,CAAC,IAAMC,EAAE,CAAC,CACb,CACH,CACF,CACF,CACF,CACF,EAIC,KACCjB,EAAI,CAAC,CAACqB,EAAMC,CAAI,KAAO,CACrB,KAAMD,EAAK,IAAI,CAAC,CAACV,CAAI,IAAMA,CAAI,EAC/B,KAAMW,EAAK,IAAI,CAAC,CAACX,CAAI,IAAMA,CAAI,CACjC,EAAE,EAGFgB,EAAU,CAAE,KAAM,CAAC,EAAG,KAAM,CAAC,CAAE,CAAC,EAChCC,GAAY,EAAG,CAAC,EAChB5B,EAAI,CAAC,CAAC,EAAGiB,CAAC,IAGJ,EAAE,KAAK,OAASA,EAAE,KAAK,OAClB,CACL,KAAMA,EAAE,KAAK,MAAM,KAAK,IAAI,EAAG,EAAE,KAAK,OAAS,CAAC,EAAGA,EAAE,KAAK,MAAM,EAChE,KAAM,CAAC,CACT,EAIO,CACL,KAAMA,EAAE,KAAK,MAAM,EAAE,EACrB,KAAMA,EAAE,KAAK,MAAM,EAAGA,EAAE,KAAK,OAAS,EAAE,KAAK,MAAM,CACrD,CAEH,CACH,CACJ,CAYO,SAASY,GACdzC,EAAiB,CAAE,UAAAC,EAAW,QAAAC,EAAS,MAAAwC,EAAO,QAAAC,CAAQ,EACd,CACxC,OAAOrB,EAAM,IAAM,CACjB,IAAMsB,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,EAAe,EAAGC,GAAQ,EAAI,CAAC,EAoBxD,GAnBAJ,EAAM,UAAU,CAAC,CAAE,KAAAX,EAAM,KAAAC,CAAK,IAAM,CAGlC,OAAW,CAAC5B,CAAM,IAAK4B,EACrB5B,EAAO,UAAU,OAAO,sBAAsB,EAC9CA,EAAO,UAAU,OAAO,sBAAsB,EAIhD,OAAW,CAACmB,EAAO,CAACnB,CAAM,CAAC,IAAK2B,EAAK,QAAQ,EAC3C3B,EAAO,UAAU,IAAI,sBAAsB,EAC3CA,EAAO,UAAU,OACf,uBACAmB,IAAUQ,EAAK,OAAS,CAC1B,CAEJ,CAAC,EAGGgB,GAAQ,YAAY,EAAG,CAGzB,IAAMC,EAAUC,EACdlD,EAAU,KAAKmD,GAAa,CAAC,EAAGxC,EAAI,IAAG,EAAY,CAAC,EACpDX,EAAU,KAAKmD,GAAa,GAAG,EAAGxC,EAAI,IAAM,QAAiB,CAAC,CAChE,EAGAgC,EACG,KACCS,EAAO,CAAC,CAAE,KAAApB,CAAK,IAAMA,EAAK,OAAS,CAAC,EACpCH,GAAkBY,EAAM,KAAKY,GAAUC,EAAc,CAAC,CAAC,EACvDC,GAAeN,CAAO,CACxB,EACG,UAAU,CAAC,CAAC,CAAC,CAAE,KAAAjB,CAAK,CAAC,EAAGwB,CAAQ,IAAM,CACrC,GAAM,CAACnD,CAAM,EAAI2B,EAAKA,EAAK,OAAS,CAAC,EACrC,GAAI3B,EAAO,aAAc,CAGvB,IAAMoD,EAAYC,GAAoBrD,CAAM,EAC5C,GAAI,OAAOoD,GAAc,YAAa,CACpC,IAAMhC,EAASpB,EAAO,UAAYoD,EAAU,UACtC,CAAE,OAAA7C,CAAO,EAAI+C,GAAeF,CAAS,EAC3CA,EAAU,SAAS,CACjB,IAAKhC,EAASb,EAAS,EACvB,SAAA4C,CACF,CAAC,CACH,CACF,CACF,CAAC,CACP,CAGA,OAAIR,GAAQ,qBAAqB,GAC/BhD,EACG,KACC4D,EAAUf,CAAK,EACfnC,GAAwB,QAAQ,EAChCyC,GAAa,GAAG,EAChBU,GAAK,CAAC,EACND,EAAUlB,EAAQ,KAAKmB,GAAK,CAAC,CAAC,CAAC,EAC/BC,GAAO,CAAE,MAAO,GAAI,CAAC,EACrBP,GAAeZ,CAAK,CACtB,EACG,UAAU,CAAC,CAAC,CAAE,CAAE,KAAAX,CAAK,CAAC,IAAM,CAC3B,IAAM+B,EAAMC,GAAY,EAGlB3D,EAAS2B,EAAKA,EAAK,OAAS,CAAC,EACnC,GAAI3B,GAAUA,EAAO,OAAQ,CAC3B,GAAM,CAAC4D,CAAM,EAAI5D,EACX,CAAE,KAAA6D,CAAK,EAAI,IAAI,IAAID,EAAO,IAAI,EAChCF,EAAI,OAASG,IACfH,EAAI,KAAOG,EACX,QAAQ,aAAa,CAAC,EAAG,GAAI,GAAGH,CAAG,EAAE,EAIzC,MACEA,EAAI,KAAO,GACX,QAAQ,aAAa,CAAC,EAAG,GAAI,GAAGA,CAAG,EAAE,CAEzC,CAAC,EAGAjE,GAAqBC,EAAI,CAAE,UAAAC,EAAW,QAAAC,CAAQ,CAAC,EACnD,KACCkE,EAAIC,GAASzB,EAAM,KAAKyB,CAAK,CAAC,EAC9BC,EAAS,IAAM1B,EAAM,SAAS,CAAC,EAC/BhC,EAAIyD,GAAUE,EAAA,CAAE,IAAKvE,GAAOqE,EAAQ,CACtC,CACJ,CAAC,CACH,CC9RO,SAASG,GACdC,EAAkB,CAAE,UAAAC,EAAW,MAAAC,EAAO,QAAAC,CAAQ,EACvB,CAGvB,IAAMC,EAAaH,EAChB,KACCI,EAAI,CAAC,CAAE,OAAQ,CAAE,EAAAC,CAAE,CAAE,IAAMA,CAAC,EAC5BC,GAAY,EAAG,CAAC,EAChBF,EAAI,CAAC,CAACG,EAAGC,CAAC,IAAMD,EAAIC,GAAKA,EAAI,CAAC,EAC9BC,EAAqB,CACvB,EAGIC,EAAUT,EACb,KACCG,EAAI,CAAC,CAAE,OAAAO,CAAO,IAAMA,CAAM,CAC5B,EAGF,OAAOC,EAAc,CAACF,EAASP,CAAU,CAAC,EACvC,KACCC,EAAI,CAAC,CAACO,EAAQE,CAAS,IAAM,EAAEF,GAAUE,EAAU,EACnDJ,EAAqB,EACrBK,EAAUZ,EAAQ,KAAKa,GAAK,CAAC,CAAC,CAAC,EAC/BC,GAAQ,EAAI,EACZC,GAAO,CAAE,MAAO,GAAI,CAAC,EACrBb,EAAIc,IAAW,CAAE,OAAAA,CAAO,EAAE,CAC5B,CACJ,CAYO,SAASC,GACdC,EAAiB,CAAE,UAAApB,EAAW,QAAAqB,EAAS,MAAApB,EAAO,QAAAC,CAAQ,EACpB,CAClC,IAAMoB,EAAQ,IAAIC,EACZC,EAAQF,EAAM,KAAKG,EAAe,EAAGT,GAAQ,EAAI,CAAC,EACxD,OAAAM,EAAM,UAAU,CAGd,KAAK,CAAE,OAAAJ,CAAO,EAAG,CACfE,EAAG,OAASF,EACRA,GACFE,EAAG,aAAa,WAAY,IAAI,EAChCA,EAAG,KAAK,GAERA,EAAG,gBAAgB,UAAU,CAEjC,EAGA,UAAW,CACTA,EAAG,MAAM,IAAM,GACfA,EAAG,OAAS,GACZA,EAAG,gBAAgB,UAAU,CAC/B,CACF,CAAC,EAGDC,EACG,KACCP,EAAUU,CAAK,EACfE,GAAwB,QAAQ,CAClC,EACG,UAAU,CAAC,CAAE,OAAAC,CAAO,IAAM,CACzBP,EAAG,MAAM,IAAM,GAAGO,EAAS,EAAE,IAC/B,CAAC,EAGLC,EAAUR,EAAI,OAAO,EAClB,UAAUS,GAAM,CACfA,EAAG,eAAe,EAClB,OAAO,SAAS,CAAE,IAAK,CAAE,CAAC,CAC5B,CAAC,EAGI/B,GAAesB,EAAI,CAAE,UAAApB,EAAW,MAAAC,EAAO,QAAAC,CAAQ,CAAC,EACpD,KACC4B,EAAIC,GAAST,EAAM,KAAKS,CAAK,CAAC,EAC9BC,EAAS,IAAMV,EAAM,SAAS,CAAC,EAC/BlB,EAAI2B,GAAUE,EAAA,CAAE,IAAKb,GAAOW,EAAQ,CACtC,CACJ,CC5HO,SAASG,GACd,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EACf,CACND,EACG,KACCE,EAAU,IAAMC,EACd,2BACF,CAAC,EACDC,EAAIC,GAAM,CACRA,EAAG,cAAgB,GACnBA,EAAG,QAAU,EACf,CAAC,EACDC,GAASD,GAAME,EAAUF,EAAI,QAAQ,EAClC,KACCG,GAAU,IAAMH,EAAG,UAAU,SAAS,0BAA0B,CAAC,EACjEI,EAAI,IAAMJ,CAAE,CACd,CACF,EACAK,GAAeT,CAAO,CACxB,EACG,UAAU,CAAC,CAACI,EAAIM,CAAM,IAAM,CAC3BN,EAAG,UAAU,OAAO,0BAA0B,EAC1CM,IACFN,EAAG,QAAU,GACjB,CAAC,CACP,CC9BA,SAASO,IAAyB,CAChC,MAAO,qBAAqB,KAAK,UAAU,SAAS,CACtD,CAiBO,SAASC,GACd,CAAE,UAAAC,CAAU,EACN,CACNA,EACG,KACCC,EAAU,IAAMC,EAAY,qBAAqB,CAAC,EAClDC,EAAIC,GAAMA,EAAG,gBAAgB,mBAAmB,CAAC,EACjDC,EAAOP,EAAa,EACpBQ,GAASF,GAAMG,EAAUH,EAAI,YAAY,EACtC,KACCI,EAAI,IAAMJ,CAAE,CACd,CACF,CACF,EACG,UAAUA,GAAM,CACf,IAAMK,EAAML,EAAG,UAGXK,IAAQ,EACVL,EAAG,UAAY,EAGNK,EAAML,EAAG,eAAiBA,EAAG,eACtCA,EAAG,UAAYK,EAAM,EAEzB,CAAC,CACP,CCpCO,SAASC,GACd,CAAE,UAAAC,EAAW,QAAAC,CAAQ,EACf,CACNC,EAAc,CAACC,GAAY,QAAQ,EAAGF,CAAO,CAAC,EAC3C,KACCG,EAAI,CAAC,CAACC,EAAQC,CAAM,IAAMD,GAAU,CAACC,CAAM,EAC3CC,EAAUF,GAAUG,EAAGH,CAAM,EAC1B,KACCI,GAAMJ,EAAS,IAAM,GAAG,CAC1B,CACF,EACAK,GAAeV,CAAS,CAC1B,EACG,UAAU,CAAC,CAACK,EAAQ,CAAE,OAAQ,CAAE,EAAAM,CAAE,CAAC,CAAC,IAAM,CACzC,GAAIN,EACF,SAAS,KAAK,aAAa,qBAAsB,EAAE,EACnD,SAAS,KAAK,MAAM,IAAM,IAAIM,CAAC,SAC1B,CACL,IAAMC,EAAQ,GAAK,SAAS,SAAS,KAAK,MAAM,IAAK,EAAE,EACvD,SAAS,KAAK,gBAAgB,oBAAoB,EAClD,SAAS,KAAK,MAAM,IAAM,GACtBA,GACF,OAAO,SAAS,EAAGA,CAAK,CAC5B,CACF,CAAC,CACP,CC7DK,OAAO,UACV,OAAO,QAAU,SAAUC,EAAa,CACtC,IAAMC,EAA2B,CAAC,EAClC,QAAWC,KAAO,OAAO,KAAKF,CAAG,EAE/BC,EAAK,KAAK,CAACC,EAAKF,EAAIE,CAAG,CAAC,CAAC,EAG3B,OAAOD,CACT,GAGG,OAAO,SACV,OAAO,OAAS,SAAUD,EAAa,CACrC,IAAMC,EAAiB,CAAC,EACxB,QAAWC,KAAO,OAAO,KAAKF,CAAG,EAE/BC,EAAK,KAAKD,EAAIE,CAAG,CAAC,EAGpB,OAAOD,CACT,GAKE,OAAO,SAAY,cAGhB,QAAQ,UAAU,WACrB,QAAQ,UAAU,SAAW,SAC3BE,EAA8BC,EACxB,CACF,OAAOD,GAAM,UACf,KAAK,WAAaA,EAAE,KACpB,KAAK,UAAYA,EAAE,MAEnB,KAAK,WAAaA,EAClB,KAAK,UAAYC,EAErB,GAGG,QAAQ,UAAU,cACrB,QAAQ,UAAU,YAAc,YAC3BC,EACG,CACN,IAAMC,EAAS,KAAK,WACpB,GAAIA,EAAQ,CACND,EAAM,SAAW,GACnBC,EAAO,YAAY,IAAI,EAGzB,QAASC,EAAIF,EAAM,OAAS,EAAGE,GAAK,EAAGA,IAAK,CAC1C,IAAIC,EAAOH,EAAME,CAAC,EACd,OAAOC,GAAS,SAClBA,EAAO,SAAS,eAAeA,CAAI,EAC5BA,EAAK,YACZA,EAAK,WAAW,YAAYA,CAAI,EAG7BD,EAGHD,EAAO,aAAa,KAAK,gBAAkBE,CAAI,EAF/CF,EAAO,aAAaE,EAAM,IAAI,CAGlC,CACF,CACF,IjMKJ,SAASC,IAA4C,CACnD,OAAI,SAAS,WAAa,QACjBC,GACL,GAAG,IAAI,IAAI,yBAA0BC,GAAO,IAAI,CAAC,EACnD,EACG,KAECC,EAAI,IAAM,OAAO,EACjBC,EAAY,CAAC,CACf,EAEKC,GACL,IAAI,IAAI,2BAA4BH,GAAO,IAAI,CACjD,CAEJ,CAOA,SAAS,gBAAgB,UAAU,OAAO,OAAO,EACjD,SAAS,gBAAgB,UAAU,IAAI,IAAI,EAG3C,IAAMI,GAAYC,GAAc,EAC1BC,GAAYC,GAAc,EAC1BC,GAAYC,GAAoBH,EAAS,EACzCI,GAAYC,GAAc,EAG1BC,GAAYC,GAAc,EAC1BC,GAAYC,GAAW,oBAAoB,EAC3CC,GAAYD,GAAW,qBAAqB,EAC5CE,GAAYC,GAAW,EAGvBlB,GAASmB,GAAc,EACvBC,GAAS,SAAS,MAAM,UAAU,QAAQ,EAC5CtB,GAAiB,EACjBuB,GAGEC,GAAS,IAAIC,EACnBC,GAAiB,CAAE,OAAAF,EAAO,CAAC,EAG3B,IAAMG,GAAY,IAAIF,EAGlBG,GAAQ,oBAAoB,GAC9BC,GAAuB,CAAE,UAAArB,GAAW,UAAAM,GAAW,UAAAa,EAAU,CAAC,EACvD,UAAUrB,EAAS,EAxJxB,IAAAwB,KA2JIA,GAAA5B,GAAO,UAAP,YAAA4B,GAAgB,YAAa,QAC/BC,GAAqB,CAAE,UAAAzB,EAAU,CAAC,EAGpC0B,EAAMxB,GAAWE,EAAO,EACrB,KACCuB,GAAM,GAAG,CACX,EACG,UAAU,IAAM,CACfC,GAAU,SAAU,EAAK,EACzBA,GAAU,SAAU,EAAK,CAC3B,CAAC,EAGLtB,GACG,KACCuB,EAAO,CAAC,CAAE,KAAAC,CAAK,IAAMA,IAAS,QAAQ,CACxC,EACG,UAAUC,GAAO,CAChB,OAAQA,EAAI,KAAM,CAGhB,IAAK,IACL,IAAK,IACH,IAAMC,EAAOC,GAAoC,gBAAgB,EAC7D,OAAOD,GAAS,aAClBE,GAAYF,CAAI,EAClB,MAGF,IAAK,IACL,IAAK,IACH,IAAMG,EAAOF,GAAoC,gBAAgB,EAC7D,OAAOE,GAAS,aAClBD,GAAYC,CAAI,EAClB,MAGF,IAAK,QACH,IAAMC,EAASC,GAAiB,EAC5BD,aAAkB,kBACpBA,EAAO,MAAM,CACnB,CACF,CAAC,EAGLE,GAAmB,CAAE,UAAAtC,GAAW,QAAAU,EAAQ,CAAC,EACzC6B,GAAe,CAAE,UAAAvC,EAAU,CAAC,EAC5BwC,GAAgB,CAAE,UAAAhC,GAAW,QAAAE,EAAQ,CAAC,EAGtC,IAAM+B,GAAUC,GAAYC,GAAoB,QAAQ,EAAG,CAAE,UAAAnC,EAAU,CAAC,EAClEoC,GAAQ5C,GACX,KACCH,EAAI,IAAM8C,GAAoB,MAAM,CAAC,EACrCE,EAAUC,GAAMC,GAAUD,EAAI,CAAE,UAAAtC,GAAW,QAAAiC,EAAQ,CAAC,CAAC,EACrD3C,EAAY,CAAC,CACf,EAGIkD,GAAWtB,EAGf,GAAGuB,GAAqB,SAAS,EAC9B,IAAIH,GAAMI,GAAaJ,EAAI,CAAE,QAAA1C,EAAQ,CAAC,CAAC,EAG1C,GAAG6C,GAAqB,QAAQ,EAC7B,IAAIH,GAAMK,GAAYL,EAAI,CAAE,OAAA5B,EAAO,CAAC,CAAC,EAGxC,GAAG+B,GAAqB,QAAQ,EAC7B,IAAIH,GAAMM,GAAYN,EAAI,CAAE,UAAAtC,GAAW,QAAAiC,GAAS,MAAAG,EAAM,CAAC,CAAC,EAG3D,GAAGK,GAAqB,SAAS,EAC9B,IAAIH,GAAMO,GAAaP,CAAE,CAAC,EAG7B,GAAGG,GAAqB,UAAU,EAC/B,IAAIH,GAAMQ,GAAcR,EAAI,CAAE,UAAAzB,EAAU,CAAC,CAAC,EAG7C,GAAG4B,GAAqB,QAAQ,EAC7B,IAAIH,GAAMS,GAAYT,EAAI,CAAE,OAAA9B,GAAQ,UAAAV,EAAU,CAAC,CAAC,EAGnD,GAAG2C,GAAqB,QAAQ,EAC7B,IAAIH,GAAMU,GAAYV,CAAE,CAAC,CAC9B,EAGMW,GAAWC,EAAM,IAAMhC,EAG3B,GAAGuB,GAAqB,UAAU,EAC/B,IAAIH,GAAMa,GAAcb,CAAE,CAAC,EAG9B,GAAGG,GAAqB,SAAS,EAC9B,IAAIH,GAAMc,GAAad,EAAI,CAAE,UAAAtC,GAAW,QAAAJ,GAAS,OAAAS,EAAO,CAAC,CAAC,EAG7D,GAAGoC,GAAqB,SAAS,EAC9B,IAAIH,GAAMxB,GAAQ,kBAAkB,EACjCuC,GAAoBf,EAAI,CAAE,OAAA9B,GAAQ,UAAAd,EAAU,CAAC,EAC7C4D,CACJ,EAGF,GAAGb,GAAqB,cAAc,EACnC,IAAIH,GAAMiB,GAAiBjB,EAAI,CAAE,UAAAtC,GAAW,QAAAiC,EAAQ,CAAC,CAAC,EAGzD,GAAGQ,GAAqB,SAAS,EAC9B,IAAIH,GAAMA,EAAG,aAAa,cAAc,IAAM,aAC3CkB,GAAGpD,GAAS,IAAMqD,GAAanB,EAAI,CAAE,UAAAtC,GAAW,QAAAiC,GAAS,MAAAG,EAAM,CAAC,CAAC,EACjEoB,GAAGtD,GAAS,IAAMuD,GAAanB,EAAI,CAAE,UAAAtC,GAAW,QAAAiC,GAAS,MAAAG,EAAM,CAAC,CAAC,CACrE,EAGF,GAAGK,GAAqB,MAAM,EAC3B,IAAIH,GAAMoB,GAAUpB,EAAI,CAAE,UAAAtC,GAAW,QAAAiC,EAAQ,CAAC,CAAC,EAGlD,GAAGQ,GAAqB,KAAK,EAC1B,IAAIH,GAAMqB,GAAqBrB,EAAI,CAClC,UAAAtC,GAAW,QAAAiC,GAAS,MAAAG,GAAO,QAAAxC,EAC7B,CAAC,CAAC,EAGJ,GAAG6C,GAAqB,KAAK,EAC1B,IAAIH,GAAMsB,GAAetB,EAAI,CAAE,UAAAtC,GAAW,QAAAiC,GAAS,MAAAG,GAAO,QAAAxC,EAAQ,CAAC,CAAC,CACzE,CAAC,EAGKiE,GAAarE,GAChB,KACC6C,EAAU,IAAMY,EAAQ,EACxBa,GAAUtB,EAAQ,EAClBlD,EAAY,CAAC,CACf,EAGFuE,GAAW,UAAU,EAMrB,OAAO,UAAarE,GACpB,OAAO,UAAaE,GACpB,OAAO,QAAaE,GACpB,OAAO,UAAaE,GACpB,OAAO,UAAaE,GACpB,OAAO,QAAaE,GACpB,OAAO,QAAaE,GACpB,OAAO,OAAaC,GACpB,OAAO,OAAaK,GACpB,OAAO,UAAaG,GACpB,OAAO,WAAagD",
-  "names": ["require_focus_visible", "__commonJSMin", "exports", "module", "global", "factory", "applyFocusVisiblePolyfill", "scope", "hadKeyboardEvent", "hadFocusVisibleRecently", "hadFocusVisibleRecentlyTimeout", "inputTypesAllowlist", "isValidFocusTarget", "el", "focusTriggersKeyboardModality", "type", "tagName", "addFocusVisibleClass", "removeFocusVisibleClass", "onKeyDown", "e", "onPointerDown", "onFocus", "onBlur", "onVisibilityChange", "addInitialPointerMoveListeners", "onInitialPointerMove", "removeInitialPointerMoveListeners", "event", "error", "require_clipboard", "__commonJSMin", "exports", "module", "root", "factory", "__webpack_modules__", "__unused_webpack_module", "__webpack_exports__", "__webpack_require__", "clipboard", "tiny_emitter", "tiny_emitter_default", "listen", "listen_default", "src_select", "select_default", "command", "type", "err", "ClipboardActionCut", "target", "selectedText", "actions_cut", "createFakeElement", "value", "isRTL", "fakeElement", "yPosition", "fakeCopyAction", "options", "ClipboardActionCopy", "actions_copy", "_typeof", "obj", "ClipboardActionDefault", "_options$action", "action", "container", "text", "actions_default", "clipboard_typeof", "_classCallCheck", "instance", "Constructor", "_defineProperties", "props", "i", "descriptor", "_createClass", "protoProps", "staticProps", "_inherits", "subClass", "superClass", "_setPrototypeOf", "o", "p", "_createSuper", "Derived", "hasNativeReflectConstruct", "_isNativeReflectConstruct", "Super", "_getPrototypeOf", "result", "NewTarget", "_possibleConstructorReturn", "self", "call", "_assertThisInitialized", "e", "getAttributeValue", "suffix", "element", "attribute", "Clipboard", "_Emitter", "_super", "trigger", "_this", "_this2", "selector", "actions", "support", "DOCUMENT_NODE_TYPE", "proto", "closest", "__unused_webpack_exports", "_delegate", "callback", "useCapture", "listenerFn", "listener", "delegate", "elements", "is", "listenNode", "listenNodeList", "listenSelector", "node", "nodeList", "select", "isReadOnly", "selection", "range", "E", "name", "ctx", "data", "evtArr", "len", "evts", "liveEvents", "__webpack_module_cache__", "moduleId", "getter", "definition", "key", "prop", "require_escape_html", "__commonJSMin", "exports", "module", "matchHtmlRegExp", "escapeHtml", "string", "str", "match", "escape", "html", "index", "lastIndex", "import_focus_visible", "extendStatics", "d", "b", "p", "__extends", "__", "__awaiter", "thisArg", "_arguments", "P", "generator", "adopt", "value", "resolve", "reject", "fulfilled", "step", "e", "rejected", "result", "__generator", "body", "_", "t", "f", "y", "g", "verb", "n", "v", "op", "__values", "o", "s", "m", "i", "__read", "n", "r", "ar", "e", "error", "__spreadArray", "to", "from", "pack", "i", "l", "ar", "__await", "v", "__asyncGenerator", "thisArg", "_arguments", "generator", "g", "q", "verb", "n", "a", "resume", "step", "e", "settle", "r", "fulfill", "reject", "value", "f", "__asyncValues", "o", "m", "i", "__values", "verb", "n", "v", "resolve", "reject", "settle", "d", "isFunction", "value", "createErrorClass", "createImpl", "_super", "instance", "ctorFunc", "UnsubscriptionError", "createErrorClass", "_super", "errors", "err", "i", "arrRemove", "arr", "item", "index", "Subscription", "initialTeardown", "errors", "_parentage", "_parentage_1", "__values", "_parentage_1_1", "parent_1", "initialFinalizer", "isFunction", "e", "UnsubscriptionError", "_finalizers", "_finalizers_1", "_finalizers_1_1", "finalizer", "execFinalizer", "err", "__spreadArray", "__read", "teardown", "_a", "parent", "arrRemove", "empty", "EMPTY_SUBSCRIPTION", "Subscription", "isSubscription", "value", "isFunction", "execFinalizer", "finalizer", "config", "timeoutProvider", "handler", "timeout", "args", "_i", "delegate", "__spreadArray", "__read", "handle", "reportUnhandledError", "err", "timeoutProvider", "onUnhandledError", "config", "noop", "COMPLETE_NOTIFICATION", "createNotification", "errorNotification", "error", "nextNotification", "value", "kind", "context", "errorContext", "cb", "config", "isRoot", "_a", "errorThrown", "error", "captureError", "err", "Subscriber", "_super", "__extends", "destination", "_this", "isSubscription", "EMPTY_OBSERVER", "next", "error", "complete", "SafeSubscriber", "value", "handleStoppedNotification", "nextNotification", "err", "errorNotification", "COMPLETE_NOTIFICATION", "Subscription", "_bind", "bind", "fn", "thisArg", "ConsumerObserver", "partialObserver", "value", "error", "handleUnhandledError", "err", "SafeSubscriber", "_super", "__extends", "observerOrNext", "complete", "_this", "isFunction", "context_1", "config", "Subscriber", "handleUnhandledError", "error", "config", "captureError", "reportUnhandledError", "defaultErrorHandler", "err", "handleStoppedNotification", "notification", "subscriber", "onStoppedNotification", "timeoutProvider", "EMPTY_OBSERVER", "noop", "observable", "identity", "x", "pipe", "fns", "_i", "pipeFromArray", "identity", "input", "prev", "fn", "Observable", "subscribe", "operator", "observable", "observerOrNext", "error", "complete", "_this", "subscriber", "isSubscriber", "SafeSubscriber", "errorContext", "_a", "source", "sink", "err", "next", "promiseCtor", "getPromiseCtor", "resolve", "reject", "value", "operations", "_i", "pipeFromArray", "x", "getPromiseCtor", "promiseCtor", "_a", "config", "isObserver", "value", "isFunction", "isSubscriber", "Subscriber", "isSubscription", "hasLift", "source", "isFunction", "operate", "init", "liftedSource", "err", "createOperatorSubscriber", "destination", "onNext", "onComplete", "onError", "onFinalize", "OperatorSubscriber", "_super", "__extends", "shouldUnsubscribe", "_this", "value", "err", "closed_1", "_a", "Subscriber", "animationFrameProvider", "callback", "request", "cancel", "delegate", "handle", "timestamp", "Subscription", "args", "_i", "__spreadArray", "__read", "ObjectUnsubscribedError", "createErrorClass", "_super", "Subject", "_super", "__extends", "_this", "operator", "subject", "AnonymousSubject", "ObjectUnsubscribedError", "value", "errorContext", "_b", "__values", "_c", "observer", "err", "observers", "_a", "subscriber", "hasError", "isStopped", "EMPTY_SUBSCRIPTION", "Subscription", "arrRemove", "thrownError", "observable", "Observable", "destination", "source", "AnonymousSubject", "_super", "__extends", "destination", "source", "_this", "value", "_b", "_a", "err", "subscriber", "EMPTY_SUBSCRIPTION", "Subject", "dateTimestampProvider", "ReplaySubject", "_super", "__extends", "_bufferSize", "_windowTime", "_timestampProvider", "dateTimestampProvider", "_this", "value", "_a", "isStopped", "_buffer", "_infiniteTimeWindow", "subscriber", "subscription", "copy", "i", "adjustedBufferSize", "now", "last", "Subject", "Action", "_super", "__extends", "scheduler", "work", "state", "delay", "Subscription", "intervalProvider", "handler", "timeout", "args", "_i", "delegate", "__spreadArray", "__read", "handle", "AsyncAction", "_super", "__extends", "scheduler", "work", "_this", "state", "delay", "id", "_a", "_id", "intervalProvider", "_scheduler", "error", "_delay", "errored", "errorValue", "e", "actions", "arrRemove", "Action", "Scheduler", "schedulerActionCtor", "now", "work", "delay", "state", "dateTimestampProvider", "AsyncScheduler", "_super", "__extends", "SchedulerAction", "now", "Scheduler", "_this", "action", "actions", "error", "asyncScheduler", "AsyncScheduler", "AsyncAction", "async", "AnimationFrameAction", "_super", "__extends", "scheduler", "work", "_this", "id", "delay", "animationFrameProvider", "actions", "_a", "AsyncAction", "AnimationFrameScheduler", "_super", "__extends", "action", "flushId", "actions", "error", "AsyncScheduler", "animationFrameScheduler", "AnimationFrameScheduler", "AnimationFrameAction", "EMPTY", "Observable", "subscriber", "isScheduler", "value", "isFunction", "last", "arr", "popResultSelector", "args", "isFunction", "popScheduler", "isScheduler", "popNumber", "defaultValue", "isArrayLike", "x", "isPromise", "value", "isFunction", "isInteropObservable", "input", "isFunction", "observable", "isAsyncIterable", "obj", "isFunction", "createInvalidObservableTypeError", "input", "getSymbolIterator", "iterator", "isIterable", "input", "isFunction", "iterator", "readableStreamLikeToAsyncGenerator", "readableStream", "reader", "__await", "_a", "_b", "value", "done", "isReadableStreamLike", "obj", "isFunction", "innerFrom", "input", "Observable", "isInteropObservable", "fromInteropObservable", "isArrayLike", "fromArrayLike", "isPromise", "fromPromise", "isAsyncIterable", "fromAsyncIterable", "isIterable", "fromIterable", "isReadableStreamLike", "fromReadableStreamLike", "createInvalidObservableTypeError", "obj", "subscriber", "obs", "observable", "isFunction", "array", "i", "promise", "value", "err", "reportUnhandledError", "iterable", "iterable_1", "__values", "iterable_1_1", "asyncIterable", "process", "readableStream", "readableStreamLikeToAsyncGenerator", "asyncIterable_1", "__asyncValues", "asyncIterable_1_1", "executeSchedule", "parentSubscription", "scheduler", "work", "delay", "repeat", "scheduleSubscription", "observeOn", "scheduler", "delay", "operate", "source", "subscriber", "createOperatorSubscriber", "value", "executeSchedule", "err", "subscribeOn", "scheduler", "delay", "operate", "source", "subscriber", "scheduleObservable", "input", "scheduler", "innerFrom", "subscribeOn", "observeOn", "schedulePromise", "input", "scheduler", "innerFrom", "subscribeOn", "observeOn", "scheduleArray", "input", "scheduler", "Observable", "subscriber", "i", "scheduleIterable", "input", "scheduler", "Observable", "subscriber", "iterator", "executeSchedule", "value", "done", "_a", "err", "isFunction", "scheduleAsyncIterable", "input", "scheduler", "Observable", "subscriber", "executeSchedule", "iterator", "result", "scheduleReadableStreamLike", "input", "scheduler", "scheduleAsyncIterable", "readableStreamLikeToAsyncGenerator", "scheduled", "input", "scheduler", "isInteropObservable", "scheduleObservable", "isArrayLike", "scheduleArray", "isPromise", "schedulePromise", "isAsyncIterable", "scheduleAsyncIterable", "isIterable", "scheduleIterable", "isReadableStreamLike", "scheduleReadableStreamLike", "createInvalidObservableTypeError", "from", "input", "scheduler", "scheduled", "innerFrom", "of", "args", "_i", "scheduler", "popScheduler", "from", "throwError", "errorOrErrorFactory", "scheduler", "errorFactory", "isFunction", "init", "subscriber", "Observable", "EmptyError", "createErrorClass", "_super", "isValidDate", "value", "map", "project", "thisArg", "operate", "source", "subscriber", "index", "createOperatorSubscriber", "value", "isArray", "callOrApply", "fn", "args", "__spreadArray", "__read", "mapOneOrManyArgs", "map", "isArray", "getPrototypeOf", "objectProto", "getKeys", "argsArgArrayOrObject", "args", "first_1", "isPOJO", "keys", "key", "obj", "createObject", "keys", "values", "result", "key", "i", "combineLatest", "args", "_i", "scheduler", "popScheduler", "resultSelector", "popResultSelector", "_a", "argsArgArrayOrObject", "observables", "keys", "from", "result", "Observable", "combineLatestInit", "values", "createObject", "identity", "mapOneOrManyArgs", "valueTransform", "subscriber", "maybeSchedule", "length", "active", "remainingFirstValues", "i", "source", "hasFirstValue", "createOperatorSubscriber", "value", "execute", "subscription", "executeSchedule", "mergeInternals", "source", "subscriber", "project", "concurrent", "onBeforeNext", "expand", "innerSubScheduler", "additionalFinalizer", "buffer", "active", "index", "isComplete", "checkComplete", "outerNext", "value", "doInnerSub", "innerComplete", "innerFrom", "createOperatorSubscriber", "innerValue", "bufferedValue", "executeSchedule", "err", "mergeMap", "project", "resultSelector", "concurrent", "isFunction", "a", "i", "map", "b", "ii", "innerFrom", "operate", "source", "subscriber", "mergeInternals", "mergeAll", "concurrent", "mergeMap", "identity", "concatAll", "mergeAll", "concat", "args", "_i", "concatAll", "from", "popScheduler", "defer", "observableFactory", "Observable", "subscriber", "innerFrom", "nodeEventEmitterMethods", "eventTargetMethods", "jqueryMethods", "fromEvent", "target", "eventName", "options", "resultSelector", "isFunction", "mapOneOrManyArgs", "_a", "__read", "isEventTarget", "methodName", "handler", "isNodeStyleEventEmitter", "toCommonHandlerRegistry", "isJQueryStyleEventEmitter", "add", "remove", "isArrayLike", "mergeMap", "subTarget", "innerFrom", "Observable", "subscriber", "args", "_i", "fromEventPattern", "addHandler", "removeHandler", "resultSelector", "mapOneOrManyArgs", "Observable", "subscriber", "handler", "e", "_i", "retValue", "isFunction", "timer", "dueTime", "intervalOrScheduler", "scheduler", "async", "intervalDuration", "isScheduler", "Observable", "subscriber", "due", "isValidDate", "n", "merge", "args", "_i", "scheduler", "popScheduler", "concurrent", "popNumber", "sources", "innerFrom", "mergeAll", "from", "EMPTY", "NEVER", "Observable", "noop", "isArray", "argsOrArgArray", "args", "filter", "predicate", "thisArg", "operate", "source", "subscriber", "index", "createOperatorSubscriber", "value", "zip", "args", "_i", "resultSelector", "popResultSelector", "sources", "argsOrArgArray", "Observable", "subscriber", "buffers", "completed", "sourceIndex", "innerFrom", "createOperatorSubscriber", "value", "buffer", "result", "__spreadArray", "__read", "i", "EMPTY", "audit", "durationSelector", "operate", "source", "subscriber", "hasValue", "lastValue", "durationSubscriber", "isComplete", "endDuration", "value", "cleanupDuration", "createOperatorSubscriber", "innerFrom", "auditTime", "duration", "scheduler", "asyncScheduler", "audit", "timer", "bufferCount", "bufferSize", "startBufferEvery", "operate", "source", "subscriber", "buffers", "count", "createOperatorSubscriber", "value", "toEmit", "buffers_1", "__values", "buffers_1_1", "buffer", "toEmit_1", "toEmit_1_1", "arrRemove", "buffers_2", "buffers_2_1", "catchError", "selector", "operate", "source", "subscriber", "innerSub", "syncUnsub", "handledResult", "createOperatorSubscriber", "err", "innerFrom", "scanInternals", "accumulator", "seed", "hasSeed", "emitOnNext", "emitBeforeComplete", "source", "subscriber", "hasState", "state", "index", "createOperatorSubscriber", "value", "i", "combineLatest", "args", "_i", "resultSelector", "popResultSelector", "pipe", "__spreadArray", "__read", "mapOneOrManyArgs", "operate", "source", "subscriber", "combineLatestInit", "argsOrArgArray", "combineLatestWith", "otherSources", "_i", "combineLatest", "__spreadArray", "__read", "debounceTime", "dueTime", "scheduler", "asyncScheduler", "operate", "source", "subscriber", "activeTask", "lastValue", "lastTime", "emit", "value", "emitWhenIdle", "targetTime", "now", "createOperatorSubscriber", "defaultIfEmpty", "defaultValue", "operate", "source", "subscriber", "hasValue", "createOperatorSubscriber", "value", "take", "count", "EMPTY", "operate", "source", "subscriber", "seen", "createOperatorSubscriber", "value", "ignoreElements", "operate", "source", "subscriber", "createOperatorSubscriber", "noop", "mapTo", "value", "map", "delayWhen", "delayDurationSelector", "subscriptionDelay", "source", "concat", "take", "ignoreElements", "mergeMap", "value", "index", "innerFrom", "mapTo", "delay", "due", "scheduler", "asyncScheduler", "duration", "timer", "delayWhen", "distinctUntilChanged", "comparator", "keySelector", "identity", "defaultCompare", "operate", "source", "subscriber", "previousKey", "first", "createOperatorSubscriber", "value", "currentKey", "a", "b", "distinctUntilKeyChanged", "key", "compare", "distinctUntilChanged", "x", "y", "throwIfEmpty", "errorFactory", "defaultErrorFactory", "operate", "source", "subscriber", "hasValue", "createOperatorSubscriber", "value", "EmptyError", "endWith", "values", "_i", "source", "concat", "of", "__spreadArray", "__read", "finalize", "callback", "operate", "source", "subscriber", "first", "predicate", "defaultValue", "hasDefaultValue", "source", "filter", "v", "identity", "take", "defaultIfEmpty", "throwIfEmpty", "EmptyError", "merge", "args", "_i", "scheduler", "popScheduler", "concurrent", "popNumber", "argsOrArgArray", "operate", "source", "subscriber", "mergeAll", "from", "__spreadArray", "__read", "mergeWith", "otherSources", "_i", "merge", "__spreadArray", "__read", "repeat", "countOrConfig", "count", "delay", "_a", "EMPTY", "operate", "source", "subscriber", "soFar", "sourceSub", "resubscribe", "notifier", "timer", "innerFrom", "notifierSubscriber_1", "createOperatorSubscriber", "subscribeToSource", "syncUnsub", "sample", "notifier", "operate", "source", "subscriber", "hasValue", "lastValue", "createOperatorSubscriber", "value", "innerFrom", "noop", "scan", "accumulator", "seed", "operate", "scanInternals", "share", "options", "_a", "connector", "Subject", "_b", "resetOnError", "_c", "resetOnComplete", "_d", "resetOnRefCountZero", "wrapperSource", "connection", "resetConnection", "subject", "refCount", "hasCompleted", "hasErrored", "cancelReset", "reset", "resetAndUnsubscribe", "conn", "operate", "source", "subscriber", "dest", "handleReset", "SafeSubscriber", "value", "err", "innerFrom", "on", "args", "_i", "onSubscriber", "__spreadArray", "__read", "shareReplay", "configOrBufferSize", "windowTime", "scheduler", "bufferSize", "refCount", "_a", "_b", "_c", "share", "ReplaySubject", "skip", "count", "filter", "_", "index", "skipUntil", "notifier", "operate", "source", "subscriber", "taking", "skipSubscriber", "createOperatorSubscriber", "noop", "innerFrom", "value", "startWith", "values", "_i", "scheduler", "popScheduler", "operate", "source", "subscriber", "concat", "switchMap", "project", "resultSelector", "operate", "source", "subscriber", "innerSubscriber", "index", "isComplete", "checkComplete", "createOperatorSubscriber", "value", "innerIndex", "outerIndex", "innerFrom", "innerValue", "takeUntil", "notifier", "operate", "source", "subscriber", "innerFrom", "createOperatorSubscriber", "noop", "takeWhile", "predicate", "inclusive", "operate", "source", "subscriber", "index", "createOperatorSubscriber", "value", "result", "tap", "observerOrNext", "error", "complete", "tapObserver", "isFunction", "operate", "source", "subscriber", "_a", "isUnsub", "createOperatorSubscriber", "value", "err", "_b", "identity", "throttle", "durationSelector", "config", "operate", "source", "subscriber", "_a", "_b", "leading", "_c", "trailing", "hasValue", "sendValue", "throttled", "isComplete", "endThrottling", "send", "cleanupThrottling", "startThrottle", "value", "innerFrom", "createOperatorSubscriber", "throttleTime", "duration", "scheduler", "config", "asyncScheduler", "duration$", "timer", "throttle", "withLatestFrom", "inputs", "_i", "project", "popResultSelector", "operate", "source", "subscriber", "len", "otherValues", "hasValue", "ready", "i", "innerFrom", "createOperatorSubscriber", "value", "identity", "noop", "values", "__spreadArray", "__read", "zip", "sources", "_i", "operate", "source", "subscriber", "__spreadArray", "__read", "zipWith", "otherInputs", "_i", "zip", "__spreadArray", "__read", "watchDocument", "document$", "ReplaySubject", "fromEvent", "getElements", "selector", "node", "getElement", "el", "getOptionalElement", "getActiveElement", "observer$", "merge", "fromEvent", "debounceTime", "startWith", "map", "getActiveElement", "shareReplay", "watchElementFocus", "el", "active", "distinctUntilChanged", "getElementOffset", "el", "watchElementOffset", "merge", "fromEvent", "auditTime", "animationFrameScheduler", "map", "startWith", "getElementContentOffset", "el", "watchElementContentOffset", "merge", "fromEvent", "auditTime", "animationFrameScheduler", "map", "startWith", "appendChild", "el", "child", "node", "h", "tag", "attributes", "children", "attr", "round", "value", "digits", "watchScript", "src", "script", "h", "defer", "merge", "fromEvent", "switchMap", "throwError", "map", "finalize", "take", "entry$", "Subject", "observer$", "defer", "watchScript", "of", "map", "entries", "entry", "switchMap", "observer", "merge", "NEVER", "finalize", "shareReplay", "getElementSize", "el", "watchElementSize", "tap", "filter", "target", "startWith", "getElementContentSize", "el", "getElementContainer", "parent", "entry$", "Subject", "observer$", "defer", "of", "entries", "entry", "switchMap", "observer", "merge", "NEVER", "finalize", "shareReplay", "watchElementVisibility", "el", "tap", "filter", "target", "map", "isIntersecting", "watchElementBoundary", "threshold", "watchElementContentOffset", "y", "visible", "getElementSize", "content", "getElementContentSize", "distinctUntilChanged", "toggles", "getElement", "getToggle", "name", "setToggle", "value", "watchToggle", "el", "fromEvent", "map", "startWith", "isSusceptibleToKeyboard", "el", "type", "watchComposition", "merge", "fromEvent", "map", "startWith", "watchKeyboard", "keyboard$", "filter", "ev", "getToggle", "mode", "active", "getActiveElement", "share", "switchMap", "EMPTY", "getLocation", "setLocation", "url", "navigate", "feature", "el", "h", "watchLocation", "Subject", "getLocationHash", "setLocationHash", "hash", "el", "h", "ev", "watchLocationHash", "location$", "merge", "fromEvent", "map", "startWith", "filter", "shareReplay", "watchLocationTarget", "id", "getOptionalElement", "watchMedia", "query", "media", "fromEventPattern", "next", "startWith", "watchPrint", "merge", "fromEvent", "map", "at", "query$", "factory", "switchMap", "active", "EMPTY", "request", "url", "options", "Observable", "observer", "req", "event", "totalFromHeader", "requestJSON", "switchMap", "res", "map", "body", "shareReplay", "requestXML", "dom", "getViewportOffset", "watchViewportOffset", "merge", "fromEvent", "map", "startWith", "getViewportSize", "watchViewportSize", "fromEvent", "map", "startWith", "watchViewport", "combineLatest", "watchViewportOffset", "watchViewportSize", "map", "offset", "size", "shareReplay", "watchViewportAt", "el", "viewport$", "header$", "size$", "distinctUntilKeyChanged", "offset$", "combineLatest", "map", "getElementOffset", "height", "offset", "size", "x", "y", "recv", "worker", "fromEvent", "ev", "send", "send$", "Subject", "data", "watchWorker", "url", "recv$", "worker$", "done$", "ignoreElements", "endWith", "mergeWith", "takeUntil", "share", "script", "getElement", "config", "getLocation", "configuration", "feature", "flag", "translation", "key", "value", "getComponentElement", "type", "node", "getElement", "getComponentElements", "getElements", "watchAnnounce", "el", "button", "getElement", "fromEvent", "map", "content", "mountAnnounce", "feature", "EMPTY", "defer", "push$", "Subject", "hash", "tap", "state", "finalize", "__spreadValues", "watchConsent", "el", "target$", "map", "target", "mountConsent", "options", "internal$", "Subject", "hidden", "tap", "state", "finalize", "__spreadValues", "watchAnnotation", "el", "container", "offset$", "defer", "combineLatest", "watchElementOffset", "watchElementContentOffset", "map", "x", "y", "scroll", "width", "height", "getElementSize", "watchElementFocus", "switchMap", "active", "offset", "take", "mountAnnotation", "target$", "tooltip", "index", "push$", "Subject", "done$", "ignoreElements", "endWith", "watchElementVisibility", "takeUntil", "visible", "merge", "filter", "debounceTime", "auditTime", "animationFrameScheduler", "throttleTime", "origin", "fromEvent", "ev", "withLatestFrom", "_a", "parent", "getActiveElement", "target", "delay", "tap", "state", "finalize", "__spreadValues", "renderTooltip", "id", "h", "renderAnnotation", "id", "prefix", "anchor", "h", "renderTooltip", "renderClipboardButton", "id", "h", "translation", "renderSearchDocument", "document", "flag", "parent", "teaser", "missing", "key", "list", "h", "config", "configuration", "url", "feature", "match", "highlight", "value", "tags", "tag", "type", "translation", "renderSearchResultItem", "result", "threshold", "docs", "doc", "article", "index", "best", "more", "children", "section", "renderSourceFacts", "facts", "h", "key", "value", "round", "renderTabbedControl", "type", "classes", "h", "renderTable", "table", "h", "renderVersion", "version", "config", "configuration", "url", "h", "renderVersionSelector", "versions", "active", "translation", "findHosts", "container", "getElements", "findMarkers", "markers", "el", "nodes", "it", "node", "text", "match", "id", "force", "marker", "swap", "source", "target", "mountAnnotationList", "target$", "print$", "parent", "prefix", "annotations", "getOptionalElement", "renderAnnotation", "EMPTY", "defer", "push$", "Subject", "done$", "ignoreElements", "endWith", "pairs", "annotation", "getElement", "takeUntil", "active", "inner", "child", "merge", "mountAnnotation", "finalize", "share", "findList", "el", "sibling", "mountAnnotationBlock", "options", "defer", "list", "mountAnnotationList", "EMPTY", "import_clipboard", "sequence", "findCandidateList", "el", "sibling", "watchCodeBlock", "watchElementSize", "map", "width", "getElementContentSize", "distinctUntilKeyChanged", "mountCodeBlock", "options", "hover", "factory$", "defer", "push$", "Subject", "scrollable", "ClipboardJS", "feature", "parent", "renderClipboardButton", "container", "list", "annotations$", "mountAnnotationList", "tap", "state", "finalize", "__spreadValues", "mergeWith", "height", "distinctUntilChanged", "switchMap", "active", "EMPTY", "watchElementVisibility", "filter", "visible", "take", "watchDetails", "el", "target$", "print$", "open", "merge", "map", "target", "filter", "details", "active", "tap", "mountDetails", "options", "defer", "push$", "Subject", "action", "reveal", "state", "finalize", "__spreadValues", "mermaid_default", "mermaid$", "sequence", "fetchScripts", "watchScript", "of", "mountMermaid", "el", "tap", "mermaid_default", "map", "shareReplay", "id", "host", "h", "text", "svg", "fn", "shadow", "sentinel", "h", "mountDataTable", "el", "renderTable", "of", "watchContentTabs", "el", "inputs", "getElements", "initial", "input", "merge", "fromEvent", "map", "getElement", "startWith", "active", "mountContentTabs", "viewport$", "prev", "renderTabbedControl", "next", "container", "defer", "push$", "Subject", "done$", "ignoreElements", "endWith", "combineLatest", "watchElementSize", "auditTime", "animationFrameScheduler", "takeUntil", "size", "offset", "getElementOffset", "width", "getElementSize", "content", "getElementContentOffset", "watchElementContentOffset", "getElementContentSize", "direction", "feature", "skip", "withLatestFrom", "tab", "y", "set", "label", "tabs", "media", "tap", "state", "finalize", "__spreadValues", "subscribeOn", "asyncScheduler", "mountContent", "el", "viewport$", "target$", "print$", "merge", "getElements", "child", "mountAnnotationBlock", "mountCodeBlock", "mountMermaid", "mountDataTable", "mountDetails", "mountContentTabs", "watchDialog", "_el", "alert$", "switchMap", "message", "merge", "of", "delay", "map", "active", "mountDialog", "el", "options", "inner", "getElement", "defer", "push$", "Subject", "tap", "state", "finalize", "__spreadValues", "isHidden", "viewport$", "feature", "of", "direction$", "map", "y", "bufferCount", "a", "b", "distinctUntilKeyChanged", "hidden$", "combineLatest", "filter", "offset", "direction", "distinctUntilChanged", "search$", "watchToggle", "search", "switchMap", "active", "startWith", "watchHeader", "el", "options", "defer", "watchElementSize", "height", "hidden", "shareReplay", "mountHeader", "header$", "main$", "push$", "Subject", "done$", "ignoreElements", "endWith", "combineLatestWith", "takeUntil", "state", "__spreadValues", "watchHeaderTitle", "el", "viewport$", "header$", "watchViewportAt", "map", "y", "height", "getElementSize", "distinctUntilKeyChanged", "mountHeaderTitle", "options", "defer", "push$", "Subject", "active", "heading", "getOptionalElement", "EMPTY", "tap", "state", "finalize", "__spreadValues", "watchMain", "el", "viewport$", "header$", "adjust$", "map", "height", "distinctUntilChanged", "border$", "switchMap", "watchElementSize", "distinctUntilKeyChanged", "combineLatest", "header", "top", "bottom", "y", "a", "b", "watchPalette", "inputs", "current", "input", "of", "mergeMap", "fromEvent", "map", "startWith", "shareReplay", "mountPalette", "el", "meta", "h", "scheme", "defer", "push$", "Subject", "palette", "key", "value", "index", "label", "header", "getComponentElement", "style", "color", "observeOn", "asyncScheduler", "getElements", "tap", "state", "finalize", "__spreadValues", "mountProgress", "el", "progress$", "defer", "push$", "Subject", "value", "tap", "finalize", "map", "import_clipboard", "extract", "el", "copy", "text", "setupClipboardJS", "alert$", "ClipboardJS", "Observable", "subscriber", "getElement", "ev", "tap", "map", "translation", "preprocess", "urls", "root", "next", "a", "b", "url", "index", "fetchSitemap", "base", "cached", "of", "config", "configuration", "requestXML", "map", "sitemap", "getElements", "node", "catchError", "EMPTY", "defaultIfEmpty", "tap", "lookup", "head", "canonical", "getOptionalElement", "tags", "el", "getElements", "html", "key", "value", "url", "ref", "setupInstantNavigation", "location$", "viewport$", "progress$", "config", "configuration", "EMPTY", "sitemap$", "fetchSitemap", "map", "paths", "path", "instant$", "fromEvent", "withLatestFrom", "switchMap", "ev", "sitemap", "of", "share", "take", "favicon", "offset", "response$", "startWith", "getLocation", "distinctUntilKeyChanged", "skip", "request", "catchError", "setLocation", "dom", "document$", "res", "next", "selector", "feature", "source", "target", "container", "getComponentElement", "concat", "script", "name", "Observable", "observer", "ignoreElements", "endWith", "bufferCount", "filter", "prev", "_a", "_b", "setLocationHash", "sample", "debounceTime", "import_escape_html", "setupSearchHighlighter", "config", "regex", "term", "separator", "highlight", "_", "data", "query", "match", "value", "escapeHTML", "isSearchReadyMessage", "message", "isSearchResultMessage", "setupSearchWorker", "url", "index$", "worker$", "watchWorker", "merge", "of", "watchToggle", "first", "active", "switchMap", "config", "docs", "feature", "setupVersionSelector", "document$", "config", "configuration", "versions$", "requestJSON", "catchError", "EMPTY", "current$", "map", "versions", "current", "version", "aliases", "switchMap", "urls", "fromEvent", "filter", "ev", "withLatestFrom", "el", "url", "of", "fetchSitemap", "sitemap", "path", "getLocation", "setLocation", "combineLatest", "getElement", "renderVersionSelector", "_a", "outdated", "ignored", "main", "ignore", "warning", "getComponentElements", "watchSearchQuery", "el", "worker$", "searchParams", "getLocation", "setToggle", "watchToggle", "first", "active", "url", "focus$", "watchElementFocus", "value$", "merge", "isSearchReadyMessage", "fromEvent", "map", "distinctUntilChanged", "combineLatest", "value", "focus", "shareReplay", "mountSearchQuery", "push$", "Subject", "done$", "ignoreElements", "endWith", "_", "query", "distinctUntilKeyChanged", "takeUntil", "label", "getElement", "tap", "state", "finalize", "__spreadValues", "mountSearchResult", "el", "worker$", "query$", "push$", "Subject", "boundary$", "watchElementBoundary", "filter", "container", "meta", "getElement", "list", "watchToggle", "active", "withLatestFrom", "skipUntil", "first", "isSearchReadyMessage", "items", "value", "translation", "count", "round", "render$", "tap", "switchMap", "merge", "of", "bufferCount", "zipWith", "chunk", "map", "renderSearchResultItem", "share", "item", "mergeMap", "details", "getOptionalElement", "EMPTY", "fromEvent", "takeUntil", "isSearchResultMessage", "data", "state", "finalize", "__spreadValues", "watchSearchShare", "_el", "query$", "map", "value", "url", "getLocation", "mountSearchShare", "el", "options", "push$", "Subject", "done$", "ignoreElements", "endWith", "fromEvent", "takeUntil", "ev", "tap", "state", "finalize", "__spreadValues", "mountSearchSuggest", "el", "worker$", "keyboard$", "push$", "Subject", "query", "getComponentElement", "query$", "merge", "fromEvent", "observeOn", "asyncScheduler", "map", "distinctUntilChanged", "combineLatestWith", "suggest", "value", "words", "last", "filter", "mode", "key", "isSearchResultMessage", "data", "tap", "state", "finalize", "mountSearch", "el", "index$", "keyboard$", "config", "configuration", "worker$", "setupSearchWorker", "query", "getComponentElement", "result", "fromEvent", "filter", "target", "setToggle", "mode", "key", "active", "getActiveElement", "anchors", "anchor", "getElements", "article", "best", "a", "b", "els", "i", "query$", "mountSearchQuery", "merge", "mountSearchResult", "mergeWith", "getComponentElements", "child", "mountSearchShare", "mountSearchSuggest", "err", "NEVER", "mountSearchHiglight", "el", "index$", "location$", "combineLatest", "startWith", "getLocation", "filter", "url", "map", "index", "setupSearchHighlighter", "fn", "_a", "nodes", "it", "node", "original", "replaced", "text", "childNodes", "h", "watchSidebar", "el", "viewport$", "main$", "parent", "adjust", "combineLatest", "map", "offset", "height", "y", "distinctUntilChanged", "a", "b", "mountSidebar", "_a", "_b", "header$", "options", "__objRest", "inner", "getElement", "getElementOffset", "defer", "push$", "Subject", "done$", "ignoreElements", "endWith", "next$", "auditTime", "animationFrameScheduler", "withLatestFrom", "first", "item", "getElements", "container", "getElementSize", "from", "mergeMap", "label", "fromEvent", "observeOn", "asyncScheduler", "takeUntil", "input", "tap", "state", "finalize", "__spreadValues", "fetchSourceFactsFromGitHub", "user", "repo", "url", "zip", "requestJSON", "catchError", "EMPTY", "map", "release", "defaultIfEmpty", "info", "__spreadValues", "fetchSourceFactsFromGitLab", "base", "project", "url", "requestJSON", "catchError", "EMPTY", "map", "star_count", "forks_count", "defaultIfEmpty", "fetchSourceFacts", "url", "match", "user", "repo", "fetchSourceFactsFromGitHub", "base", "slug", "fetchSourceFactsFromGitLab", "EMPTY", "fetch$", "watchSource", "el", "defer", "cached", "of", "getComponentElements", "consent", "EMPTY", "fetchSourceFacts", "tap", "facts", "catchError", "filter", "map", "shareReplay", "mountSource", "inner", "getElement", "push$", "Subject", "renderSourceFacts", "state", "finalize", "__spreadValues", "watchTabs", "el", "viewport$", "header$", "watchElementSize", "switchMap", "watchViewportAt", "map", "y", "distinctUntilKeyChanged", "mountTabs", "options", "defer", "push$", "Subject", "hidden", "feature", "of", "tap", "state", "finalize", "__spreadValues", "watchTableOfContents", "el", "viewport$", "header$", "table", "anchors", "getElements", "anchor", "id", "target", "getOptionalElement", "adjust$", "distinctUntilKeyChanged", "map", "height", "main", "getComponentElement", "grid", "getElement", "share", "watchElementSize", "switchMap", "body", "defer", "path", "of", "index", "offset", "parent", "a", "b", "combineLatestWith", "adjust", "scan", "prev", "next", "y", "size", "last", "distinctUntilChanged", "startWith", "bufferCount", "mountTableOfContents", "main$", "target$", "push$", "Subject", "done$", "ignoreElements", "endWith", "feature", "smooth$", "merge", "debounceTime", "filter", "observeOn", "asyncScheduler", "withLatestFrom", "behavior", "container", "getElementContainer", "getElementSize", "takeUntil", "skip", "repeat", "url", "getLocation", "active", "hash", "tap", "state", "finalize", "__spreadValues", "watchBackToTop", "_el", "viewport$", "main$", "target$", "direction$", "map", "y", "bufferCount", "a", "b", "distinctUntilChanged", "active$", "active", "combineLatest", "direction", "takeUntil", "skip", "endWith", "repeat", "hidden", "mountBackToTop", "el", "header$", "push$", "Subject", "done$", "ignoreElements", "distinctUntilKeyChanged", "height", "fromEvent", "ev", "tap", "state", "finalize", "__spreadValues", "patchIndeterminate", "document$", "tablet$", "switchMap", "getElements", "tap", "el", "mergeMap", "fromEvent", "takeWhile", "map", "withLatestFrom", "tablet", "isAppleDevice", "patchScrollfix", "document$", "switchMap", "getElements", "tap", "el", "filter", "mergeMap", "fromEvent", "map", "top", "patchScrolllock", "viewport$", "tablet$", "combineLatest", "watchToggle", "map", "active", "tablet", "switchMap", "of", "delay", "withLatestFrom", "y", "value", "obj", "data", "key", "x", "y", "nodes", "parent", "i", "node", "fetchSearchIndex", "watchScript", "config", "map", "shareReplay", "requestJSON", "document$", "watchDocument", "location$", "watchLocation", "target$", "watchLocationTarget", "keyboard$", "watchKeyboard", "viewport$", "watchViewport", "tablet$", "watchMedia", "screen$", "print$", "watchPrint", "configuration", "index$", "NEVER", "alert$", "Subject", "setupClipboardJS", "progress$", "feature", "setupInstantNavigation", "_a", "setupVersionSelector", "merge", "delay", "setToggle", "filter", "mode", "key", "prev", "getOptionalElement", "setLocation", "next", "active", "getActiveElement", "patchIndeterminate", "patchScrollfix", "patchScrolllock", "header$", "watchHeader", "getComponentElement", "main$", "switchMap", "el", "watchMain", "control$", "getComponentElements", "mountConsent", "mountDialog", "mountHeader", "mountPalette", "mountProgress", "mountSearch", "mountSource", "content$", "defer", "mountAnnounce", "mountContent", "mountSearchHiglight", "EMPTY", "mountHeaderTitle", "at", "mountSidebar", "mountTabs", "mountTableOfContents", "mountBackToTop", "component$", "mergeWith"]
-}
diff --git a/assets/javascripts/workers/search.f886a092.min.js b/assets/javascripts/workers/search.b8dbb3d2.min.js
similarity index 99%
rename from assets/javascripts/workers/search.f886a092.min.js
rename to assets/javascripts/workers/search.b8dbb3d2.min.js
index 283b855c33..c8a1ec8d8c 100644
--- a/assets/javascripts/workers/search.f886a092.min.js
+++ b/assets/javascripts/workers/search.b8dbb3d2.min.js
@@ -38,5 +38,5 @@
  * lunr.Builder
  * Copyright (C) 2020 Oliver Nightingale
  */t.Builder=function(){this._ref="id",this._fields=Object.create(null),this._documents=Object.create(null),this.invertedIndex=Object.create(null),this.fieldTermFrequencies={},this.fieldLengths={},this.tokenizer=t.tokenizer,this.pipeline=new t.Pipeline,this.searchPipeline=new t.Pipeline,this.documentCount=0,this._b=.75,this._k1=1.2,this.termIndex=0,this.metadataWhitelist=[]},t.Builder.prototype.ref=function(e){this._ref=e},t.Builder.prototype.field=function(e,r){if(/\//.test(e))throw new RangeError("Field '"+e+"' contains illegal character '/'");this._fields[e]=r||{}},t.Builder.prototype.b=function(e){e<0?this._b=0:e>1?this._b=1:this._b=e},t.Builder.prototype.k1=function(e){this._k1=e},t.Builder.prototype.add=function(e,r){var n=e[this._ref],i=Object.keys(this._fields);this._documents[n]=r||{},this.documentCount+=1;for(var s=0;s<i.length;s++){var o=i[s],a=this._fields[o].extractor,u=a?a(e):e[o],c=this.tokenizer(u,{fields:[o]}),f=this.pipeline.run(c),g=new t.FieldRef(n,o),l=Object.create(null);this.fieldTermFrequencies[g]=l,this.fieldLengths[g]=0,this.fieldLengths[g]+=f.length;for(var m=0;m<f.length;m++){var x=f[m];if(l[x]==null&&(l[x]=0),l[x]+=1,this.invertedIndex[x]==null){var v=Object.create(null);v._index=this.termIndex,this.termIndex+=1;for(var d=0;d<i.length;d++)v[i[d]]=Object.create(null);this.invertedIndex[x]=v}this.invertedIndex[x][o][n]==null&&(this.invertedIndex[x][o][n]=Object.create(null));for(var y=0;y<this.metadataWhitelist.length;y++){var b=this.metadataWhitelist[y],E=x.metadata[b];this.invertedIndex[x][o][n][b]==null&&(this.invertedIndex[x][o][n][b]=[]),this.invertedIndex[x][o][n][b].push(E)}}}},t.Builder.prototype.calculateAverageFieldLengths=function(){for(var e=Object.keys(this.fieldLengths),r=e.length,n={},i={},s=0;s<r;s++){var o=t.FieldRef.fromString(e[s]),a=o.fieldName;i[a]||(i[a]=0),i[a]+=1,n[a]||(n[a]=0),n[a]+=this.fieldLengths[o]}for(var u=Object.keys(this._fields),s=0;s<u.length;s++){var c=u[s];n[c]=n[c]/i[c]}this.averageFieldLength=n},t.Builder.prototype.createFieldVectors=function(){for(var e={},r=Object.keys(this.fieldTermFrequencies),n=r.length,i=Object.create(null),s=0;s<n;s++){for(var o=t.FieldRef.fromString(r[s]),a=o.fieldName,u=this.fieldLengths[o],c=new t.Vector,f=this.fieldTermFrequencies[o],g=Object.keys(f),l=g.length,m=this._fields[a].boost||1,x=this._documents[o.docRef].boost||1,v=0;v<l;v++){var d=g[v],y=f[d],b=this.invertedIndex[d]._index,E,w,R;i[d]===void 0?(E=t.idf(this.invertedIndex[d],this.documentCount),i[d]=E):E=i[d],w=E*((this._k1+1)*y)/(this._k1*(1-this._b+this._b*(u/this.averageFieldLength[a]))+y),w*=m,w*=x,R=Math.round(w*1e3)/1e3,c.insert(b,R)}e[o]=c}this.fieldVectors=e},t.Builder.prototype.createTokenSet=function(){this.tokenSet=t.TokenSet.fromArray(Object.keys(this.invertedIndex).sort())},t.Builder.prototype.build=function(){return this.calculateAverageFieldLengths(),this.createFieldVectors(),this.createTokenSet(),new t.Index({invertedIndex:this.invertedIndex,fieldVectors:this.fieldVectors,tokenSet:this.tokenSet,fields:Object.keys(this._fields),pipeline:this.searchPipeline})},t.Builder.prototype.use=function(e){var r=Array.prototype.slice.call(arguments,1);r.unshift(this),e.apply(this,r)},t.MatchData=function(e,r,n){for(var i=Object.create(null),s=Object.keys(n||{}),o=0;o<s.length;o++){var a=s[o];i[a]=n[a].slice()}this.metadata=Object.create(null),e!==void 0&&(this.metadata[e]=Object.create(null),this.metadata[e][r]=i)},t.MatchData.prototype.combine=function(e){for(var r=Object.keys(e.metadata),n=0;n<r.length;n++){var i=r[n],s=Object.keys(e.metadata[i]);this.metadata[i]==null&&(this.metadata[i]=Object.create(null));for(var o=0;o<s.length;o++){var a=s[o],u=Object.keys(e.metadata[i][a]);this.metadata[i][a]==null&&(this.metadata[i][a]=Object.create(null));for(var c=0;c<u.length;c++){var f=u[c];this.metadata[i][a][f]==null?this.metadata[i][a][f]=e.metadata[i][a][f]:this.metadata[i][a][f]=this.metadata[i][a][f].concat(e.metadata[i][a][f])}}}},t.MatchData.prototype.add=function(e,r,n){if(!(e in this.metadata)){this.metadata[e]=Object.create(null),this.metadata[e][r]=n;return}if(!(r in this.metadata[e])){this.metadata[e][r]=n;return}for(var i=Object.keys(n),s=0;s<i.length;s++){var o=i[s];o in this.metadata[e][r]?this.metadata[e][r][o]=this.metadata[e][r][o].concat(n[o]):this.metadata[e][r][o]=n[o]}},t.Query=function(e){this.clauses=[],this.allFields=e},t.Query.wildcard=new String("*"),t.Query.wildcard.NONE=0,t.Query.wildcard.LEADING=1,t.Query.wildcard.TRAILING=2,t.Query.presence={OPTIONAL:1,REQUIRED:2,PROHIBITED:3},t.Query.prototype.clause=function(e){return"fields"in e||(e.fields=this.allFields),"boost"in e||(e.boost=1),"usePipeline"in e||(e.usePipeline=!0),"wildcard"in e||(e.wildcard=t.Query.wildcard.NONE),e.wildcard&t.Query.wildcard.LEADING&&e.term.charAt(0)!=t.Query.wildcard&&(e.term="*"+e.term),e.wildcard&t.Query.wildcard.TRAILING&&e.term.slice(-1)!=t.Query.wildcard&&(e.term=""+e.term+"*"),"presence"in e||(e.presence=t.Query.presence.OPTIONAL),this.clauses.push(e),this},t.Query.prototype.isNegated=function(){for(var e=0;e<this.clauses.length;e++)if(this.clauses[e].presence!=t.Query.presence.PROHIBITED)return!1;return!0},t.Query.prototype.term=function(e,r){if(Array.isArray(e))return e.forEach(function(i){this.term(i,t.utils.clone(r))},this),this;var n=r||{};return n.term=e.toString(),this.clause(n),this},t.QueryParseError=function(e,r,n){this.name="QueryParseError",this.message=e,this.start=r,this.end=n},t.QueryParseError.prototype=new Error,t.QueryLexer=function(e){this.lexemes=[],this.str=e,this.length=e.length,this.pos=0,this.start=0,this.escapeCharPositions=[]},t.QueryLexer.prototype.run=function(){for(var e=t.QueryLexer.lexText;e;)e=e(this)},t.QueryLexer.prototype.sliceString=function(){for(var e=[],r=this.start,n=this.pos,i=0;i<this.escapeCharPositions.length;i++)n=this.escapeCharPositions[i],e.push(this.str.slice(r,n)),r=n+1;return e.push(this.str.slice(r,this.pos)),this.escapeCharPositions.length=0,e.join("")},t.QueryLexer.prototype.emit=function(e){this.lexemes.push({type:e,str:this.sliceString(),start:this.start,end:this.pos}),this.start=this.pos},t.QueryLexer.prototype.escapeCharacter=function(){this.escapeCharPositions.push(this.pos-1),this.pos+=1},t.QueryLexer.prototype.next=function(){if(this.pos>=this.length)return t.QueryLexer.EOS;var e=this.str.charAt(this.pos);return this.pos+=1,e},t.QueryLexer.prototype.width=function(){return this.pos-this.start},t.QueryLexer.prototype.ignore=function(){this.start==this.pos&&(this.pos+=1),this.start=this.pos},t.QueryLexer.prototype.backup=function(){this.pos-=1},t.QueryLexer.prototype.acceptDigitRun=function(){var e,r;do e=this.next(),r=e.charCodeAt(0);while(r>47&&r<58);e!=t.QueryLexer.EOS&&this.backup()},t.QueryLexer.prototype.more=function(){return this.pos<this.length},t.QueryLexer.EOS="EOS",t.QueryLexer.FIELD="FIELD",t.QueryLexer.TERM="TERM",t.QueryLexer.EDIT_DISTANCE="EDIT_DISTANCE",t.QueryLexer.BOOST="BOOST",t.QueryLexer.PRESENCE="PRESENCE",t.QueryLexer.lexField=function(e){return e.backup(),e.emit(t.QueryLexer.FIELD),e.ignore(),t.QueryLexer.lexText},t.QueryLexer.lexTerm=function(e){if(e.width()>1&&(e.backup(),e.emit(t.QueryLexer.TERM)),e.ignore(),e.more())return t.QueryLexer.lexText},t.QueryLexer.lexEditDistance=function(e){return e.ignore(),e.acceptDigitRun(),e.emit(t.QueryLexer.EDIT_DISTANCE),t.QueryLexer.lexText},t.QueryLexer.lexBoost=function(e){return e.ignore(),e.acceptDigitRun(),e.emit(t.QueryLexer.BOOST),t.QueryLexer.lexText},t.QueryLexer.lexEOS=function(e){e.width()>0&&e.emit(t.QueryLexer.TERM)},t.QueryLexer.termSeparator=t.tokenizer.separator,t.QueryLexer.lexText=function(e){for(;;){var r=e.next();if(r==t.QueryLexer.EOS)return t.QueryLexer.lexEOS;if(r.charCodeAt(0)==92){e.escapeCharacter();continue}if(r==":")return t.QueryLexer.lexField;if(r=="~")return e.backup(),e.width()>0&&e.emit(t.QueryLexer.TERM),t.QueryLexer.lexEditDistance;if(r=="^")return e.backup(),e.width()>0&&e.emit(t.QueryLexer.TERM),t.QueryLexer.lexBoost;if(r=="+"&&e.width()===1||r=="-"&&e.width()===1)return e.emit(t.QueryLexer.PRESENCE),t.QueryLexer.lexText;if(r.match(t.QueryLexer.termSeparator))return t.QueryLexer.lexTerm}},t.QueryParser=function(e,r){this.lexer=new t.QueryLexer(e),this.query=r,this.currentClause={},this.lexemeIdx=0},t.QueryParser.prototype.parse=function(){this.lexer.run(),this.lexemes=this.lexer.lexemes;for(var e=t.QueryParser.parseClause;e;)e=e(this);return this.query},t.QueryParser.prototype.peekLexeme=function(){return this.lexemes[this.lexemeIdx]},t.QueryParser.prototype.consumeLexeme=function(){var e=this.peekLexeme();return this.lexemeIdx+=1,e},t.QueryParser.prototype.nextClause=function(){var e=this.currentClause;this.query.clause(e),this.currentClause={}},t.QueryParser.parseClause=function(e){var r=e.peekLexeme();if(r!=null)switch(r.type){case t.QueryLexer.PRESENCE:return t.QueryParser.parsePresence;case t.QueryLexer.FIELD:return t.QueryParser.parseField;case t.QueryLexer.TERM:return t.QueryParser.parseTerm;default:var n="expected either a field or a term, found "+r.type;throw r.str.length>=1&&(n+=" with value '"+r.str+"'"),new t.QueryParseError(n,r.start,r.end)}},t.QueryParser.parsePresence=function(e){var r=e.consumeLexeme();if(r!=null){switch(r.str){case"-":e.currentClause.presence=t.Query.presence.PROHIBITED;break;case"+":e.currentClause.presence=t.Query.presence.REQUIRED;break;default:var n="unrecognised presence operator'"+r.str+"'";throw new t.QueryParseError(n,r.start,r.end)}var i=e.peekLexeme();if(i==null){var n="expecting term or field, found nothing";throw new t.QueryParseError(n,r.start,r.end)}switch(i.type){case t.QueryLexer.FIELD:return t.QueryParser.parseField;case t.QueryLexer.TERM:return t.QueryParser.parseTerm;default:var n="expecting term or field, found '"+i.type+"'";throw new t.QueryParseError(n,i.start,i.end)}}},t.QueryParser.parseField=function(e){var r=e.consumeLexeme();if(r!=null){if(e.query.allFields.indexOf(r.str)==-1){var n=e.query.allFields.map(function(o){return"'"+o+"'"}).join(", "),i="unrecognised field '"+r.str+"', possible fields: "+n;throw new t.QueryParseError(i,r.start,r.end)}e.currentClause.fields=[r.str];var s=e.peekLexeme();if(s==null){var i="expecting term, found nothing";throw new t.QueryParseError(i,r.start,r.end)}switch(s.type){case t.QueryLexer.TERM:return t.QueryParser.parseTerm;default:var i="expecting term, found '"+s.type+"'";throw new t.QueryParseError(i,s.start,s.end)}}},t.QueryParser.parseTerm=function(e){var r=e.consumeLexeme();if(r!=null){e.currentClause.term=r.str.toLowerCase(),r.str.indexOf("*")!=-1&&(e.currentClause.usePipeline=!1);var n=e.peekLexeme();if(n==null){e.nextClause();return}switch(n.type){case t.QueryLexer.TERM:return e.nextClause(),t.QueryParser.parseTerm;case t.QueryLexer.FIELD:return e.nextClause(),t.QueryParser.parseField;case t.QueryLexer.EDIT_DISTANCE:return t.QueryParser.parseEditDistance;case t.QueryLexer.BOOST:return t.QueryParser.parseBoost;case t.QueryLexer.PRESENCE:return e.nextClause(),t.QueryParser.parsePresence;default:var i="Unexpected lexeme type '"+n.type+"'";throw new t.QueryParseError(i,n.start,n.end)}}},t.QueryParser.parseEditDistance=function(e){var r=e.consumeLexeme();if(r!=null){var n=parseInt(r.str,10);if(isNaN(n)){var i="edit distance must be numeric";throw new t.QueryParseError(i,r.start,r.end)}e.currentClause.editDistance=n;var s=e.peekLexeme();if(s==null){e.nextClause();return}switch(s.type){case t.QueryLexer.TERM:return e.nextClause(),t.QueryParser.parseTerm;case t.QueryLexer.FIELD:return e.nextClause(),t.QueryParser.parseField;case t.QueryLexer.EDIT_DISTANCE:return t.QueryParser.parseEditDistance;case t.QueryLexer.BOOST:return t.QueryParser.parseBoost;case t.QueryLexer.PRESENCE:return e.nextClause(),t.QueryParser.parsePresence;default:var i="Unexpected lexeme type '"+s.type+"'";throw new t.QueryParseError(i,s.start,s.end)}}},t.QueryParser.parseBoost=function(e){var r=e.consumeLexeme();if(r!=null){var n=parseInt(r.str,10);if(isNaN(n)){var i="boost must be numeric";throw new t.QueryParseError(i,r.start,r.end)}e.currentClause.boost=n;var s=e.peekLexeme();if(s==null){e.nextClause();return}switch(s.type){case t.QueryLexer.TERM:return e.nextClause(),t.QueryParser.parseTerm;case t.QueryLexer.FIELD:return e.nextClause(),t.QueryParser.parseField;case t.QueryLexer.EDIT_DISTANCE:return t.QueryParser.parseEditDistance;case t.QueryLexer.BOOST:return t.QueryParser.parseBoost;case t.QueryLexer.PRESENCE:return e.nextClause(),t.QueryParser.parsePresence;default:var i="Unexpected lexeme type '"+s.type+"'";throw new t.QueryParseError(i,s.start,s.end)}}},function(e,r){typeof define=="function"&&define.amd?define(r):typeof K=="object"?ee.exports=r():e.lunr=r()}(this,function(){return t})})()});var de=Pe(te());function re(t,e=document){let r=ke(t,e);if(typeof r=="undefined")throw new ReferenceError(`Missing element: expected "${t}" to be present`);return r}function ke(t,e=document){return e.querySelector(t)||void 0}Object.entries||(Object.entries=function(t){let e=[];for(let r of Object.keys(t))e.push([r,t[r]]);return e});Object.values||(Object.values=function(t){let e=[];for(let r of Object.keys(t))e.push(t[r]);return e});typeof Element!="undefined"&&(Element.prototype.scrollTo||(Element.prototype.scrollTo=function(t,e){typeof t=="object"?(this.scrollLeft=t.left,this.scrollTop=t.top):(this.scrollLeft=t,this.scrollTop=e)}),Element.prototype.replaceWith||(Element.prototype.replaceWith=function(...t){let e=this.parentNode;if(e){t.length===0&&e.removeChild(this);for(let r=t.length-1;r>=0;r--){let n=t[r];typeof n=="string"?n=document.createTextNode(n):n.parentNode&&n.parentNode.removeChild(n),r?e.insertBefore(this.previousSibling,n):e.replaceChild(n,this)}}}));function ne(t){let e=new Map;for(let r of t){let[n]=r.location.split("#"),i=e.get(n);typeof i=="undefined"?e.set(n,r):(e.set(r.location,r),r.parent=i)}return e}function W(t,e,r){var s;e=new RegExp(e,"g");let n,i=0;do{n=e.exec(t);let o=(s=n==null?void 0:n.index)!=null?s:t.length;if(i<o&&r(i,o),n){let[a]=n;i=n.index+a.length,a.length===0&&(e.lastIndex=n.index+1)}}while(n)}function ie(t,e){let r=0,n=0,i=0;for(let s=0;i<t.length;i++)t.charAt(i)==="<"&&i>n?e(r,1,n,n=i):t.charAt(i)===">"&&(t.charAt(n+1)==="/"?--s===0&&e(r++,2,n,i+1):t.charAt(i-1)!=="/"&&s++===0&&e(r,0,n,i+1),n=i+1);i>n&&e(r,1,n,i)}function se(t,e,r,n=!1){return q([t],e,r,n).pop()}function q(t,e,r,n=!1){let i=[0];for(let s=1;s<e.length;s++){let o=e[s-1],a=e[s],u=o[o.length-1]>>>2&1023,c=a[0]>>>12;i.push(+(u>c)+i[i.length-1])}return t.map((s,o)=>{let a=0,u=new Map;for(let f of r.sort((g,l)=>g-l)){let g=f&1048575,l=f>>>20;if(i[l]!==o)continue;let m=u.get(l);typeof m=="undefined"&&u.set(l,m=[]),m.push(g)}if(u.size===0)return s;let c=[];for(let[f,g]of u){let l=e[f],m=l[0]>>>12,x=l[l.length-1]>>>12,v=l[l.length-1]>>>2&1023;n&&m>a&&c.push(s.slice(a,m));let d=s.slice(m,x+v);for(let y of g.sort((b,E)=>E-b)){let b=(l[y]>>>12)-m,E=(l[y]>>>2&1023)+b;d=[d.slice(0,b),"<mark>",d.slice(b,E),"</mark>",d.slice(E)].join("")}if(a=x+v,c.push(d)===2)break}return n&&a<s.length&&c.push(s.slice(a)),c.join("")})}function oe(t){let e=[];if(typeof t=="undefined")return e;let r=Array.isArray(t)?t:[t];for(let n=0;n<r.length;n++){let i=lunr.tokenizer.table,s=i.length;ie(r[n],(o,a,u,c)=>{var f;switch(i[f=o+=s]||(i[f]=[]),a){case 0:case 2:i[o].push(u<<12|c-u<<2|a);break;case 1:let g=r[n].slice(u,c);W(g,lunr.tokenizer.separator,(l,m)=>{if(typeof lunr.segmenter!="undefined"){let x=g.slice(l,m);if(/^[MHIK]$/.test(lunr.segmenter.ctype_(x))){let v=lunr.segmenter.segment(x);for(let d=0,y=0;d<v.length;d++)i[o]||(i[o]=[]),i[o].push(u+l+y<<12|v[d].length<<2|a),e.push(new lunr.Token(v[d].toLowerCase(),{position:o<<20|i[o].length-1})),y+=v[d].length;return}}i[o].push(u+l<<12|m-l<<2|a),e.push(new lunr.Token(g.slice(l,m).toLowerCase(),{position:o<<20|i[o].length-1}))})}})}return e}function ae(t,e=r=>r){return t.trim().split(/"([^"]+)"/g).map((r,n)=>n&1?r.replace(/^\b|^(?![^\x00-\x7F]|$)|\s+/g," +"):r).join("").replace(/"|(?:^|\s+)[*+\-:^~]+(?=\s+|$)/g,"").split(/\s+/g).reduce((r,n)=>{let i=e(n);return[...r,...Array.isArray(i)?i:[i]]},[]).map(r=>/([~^]$)/.test(r)?`${r}1`:r).map(r=>/(^[+-]|[~^]\d+$)/.test(r)?r:`${r}*`).join(" ")}function ue(t){return ae(t,e=>{let r=[],n=new lunr.QueryLexer(e);n.run();for(let{type:i,str:s,start:o,end:a}of n.lexemes)switch(i){case"FIELD":["title","text","tags"].includes(s)||(e=[e.slice(0,a)," ",e.slice(a+1)].join(""));break;case"TERM":W(s,lunr.tokenizer.separator,(...u)=>{r.push([e.slice(0,o),s.slice(...u),e.slice(a)].join(""))})}return r})}function ce(t){let e=new lunr.Query(["title","text","tags"]);new lunr.QueryParser(t,e).parse();for(let n of e.clauses)n.usePipeline=!0,n.term.startsWith("*")&&(n.wildcard=lunr.Query.wildcard.LEADING,n.term=n.term.slice(1)),n.term.endsWith("*")&&(n.wildcard=lunr.Query.wildcard.TRAILING,n.term=n.term.slice(0,-1));return e.clauses}function le(t,e){var i;let r=new Set(t),n={};for(let s=0;s<e.length;s++)for(let o of r)e[s].startsWith(o.term)&&(n[o.term]=!0,r.delete(o));for(let s of r)(i=lunr.stopWordFilter)!=null&&i.call(lunr,s.term)&&(n[s.term]=!1);return n}function he(t,e){let r=new Set,n=new Uint16Array(t.length);for(let s=0;s<t.length;s++)for(let o=s+1;o<t.length;o++)t.slice(s,o)in e&&(n[s]=o-s);let i=[0];for(let s=i.length;s>0;){let o=i[--s];for(let u=1;u<n[o];u++)n[o+u]>n[o]-u&&(r.add(t.slice(o,o+u)),i[s++]=o+u);let a=o+n[o];n[a]&&a<t.length-1&&(i[s++]=a),r.add(t.slice(o,a))}return r.has("")?new Set([t]):r}function Oe(t){return e=>r=>{if(typeof r[e]=="undefined")return;let n=[r.location,e].join(":");return t.set(n,lunr.tokenizer.table=[]),r[e]}}function Re(t,e){let[r,n]=[new Set(t),new Set(e)];return[...new Set([...r].filter(i=>!n.has(i)))]}var H=class{constructor({config:e,docs:r,options:n}){let i=Oe(this.table=new Map);this.map=ne(r),this.options=n,this.index=lunr(function(){this.metadataWhitelist=["position"],this.b(0),e.lang.length===1&&e.lang[0]!=="en"?this.use(lunr[e.lang[0]]):e.lang.length>1&&this.use(lunr.multiLanguage(...e.lang)),this.tokenizer=oe,lunr.tokenizer.separator=new RegExp(e.separator),lunr.segmenter="TinySegmenter"in lunr?new lunr.TinySegmenter:void 0;let s=Re(["trimmer","stopWordFilter","stemmer"],e.pipeline);for(let o of e.lang.map(a=>a==="en"?lunr:lunr[a]))for(let a of s)this.pipeline.remove(o[a]),this.searchPipeline.remove(o[a]);this.ref("location"),this.field("title",{boost:1e3,extractor:i("title")}),this.field("text",{boost:1,extractor:i("text")}),this.field("tags",{boost:1e6,extractor:i("tags")});for(let o of r)this.add(o,{boost:o.boost})})}search(e){if(e=e.replace(new RegExp("\\p{sc=Han}+","gu"),s=>[...he(s,this.index.invertedIndex)].join("* ")),e=ue(e),!e)return{items:[]};let r=ce(e).filter(s=>s.presence!==lunr.Query.presence.PROHIBITED),n=this.index.search(e).reduce((s,{ref:o,score:a,matchData:u})=>{let c=this.map.get(o);if(typeof c!="undefined"){c=A({},c),c.tags&&(c.tags=[...c.tags]);let f=le(r,Object.keys(u.metadata));for(let l of this.index.fields){if(typeof c[l]=="undefined")continue;let m=[];for(let d of Object.values(u.metadata))typeof d[l]!="undefined"&&m.push(...d[l].position);if(!m.length)continue;let x=this.table.get([c.location,l].join(":")),v=Array.isArray(c[l])?q:se;c[l]=v(c[l],x,m,l!=="text")}let g=+!c.parent+Object.values(f).filter(l=>l).length/Object.keys(f).length;s.push(G(A({},c),{score:a*(1+Z(g,2)),terms:f}))}return s},[]).sort((s,o)=>o.score-s.score).reduce((s,o)=>{let a=this.map.get(o.location);if(typeof a!="undefined"){let u=a.parent?a.parent.location:a.location;s.set(u,[...s.get(u)||[],o])}return s},new Map);for(let[s,o]of n)if(!o.find(a=>a.location===s)){let a=this.map.get(s);o.push(G(A({},a),{score:0,terms:{}}))}let i;if(this.options.suggest){let s=this.index.query(o=>{for(let a of r)o.term(a.term,{fields:["title"],presence:lunr.Query.presence.REQUIRED,wildcard:lunr.Query.wildcard.TRAILING})});i=s.length?Object.keys(s[0].matchData.metadata):[]}return A({items:[...n.values()]},typeof i!="undefined"&&{suggest:i})}};var fe;function Ie(t){return B(this,null,function*(){let e="../lunr";if(typeof parent!="undefined"&&"IFrameWorker"in parent){let n=re("script[src]"),[i]=n.src.split("/worker");e=e.replace("..",i)}let r=[];for(let n of t.lang){switch(n){case"ja":r.push(`${e}/tinyseg.js`);break;case"hi":case"th":r.push(`${e}/wordcut.js`);break}n!=="en"&&r.push(`${e}/min/lunr.${n}.min.js`)}t.lang.length>1&&r.push(`${e}/min/lunr.multi.min.js`),r.length&&(yield importScripts(`${e}/min/lunr.stemmer.support.min.js`,...r))})}function Fe(t){return B(this,null,function*(){switch(t.type){case 0:return yield Ie(t.data.config),fe=new H(t.data),{type:1};case 2:let e=t.data;try{return{type:3,data:fe.search(e)}}catch(r){return console.warn(`Invalid query: ${e} \u2013 see https://bit.ly/2s3ChXG`),console.warn(r),{type:3,data:{items:[]}}}default:throw new TypeError("Invalid message type")}})}self.lunr=de.default;addEventListener("message",t=>B(void 0,null,function*(){postMessage(yield Fe(t.data))}));})();
-//# sourceMappingURL=search.f886a092.min.js.map
+//# sourceMappingURL=search.b8dbb3d2.min.js.map
 
diff --git a/assets/javascripts/workers/search.f886a092.min.js.map b/assets/javascripts/workers/search.b8dbb3d2.min.js.map
similarity index 79%
rename from assets/javascripts/workers/search.f886a092.min.js.map
rename to assets/javascripts/workers/search.b8dbb3d2.min.js.map
index 9115984a7e..7ef0846c06 100644
--- a/assets/javascripts/workers/search.f886a092.min.js.map
+++ b/assets/javascripts/workers/search.b8dbb3d2.min.js.map
@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["node_modules/lunr/lunr.js", "src/templates/assets/javascripts/integrations/search/worker/main/index.ts", "src/templates/assets/javascripts/browser/element/_/index.ts", "src/templates/assets/javascripts/polyfills/index.ts", "src/templates/assets/javascripts/integrations/search/config/index.ts", "src/templates/assets/javascripts/integrations/search/internal/_/index.ts", "src/templates/assets/javascripts/integrations/search/internal/extract/index.ts", "src/templates/assets/javascripts/integrations/search/internal/highlight/index.ts", "src/templates/assets/javascripts/integrations/search/internal/tokenize/index.ts", "src/templates/assets/javascripts/integrations/search/query/transform/index.ts", "src/templates/assets/javascripts/integrations/search/query/_/index.ts", "src/templates/assets/javascripts/integrations/search/query/segment/index.ts", "src/templates/assets/javascripts/integrations/search/_/index.ts"],
-  "sourcesContent": ["/**\n * lunr - http://lunrjs.com - A bit like Solr, but much smaller and not as bright - 2.3.9\n * Copyright (C) 2020 Oliver Nightingale\n * @license MIT\n */\n\n;(function(){\n\n/**\n * A convenience function for configuring and constructing\n * a new lunr Index.\n *\n * A lunr.Builder instance is created and the pipeline setup\n * with a trimmer, stop word filter and stemmer.\n *\n * This builder object is yielded to the configuration function\n * that is passed as a parameter, allowing the list of fields\n * and other builder parameters to be customised.\n *\n * All documents _must_ be added within the passed config function.\n *\n * @example\n * var idx = lunr(function () {\n *   this.field('title')\n *   this.field('body')\n *   this.ref('id')\n *\n *   documents.forEach(function (doc) {\n *     this.add(doc)\n *   }, this)\n * })\n *\n * @see {@link lunr.Builder}\n * @see {@link lunr.Pipeline}\n * @see {@link lunr.trimmer}\n * @see {@link lunr.stopWordFilter}\n * @see {@link lunr.stemmer}\n * @namespace {function} lunr\n */\nvar lunr = function (config) {\n  var builder = new lunr.Builder\n\n  builder.pipeline.add(\n    lunr.trimmer,\n    lunr.stopWordFilter,\n    lunr.stemmer\n  )\n\n  builder.searchPipeline.add(\n    lunr.stemmer\n  )\n\n  config.call(builder, builder)\n  return builder.build()\n}\n\nlunr.version = \"2.3.9\"\n/*!\n * lunr.utils\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * A namespace containing utils for the rest of the lunr library\n * @namespace lunr.utils\n */\nlunr.utils = {}\n\n/**\n * Print a warning message to the console.\n *\n * @param {String} message The message to be printed.\n * @memberOf lunr.utils\n * @function\n */\nlunr.utils.warn = (function (global) {\n  /* eslint-disable no-console */\n  return function (message) {\n    if (global.console && console.warn) {\n      console.warn(message)\n    }\n  }\n  /* eslint-enable no-console */\n})(this)\n\n/**\n * Convert an object to a string.\n *\n * In the case of `null` and `undefined` the function returns\n * the empty string, in all other cases the result of calling\n * `toString` on the passed object is returned.\n *\n * @param {Any} obj The object to convert to a string.\n * @return {String} string representation of the passed object.\n * @memberOf lunr.utils\n */\nlunr.utils.asString = function (obj) {\n  if (obj === void 0 || obj === null) {\n    return \"\"\n  } else {\n    return obj.toString()\n  }\n}\n\n/**\n * Clones an object.\n *\n * Will create a copy of an existing object such that any mutations\n * on the copy cannot affect the original.\n *\n * Only shallow objects are supported, passing a nested object to this\n * function will cause a TypeError.\n *\n * Objects with primitives, and arrays of primitives are supported.\n *\n * @param {Object} obj The object to clone.\n * @return {Object} a clone of the passed object.\n * @throws {TypeError} when a nested object is passed.\n * @memberOf Utils\n */\nlunr.utils.clone = function (obj) {\n  if (obj === null || obj === undefined) {\n    return obj\n  }\n\n  var clone = Object.create(null),\n      keys = Object.keys(obj)\n\n  for (var i = 0; i < keys.length; i++) {\n    var key = keys[i],\n        val = obj[key]\n\n    if (Array.isArray(val)) {\n      clone[key] = val.slice()\n      continue\n    }\n\n    if (typeof val === 'string' ||\n        typeof val === 'number' ||\n        typeof val === 'boolean') {\n      clone[key] = val\n      continue\n    }\n\n    throw new TypeError(\"clone is not deep and does not support nested objects\")\n  }\n\n  return clone\n}\nlunr.FieldRef = function (docRef, fieldName, stringValue) {\n  this.docRef = docRef\n  this.fieldName = fieldName\n  this._stringValue = stringValue\n}\n\nlunr.FieldRef.joiner = \"/\"\n\nlunr.FieldRef.fromString = function (s) {\n  var n = s.indexOf(lunr.FieldRef.joiner)\n\n  if (n === -1) {\n    throw \"malformed field ref string\"\n  }\n\n  var fieldRef = s.slice(0, n),\n      docRef = s.slice(n + 1)\n\n  return new lunr.FieldRef (docRef, fieldRef, s)\n}\n\nlunr.FieldRef.prototype.toString = function () {\n  if (this._stringValue == undefined) {\n    this._stringValue = this.fieldName + lunr.FieldRef.joiner + this.docRef\n  }\n\n  return this._stringValue\n}\n/*!\n * lunr.Set\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * A lunr set.\n *\n * @constructor\n */\nlunr.Set = function (elements) {\n  this.elements = Object.create(null)\n\n  if (elements) {\n    this.length = elements.length\n\n    for (var i = 0; i < this.length; i++) {\n      this.elements[elements[i]] = true\n    }\n  } else {\n    this.length = 0\n  }\n}\n\n/**\n * A complete set that contains all elements.\n *\n * @static\n * @readonly\n * @type {lunr.Set}\n */\nlunr.Set.complete = {\n  intersect: function (other) {\n    return other\n  },\n\n  union: function () {\n    return this\n  },\n\n  contains: function () {\n    return true\n  }\n}\n\n/**\n * An empty set that contains no elements.\n *\n * @static\n * @readonly\n * @type {lunr.Set}\n */\nlunr.Set.empty = {\n  intersect: function () {\n    return this\n  },\n\n  union: function (other) {\n    return other\n  },\n\n  contains: function () {\n    return false\n  }\n}\n\n/**\n * Returns true if this set contains the specified object.\n *\n * @param {object} object - Object whose presence in this set is to be tested.\n * @returns {boolean} - True if this set contains the specified object.\n */\nlunr.Set.prototype.contains = function (object) {\n  return !!this.elements[object]\n}\n\n/**\n * Returns a new set containing only the elements that are present in both\n * this set and the specified set.\n *\n * @param {lunr.Set} other - set to intersect with this set.\n * @returns {lunr.Set} a new set that is the intersection of this and the specified set.\n */\n\nlunr.Set.prototype.intersect = function (other) {\n  var a, b, elements, intersection = []\n\n  if (other === lunr.Set.complete) {\n    return this\n  }\n\n  if (other === lunr.Set.empty) {\n    return other\n  }\n\n  if (this.length < other.length) {\n    a = this\n    b = other\n  } else {\n    a = other\n    b = this\n  }\n\n  elements = Object.keys(a.elements)\n\n  for (var i = 0; i < elements.length; i++) {\n    var element = elements[i]\n    if (element in b.elements) {\n      intersection.push(element)\n    }\n  }\n\n  return new lunr.Set (intersection)\n}\n\n/**\n * Returns a new set combining the elements of this and the specified set.\n *\n * @param {lunr.Set} other - set to union with this set.\n * @return {lunr.Set} a new set that is the union of this and the specified set.\n */\n\nlunr.Set.prototype.union = function (other) {\n  if (other === lunr.Set.complete) {\n    return lunr.Set.complete\n  }\n\n  if (other === lunr.Set.empty) {\n    return this\n  }\n\n  return new lunr.Set(Object.keys(this.elements).concat(Object.keys(other.elements)))\n}\n/**\n * A function to calculate the inverse document frequency for\n * a posting. This is shared between the builder and the index\n *\n * @private\n * @param {object} posting - The posting for a given term\n * @param {number} documentCount - The total number of documents.\n */\nlunr.idf = function (posting, documentCount) {\n  var documentsWithTerm = 0\n\n  for (var fieldName in posting) {\n    if (fieldName == '_index') continue // Ignore the term index, its not a field\n    documentsWithTerm += Object.keys(posting[fieldName]).length\n  }\n\n  var x = (documentCount - documentsWithTerm + 0.5) / (documentsWithTerm + 0.5)\n\n  return Math.log(1 + Math.abs(x))\n}\n\n/**\n * A token wraps a string representation of a token\n * as it is passed through the text processing pipeline.\n *\n * @constructor\n * @param {string} [str=''] - The string token being wrapped.\n * @param {object} [metadata={}] - Metadata associated with this token.\n */\nlunr.Token = function (str, metadata) {\n  this.str = str || \"\"\n  this.metadata = metadata || {}\n}\n\n/**\n * Returns the token string that is being wrapped by this object.\n *\n * @returns {string}\n */\nlunr.Token.prototype.toString = function () {\n  return this.str\n}\n\n/**\n * A token update function is used when updating or optionally\n * when cloning a token.\n *\n * @callback lunr.Token~updateFunction\n * @param {string} str - The string representation of the token.\n * @param {Object} metadata - All metadata associated with this token.\n */\n\n/**\n * Applies the given function to the wrapped string token.\n *\n * @example\n * token.update(function (str, metadata) {\n *   return str.toUpperCase()\n * })\n *\n * @param {lunr.Token~updateFunction} fn - A function to apply to the token string.\n * @returns {lunr.Token}\n */\nlunr.Token.prototype.update = function (fn) {\n  this.str = fn(this.str, this.metadata)\n  return this\n}\n\n/**\n * Creates a clone of this token. Optionally a function can be\n * applied to the cloned token.\n *\n * @param {lunr.Token~updateFunction} [fn] - An optional function to apply to the cloned token.\n * @returns {lunr.Token}\n */\nlunr.Token.prototype.clone = function (fn) {\n  fn = fn || function (s) { return s }\n  return new lunr.Token (fn(this.str, this.metadata), this.metadata)\n}\n/*!\n * lunr.tokenizer\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * A function for splitting a string into tokens ready to be inserted into\n * the search index. Uses `lunr.tokenizer.separator` to split strings, change\n * the value of this property to change how strings are split into tokens.\n *\n * This tokenizer will convert its parameter to a string by calling `toString` and\n * then will split this string on the character in `lunr.tokenizer.separator`.\n * Arrays will have their elements converted to strings and wrapped in a lunr.Token.\n *\n * Optional metadata can be passed to the tokenizer, this metadata will be cloned and\n * added as metadata to every token that is created from the object to be tokenized.\n *\n * @static\n * @param {?(string|object|object[])} obj - The object to convert into tokens\n * @param {?object} metadata - Optional metadata to associate with every token\n * @returns {lunr.Token[]}\n * @see {@link lunr.Pipeline}\n */\nlunr.tokenizer = function (obj, metadata) {\n  if (obj == null || obj == undefined) {\n    return []\n  }\n\n  if (Array.isArray(obj)) {\n    return obj.map(function (t) {\n      return new lunr.Token(\n        lunr.utils.asString(t).toLowerCase(),\n        lunr.utils.clone(metadata)\n      )\n    })\n  }\n\n  var str = obj.toString().toLowerCase(),\n      len = str.length,\n      tokens = []\n\n  for (var sliceEnd = 0, sliceStart = 0; sliceEnd <= len; sliceEnd++) {\n    var char = str.charAt(sliceEnd),\n        sliceLength = sliceEnd - sliceStart\n\n    if ((char.match(lunr.tokenizer.separator) || sliceEnd == len)) {\n\n      if (sliceLength > 0) {\n        var tokenMetadata = lunr.utils.clone(metadata) || {}\n        tokenMetadata[\"position\"] = [sliceStart, sliceLength]\n        tokenMetadata[\"index\"] = tokens.length\n\n        tokens.push(\n          new lunr.Token (\n            str.slice(sliceStart, sliceEnd),\n            tokenMetadata\n          )\n        )\n      }\n\n      sliceStart = sliceEnd + 1\n    }\n\n  }\n\n  return tokens\n}\n\n/**\n * The separator used to split a string into tokens. Override this property to change the behaviour of\n * `lunr.tokenizer` behaviour when tokenizing strings. By default this splits on whitespace and hyphens.\n *\n * @static\n * @see lunr.tokenizer\n */\nlunr.tokenizer.separator = /[\\s\\-]+/\n/*!\n * lunr.Pipeline\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * lunr.Pipelines maintain an ordered list of functions to be applied to all\n * tokens in documents entering the search index and queries being ran against\n * the index.\n *\n * An instance of lunr.Index created with the lunr shortcut will contain a\n * pipeline with a stop word filter and an English language stemmer. Extra\n * functions can be added before or after either of these functions or these\n * default functions can be removed.\n *\n * When run the pipeline will call each function in turn, passing a token, the\n * index of that token in the original list of all tokens and finally a list of\n * all the original tokens.\n *\n * The output of functions in the pipeline will be passed to the next function\n * in the pipeline. To exclude a token from entering the index the function\n * should return undefined, the rest of the pipeline will not be called with\n * this token.\n *\n * For serialisation of pipelines to work, all functions used in an instance of\n * a pipeline should be registered with lunr.Pipeline. Registered functions can\n * then be loaded. If trying to load a serialised pipeline that uses functions\n * that are not registered an error will be thrown.\n *\n * If not planning on serialising the pipeline then registering pipeline functions\n * is not necessary.\n *\n * @constructor\n */\nlunr.Pipeline = function () {\n  this._stack = []\n}\n\nlunr.Pipeline.registeredFunctions = Object.create(null)\n\n/**\n * A pipeline function maps lunr.Token to lunr.Token. A lunr.Token contains the token\n * string as well as all known metadata. A pipeline function can mutate the token string\n * or mutate (or add) metadata for a given token.\n *\n * A pipeline function can indicate that the passed token should be discarded by returning\n * null, undefined or an empty string. This token will not be passed to any downstream pipeline\n * functions and will not be added to the index.\n *\n * Multiple tokens can be returned by returning an array of tokens. Each token will be passed\n * to any downstream pipeline functions and all will returned tokens will be added to the index.\n *\n * Any number of pipeline functions may be chained together using a lunr.Pipeline.\n *\n * @interface lunr.PipelineFunction\n * @param {lunr.Token} token - A token from the document being processed.\n * @param {number} i - The index of this token in the complete list of tokens for this document/field.\n * @param {lunr.Token[]} tokens - All tokens for this document/field.\n * @returns {(?lunr.Token|lunr.Token[])}\n */\n\n/**\n * Register a function with the pipeline.\n *\n * Functions that are used in the pipeline should be registered if the pipeline\n * needs to be serialised, or a serialised pipeline needs to be loaded.\n *\n * Registering a function does not add it to a pipeline, functions must still be\n * added to instances of the pipeline for them to be used when running a pipeline.\n *\n * @param {lunr.PipelineFunction} fn - The function to check for.\n * @param {String} label - The label to register this function with\n */\nlunr.Pipeline.registerFunction = function (fn, label) {\n  if (label in this.registeredFunctions) {\n    lunr.utils.warn('Overwriting existing registered function: ' + label)\n  }\n\n  fn.label = label\n  lunr.Pipeline.registeredFunctions[fn.label] = fn\n}\n\n/**\n * Warns if the function is not registered as a Pipeline function.\n *\n * @param {lunr.PipelineFunction} fn - The function to check for.\n * @private\n */\nlunr.Pipeline.warnIfFunctionNotRegistered = function (fn) {\n  var isRegistered = fn.label && (fn.label in this.registeredFunctions)\n\n  if (!isRegistered) {\n    lunr.utils.warn('Function is not registered with pipeline. This may cause problems when serialising the index.\\n', fn)\n  }\n}\n\n/**\n * Loads a previously serialised pipeline.\n *\n * All functions to be loaded must already be registered with lunr.Pipeline.\n * If any function from the serialised data has not been registered then an\n * error will be thrown.\n *\n * @param {Object} serialised - The serialised pipeline to load.\n * @returns {lunr.Pipeline}\n */\nlunr.Pipeline.load = function (serialised) {\n  var pipeline = new lunr.Pipeline\n\n  serialised.forEach(function (fnName) {\n    var fn = lunr.Pipeline.registeredFunctions[fnName]\n\n    if (fn) {\n      pipeline.add(fn)\n    } else {\n      throw new Error('Cannot load unregistered function: ' + fnName)\n    }\n  })\n\n  return pipeline\n}\n\n/**\n * Adds new functions to the end of the pipeline.\n *\n * Logs a warning if the function has not been registered.\n *\n * @param {lunr.PipelineFunction[]} functions - Any number of functions to add to the pipeline.\n */\nlunr.Pipeline.prototype.add = function () {\n  var fns = Array.prototype.slice.call(arguments)\n\n  fns.forEach(function (fn) {\n    lunr.Pipeline.warnIfFunctionNotRegistered(fn)\n    this._stack.push(fn)\n  }, this)\n}\n\n/**\n * Adds a single function after a function that already exists in the\n * pipeline.\n *\n * Logs a warning if the function has not been registered.\n *\n * @param {lunr.PipelineFunction} existingFn - A function that already exists in the pipeline.\n * @param {lunr.PipelineFunction} newFn - The new function to add to the pipeline.\n */\nlunr.Pipeline.prototype.after = function (existingFn, newFn) {\n  lunr.Pipeline.warnIfFunctionNotRegistered(newFn)\n\n  var pos = this._stack.indexOf(existingFn)\n  if (pos == -1) {\n    throw new Error('Cannot find existingFn')\n  }\n\n  pos = pos + 1\n  this._stack.splice(pos, 0, newFn)\n}\n\n/**\n * Adds a single function before a function that already exists in the\n * pipeline.\n *\n * Logs a warning if the function has not been registered.\n *\n * @param {lunr.PipelineFunction} existingFn - A function that already exists in the pipeline.\n * @param {lunr.PipelineFunction} newFn - The new function to add to the pipeline.\n */\nlunr.Pipeline.prototype.before = function (existingFn, newFn) {\n  lunr.Pipeline.warnIfFunctionNotRegistered(newFn)\n\n  var pos = this._stack.indexOf(existingFn)\n  if (pos == -1) {\n    throw new Error('Cannot find existingFn')\n  }\n\n  this._stack.splice(pos, 0, newFn)\n}\n\n/**\n * Removes a function from the pipeline.\n *\n * @param {lunr.PipelineFunction} fn The function to remove from the pipeline.\n */\nlunr.Pipeline.prototype.remove = function (fn) {\n  var pos = this._stack.indexOf(fn)\n  if (pos == -1) {\n    return\n  }\n\n  this._stack.splice(pos, 1)\n}\n\n/**\n * Runs the current list of functions that make up the pipeline against the\n * passed tokens.\n *\n * @param {Array} tokens The tokens to run through the pipeline.\n * @returns {Array}\n */\nlunr.Pipeline.prototype.run = function (tokens) {\n  var stackLength = this._stack.length\n\n  for (var i = 0; i < stackLength; i++) {\n    var fn = this._stack[i]\n    var memo = []\n\n    for (var j = 0; j < tokens.length; j++) {\n      var result = fn(tokens[j], j, tokens)\n\n      if (result === null || result === void 0 || result === '') continue\n\n      if (Array.isArray(result)) {\n        for (var k = 0; k < result.length; k++) {\n          memo.push(result[k])\n        }\n      } else {\n        memo.push(result)\n      }\n    }\n\n    tokens = memo\n  }\n\n  return tokens\n}\n\n/**\n * Convenience method for passing a string through a pipeline and getting\n * strings out. This method takes care of wrapping the passed string in a\n * token and mapping the resulting tokens back to strings.\n *\n * @param {string} str - The string to pass through the pipeline.\n * @param {?object} metadata - Optional metadata to associate with the token\n * passed to the pipeline.\n * @returns {string[]}\n */\nlunr.Pipeline.prototype.runString = function (str, metadata) {\n  var token = new lunr.Token (str, metadata)\n\n  return this.run([token]).map(function (t) {\n    return t.toString()\n  })\n}\n\n/**\n * Resets the pipeline by removing any existing processors.\n *\n */\nlunr.Pipeline.prototype.reset = function () {\n  this._stack = []\n}\n\n/**\n * Returns a representation of the pipeline ready for serialisation.\n *\n * Logs a warning if the function has not been registered.\n *\n * @returns {Array}\n */\nlunr.Pipeline.prototype.toJSON = function () {\n  return this._stack.map(function (fn) {\n    lunr.Pipeline.warnIfFunctionNotRegistered(fn)\n\n    return fn.label\n  })\n}\n/*!\n * lunr.Vector\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * A vector is used to construct the vector space of documents and queries. These\n * vectors support operations to determine the similarity between two documents or\n * a document and a query.\n *\n * Normally no parameters are required for initializing a vector, but in the case of\n * loading a previously dumped vector the raw elements can be provided to the constructor.\n *\n * For performance reasons vectors are implemented with a flat array, where an elements\n * index is immediately followed by its value. E.g. [index, value, index, value]. This\n * allows the underlying array to be as sparse as possible and still offer decent\n * performance when being used for vector calculations.\n *\n * @constructor\n * @param {Number[]} [elements] - The flat list of element index and element value pairs.\n */\nlunr.Vector = function (elements) {\n  this._magnitude = 0\n  this.elements = elements || []\n}\n\n\n/**\n * Calculates the position within the vector to insert a given index.\n *\n * This is used internally by insert and upsert. If there are duplicate indexes then\n * the position is returned as if the value for that index were to be updated, but it\n * is the callers responsibility to check whether there is a duplicate at that index\n *\n * @param {Number} insertIdx - The index at which the element should be inserted.\n * @returns {Number}\n */\nlunr.Vector.prototype.positionForIndex = function (index) {\n  // For an empty vector the tuple can be inserted at the beginning\n  if (this.elements.length == 0) {\n    return 0\n  }\n\n  var start = 0,\n      end = this.elements.length / 2,\n      sliceLength = end - start,\n      pivotPoint = Math.floor(sliceLength / 2),\n      pivotIndex = this.elements[pivotPoint * 2]\n\n  while (sliceLength > 1) {\n    if (pivotIndex < index) {\n      start = pivotPoint\n    }\n\n    if (pivotIndex > index) {\n      end = pivotPoint\n    }\n\n    if (pivotIndex == index) {\n      break\n    }\n\n    sliceLength = end - start\n    pivotPoint = start + Math.floor(sliceLength / 2)\n    pivotIndex = this.elements[pivotPoint * 2]\n  }\n\n  if (pivotIndex == index) {\n    return pivotPoint * 2\n  }\n\n  if (pivotIndex > index) {\n    return pivotPoint * 2\n  }\n\n  if (pivotIndex < index) {\n    return (pivotPoint + 1) * 2\n  }\n}\n\n/**\n * Inserts an element at an index within the vector.\n *\n * Does not allow duplicates, will throw an error if there is already an entry\n * for this index.\n *\n * @param {Number} insertIdx - The index at which the element should be inserted.\n * @param {Number} val - The value to be inserted into the vector.\n */\nlunr.Vector.prototype.insert = function (insertIdx, val) {\n  this.upsert(insertIdx, val, function () {\n    throw \"duplicate index\"\n  })\n}\n\n/**\n * Inserts or updates an existing index within the vector.\n *\n * @param {Number} insertIdx - The index at which the element should be inserted.\n * @param {Number} val - The value to be inserted into the vector.\n * @param {function} fn - A function that is called for updates, the existing value and the\n * requested value are passed as arguments\n */\nlunr.Vector.prototype.upsert = function (insertIdx, val, fn) {\n  this._magnitude = 0\n  var position = this.positionForIndex(insertIdx)\n\n  if (this.elements[position] == insertIdx) {\n    this.elements[position + 1] = fn(this.elements[position + 1], val)\n  } else {\n    this.elements.splice(position, 0, insertIdx, val)\n  }\n}\n\n/**\n * Calculates the magnitude of this vector.\n *\n * @returns {Number}\n */\nlunr.Vector.prototype.magnitude = function () {\n  if (this._magnitude) return this._magnitude\n\n  var sumOfSquares = 0,\n      elementsLength = this.elements.length\n\n  for (var i = 1; i < elementsLength; i += 2) {\n    var val = this.elements[i]\n    sumOfSquares += val * val\n  }\n\n  return this._magnitude = Math.sqrt(sumOfSquares)\n}\n\n/**\n * Calculates the dot product of this vector and another vector.\n *\n * @param {lunr.Vector} otherVector - The vector to compute the dot product with.\n * @returns {Number}\n */\nlunr.Vector.prototype.dot = function (otherVector) {\n  var dotProduct = 0,\n      a = this.elements, b = otherVector.elements,\n      aLen = a.length, bLen = b.length,\n      aVal = 0, bVal = 0,\n      i = 0, j = 0\n\n  while (i < aLen && j < bLen) {\n    aVal = a[i], bVal = b[j]\n    if (aVal < bVal) {\n      i += 2\n    } else if (aVal > bVal) {\n      j += 2\n    } else if (aVal == bVal) {\n      dotProduct += a[i + 1] * b[j + 1]\n      i += 2\n      j += 2\n    }\n  }\n\n  return dotProduct\n}\n\n/**\n * Calculates the similarity between this vector and another vector.\n *\n * @param {lunr.Vector} otherVector - The other vector to calculate the\n * similarity with.\n * @returns {Number}\n */\nlunr.Vector.prototype.similarity = function (otherVector) {\n  return this.dot(otherVector) / this.magnitude() || 0\n}\n\n/**\n * Converts the vector to an array of the elements within the vector.\n *\n * @returns {Number[]}\n */\nlunr.Vector.prototype.toArray = function () {\n  var output = new Array (this.elements.length / 2)\n\n  for (var i = 1, j = 0; i < this.elements.length; i += 2, j++) {\n    output[j] = this.elements[i]\n  }\n\n  return output\n}\n\n/**\n * A JSON serializable representation of the vector.\n *\n * @returns {Number[]}\n */\nlunr.Vector.prototype.toJSON = function () {\n  return this.elements\n}\n/* eslint-disable */\n/*!\n * lunr.stemmer\n * Copyright (C) 2020 Oliver Nightingale\n * Includes code from - http://tartarus.org/~martin/PorterStemmer/js.txt\n */\n\n/**\n * lunr.stemmer is an english language stemmer, this is a JavaScript\n * implementation of the PorterStemmer taken from http://tartarus.org/~martin\n *\n * @static\n * @implements {lunr.PipelineFunction}\n * @param {lunr.Token} token - The string to stem\n * @returns {lunr.Token}\n * @see {@link lunr.Pipeline}\n * @function\n */\nlunr.stemmer = (function(){\n  var step2list = {\n      \"ational\" : \"ate\",\n      \"tional\" : \"tion\",\n      \"enci\" : \"ence\",\n      \"anci\" : \"ance\",\n      \"izer\" : \"ize\",\n      \"bli\" : \"ble\",\n      \"alli\" : \"al\",\n      \"entli\" : \"ent\",\n      \"eli\" : \"e\",\n      \"ousli\" : \"ous\",\n      \"ization\" : \"ize\",\n      \"ation\" : \"ate\",\n      \"ator\" : \"ate\",\n      \"alism\" : \"al\",\n      \"iveness\" : \"ive\",\n      \"fulness\" : \"ful\",\n      \"ousness\" : \"ous\",\n      \"aliti\" : \"al\",\n      \"iviti\" : \"ive\",\n      \"biliti\" : \"ble\",\n      \"logi\" : \"log\"\n    },\n\n    step3list = {\n      \"icate\" : \"ic\",\n      \"ative\" : \"\",\n      \"alize\" : \"al\",\n      \"iciti\" : \"ic\",\n      \"ical\" : \"ic\",\n      \"ful\" : \"\",\n      \"ness\" : \"\"\n    },\n\n    c = \"[^aeiou]\",          // consonant\n    v = \"[aeiouy]\",          // vowel\n    C = c + \"[^aeiouy]*\",    // consonant sequence\n    V = v + \"[aeiou]*\",      // vowel sequence\n\n    mgr0 = \"^(\" + C + \")?\" + V + C,               // [C]VC... is m>0\n    meq1 = \"^(\" + C + \")?\" + V + C + \"(\" + V + \")?$\",  // [C]VC[V] is m=1\n    mgr1 = \"^(\" + C + \")?\" + V + C + V + C,       // [C]VCVC... is m>1\n    s_v = \"^(\" + C + \")?\" + v;                   // vowel in stem\n\n  var re_mgr0 = new RegExp(mgr0);\n  var re_mgr1 = new RegExp(mgr1);\n  var re_meq1 = new RegExp(meq1);\n  var re_s_v = new RegExp(s_v);\n\n  var re_1a = /^(.+?)(ss|i)es$/;\n  var re2_1a = /^(.+?)([^s])s$/;\n  var re_1b = /^(.+?)eed$/;\n  var re2_1b = /^(.+?)(ed|ing)$/;\n  var re_1b_2 = /.$/;\n  var re2_1b_2 = /(at|bl|iz)$/;\n  var re3_1b_2 = new RegExp(\"([^aeiouylsz])\\\\1$\");\n  var re4_1b_2 = new RegExp(\"^\" + C + v + \"[^aeiouwxy]$\");\n\n  var re_1c = /^(.+?[^aeiou])y$/;\n  var re_2 = /^(.+?)(ational|tional|enci|anci|izer|bli|alli|entli|eli|ousli|ization|ation|ator|alism|iveness|fulness|ousness|aliti|iviti|biliti|logi)$/;\n\n  var re_3 = /^(.+?)(icate|ative|alize|iciti|ical|ful|ness)$/;\n\n  var re_4 = /^(.+?)(al|ance|ence|er|ic|able|ible|ant|ement|ment|ent|ou|ism|ate|iti|ous|ive|ize)$/;\n  var re2_4 = /^(.+?)(s|t)(ion)$/;\n\n  var re_5 = /^(.+?)e$/;\n  var re_5_1 = /ll$/;\n  var re3_5 = new RegExp(\"^\" + C + v + \"[^aeiouwxy]$\");\n\n  var porterStemmer = function porterStemmer(w) {\n    var stem,\n      suffix,\n      firstch,\n      re,\n      re2,\n      re3,\n      re4;\n\n    if (w.length < 3) { return w; }\n\n    firstch = w.substr(0,1);\n    if (firstch == \"y\") {\n      w = firstch.toUpperCase() + w.substr(1);\n    }\n\n    // Step 1a\n    re = re_1a\n    re2 = re2_1a;\n\n    if (re.test(w)) { w = w.replace(re,\"$1$2\"); }\n    else if (re2.test(w)) { w = w.replace(re2,\"$1$2\"); }\n\n    // Step 1b\n    re = re_1b;\n    re2 = re2_1b;\n    if (re.test(w)) {\n      var fp = re.exec(w);\n      re = re_mgr0;\n      if (re.test(fp[1])) {\n        re = re_1b_2;\n        w = w.replace(re,\"\");\n      }\n    } else if (re2.test(w)) {\n      var fp = re2.exec(w);\n      stem = fp[1];\n      re2 = re_s_v;\n      if (re2.test(stem)) {\n        w = stem;\n        re2 = re2_1b_2;\n        re3 = re3_1b_2;\n        re4 = re4_1b_2;\n        if (re2.test(w)) { w = w + \"e\"; }\n        else if (re3.test(w)) { re = re_1b_2; w = w.replace(re,\"\"); }\n        else if (re4.test(w)) { w = w + \"e\"; }\n      }\n    }\n\n    // Step 1c - replace suffix y or Y by i if preceded by a non-vowel which is not the first letter of the word (so cry -> cri, by -> by, say -> say)\n    re = re_1c;\n    if (re.test(w)) {\n      var fp = re.exec(w);\n      stem = fp[1];\n      w = stem + \"i\";\n    }\n\n    // Step 2\n    re = re_2;\n    if (re.test(w)) {\n      var fp = re.exec(w);\n      stem = fp[1];\n      suffix = fp[2];\n      re = re_mgr0;\n      if (re.test(stem)) {\n        w = stem + step2list[suffix];\n      }\n    }\n\n    // Step 3\n    re = re_3;\n    if (re.test(w)) {\n      var fp = re.exec(w);\n      stem = fp[1];\n      suffix = fp[2];\n      re = re_mgr0;\n      if (re.test(stem)) {\n        w = stem + step3list[suffix];\n      }\n    }\n\n    // Step 4\n    re = re_4;\n    re2 = re2_4;\n    if (re.test(w)) {\n      var fp = re.exec(w);\n      stem = fp[1];\n      re = re_mgr1;\n      if (re.test(stem)) {\n        w = stem;\n      }\n    } else if (re2.test(w)) {\n      var fp = re2.exec(w);\n      stem = fp[1] + fp[2];\n      re2 = re_mgr1;\n      if (re2.test(stem)) {\n        w = stem;\n      }\n    }\n\n    // Step 5\n    re = re_5;\n    if (re.test(w)) {\n      var fp = re.exec(w);\n      stem = fp[1];\n      re = re_mgr1;\n      re2 = re_meq1;\n      re3 = re3_5;\n      if (re.test(stem) || (re2.test(stem) && !(re3.test(stem)))) {\n        w = stem;\n      }\n    }\n\n    re = re_5_1;\n    re2 = re_mgr1;\n    if (re.test(w) && re2.test(w)) {\n      re = re_1b_2;\n      w = w.replace(re,\"\");\n    }\n\n    // and turn initial Y back to y\n\n    if (firstch == \"y\") {\n      w = firstch.toLowerCase() + w.substr(1);\n    }\n\n    return w;\n  };\n\n  return function (token) {\n    return token.update(porterStemmer);\n  }\n})();\n\nlunr.Pipeline.registerFunction(lunr.stemmer, 'stemmer')\n/*!\n * lunr.stopWordFilter\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * lunr.generateStopWordFilter builds a stopWordFilter function from the provided\n * list of stop words.\n *\n * The built in lunr.stopWordFilter is built using this generator and can be used\n * to generate custom stopWordFilters for applications or non English languages.\n *\n * @function\n * @param {Array} token The token to pass through the filter\n * @returns {lunr.PipelineFunction}\n * @see lunr.Pipeline\n * @see lunr.stopWordFilter\n */\nlunr.generateStopWordFilter = function (stopWords) {\n  var words = stopWords.reduce(function (memo, stopWord) {\n    memo[stopWord] = stopWord\n    return memo\n  }, {})\n\n  return function (token) {\n    if (token && words[token.toString()] !== token.toString()) return token\n  }\n}\n\n/**\n * lunr.stopWordFilter is an English language stop word list filter, any words\n * contained in the list will not be passed through the filter.\n *\n * This is intended to be used in the Pipeline. If the token does not pass the\n * filter then undefined will be returned.\n *\n * @function\n * @implements {lunr.PipelineFunction}\n * @params {lunr.Token} token - A token to check for being a stop word.\n * @returns {lunr.Token}\n * @see {@link lunr.Pipeline}\n */\nlunr.stopWordFilter = lunr.generateStopWordFilter([\n  'a',\n  'able',\n  'about',\n  'across',\n  'after',\n  'all',\n  'almost',\n  'also',\n  'am',\n  'among',\n  'an',\n  'and',\n  'any',\n  'are',\n  'as',\n  'at',\n  'be',\n  'because',\n  'been',\n  'but',\n  'by',\n  'can',\n  'cannot',\n  'could',\n  'dear',\n  'did',\n  'do',\n  'does',\n  'either',\n  'else',\n  'ever',\n  'every',\n  'for',\n  'from',\n  'get',\n  'got',\n  'had',\n  'has',\n  'have',\n  'he',\n  'her',\n  'hers',\n  'him',\n  'his',\n  'how',\n  'however',\n  'i',\n  'if',\n  'in',\n  'into',\n  'is',\n  'it',\n  'its',\n  'just',\n  'least',\n  'let',\n  'like',\n  'likely',\n  'may',\n  'me',\n  'might',\n  'most',\n  'must',\n  'my',\n  'neither',\n  'no',\n  'nor',\n  'not',\n  'of',\n  'off',\n  'often',\n  'on',\n  'only',\n  'or',\n  'other',\n  'our',\n  'own',\n  'rather',\n  'said',\n  'say',\n  'says',\n  'she',\n  'should',\n  'since',\n  'so',\n  'some',\n  'than',\n  'that',\n  'the',\n  'their',\n  'them',\n  'then',\n  'there',\n  'these',\n  'they',\n  'this',\n  'tis',\n  'to',\n  'too',\n  'twas',\n  'us',\n  'wants',\n  'was',\n  'we',\n  'were',\n  'what',\n  'when',\n  'where',\n  'which',\n  'while',\n  'who',\n  'whom',\n  'why',\n  'will',\n  'with',\n  'would',\n  'yet',\n  'you',\n  'your'\n])\n\nlunr.Pipeline.registerFunction(lunr.stopWordFilter, 'stopWordFilter')\n/*!\n * lunr.trimmer\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * lunr.trimmer is a pipeline function for trimming non word\n * characters from the beginning and end of tokens before they\n * enter the index.\n *\n * This implementation may not work correctly for non latin\n * characters and should either be removed or adapted for use\n * with languages with non-latin characters.\n *\n * @static\n * @implements {lunr.PipelineFunction}\n * @param {lunr.Token} token The token to pass through the filter\n * @returns {lunr.Token}\n * @see lunr.Pipeline\n */\nlunr.trimmer = function (token) {\n  return token.update(function (s) {\n    return s.replace(/^\\W+/, '').replace(/\\W+$/, '')\n  })\n}\n\nlunr.Pipeline.registerFunction(lunr.trimmer, 'trimmer')\n/*!\n * lunr.TokenSet\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * A token set is used to store the unique list of all tokens\n * within an index. Token sets are also used to represent an\n * incoming query to the index, this query token set and index\n * token set are then intersected to find which tokens to look\n * up in the inverted index.\n *\n * A token set can hold multiple tokens, as in the case of the\n * index token set, or it can hold a single token as in the\n * case of a simple query token set.\n *\n * Additionally token sets are used to perform wildcard matching.\n * Leading, contained and trailing wildcards are supported, and\n * from this edit distance matching can also be provided.\n *\n * Token sets are implemented as a minimal finite state automata,\n * where both common prefixes and suffixes are shared between tokens.\n * This helps to reduce the space used for storing the token set.\n *\n * @constructor\n */\nlunr.TokenSet = function () {\n  this.final = false\n  this.edges = {}\n  this.id = lunr.TokenSet._nextId\n  lunr.TokenSet._nextId += 1\n}\n\n/**\n * Keeps track of the next, auto increment, identifier to assign\n * to a new tokenSet.\n *\n * TokenSets require a unique identifier to be correctly minimised.\n *\n * @private\n */\nlunr.TokenSet._nextId = 1\n\n/**\n * Creates a TokenSet instance from the given sorted array of words.\n *\n * @param {String[]} arr - A sorted array of strings to create the set from.\n * @returns {lunr.TokenSet}\n * @throws Will throw an error if the input array is not sorted.\n */\nlunr.TokenSet.fromArray = function (arr) {\n  var builder = new lunr.TokenSet.Builder\n\n  for (var i = 0, len = arr.length; i < len; i++) {\n    builder.insert(arr[i])\n  }\n\n  builder.finish()\n  return builder.root\n}\n\n/**\n * Creates a token set from a query clause.\n *\n * @private\n * @param {Object} clause - A single clause from lunr.Query.\n * @param {string} clause.term - The query clause term.\n * @param {number} [clause.editDistance] - The optional edit distance for the term.\n * @returns {lunr.TokenSet}\n */\nlunr.TokenSet.fromClause = function (clause) {\n  if ('editDistance' in clause) {\n    return lunr.TokenSet.fromFuzzyString(clause.term, clause.editDistance)\n  } else {\n    return lunr.TokenSet.fromString(clause.term)\n  }\n}\n\n/**\n * Creates a token set representing a single string with a specified\n * edit distance.\n *\n * Insertions, deletions, substitutions and transpositions are each\n * treated as an edit distance of 1.\n *\n * Increasing the allowed edit distance will have a dramatic impact\n * on the performance of both creating and intersecting these TokenSets.\n * It is advised to keep the edit distance less than 3.\n *\n * @param {string} str - The string to create the token set from.\n * @param {number} editDistance - The allowed edit distance to match.\n * @returns {lunr.Vector}\n */\nlunr.TokenSet.fromFuzzyString = function (str, editDistance) {\n  var root = new lunr.TokenSet\n\n  var stack = [{\n    node: root,\n    editsRemaining: editDistance,\n    str: str\n  }]\n\n  while (stack.length) {\n    var frame = stack.pop()\n\n    // no edit\n    if (frame.str.length > 0) {\n      var char = frame.str.charAt(0),\n          noEditNode\n\n      if (char in frame.node.edges) {\n        noEditNode = frame.node.edges[char]\n      } else {\n        noEditNode = new lunr.TokenSet\n        frame.node.edges[char] = noEditNode\n      }\n\n      if (frame.str.length == 1) {\n        noEditNode.final = true\n      }\n\n      stack.push({\n        node: noEditNode,\n        editsRemaining: frame.editsRemaining,\n        str: frame.str.slice(1)\n      })\n    }\n\n    if (frame.editsRemaining == 0) {\n      continue\n    }\n\n    // insertion\n    if (\"*\" in frame.node.edges) {\n      var insertionNode = frame.node.edges[\"*\"]\n    } else {\n      var insertionNode = new lunr.TokenSet\n      frame.node.edges[\"*\"] = insertionNode\n    }\n\n    if (frame.str.length == 0) {\n      insertionNode.final = true\n    }\n\n    stack.push({\n      node: insertionNode,\n      editsRemaining: frame.editsRemaining - 1,\n      str: frame.str\n    })\n\n    // deletion\n    // can only do a deletion if we have enough edits remaining\n    // and if there are characters left to delete in the string\n    if (frame.str.length > 1) {\n      stack.push({\n        node: frame.node,\n        editsRemaining: frame.editsRemaining - 1,\n        str: frame.str.slice(1)\n      })\n    }\n\n    // deletion\n    // just removing the last character from the str\n    if (frame.str.length == 1) {\n      frame.node.final = true\n    }\n\n    // substitution\n    // can only do a substitution if we have enough edits remaining\n    // and if there are characters left to substitute\n    if (frame.str.length >= 1) {\n      if (\"*\" in frame.node.edges) {\n        var substitutionNode = frame.node.edges[\"*\"]\n      } else {\n        var substitutionNode = new lunr.TokenSet\n        frame.node.edges[\"*\"] = substitutionNode\n      }\n\n      if (frame.str.length == 1) {\n        substitutionNode.final = true\n      }\n\n      stack.push({\n        node: substitutionNode,\n        editsRemaining: frame.editsRemaining - 1,\n        str: frame.str.slice(1)\n      })\n    }\n\n    // transposition\n    // can only do a transposition if there are edits remaining\n    // and there are enough characters to transpose\n    if (frame.str.length > 1) {\n      var charA = frame.str.charAt(0),\n          charB = frame.str.charAt(1),\n          transposeNode\n\n      if (charB in frame.node.edges) {\n        transposeNode = frame.node.edges[charB]\n      } else {\n        transposeNode = new lunr.TokenSet\n        frame.node.edges[charB] = transposeNode\n      }\n\n      if (frame.str.length == 1) {\n        transposeNode.final = true\n      }\n\n      stack.push({\n        node: transposeNode,\n        editsRemaining: frame.editsRemaining - 1,\n        str: charA + frame.str.slice(2)\n      })\n    }\n  }\n\n  return root\n}\n\n/**\n * Creates a TokenSet from a string.\n *\n * The string may contain one or more wildcard characters (*)\n * that will allow wildcard matching when intersecting with\n * another TokenSet.\n *\n * @param {string} str - The string to create a TokenSet from.\n * @returns {lunr.TokenSet}\n */\nlunr.TokenSet.fromString = function (str) {\n  var node = new lunr.TokenSet,\n      root = node\n\n  /*\n   * Iterates through all characters within the passed string\n   * appending a node for each character.\n   *\n   * When a wildcard character is found then a self\n   * referencing edge is introduced to continually match\n   * any number of any characters.\n   */\n  for (var i = 0, len = str.length; i < len; i++) {\n    var char = str[i],\n        final = (i == len - 1)\n\n    if (char == \"*\") {\n      node.edges[char] = node\n      node.final = final\n\n    } else {\n      var next = new lunr.TokenSet\n      next.final = final\n\n      node.edges[char] = next\n      node = next\n    }\n  }\n\n  return root\n}\n\n/**\n * Converts this TokenSet into an array of strings\n * contained within the TokenSet.\n *\n * This is not intended to be used on a TokenSet that\n * contains wildcards, in these cases the results are\n * undefined and are likely to cause an infinite loop.\n *\n * @returns {string[]}\n */\nlunr.TokenSet.prototype.toArray = function () {\n  var words = []\n\n  var stack = [{\n    prefix: \"\",\n    node: this\n  }]\n\n  while (stack.length) {\n    var frame = stack.pop(),\n        edges = Object.keys(frame.node.edges),\n        len = edges.length\n\n    if (frame.node.final) {\n      /* In Safari, at this point the prefix is sometimes corrupted, see:\n       * https://github.com/olivernn/lunr.js/issues/279 Calling any\n       * String.prototype method forces Safari to \"cast\" this string to what\n       * it's supposed to be, fixing the bug. */\n      frame.prefix.charAt(0)\n      words.push(frame.prefix)\n    }\n\n    for (var i = 0; i < len; i++) {\n      var edge = edges[i]\n\n      stack.push({\n        prefix: frame.prefix.concat(edge),\n        node: frame.node.edges[edge]\n      })\n    }\n  }\n\n  return words\n}\n\n/**\n * Generates a string representation of a TokenSet.\n *\n * This is intended to allow TokenSets to be used as keys\n * in objects, largely to aid the construction and minimisation\n * of a TokenSet. As such it is not designed to be a human\n * friendly representation of the TokenSet.\n *\n * @returns {string}\n */\nlunr.TokenSet.prototype.toString = function () {\n  // NOTE: Using Object.keys here as this.edges is very likely\n  // to enter 'hash-mode' with many keys being added\n  //\n  // avoiding a for-in loop here as it leads to the function\n  // being de-optimised (at least in V8). From some simple\n  // benchmarks the performance is comparable, but allowing\n  // V8 to optimize may mean easy performance wins in the future.\n\n  if (this._str) {\n    return this._str\n  }\n\n  var str = this.final ? '1' : '0',\n      labels = Object.keys(this.edges).sort(),\n      len = labels.length\n\n  for (var i = 0; i < len; i++) {\n    var label = labels[i],\n        node = this.edges[label]\n\n    str = str + label + node.id\n  }\n\n  return str\n}\n\n/**\n * Returns a new TokenSet that is the intersection of\n * this TokenSet and the passed TokenSet.\n *\n * This intersection will take into account any wildcards\n * contained within the TokenSet.\n *\n * @param {lunr.TokenSet} b - An other TokenSet to intersect with.\n * @returns {lunr.TokenSet}\n */\nlunr.TokenSet.prototype.intersect = function (b) {\n  var output = new lunr.TokenSet,\n      frame = undefined\n\n  var stack = [{\n    qNode: b,\n    output: output,\n    node: this\n  }]\n\n  while (stack.length) {\n    frame = stack.pop()\n\n    // NOTE: As with the #toString method, we are using\n    // Object.keys and a for loop instead of a for-in loop\n    // as both of these objects enter 'hash' mode, causing\n    // the function to be de-optimised in V8\n    var qEdges = Object.keys(frame.qNode.edges),\n        qLen = qEdges.length,\n        nEdges = Object.keys(frame.node.edges),\n        nLen = nEdges.length\n\n    for (var q = 0; q < qLen; q++) {\n      var qEdge = qEdges[q]\n\n      for (var n = 0; n < nLen; n++) {\n        var nEdge = nEdges[n]\n\n        if (nEdge == qEdge || qEdge == '*') {\n          var node = frame.node.edges[nEdge],\n              qNode = frame.qNode.edges[qEdge],\n              final = node.final && qNode.final,\n              next = undefined\n\n          if (nEdge in frame.output.edges) {\n            // an edge already exists for this character\n            // no need to create a new node, just set the finality\n            // bit unless this node is already final\n            next = frame.output.edges[nEdge]\n            next.final = next.final || final\n\n          } else {\n            // no edge exists yet, must create one\n            // set the finality bit and insert it\n            // into the output\n            next = new lunr.TokenSet\n            next.final = final\n            frame.output.edges[nEdge] = next\n          }\n\n          stack.push({\n            qNode: qNode,\n            output: next,\n            node: node\n          })\n        }\n      }\n    }\n  }\n\n  return output\n}\nlunr.TokenSet.Builder = function () {\n  this.previousWord = \"\"\n  this.root = new lunr.TokenSet\n  this.uncheckedNodes = []\n  this.minimizedNodes = {}\n}\n\nlunr.TokenSet.Builder.prototype.insert = function (word) {\n  var node,\n      commonPrefix = 0\n\n  if (word < this.previousWord) {\n    throw new Error (\"Out of order word insertion\")\n  }\n\n  for (var i = 0; i < word.length && i < this.previousWord.length; i++) {\n    if (word[i] != this.previousWord[i]) break\n    commonPrefix++\n  }\n\n  this.minimize(commonPrefix)\n\n  if (this.uncheckedNodes.length == 0) {\n    node = this.root\n  } else {\n    node = this.uncheckedNodes[this.uncheckedNodes.length - 1].child\n  }\n\n  for (var i = commonPrefix; i < word.length; i++) {\n    var nextNode = new lunr.TokenSet,\n        char = word[i]\n\n    node.edges[char] = nextNode\n\n    this.uncheckedNodes.push({\n      parent: node,\n      char: char,\n      child: nextNode\n    })\n\n    node = nextNode\n  }\n\n  node.final = true\n  this.previousWord = word\n}\n\nlunr.TokenSet.Builder.prototype.finish = function () {\n  this.minimize(0)\n}\n\nlunr.TokenSet.Builder.prototype.minimize = function (downTo) {\n  for (var i = this.uncheckedNodes.length - 1; i >= downTo; i--) {\n    var node = this.uncheckedNodes[i],\n        childKey = node.child.toString()\n\n    if (childKey in this.minimizedNodes) {\n      node.parent.edges[node.char] = this.minimizedNodes[childKey]\n    } else {\n      // Cache the key for this node since\n      // we know it can't change anymore\n      node.child._str = childKey\n\n      this.minimizedNodes[childKey] = node.child\n    }\n\n    this.uncheckedNodes.pop()\n  }\n}\n/*!\n * lunr.Index\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * An index contains the built index of all documents and provides a query interface\n * to the index.\n *\n * Usually instances of lunr.Index will not be created using this constructor, instead\n * lunr.Builder should be used to construct new indexes, or lunr.Index.load should be\n * used to load previously built and serialized indexes.\n *\n * @constructor\n * @param {Object} attrs - The attributes of the built search index.\n * @param {Object} attrs.invertedIndex - An index of term/field to document reference.\n * @param {Object<string, lunr.Vector>} attrs.fieldVectors - Field vectors\n * @param {lunr.TokenSet} attrs.tokenSet - An set of all corpus tokens.\n * @param {string[]} attrs.fields - The names of indexed document fields.\n * @param {lunr.Pipeline} attrs.pipeline - The pipeline to use for search terms.\n */\nlunr.Index = function (attrs) {\n  this.invertedIndex = attrs.invertedIndex\n  this.fieldVectors = attrs.fieldVectors\n  this.tokenSet = attrs.tokenSet\n  this.fields = attrs.fields\n  this.pipeline = attrs.pipeline\n}\n\n/**\n * A result contains details of a document matching a search query.\n * @typedef {Object} lunr.Index~Result\n * @property {string} ref - The reference of the document this result represents.\n * @property {number} score - A number between 0 and 1 representing how similar this document is to the query.\n * @property {lunr.MatchData} matchData - Contains metadata about this match including which term(s) caused the match.\n */\n\n/**\n * Although lunr provides the ability to create queries using lunr.Query, it also provides a simple\n * query language which itself is parsed into an instance of lunr.Query.\n *\n * For programmatically building queries it is advised to directly use lunr.Query, the query language\n * is best used for human entered text rather than program generated text.\n *\n * At its simplest queries can just be a single term, e.g. `hello`, multiple terms are also supported\n * and will be combined with OR, e.g `hello world` will match documents that contain either 'hello'\n * or 'world', though those that contain both will rank higher in the results.\n *\n * Wildcards can be included in terms to match one or more unspecified characters, these wildcards can\n * be inserted anywhere within the term, and more than one wildcard can exist in a single term. Adding\n * wildcards will increase the number of documents that will be found but can also have a negative\n * impact on query performance, especially with wildcards at the beginning of a term.\n *\n * Terms can be restricted to specific fields, e.g. `title:hello`, only documents with the term\n * hello in the title field will match this query. Using a field not present in the index will lead\n * to an error being thrown.\n *\n * Modifiers can also be added to terms, lunr supports edit distance and boost modifiers on terms. A term\n * boost will make documents matching that term score higher, e.g. `foo^5`. Edit distance is also supported\n * to provide fuzzy matching, e.g. 'hello~2' will match documents with hello with an edit distance of 2.\n * Avoid large values for edit distance to improve query performance.\n *\n * Each term also supports a presence modifier. By default a term's presence in document is optional, however\n * this can be changed to either required or prohibited. For a term's presence to be required in a document the\n * term should be prefixed with a '+', e.g. `+foo bar` is a search for documents that must contain 'foo' and\n * optionally contain 'bar'. Conversely a leading '-' sets the terms presence to prohibited, i.e. it must not\n * appear in a document, e.g. `-foo bar` is a search for documents that do not contain 'foo' but may contain 'bar'.\n *\n * To escape special characters the backslash character '\\' can be used, this allows searches to include\n * characters that would normally be considered modifiers, e.g. `foo\\~2` will search for a term \"foo~2\" instead\n * of attempting to apply a boost of 2 to the search term \"foo\".\n *\n * @typedef {string} lunr.Index~QueryString\n * @example <caption>Simple single term query</caption>\n * hello\n * @example <caption>Multiple term query</caption>\n * hello world\n * @example <caption>term scoped to a field</caption>\n * title:hello\n * @example <caption>term with a boost of 10</caption>\n * hello^10\n * @example <caption>term with an edit distance of 2</caption>\n * hello~2\n * @example <caption>terms with presence modifiers</caption>\n * -foo +bar baz\n */\n\n/**\n * Performs a search against the index using lunr query syntax.\n *\n * Results will be returned sorted by their score, the most relevant results\n * will be returned first.  For details on how the score is calculated, please see\n * the {@link https://lunrjs.com/guides/searching.html#scoring|guide}.\n *\n * For more programmatic querying use lunr.Index#query.\n *\n * @param {lunr.Index~QueryString} queryString - A string containing a lunr query.\n * @throws {lunr.QueryParseError} If the passed query string cannot be parsed.\n * @returns {lunr.Index~Result[]}\n */\nlunr.Index.prototype.search = function (queryString) {\n  return this.query(function (query) {\n    var parser = new lunr.QueryParser(queryString, query)\n    parser.parse()\n  })\n}\n\n/**\n * A query builder callback provides a query object to be used to express\n * the query to perform on the index.\n *\n * @callback lunr.Index~queryBuilder\n * @param {lunr.Query} query - The query object to build up.\n * @this lunr.Query\n */\n\n/**\n * Performs a query against the index using the yielded lunr.Query object.\n *\n * If performing programmatic queries against the index, this method is preferred\n * over lunr.Index#search so as to avoid the additional query parsing overhead.\n *\n * A query object is yielded to the supplied function which should be used to\n * express the query to be run against the index.\n *\n * Note that although this function takes a callback parameter it is _not_ an\n * asynchronous operation, the callback is just yielded a query object to be\n * customized.\n *\n * @param {lunr.Index~queryBuilder} fn - A function that is used to build the query.\n * @returns {lunr.Index~Result[]}\n */\nlunr.Index.prototype.query = function (fn) {\n  // for each query clause\n  // * process terms\n  // * expand terms from token set\n  // * find matching documents and metadata\n  // * get document vectors\n  // * score documents\n\n  var query = new lunr.Query(this.fields),\n      matchingFields = Object.create(null),\n      queryVectors = Object.create(null),\n      termFieldCache = Object.create(null),\n      requiredMatches = Object.create(null),\n      prohibitedMatches = Object.create(null)\n\n  /*\n   * To support field level boosts a query vector is created per\n   * field. An empty vector is eagerly created to support negated\n   * queries.\n   */\n  for (var i = 0; i < this.fields.length; i++) {\n    queryVectors[this.fields[i]] = new lunr.Vector\n  }\n\n  fn.call(query, query)\n\n  for (var i = 0; i < query.clauses.length; i++) {\n    /*\n     * Unless the pipeline has been disabled for this term, which is\n     * the case for terms with wildcards, we need to pass the clause\n     * term through the search pipeline. A pipeline returns an array\n     * of processed terms. Pipeline functions may expand the passed\n     * term, which means we may end up performing multiple index lookups\n     * for a single query term.\n     */\n    var clause = query.clauses[i],\n        terms = null,\n        clauseMatches = lunr.Set.empty\n\n    if (clause.usePipeline) {\n      terms = this.pipeline.runString(clause.term, {\n        fields: clause.fields\n      })\n    } else {\n      terms = [clause.term]\n    }\n\n    for (var m = 0; m < terms.length; m++) {\n      var term = terms[m]\n\n      /*\n       * Each term returned from the pipeline needs to use the same query\n       * clause object, e.g. the same boost and or edit distance. The\n       * simplest way to do this is to re-use the clause object but mutate\n       * its term property.\n       */\n      clause.term = term\n\n      /*\n       * From the term in the clause we create a token set which will then\n       * be used to intersect the indexes token set to get a list of terms\n       * to lookup in the inverted index\n       */\n      var termTokenSet = lunr.TokenSet.fromClause(clause),\n          expandedTerms = this.tokenSet.intersect(termTokenSet).toArray()\n\n      /*\n       * If a term marked as required does not exist in the tokenSet it is\n       * impossible for the search to return any matches. We set all the field\n       * scoped required matches set to empty and stop examining any further\n       * clauses.\n       */\n      if (expandedTerms.length === 0 && clause.presence === lunr.Query.presence.REQUIRED) {\n        for (var k = 0; k < clause.fields.length; k++) {\n          var field = clause.fields[k]\n          requiredMatches[field] = lunr.Set.empty\n        }\n\n        break\n      }\n\n      for (var j = 0; j < expandedTerms.length; j++) {\n        /*\n         * For each term get the posting and termIndex, this is required for\n         * building the query vector.\n         */\n        var expandedTerm = expandedTerms[j],\n            posting = this.invertedIndex[expandedTerm],\n            termIndex = posting._index\n\n        for (var k = 0; k < clause.fields.length; k++) {\n          /*\n           * For each field that this query term is scoped by (by default\n           * all fields are in scope) we need to get all the document refs\n           * that have this term in that field.\n           *\n           * The posting is the entry in the invertedIndex for the matching\n           * term from above.\n           */\n          var field = clause.fields[k],\n              fieldPosting = posting[field],\n              matchingDocumentRefs = Object.keys(fieldPosting),\n              termField = expandedTerm + \"/\" + field,\n              matchingDocumentsSet = new lunr.Set(matchingDocumentRefs)\n\n          /*\n           * if the presence of this term is required ensure that the matching\n           * documents are added to the set of required matches for this clause.\n           *\n           */\n          if (clause.presence == lunr.Query.presence.REQUIRED) {\n            clauseMatches = clauseMatches.union(matchingDocumentsSet)\n\n            if (requiredMatches[field] === undefined) {\n              requiredMatches[field] = lunr.Set.complete\n            }\n          }\n\n          /*\n           * if the presence of this term is prohibited ensure that the matching\n           * documents are added to the set of prohibited matches for this field,\n           * creating that set if it does not yet exist.\n           */\n          if (clause.presence == lunr.Query.presence.PROHIBITED) {\n            if (prohibitedMatches[field] === undefined) {\n              prohibitedMatches[field] = lunr.Set.empty\n            }\n\n            prohibitedMatches[field] = prohibitedMatches[field].union(matchingDocumentsSet)\n\n            /*\n             * Prohibited matches should not be part of the query vector used for\n             * similarity scoring and no metadata should be extracted so we continue\n             * to the next field\n             */\n            continue\n          }\n\n          /*\n           * The query field vector is populated using the termIndex found for\n           * the term and a unit value with the appropriate boost applied.\n           * Using upsert because there could already be an entry in the vector\n           * for the term we are working with. In that case we just add the scores\n           * together.\n           */\n          queryVectors[field].upsert(termIndex, clause.boost, function (a, b) { return a + b })\n\n          /**\n           * If we've already seen this term, field combo then we've already collected\n           * the matching documents and metadata, no need to go through all that again\n           */\n          if (termFieldCache[termField]) {\n            continue\n          }\n\n          for (var l = 0; l < matchingDocumentRefs.length; l++) {\n            /*\n             * All metadata for this term/field/document triple\n             * are then extracted and collected into an instance\n             * of lunr.MatchData ready to be returned in the query\n             * results\n             */\n            var matchingDocumentRef = matchingDocumentRefs[l],\n                matchingFieldRef = new lunr.FieldRef (matchingDocumentRef, field),\n                metadata = fieldPosting[matchingDocumentRef],\n                fieldMatch\n\n            if ((fieldMatch = matchingFields[matchingFieldRef]) === undefined) {\n              matchingFields[matchingFieldRef] = new lunr.MatchData (expandedTerm, field, metadata)\n            } else {\n              fieldMatch.add(expandedTerm, field, metadata)\n            }\n\n          }\n\n          termFieldCache[termField] = true\n        }\n      }\n    }\n\n    /**\n     * If the presence was required we need to update the requiredMatches field sets.\n     * We do this after all fields for the term have collected their matches because\n     * the clause terms presence is required in _any_ of the fields not _all_ of the\n     * fields.\n     */\n    if (clause.presence === lunr.Query.presence.REQUIRED) {\n      for (var k = 0; k < clause.fields.length; k++) {\n        var field = clause.fields[k]\n        requiredMatches[field] = requiredMatches[field].intersect(clauseMatches)\n      }\n    }\n  }\n\n  /**\n   * Need to combine the field scoped required and prohibited\n   * matching documents into a global set of required and prohibited\n   * matches\n   */\n  var allRequiredMatches = lunr.Set.complete,\n      allProhibitedMatches = lunr.Set.empty\n\n  for (var i = 0; i < this.fields.length; i++) {\n    var field = this.fields[i]\n\n    if (requiredMatches[field]) {\n      allRequiredMatches = allRequiredMatches.intersect(requiredMatches[field])\n    }\n\n    if (prohibitedMatches[field]) {\n      allProhibitedMatches = allProhibitedMatches.union(prohibitedMatches[field])\n    }\n  }\n\n  var matchingFieldRefs = Object.keys(matchingFields),\n      results = [],\n      matches = Object.create(null)\n\n  /*\n   * If the query is negated (contains only prohibited terms)\n   * we need to get _all_ fieldRefs currently existing in the\n   * index. This is only done when we know that the query is\n   * entirely prohibited terms to avoid any cost of getting all\n   * fieldRefs unnecessarily.\n   *\n   * Additionally, blank MatchData must be created to correctly\n   * populate the results.\n   */\n  if (query.isNegated()) {\n    matchingFieldRefs = Object.keys(this.fieldVectors)\n\n    for (var i = 0; i < matchingFieldRefs.length; i++) {\n      var matchingFieldRef = matchingFieldRefs[i]\n      var fieldRef = lunr.FieldRef.fromString(matchingFieldRef)\n      matchingFields[matchingFieldRef] = new lunr.MatchData\n    }\n  }\n\n  for (var i = 0; i < matchingFieldRefs.length; i++) {\n    /*\n     * Currently we have document fields that match the query, but we\n     * need to return documents. The matchData and scores are combined\n     * from multiple fields belonging to the same document.\n     *\n     * Scores are calculated by field, using the query vectors created\n     * above, and combined into a final document score using addition.\n     */\n    var fieldRef = lunr.FieldRef.fromString(matchingFieldRefs[i]),\n        docRef = fieldRef.docRef\n\n    if (!allRequiredMatches.contains(docRef)) {\n      continue\n    }\n\n    if (allProhibitedMatches.contains(docRef)) {\n      continue\n    }\n\n    var fieldVector = this.fieldVectors[fieldRef],\n        score = queryVectors[fieldRef.fieldName].similarity(fieldVector),\n        docMatch\n\n    if ((docMatch = matches[docRef]) !== undefined) {\n      docMatch.score += score\n      docMatch.matchData.combine(matchingFields[fieldRef])\n    } else {\n      var match = {\n        ref: docRef,\n        score: score,\n        matchData: matchingFields[fieldRef]\n      }\n      matches[docRef] = match\n      results.push(match)\n    }\n  }\n\n  /*\n   * Sort the results objects by score, highest first.\n   */\n  return results.sort(function (a, b) {\n    return b.score - a.score\n  })\n}\n\n/**\n * Prepares the index for JSON serialization.\n *\n * The schema for this JSON blob will be described in a\n * separate JSON schema file.\n *\n * @returns {Object}\n */\nlunr.Index.prototype.toJSON = function () {\n  var invertedIndex = Object.keys(this.invertedIndex)\n    .sort()\n    .map(function (term) {\n      return [term, this.invertedIndex[term]]\n    }, this)\n\n  var fieldVectors = Object.keys(this.fieldVectors)\n    .map(function (ref) {\n      return [ref, this.fieldVectors[ref].toJSON()]\n    }, this)\n\n  return {\n    version: lunr.version,\n    fields: this.fields,\n    fieldVectors: fieldVectors,\n    invertedIndex: invertedIndex,\n    pipeline: this.pipeline.toJSON()\n  }\n}\n\n/**\n * Loads a previously serialized lunr.Index\n *\n * @param {Object} serializedIndex - A previously serialized lunr.Index\n * @returns {lunr.Index}\n */\nlunr.Index.load = function (serializedIndex) {\n  var attrs = {},\n      fieldVectors = {},\n      serializedVectors = serializedIndex.fieldVectors,\n      invertedIndex = Object.create(null),\n      serializedInvertedIndex = serializedIndex.invertedIndex,\n      tokenSetBuilder = new lunr.TokenSet.Builder,\n      pipeline = lunr.Pipeline.load(serializedIndex.pipeline)\n\n  if (serializedIndex.version != lunr.version) {\n    lunr.utils.warn(\"Version mismatch when loading serialised index. Current version of lunr '\" + lunr.version + \"' does not match serialized index '\" + serializedIndex.version + \"'\")\n  }\n\n  for (var i = 0; i < serializedVectors.length; i++) {\n    var tuple = serializedVectors[i],\n        ref = tuple[0],\n        elements = tuple[1]\n\n    fieldVectors[ref] = new lunr.Vector(elements)\n  }\n\n  for (var i = 0; i < serializedInvertedIndex.length; i++) {\n    var tuple = serializedInvertedIndex[i],\n        term = tuple[0],\n        posting = tuple[1]\n\n    tokenSetBuilder.insert(term)\n    invertedIndex[term] = posting\n  }\n\n  tokenSetBuilder.finish()\n\n  attrs.fields = serializedIndex.fields\n\n  attrs.fieldVectors = fieldVectors\n  attrs.invertedIndex = invertedIndex\n  attrs.tokenSet = tokenSetBuilder.root\n  attrs.pipeline = pipeline\n\n  return new lunr.Index(attrs)\n}\n/*!\n * lunr.Builder\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * lunr.Builder performs indexing on a set of documents and\n * returns instances of lunr.Index ready for querying.\n *\n * All configuration of the index is done via the builder, the\n * fields to index, the document reference, the text processing\n * pipeline and document scoring parameters are all set on the\n * builder before indexing.\n *\n * @constructor\n * @property {string} _ref - Internal reference to the document reference field.\n * @property {string[]} _fields - Internal reference to the document fields to index.\n * @property {object} invertedIndex - The inverted index maps terms to document fields.\n * @property {object} documentTermFrequencies - Keeps track of document term frequencies.\n * @property {object} documentLengths - Keeps track of the length of documents added to the index.\n * @property {lunr.tokenizer} tokenizer - Function for splitting strings into tokens for indexing.\n * @property {lunr.Pipeline} pipeline - The pipeline performs text processing on tokens before indexing.\n * @property {lunr.Pipeline} searchPipeline - A pipeline for processing search terms before querying the index.\n * @property {number} documentCount - Keeps track of the total number of documents indexed.\n * @property {number} _b - A parameter to control field length normalization, setting this to 0 disabled normalization, 1 fully normalizes field lengths, the default value is 0.75.\n * @property {number} _k1 - A parameter to control how quickly an increase in term frequency results in term frequency saturation, the default value is 1.2.\n * @property {number} termIndex - A counter incremented for each unique term, used to identify a terms position in the vector space.\n * @property {array} metadataWhitelist - A list of metadata keys that have been whitelisted for entry in the index.\n */\nlunr.Builder = function () {\n  this._ref = \"id\"\n  this._fields = Object.create(null)\n  this._documents = Object.create(null)\n  this.invertedIndex = Object.create(null)\n  this.fieldTermFrequencies = {}\n  this.fieldLengths = {}\n  this.tokenizer = lunr.tokenizer\n  this.pipeline = new lunr.Pipeline\n  this.searchPipeline = new lunr.Pipeline\n  this.documentCount = 0\n  this._b = 0.75\n  this._k1 = 1.2\n  this.termIndex = 0\n  this.metadataWhitelist = []\n}\n\n/**\n * Sets the document field used as the document reference. Every document must have this field.\n * The type of this field in the document should be a string, if it is not a string it will be\n * coerced into a string by calling toString.\n *\n * The default ref is 'id'.\n *\n * The ref should _not_ be changed during indexing, it should be set before any documents are\n * added to the index. Changing it during indexing can lead to inconsistent results.\n *\n * @param {string} ref - The name of the reference field in the document.\n */\nlunr.Builder.prototype.ref = function (ref) {\n  this._ref = ref\n}\n\n/**\n * A function that is used to extract a field from a document.\n *\n * Lunr expects a field to be at the top level of a document, if however the field\n * is deeply nested within a document an extractor function can be used to extract\n * the right field for indexing.\n *\n * @callback fieldExtractor\n * @param {object} doc - The document being added to the index.\n * @returns {?(string|object|object[])} obj - The object that will be indexed for this field.\n * @example <caption>Extracting a nested field</caption>\n * function (doc) { return doc.nested.field }\n */\n\n/**\n * Adds a field to the list of document fields that will be indexed. Every document being\n * indexed should have this field. Null values for this field in indexed documents will\n * not cause errors but will limit the chance of that document being retrieved by searches.\n *\n * All fields should be added before adding documents to the index. Adding fields after\n * a document has been indexed will have no effect on already indexed documents.\n *\n * Fields can be boosted at build time. This allows terms within that field to have more\n * importance when ranking search results. Use a field boost to specify that matches within\n * one field are more important than other fields.\n *\n * @param {string} fieldName - The name of a field to index in all documents.\n * @param {object} attributes - Optional attributes associated with this field.\n * @param {number} [attributes.boost=1] - Boost applied to all terms within this field.\n * @param {fieldExtractor} [attributes.extractor] - Function to extract a field from a document.\n * @throws {RangeError} fieldName cannot contain unsupported characters '/'\n */\nlunr.Builder.prototype.field = function (fieldName, attributes) {\n  if (/\\//.test(fieldName)) {\n    throw new RangeError (\"Field '\" + fieldName + \"' contains illegal character '/'\")\n  }\n\n  this._fields[fieldName] = attributes || {}\n}\n\n/**\n * A parameter to tune the amount of field length normalisation that is applied when\n * calculating relevance scores. A value of 0 will completely disable any normalisation\n * and a value of 1 will fully normalise field lengths. The default is 0.75. Values of b\n * will be clamped to the range 0 - 1.\n *\n * @param {number} number - The value to set for this tuning parameter.\n */\nlunr.Builder.prototype.b = function (number) {\n  if (number < 0) {\n    this._b = 0\n  } else if (number > 1) {\n    this._b = 1\n  } else {\n    this._b = number\n  }\n}\n\n/**\n * A parameter that controls the speed at which a rise in term frequency results in term\n * frequency saturation. The default value is 1.2. Setting this to a higher value will give\n * slower saturation levels, a lower value will result in quicker saturation.\n *\n * @param {number} number - The value to set for this tuning parameter.\n */\nlunr.Builder.prototype.k1 = function (number) {\n  this._k1 = number\n}\n\n/**\n * Adds a document to the index.\n *\n * Before adding fields to the index the index should have been fully setup, with the document\n * ref and all fields to index already having been specified.\n *\n * The document must have a field name as specified by the ref (by default this is 'id') and\n * it should have all fields defined for indexing, though null or undefined values will not\n * cause errors.\n *\n * Entire documents can be boosted at build time. Applying a boost to a document indicates that\n * this document should rank higher in search results than other documents.\n *\n * @param {object} doc - The document to add to the index.\n * @param {object} attributes - Optional attributes associated with this document.\n * @param {number} [attributes.boost=1] - Boost applied to all terms within this document.\n */\nlunr.Builder.prototype.add = function (doc, attributes) {\n  var docRef = doc[this._ref],\n      fields = Object.keys(this._fields)\n\n  this._documents[docRef] = attributes || {}\n  this.documentCount += 1\n\n  for (var i = 0; i < fields.length; i++) {\n    var fieldName = fields[i],\n        extractor = this._fields[fieldName].extractor,\n        field = extractor ? extractor(doc) : doc[fieldName],\n        tokens = this.tokenizer(field, {\n          fields: [fieldName]\n        }),\n        terms = this.pipeline.run(tokens),\n        fieldRef = new lunr.FieldRef (docRef, fieldName),\n        fieldTerms = Object.create(null)\n\n    this.fieldTermFrequencies[fieldRef] = fieldTerms\n    this.fieldLengths[fieldRef] = 0\n\n    // store the length of this field for this document\n    this.fieldLengths[fieldRef] += terms.length\n\n    // calculate term frequencies for this field\n    for (var j = 0; j < terms.length; j++) {\n      var term = terms[j]\n\n      if (fieldTerms[term] == undefined) {\n        fieldTerms[term] = 0\n      }\n\n      fieldTerms[term] += 1\n\n      // add to inverted index\n      // create an initial posting if one doesn't exist\n      if (this.invertedIndex[term] == undefined) {\n        var posting = Object.create(null)\n        posting[\"_index\"] = this.termIndex\n        this.termIndex += 1\n\n        for (var k = 0; k < fields.length; k++) {\n          posting[fields[k]] = Object.create(null)\n        }\n\n        this.invertedIndex[term] = posting\n      }\n\n      // add an entry for this term/fieldName/docRef to the invertedIndex\n      if (this.invertedIndex[term][fieldName][docRef] == undefined) {\n        this.invertedIndex[term][fieldName][docRef] = Object.create(null)\n      }\n\n      // store all whitelisted metadata about this token in the\n      // inverted index\n      for (var l = 0; l < this.metadataWhitelist.length; l++) {\n        var metadataKey = this.metadataWhitelist[l],\n            metadata = term.metadata[metadataKey]\n\n        if (this.invertedIndex[term][fieldName][docRef][metadataKey] == undefined) {\n          this.invertedIndex[term][fieldName][docRef][metadataKey] = []\n        }\n\n        this.invertedIndex[term][fieldName][docRef][metadataKey].push(metadata)\n      }\n    }\n\n  }\n}\n\n/**\n * Calculates the average document length for this index\n *\n * @private\n */\nlunr.Builder.prototype.calculateAverageFieldLengths = function () {\n\n  var fieldRefs = Object.keys(this.fieldLengths),\n      numberOfFields = fieldRefs.length,\n      accumulator = {},\n      documentsWithField = {}\n\n  for (var i = 0; i < numberOfFields; i++) {\n    var fieldRef = lunr.FieldRef.fromString(fieldRefs[i]),\n        field = fieldRef.fieldName\n\n    documentsWithField[field] || (documentsWithField[field] = 0)\n    documentsWithField[field] += 1\n\n    accumulator[field] || (accumulator[field] = 0)\n    accumulator[field] += this.fieldLengths[fieldRef]\n  }\n\n  var fields = Object.keys(this._fields)\n\n  for (var i = 0; i < fields.length; i++) {\n    var fieldName = fields[i]\n    accumulator[fieldName] = accumulator[fieldName] / documentsWithField[fieldName]\n  }\n\n  this.averageFieldLength = accumulator\n}\n\n/**\n * Builds a vector space model of every document using lunr.Vector\n *\n * @private\n */\nlunr.Builder.prototype.createFieldVectors = function () {\n  var fieldVectors = {},\n      fieldRefs = Object.keys(this.fieldTermFrequencies),\n      fieldRefsLength = fieldRefs.length,\n      termIdfCache = Object.create(null)\n\n  for (var i = 0; i < fieldRefsLength; i++) {\n    var fieldRef = lunr.FieldRef.fromString(fieldRefs[i]),\n        fieldName = fieldRef.fieldName,\n        fieldLength = this.fieldLengths[fieldRef],\n        fieldVector = new lunr.Vector,\n        termFrequencies = this.fieldTermFrequencies[fieldRef],\n        terms = Object.keys(termFrequencies),\n        termsLength = terms.length\n\n\n    var fieldBoost = this._fields[fieldName].boost || 1,\n        docBoost = this._documents[fieldRef.docRef].boost || 1\n\n    for (var j = 0; j < termsLength; j++) {\n      var term = terms[j],\n          tf = termFrequencies[term],\n          termIndex = this.invertedIndex[term]._index,\n          idf, score, scoreWithPrecision\n\n      if (termIdfCache[term] === undefined) {\n        idf = lunr.idf(this.invertedIndex[term], this.documentCount)\n        termIdfCache[term] = idf\n      } else {\n        idf = termIdfCache[term]\n      }\n\n      score = idf * ((this._k1 + 1) * tf) / (this._k1 * (1 - this._b + this._b * (fieldLength / this.averageFieldLength[fieldName])) + tf)\n      score *= fieldBoost\n      score *= docBoost\n      scoreWithPrecision = Math.round(score * 1000) / 1000\n      // Converts 1.23456789 to 1.234.\n      // Reducing the precision so that the vectors take up less\n      // space when serialised. Doing it now so that they behave\n      // the same before and after serialisation. Also, this is\n      // the fastest approach to reducing a number's precision in\n      // JavaScript.\n\n      fieldVector.insert(termIndex, scoreWithPrecision)\n    }\n\n    fieldVectors[fieldRef] = fieldVector\n  }\n\n  this.fieldVectors = fieldVectors\n}\n\n/**\n * Creates a token set of all tokens in the index using lunr.TokenSet\n *\n * @private\n */\nlunr.Builder.prototype.createTokenSet = function () {\n  this.tokenSet = lunr.TokenSet.fromArray(\n    Object.keys(this.invertedIndex).sort()\n  )\n}\n\n/**\n * Builds the index, creating an instance of lunr.Index.\n *\n * This completes the indexing process and should only be called\n * once all documents have been added to the index.\n *\n * @returns {lunr.Index}\n */\nlunr.Builder.prototype.build = function () {\n  this.calculateAverageFieldLengths()\n  this.createFieldVectors()\n  this.createTokenSet()\n\n  return new lunr.Index({\n    invertedIndex: this.invertedIndex,\n    fieldVectors: this.fieldVectors,\n    tokenSet: this.tokenSet,\n    fields: Object.keys(this._fields),\n    pipeline: this.searchPipeline\n  })\n}\n\n/**\n * Applies a plugin to the index builder.\n *\n * A plugin is a function that is called with the index builder as its context.\n * Plugins can be used to customise or extend the behaviour of the index\n * in some way. A plugin is just a function, that encapsulated the custom\n * behaviour that should be applied when building the index.\n *\n * The plugin function will be called with the index builder as its argument, additional\n * arguments can also be passed when calling use. The function will be called\n * with the index builder as its context.\n *\n * @param {Function} plugin The plugin to apply.\n */\nlunr.Builder.prototype.use = function (fn) {\n  var args = Array.prototype.slice.call(arguments, 1)\n  args.unshift(this)\n  fn.apply(this, args)\n}\n/**\n * Contains and collects metadata about a matching document.\n * A single instance of lunr.MatchData is returned as part of every\n * lunr.Index~Result.\n *\n * @constructor\n * @param {string} term - The term this match data is associated with\n * @param {string} field - The field in which the term was found\n * @param {object} metadata - The metadata recorded about this term in this field\n * @property {object} metadata - A cloned collection of metadata associated with this document.\n * @see {@link lunr.Index~Result}\n */\nlunr.MatchData = function (term, field, metadata) {\n  var clonedMetadata = Object.create(null),\n      metadataKeys = Object.keys(metadata || {})\n\n  // Cloning the metadata to prevent the original\n  // being mutated during match data combination.\n  // Metadata is kept in an array within the inverted\n  // index so cloning the data can be done with\n  // Array#slice\n  for (var i = 0; i < metadataKeys.length; i++) {\n    var key = metadataKeys[i]\n    clonedMetadata[key] = metadata[key].slice()\n  }\n\n  this.metadata = Object.create(null)\n\n  if (term !== undefined) {\n    this.metadata[term] = Object.create(null)\n    this.metadata[term][field] = clonedMetadata\n  }\n}\n\n/**\n * An instance of lunr.MatchData will be created for every term that matches a\n * document. However only one instance is required in a lunr.Index~Result. This\n * method combines metadata from another instance of lunr.MatchData with this\n * objects metadata.\n *\n * @param {lunr.MatchData} otherMatchData - Another instance of match data to merge with this one.\n * @see {@link lunr.Index~Result}\n */\nlunr.MatchData.prototype.combine = function (otherMatchData) {\n  var terms = Object.keys(otherMatchData.metadata)\n\n  for (var i = 0; i < terms.length; i++) {\n    var term = terms[i],\n        fields = Object.keys(otherMatchData.metadata[term])\n\n    if (this.metadata[term] == undefined) {\n      this.metadata[term] = Object.create(null)\n    }\n\n    for (var j = 0; j < fields.length; j++) {\n      var field = fields[j],\n          keys = Object.keys(otherMatchData.metadata[term][field])\n\n      if (this.metadata[term][field] == undefined) {\n        this.metadata[term][field] = Object.create(null)\n      }\n\n      for (var k = 0; k < keys.length; k++) {\n        var key = keys[k]\n\n        if (this.metadata[term][field][key] == undefined) {\n          this.metadata[term][field][key] = otherMatchData.metadata[term][field][key]\n        } else {\n          this.metadata[term][field][key] = this.metadata[term][field][key].concat(otherMatchData.metadata[term][field][key])\n        }\n\n      }\n    }\n  }\n}\n\n/**\n * Add metadata for a term/field pair to this instance of match data.\n *\n * @param {string} term - The term this match data is associated with\n * @param {string} field - The field in which the term was found\n * @param {object} metadata - The metadata recorded about this term in this field\n */\nlunr.MatchData.prototype.add = function (term, field, metadata) {\n  if (!(term in this.metadata)) {\n    this.metadata[term] = Object.create(null)\n    this.metadata[term][field] = metadata\n    return\n  }\n\n  if (!(field in this.metadata[term])) {\n    this.metadata[term][field] = metadata\n    return\n  }\n\n  var metadataKeys = Object.keys(metadata)\n\n  for (var i = 0; i < metadataKeys.length; i++) {\n    var key = metadataKeys[i]\n\n    if (key in this.metadata[term][field]) {\n      this.metadata[term][field][key] = this.metadata[term][field][key].concat(metadata[key])\n    } else {\n      this.metadata[term][field][key] = metadata[key]\n    }\n  }\n}\n/**\n * A lunr.Query provides a programmatic way of defining queries to be performed\n * against a {@link lunr.Index}.\n *\n * Prefer constructing a lunr.Query using the {@link lunr.Index#query} method\n * so the query object is pre-initialized with the right index fields.\n *\n * @constructor\n * @property {lunr.Query~Clause[]} clauses - An array of query clauses.\n * @property {string[]} allFields - An array of all available fields in a lunr.Index.\n */\nlunr.Query = function (allFields) {\n  this.clauses = []\n  this.allFields = allFields\n}\n\n/**\n * Constants for indicating what kind of automatic wildcard insertion will be used when constructing a query clause.\n *\n * This allows wildcards to be added to the beginning and end of a term without having to manually do any string\n * concatenation.\n *\n * The wildcard constants can be bitwise combined to select both leading and trailing wildcards.\n *\n * @constant\n * @default\n * @property {number} wildcard.NONE - The term will have no wildcards inserted, this is the default behaviour\n * @property {number} wildcard.LEADING - Prepend the term with a wildcard, unless a leading wildcard already exists\n * @property {number} wildcard.TRAILING - Append a wildcard to the term, unless a trailing wildcard already exists\n * @see lunr.Query~Clause\n * @see lunr.Query#clause\n * @see lunr.Query#term\n * @example <caption>query term with trailing wildcard</caption>\n * query.term('foo', { wildcard: lunr.Query.wildcard.TRAILING })\n * @example <caption>query term with leading and trailing wildcard</caption>\n * query.term('foo', {\n *   wildcard: lunr.Query.wildcard.LEADING | lunr.Query.wildcard.TRAILING\n * })\n */\n\nlunr.Query.wildcard = new String (\"*\")\nlunr.Query.wildcard.NONE = 0\nlunr.Query.wildcard.LEADING = 1\nlunr.Query.wildcard.TRAILING = 2\n\n/**\n * Constants for indicating what kind of presence a term must have in matching documents.\n *\n * @constant\n * @enum {number}\n * @see lunr.Query~Clause\n * @see lunr.Query#clause\n * @see lunr.Query#term\n * @example <caption>query term with required presence</caption>\n * query.term('foo', { presence: lunr.Query.presence.REQUIRED })\n */\nlunr.Query.presence = {\n  /**\n   * Term's presence in a document is optional, this is the default value.\n   */\n  OPTIONAL: 1,\n\n  /**\n   * Term's presence in a document is required, documents that do not contain\n   * this term will not be returned.\n   */\n  REQUIRED: 2,\n\n  /**\n   * Term's presence in a document is prohibited, documents that do contain\n   * this term will not be returned.\n   */\n  PROHIBITED: 3\n}\n\n/**\n * A single clause in a {@link lunr.Query} contains a term and details on how to\n * match that term against a {@link lunr.Index}.\n *\n * @typedef {Object} lunr.Query~Clause\n * @property {string[]} fields - The fields in an index this clause should be matched against.\n * @property {number} [boost=1] - Any boost that should be applied when matching this clause.\n * @property {number} [editDistance] - Whether the term should have fuzzy matching applied, and how fuzzy the match should be.\n * @property {boolean} [usePipeline] - Whether the term should be passed through the search pipeline.\n * @property {number} [wildcard=lunr.Query.wildcard.NONE] - Whether the term should have wildcards appended or prepended.\n * @property {number} [presence=lunr.Query.presence.OPTIONAL] - The terms presence in any matching documents.\n */\n\n/**\n * Adds a {@link lunr.Query~Clause} to this query.\n *\n * Unless the clause contains the fields to be matched all fields will be matched. In addition\n * a default boost of 1 is applied to the clause.\n *\n * @param {lunr.Query~Clause} clause - The clause to add to this query.\n * @see lunr.Query~Clause\n * @returns {lunr.Query}\n */\nlunr.Query.prototype.clause = function (clause) {\n  if (!('fields' in clause)) {\n    clause.fields = this.allFields\n  }\n\n  if (!('boost' in clause)) {\n    clause.boost = 1\n  }\n\n  if (!('usePipeline' in clause)) {\n    clause.usePipeline = true\n  }\n\n  if (!('wildcard' in clause)) {\n    clause.wildcard = lunr.Query.wildcard.NONE\n  }\n\n  if ((clause.wildcard & lunr.Query.wildcard.LEADING) && (clause.term.charAt(0) != lunr.Query.wildcard)) {\n    clause.term = \"*\" + clause.term\n  }\n\n  if ((clause.wildcard & lunr.Query.wildcard.TRAILING) && (clause.term.slice(-1) != lunr.Query.wildcard)) {\n    clause.term = \"\" + clause.term + \"*\"\n  }\n\n  if (!('presence' in clause)) {\n    clause.presence = lunr.Query.presence.OPTIONAL\n  }\n\n  this.clauses.push(clause)\n\n  return this\n}\n\n/**\n * A negated query is one in which every clause has a presence of\n * prohibited. These queries require some special processing to return\n * the expected results.\n *\n * @returns boolean\n */\nlunr.Query.prototype.isNegated = function () {\n  for (var i = 0; i < this.clauses.length; i++) {\n    if (this.clauses[i].presence != lunr.Query.presence.PROHIBITED) {\n      return false\n    }\n  }\n\n  return true\n}\n\n/**\n * Adds a term to the current query, under the covers this will create a {@link lunr.Query~Clause}\n * to the list of clauses that make up this query.\n *\n * The term is used as is, i.e. no tokenization will be performed by this method. Instead conversion\n * to a token or token-like string should be done before calling this method.\n *\n * The term will be converted to a string by calling `toString`. Multiple terms can be passed as an\n * array, each term in the array will share the same options.\n *\n * @param {object|object[]} term - The term(s) to add to the query.\n * @param {object} [options] - Any additional properties to add to the query clause.\n * @returns {lunr.Query}\n * @see lunr.Query#clause\n * @see lunr.Query~Clause\n * @example <caption>adding a single term to a query</caption>\n * query.term(\"foo\")\n * @example <caption>adding a single term to a query and specifying search fields, term boost and automatic trailing wildcard</caption>\n * query.term(\"foo\", {\n *   fields: [\"title\"],\n *   boost: 10,\n *   wildcard: lunr.Query.wildcard.TRAILING\n * })\n * @example <caption>using lunr.tokenizer to convert a string to tokens before using them as terms</caption>\n * query.term(lunr.tokenizer(\"foo bar\"))\n */\nlunr.Query.prototype.term = function (term, options) {\n  if (Array.isArray(term)) {\n    term.forEach(function (t) { this.term(t, lunr.utils.clone(options)) }, this)\n    return this\n  }\n\n  var clause = options || {}\n  clause.term = term.toString()\n\n  this.clause(clause)\n\n  return this\n}\nlunr.QueryParseError = function (message, start, end) {\n  this.name = \"QueryParseError\"\n  this.message = message\n  this.start = start\n  this.end = end\n}\n\nlunr.QueryParseError.prototype = new Error\nlunr.QueryLexer = function (str) {\n  this.lexemes = []\n  this.str = str\n  this.length = str.length\n  this.pos = 0\n  this.start = 0\n  this.escapeCharPositions = []\n}\n\nlunr.QueryLexer.prototype.run = function () {\n  var state = lunr.QueryLexer.lexText\n\n  while (state) {\n    state = state(this)\n  }\n}\n\nlunr.QueryLexer.prototype.sliceString = function () {\n  var subSlices = [],\n      sliceStart = this.start,\n      sliceEnd = this.pos\n\n  for (var i = 0; i < this.escapeCharPositions.length; i++) {\n    sliceEnd = this.escapeCharPositions[i]\n    subSlices.push(this.str.slice(sliceStart, sliceEnd))\n    sliceStart = sliceEnd + 1\n  }\n\n  subSlices.push(this.str.slice(sliceStart, this.pos))\n  this.escapeCharPositions.length = 0\n\n  return subSlices.join('')\n}\n\nlunr.QueryLexer.prototype.emit = function (type) {\n  this.lexemes.push({\n    type: type,\n    str: this.sliceString(),\n    start: this.start,\n    end: this.pos\n  })\n\n  this.start = this.pos\n}\n\nlunr.QueryLexer.prototype.escapeCharacter = function () {\n  this.escapeCharPositions.push(this.pos - 1)\n  this.pos += 1\n}\n\nlunr.QueryLexer.prototype.next = function () {\n  if (this.pos >= this.length) {\n    return lunr.QueryLexer.EOS\n  }\n\n  var char = this.str.charAt(this.pos)\n  this.pos += 1\n  return char\n}\n\nlunr.QueryLexer.prototype.width = function () {\n  return this.pos - this.start\n}\n\nlunr.QueryLexer.prototype.ignore = function () {\n  if (this.start == this.pos) {\n    this.pos += 1\n  }\n\n  this.start = this.pos\n}\n\nlunr.QueryLexer.prototype.backup = function () {\n  this.pos -= 1\n}\n\nlunr.QueryLexer.prototype.acceptDigitRun = function () {\n  var char, charCode\n\n  do {\n    char = this.next()\n    charCode = char.charCodeAt(0)\n  } while (charCode > 47 && charCode < 58)\n\n  if (char != lunr.QueryLexer.EOS) {\n    this.backup()\n  }\n}\n\nlunr.QueryLexer.prototype.more = function () {\n  return this.pos < this.length\n}\n\nlunr.QueryLexer.EOS = 'EOS'\nlunr.QueryLexer.FIELD = 'FIELD'\nlunr.QueryLexer.TERM = 'TERM'\nlunr.QueryLexer.EDIT_DISTANCE = 'EDIT_DISTANCE'\nlunr.QueryLexer.BOOST = 'BOOST'\nlunr.QueryLexer.PRESENCE = 'PRESENCE'\n\nlunr.QueryLexer.lexField = function (lexer) {\n  lexer.backup()\n  lexer.emit(lunr.QueryLexer.FIELD)\n  lexer.ignore()\n  return lunr.QueryLexer.lexText\n}\n\nlunr.QueryLexer.lexTerm = function (lexer) {\n  if (lexer.width() > 1) {\n    lexer.backup()\n    lexer.emit(lunr.QueryLexer.TERM)\n  }\n\n  lexer.ignore()\n\n  if (lexer.more()) {\n    return lunr.QueryLexer.lexText\n  }\n}\n\nlunr.QueryLexer.lexEditDistance = function (lexer) {\n  lexer.ignore()\n  lexer.acceptDigitRun()\n  lexer.emit(lunr.QueryLexer.EDIT_DISTANCE)\n  return lunr.QueryLexer.lexText\n}\n\nlunr.QueryLexer.lexBoost = function (lexer) {\n  lexer.ignore()\n  lexer.acceptDigitRun()\n  lexer.emit(lunr.QueryLexer.BOOST)\n  return lunr.QueryLexer.lexText\n}\n\nlunr.QueryLexer.lexEOS = function (lexer) {\n  if (lexer.width() > 0) {\n    lexer.emit(lunr.QueryLexer.TERM)\n  }\n}\n\n// This matches the separator used when tokenising fields\n// within a document. These should match otherwise it is\n// not possible to search for some tokens within a document.\n//\n// It is possible for the user to change the separator on the\n// tokenizer so it _might_ clash with any other of the special\n// characters already used within the search string, e.g. :.\n//\n// This means that it is possible to change the separator in\n// such a way that makes some words unsearchable using a search\n// string.\nlunr.QueryLexer.termSeparator = lunr.tokenizer.separator\n\nlunr.QueryLexer.lexText = function (lexer) {\n  while (true) {\n    var char = lexer.next()\n\n    if (char == lunr.QueryLexer.EOS) {\n      return lunr.QueryLexer.lexEOS\n    }\n\n    // Escape character is '\\'\n    if (char.charCodeAt(0) == 92) {\n      lexer.escapeCharacter()\n      continue\n    }\n\n    if (char == \":\") {\n      return lunr.QueryLexer.lexField\n    }\n\n    if (char == \"~\") {\n      lexer.backup()\n      if (lexer.width() > 0) {\n        lexer.emit(lunr.QueryLexer.TERM)\n      }\n      return lunr.QueryLexer.lexEditDistance\n    }\n\n    if (char == \"^\") {\n      lexer.backup()\n      if (lexer.width() > 0) {\n        lexer.emit(lunr.QueryLexer.TERM)\n      }\n      return lunr.QueryLexer.lexBoost\n    }\n\n    // \"+\" indicates term presence is required\n    // checking for length to ensure that only\n    // leading \"+\" are considered\n    if (char == \"+\" && lexer.width() === 1) {\n      lexer.emit(lunr.QueryLexer.PRESENCE)\n      return lunr.QueryLexer.lexText\n    }\n\n    // \"-\" indicates term presence is prohibited\n    // checking for length to ensure that only\n    // leading \"-\" are considered\n    if (char == \"-\" && lexer.width() === 1) {\n      lexer.emit(lunr.QueryLexer.PRESENCE)\n      return lunr.QueryLexer.lexText\n    }\n\n    if (char.match(lunr.QueryLexer.termSeparator)) {\n      return lunr.QueryLexer.lexTerm\n    }\n  }\n}\n\nlunr.QueryParser = function (str, query) {\n  this.lexer = new lunr.QueryLexer (str)\n  this.query = query\n  this.currentClause = {}\n  this.lexemeIdx = 0\n}\n\nlunr.QueryParser.prototype.parse = function () {\n  this.lexer.run()\n  this.lexemes = this.lexer.lexemes\n\n  var state = lunr.QueryParser.parseClause\n\n  while (state) {\n    state = state(this)\n  }\n\n  return this.query\n}\n\nlunr.QueryParser.prototype.peekLexeme = function () {\n  return this.lexemes[this.lexemeIdx]\n}\n\nlunr.QueryParser.prototype.consumeLexeme = function () {\n  var lexeme = this.peekLexeme()\n  this.lexemeIdx += 1\n  return lexeme\n}\n\nlunr.QueryParser.prototype.nextClause = function () {\n  var completedClause = this.currentClause\n  this.query.clause(completedClause)\n  this.currentClause = {}\n}\n\nlunr.QueryParser.parseClause = function (parser) {\n  var lexeme = parser.peekLexeme()\n\n  if (lexeme == undefined) {\n    return\n  }\n\n  switch (lexeme.type) {\n    case lunr.QueryLexer.PRESENCE:\n      return lunr.QueryParser.parsePresence\n    case lunr.QueryLexer.FIELD:\n      return lunr.QueryParser.parseField\n    case lunr.QueryLexer.TERM:\n      return lunr.QueryParser.parseTerm\n    default:\n      var errorMessage = \"expected either a field or a term, found \" + lexeme.type\n\n      if (lexeme.str.length >= 1) {\n        errorMessage += \" with value '\" + lexeme.str + \"'\"\n      }\n\n      throw new lunr.QueryParseError (errorMessage, lexeme.start, lexeme.end)\n  }\n}\n\nlunr.QueryParser.parsePresence = function (parser) {\n  var lexeme = parser.consumeLexeme()\n\n  if (lexeme == undefined) {\n    return\n  }\n\n  switch (lexeme.str) {\n    case \"-\":\n      parser.currentClause.presence = lunr.Query.presence.PROHIBITED\n      break\n    case \"+\":\n      parser.currentClause.presence = lunr.Query.presence.REQUIRED\n      break\n    default:\n      var errorMessage = \"unrecognised presence operator'\" + lexeme.str + \"'\"\n      throw new lunr.QueryParseError (errorMessage, lexeme.start, lexeme.end)\n  }\n\n  var nextLexeme = parser.peekLexeme()\n\n  if (nextLexeme == undefined) {\n    var errorMessage = \"expecting term or field, found nothing\"\n    throw new lunr.QueryParseError (errorMessage, lexeme.start, lexeme.end)\n  }\n\n  switch (nextLexeme.type) {\n    case lunr.QueryLexer.FIELD:\n      return lunr.QueryParser.parseField\n    case lunr.QueryLexer.TERM:\n      return lunr.QueryParser.parseTerm\n    default:\n      var errorMessage = \"expecting term or field, found '\" + nextLexeme.type + \"'\"\n      throw new lunr.QueryParseError (errorMessage, nextLexeme.start, nextLexeme.end)\n  }\n}\n\nlunr.QueryParser.parseField = function (parser) {\n  var lexeme = parser.consumeLexeme()\n\n  if (lexeme == undefined) {\n    return\n  }\n\n  if (parser.query.allFields.indexOf(lexeme.str) == -1) {\n    var possibleFields = parser.query.allFields.map(function (f) { return \"'\" + f + \"'\" }).join(', '),\n        errorMessage = \"unrecognised field '\" + lexeme.str + \"', possible fields: \" + possibleFields\n\n    throw new lunr.QueryParseError (errorMessage, lexeme.start, lexeme.end)\n  }\n\n  parser.currentClause.fields = [lexeme.str]\n\n  var nextLexeme = parser.peekLexeme()\n\n  if (nextLexeme == undefined) {\n    var errorMessage = \"expecting term, found nothing\"\n    throw new lunr.QueryParseError (errorMessage, lexeme.start, lexeme.end)\n  }\n\n  switch (nextLexeme.type) {\n    case lunr.QueryLexer.TERM:\n      return lunr.QueryParser.parseTerm\n    default:\n      var errorMessage = \"expecting term, found '\" + nextLexeme.type + \"'\"\n      throw new lunr.QueryParseError (errorMessage, nextLexeme.start, nextLexeme.end)\n  }\n}\n\nlunr.QueryParser.parseTerm = function (parser) {\n  var lexeme = parser.consumeLexeme()\n\n  if (lexeme == undefined) {\n    return\n  }\n\n  parser.currentClause.term = lexeme.str.toLowerCase()\n\n  if (lexeme.str.indexOf(\"*\") != -1) {\n    parser.currentClause.usePipeline = false\n  }\n\n  var nextLexeme = parser.peekLexeme()\n\n  if (nextLexeme == undefined) {\n    parser.nextClause()\n    return\n  }\n\n  switch (nextLexeme.type) {\n    case lunr.QueryLexer.TERM:\n      parser.nextClause()\n      return lunr.QueryParser.parseTerm\n    case lunr.QueryLexer.FIELD:\n      parser.nextClause()\n      return lunr.QueryParser.parseField\n    case lunr.QueryLexer.EDIT_DISTANCE:\n      return lunr.QueryParser.parseEditDistance\n    case lunr.QueryLexer.BOOST:\n      return lunr.QueryParser.parseBoost\n    case lunr.QueryLexer.PRESENCE:\n      parser.nextClause()\n      return lunr.QueryParser.parsePresence\n    default:\n      var errorMessage = \"Unexpected lexeme type '\" + nextLexeme.type + \"'\"\n      throw new lunr.QueryParseError (errorMessage, nextLexeme.start, nextLexeme.end)\n  }\n}\n\nlunr.QueryParser.parseEditDistance = function (parser) {\n  var lexeme = parser.consumeLexeme()\n\n  if (lexeme == undefined) {\n    return\n  }\n\n  var editDistance = parseInt(lexeme.str, 10)\n\n  if (isNaN(editDistance)) {\n    var errorMessage = \"edit distance must be numeric\"\n    throw new lunr.QueryParseError (errorMessage, lexeme.start, lexeme.end)\n  }\n\n  parser.currentClause.editDistance = editDistance\n\n  var nextLexeme = parser.peekLexeme()\n\n  if (nextLexeme == undefined) {\n    parser.nextClause()\n    return\n  }\n\n  switch (nextLexeme.type) {\n    case lunr.QueryLexer.TERM:\n      parser.nextClause()\n      return lunr.QueryParser.parseTerm\n    case lunr.QueryLexer.FIELD:\n      parser.nextClause()\n      return lunr.QueryParser.parseField\n    case lunr.QueryLexer.EDIT_DISTANCE:\n      return lunr.QueryParser.parseEditDistance\n    case lunr.QueryLexer.BOOST:\n      return lunr.QueryParser.parseBoost\n    case lunr.QueryLexer.PRESENCE:\n      parser.nextClause()\n      return lunr.QueryParser.parsePresence\n    default:\n      var errorMessage = \"Unexpected lexeme type '\" + nextLexeme.type + \"'\"\n      throw new lunr.QueryParseError (errorMessage, nextLexeme.start, nextLexeme.end)\n  }\n}\n\nlunr.QueryParser.parseBoost = function (parser) {\n  var lexeme = parser.consumeLexeme()\n\n  if (lexeme == undefined) {\n    return\n  }\n\n  var boost = parseInt(lexeme.str, 10)\n\n  if (isNaN(boost)) {\n    var errorMessage = \"boost must be numeric\"\n    throw new lunr.QueryParseError (errorMessage, lexeme.start, lexeme.end)\n  }\n\n  parser.currentClause.boost = boost\n\n  var nextLexeme = parser.peekLexeme()\n\n  if (nextLexeme == undefined) {\n    parser.nextClause()\n    return\n  }\n\n  switch (nextLexeme.type) {\n    case lunr.QueryLexer.TERM:\n      parser.nextClause()\n      return lunr.QueryParser.parseTerm\n    case lunr.QueryLexer.FIELD:\n      parser.nextClause()\n      return lunr.QueryParser.parseField\n    case lunr.QueryLexer.EDIT_DISTANCE:\n      return lunr.QueryParser.parseEditDistance\n    case lunr.QueryLexer.BOOST:\n      return lunr.QueryParser.parseBoost\n    case lunr.QueryLexer.PRESENCE:\n      parser.nextClause()\n      return lunr.QueryParser.parsePresence\n    default:\n      var errorMessage = \"Unexpected lexeme type '\" + nextLexeme.type + \"'\"\n      throw new lunr.QueryParseError (errorMessage, nextLexeme.start, nextLexeme.end)\n  }\n}\n\n  /**\n   * export the module via AMD, CommonJS or as a browser global\n   * Export code from https://github.com/umdjs/umd/blob/master/returnExports.js\n   */\n  ;(function (root, factory) {\n    if (typeof define === 'function' && define.amd) {\n      // AMD. Register as an anonymous module.\n      define(factory)\n    } else if (typeof exports === 'object') {\n      /**\n       * Node. Does not work with strict CommonJS, but\n       * only CommonJS-like enviroments that support module.exports,\n       * like Node.\n       */\n      module.exports = factory()\n    } else {\n      // Browser globals (root is window)\n      root.lunr = factory()\n    }\n  }(this, function () {\n    /**\n     * Just return a value to define the module export.\n     * This example returns an object, but the module\n     * can return a function as the exported value.\n     */\n    return lunr\n  }))\n})();\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A RTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport lunr from \"lunr\"\n\nimport { getElement } from \"~/browser/element/_\"\nimport \"~/polyfills\"\n\nimport { Search } from \"../../_\"\nimport { SearchConfig } from \"../../config\"\nimport {\n  SearchMessage,\n  SearchMessageType\n} from \"../message\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Add support for `iframe-worker` shim\n *\n * While `importScripts` is synchronous when executed inside of a web worker,\n * it's not possible to provide a synchronous shim implementation. The cool\n * thing is that awaiting a non-Promise will convert it into a Promise, so\n * extending the type definition to return a `Promise` shouldn't break anything.\n *\n * @see https://bit.ly/2PjDnXi - GitHub comment\n *\n * @param urls - Scripts to load\n *\n * @returns Promise resolving with no result\n */\ndeclare global {\n  function importScripts(...urls: string[]): Promise<void> | void\n}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Search index\n */\nlet index: Search\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch (= import) multi-language support through `lunr-languages`\n *\n * This function automatically imports the stemmers necessary to process the\n * languages which are defined as part of the search configuration.\n *\n * If the worker runs inside of an `iframe` (when using `iframe-worker` as\n * a shim), the base URL for the stemmers to be loaded must be determined by\n * searching for the first `script` element with a `src` attribute, which will\n * contain the contents of this script.\n *\n * @param config - Search configuration\n *\n * @returns Promise resolving with no result\n */\nasync function setupSearchLanguages(\n  config: SearchConfig\n): Promise<void> {\n  let base = \"../lunr\"\n\n  /* Detect `iframe-worker` and fix base URL */\n  if (typeof parent !== \"undefined\" && \"IFrameWorker\" in parent) {\n    const worker = getElement<HTMLScriptElement>(\"script[src]\")!\n    const [path] = worker.src.split(\"/worker\")\n\n    /* Prefix base with path */\n    base = base.replace(\"..\", path)\n  }\n\n  /* Add scripts for languages */\n  const scripts = []\n  for (const lang of config.lang) {\n    switch (lang) {\n\n      /* Add segmenter for Japanese */\n      case \"ja\":\n        scripts.push(`${base}/tinyseg.js`)\n        break\n\n      /* Add segmenter for Hindi and Thai */\n      case \"hi\":\n      case \"th\":\n        scripts.push(`${base}/wordcut.js`)\n        break\n    }\n\n    /* Add language support */\n    if (lang !== \"en\")\n      scripts.push(`${base}/min/lunr.${lang}.min.js`)\n  }\n\n  /* Add multi-language support */\n  if (config.lang.length > 1)\n    scripts.push(`${base}/min/lunr.multi.min.js`)\n\n  /* Load scripts synchronously */\n  if (scripts.length)\n    await importScripts(\n      `${base}/min/lunr.stemmer.support.min.js`,\n      ...scripts\n    )\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Message handler\n *\n * @param message - Source message\n *\n * @returns Target message\n */\nexport async function handler(\n  message: SearchMessage\n): Promise<SearchMessage> {\n  switch (message.type) {\n\n    /* Search setup message */\n    case SearchMessageType.SETUP:\n      await setupSearchLanguages(message.data.config)\n      index = new Search(message.data)\n      return {\n        type: SearchMessageType.READY\n      }\n\n    /* Search query message */\n    case SearchMessageType.QUERY:\n      const query = message.data\n      try {\n        return {\n          type: SearchMessageType.RESULT,\n          data: index.search(query)\n        }\n\n      /* Return empty result in case of error */\n      } catch (err) {\n        console.warn(`Invalid query: ${query} \u2013 see https://bit.ly/2s3ChXG`)\n        console.warn(err)\n        return {\n          type: SearchMessageType.RESULT,\n          data: { items: [] }\n        }\n      }\n\n    /* All other messages */\n    default:\n      throw new TypeError(\"Invalid message type\")\n  }\n}\n\n/* ----------------------------------------------------------------------------\n * Worker\n * ------------------------------------------------------------------------- */\n\n/* Expose Lunr.js in global scope, or stemmers won't work */\nself.lunr = lunr\n\n/* Handle messages */\naddEventListener(\"message\", async ev => {\n  postMessage(await handler(ev.data))\n})\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve all elements matching the query selector\n *\n * @template T - Element type\n *\n * @param selector - Query selector\n * @param node - Node of reference\n *\n * @returns Elements\n */\nexport function getElements<T extends keyof HTMLElementTagNameMap>(\n  selector: T, node?: ParentNode\n): HTMLElementTagNameMap[T][]\n\nexport function getElements<T extends HTMLElement>(\n  selector: string, node?: ParentNode\n): T[]\n\nexport function getElements<T extends HTMLElement>(\n  selector: string, node: ParentNode = document\n): T[] {\n  return Array.from(node.querySelectorAll<T>(selector))\n}\n\n/**\n * Retrieve an element matching a query selector or throw a reference error\n *\n * Note that this function assumes that the element is present. If unsure if an\n * element is existent, use the `getOptionalElement` function instead.\n *\n * @template T - Element type\n *\n * @param selector - Query selector\n * @param node - Node of reference\n *\n * @returns Element\n */\nexport function getElement<T extends keyof HTMLElementTagNameMap>(\n  selector: T, node?: ParentNode\n): HTMLElementTagNameMap[T]\n\nexport function getElement<T extends HTMLElement>(\n  selector: string, node?: ParentNode\n): T\n\nexport function getElement<T extends HTMLElement>(\n  selector: string, node: ParentNode = document\n): T {\n  const el = getOptionalElement<T>(selector, node)\n  if (typeof el === \"undefined\")\n    throw new ReferenceError(\n      `Missing element: expected \"${selector}\" to be present`\n    )\n\n  /* Return element */\n  return el\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Retrieve an optional element matching the query selector\n *\n * @template T - Element type\n *\n * @param selector - Query selector\n * @param node - Node of reference\n *\n * @returns Element or nothing\n */\nexport function getOptionalElement<T extends keyof HTMLElementTagNameMap>(\n  selector: T, node?: ParentNode\n): HTMLElementTagNameMap[T] | undefined\n\nexport function getOptionalElement<T extends HTMLElement>(\n  selector: string, node?: ParentNode\n): T | undefined\n\nexport function getOptionalElement<T extends HTMLElement>(\n  selector: string, node: ParentNode = document\n): T | undefined {\n  return node.querySelector<T>(selector) || undefined\n}\n\n/**\n * Retrieve the currently active element\n *\n * @returns Element or nothing\n */\nexport function getActiveElement(): HTMLElement | undefined {\n  return document.activeElement instanceof HTMLElement\n    ? document.activeElement || undefined\n    : undefined\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Polyfills\n * ------------------------------------------------------------------------- */\n\n/* Polyfill `Object.entries` */\nif (!Object.entries)\n  Object.entries = function (obj: object) {\n    const data: [string, string][] = []\n    for (const key of Object.keys(obj))\n      // @ts-expect-error - ignore property access warning\n      data.push([key, obj[key]])\n\n    /* Return entries */\n    return data\n  }\n\n/* Polyfill `Object.values` */\nif (!Object.values)\n  Object.values = function (obj: object) {\n    const data: string[] = []\n    for (const key of Object.keys(obj))\n      // @ts-expect-error - ignore property access warning\n      data.push(obj[key])\n\n    /* Return values */\n    return data\n  }\n\n/* ------------------------------------------------------------------------- */\n\n/* Polyfills for `Element` */\nif (typeof Element !== \"undefined\") {\n\n  /* Polyfill `Element.scrollTo` */\n  if (!Element.prototype.scrollTo)\n    Element.prototype.scrollTo = function (\n      x?: ScrollToOptions | number, y?: number\n    ): void {\n      if (typeof x === \"object\") {\n        this.scrollLeft = x.left!\n        this.scrollTop = x.top!\n      } else {\n        this.scrollLeft = x!\n        this.scrollTop = y!\n      }\n    }\n\n  /* Polyfill `Element.replaceWith` */\n  if (!Element.prototype.replaceWith)\n    Element.prototype.replaceWith = function (\n      ...nodes: Array<string | Node>\n    ): void {\n      const parent = this.parentNode\n      if (parent) {\n        if (nodes.length === 0)\n          parent.removeChild(this)\n\n        /* Replace children and create text nodes */\n        for (let i = nodes.length - 1; i >= 0; i--) {\n          let node = nodes[i]\n          if (typeof node === \"string\")\n            node = document.createTextNode(node)\n          else if (node.parentNode)\n            node.parentNode.removeChild(node)\n\n          /* Replace child or insert before previous sibling */\n          if (!i)\n            parent.replaceChild(node, this)\n          else\n            parent.insertBefore(this.previousSibling!, node)\n        }\n      }\n    }\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search configuration\n */\nexport interface SearchConfig {\n  lang: string[]                       /* Search languages */\n  separator: string                    /* Search separator */\n  pipeline: SearchPipelineFn[]         /* Search pipeline */\n}\n\n/**\n * Search document\n */\nexport interface SearchDocument {\n  location: string                     /* Document location */\n  title: string                        /* Document title */\n  text: string                         /* Document text */\n  tags?: string[]                      /* Document tags */\n  boost?: number                       /* Document boost */\n  parent?: SearchDocument              /* Document parent */\n}\n\n/**\n * Search options\n */\nexport interface SearchOptions {\n  suggest: boolean                     /* Search suggestions */\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Search index\n */\nexport interface SearchIndex {\n  config: SearchConfig                 /* Search configuration */\n  docs: SearchDocument[]               /* Search documents */\n  options: SearchOptions               /* Search options */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search pipeline function\n */\ntype SearchPipelineFn =\n  | \"trimmer\"                          /* Trimmer */\n  | \"stopWordFilter\"                   /* Stop word filter */\n  | \"stemmer\"                          /* Stemmer */\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Create a search document map\n *\n * This function creates a mapping of URLs (including anchors) to the actual\n * articles and sections. It relies on the invariant that the search index is\n * ordered with the main article appearing before all sections with anchors.\n * If this is not the case, the logic music be changed.\n *\n * @param docs - Search documents\n *\n * @returns Search document map\n */\nexport function setupSearchDocumentMap(\n  docs: SearchDocument[]\n): Map<string, SearchDocument> {\n  const map = new Map<string, SearchDocument>()\n  for (const doc of docs) {\n    const [path] = doc.location.split(\"#\")\n\n    /* Add document article */\n    const article = map.get(path)\n    if (typeof article === \"undefined\") {\n      map.set(path, doc)\n\n      /* Add document section */\n    } else {\n      map.set(doc.location, doc)\n      doc.parent = article\n    }\n  }\n\n  /* Return search document map */\n  return map\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Visitor function\n *\n * @param start - Start offset\n * @param end - End offset\n */\ntype VisitorFn = (\n  start: number, end: number\n) => void\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Split a string using the given separator\n *\n * @param input - Input value\n * @param separator - Separator\n * @param fn - Visitor function\n */\nexport function split(\n  input: string, separator: RegExp, fn: VisitorFn\n): void {\n  separator = new RegExp(separator, \"g\")\n\n  /* Split string using separator */\n  let match: RegExpExecArray | null\n  let index = 0\n  do {\n    match = separator.exec(input)\n\n    /* Emit non-empty range */\n    const until = match?.index ?? input.length\n    if (index < until)\n      fn(index, until)\n\n    /* Update last index */\n    if (match) {\n      const [term] = match\n      index = match.index + term.length\n\n      /* Support zero-length lookaheads */\n      if (term.length === 0)\n        separator.lastIndex = match.index + 1\n    }\n  } while (match)\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Extraction type\n *\n * This type defines the possible values that are encoded into the first two\n * bits of a section that is part of the blocks of a tokenization table. There\n * are three types of interest: HTML opening and closing tags, as well as the\n * actual text content we need to extract for indexing.\n */\nexport const enum Extract {\n  TAG_OPEN  = 0,                       /* HTML opening tag */\n  TEXT      = 1,                       /* Text content */\n  TAG_CLOSE = 2                        /* HTML closing tag */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Visitor function\n *\n * @param block - Block index\n * @param type - Extraction type\n * @param start - Start offset\n * @param end - End offset\n */\ntype VisitorFn = (\n  block: number, type: Extract, start: number, end: number\n) => void\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Split a string into markup and text sections\n *\n * This function scans a string and divides it up into sections of markup and\n * text. For each section, it invokes the given visitor function with the block\n * index, extraction type, as well as start and end offsets. Using a visitor\n * function (= streaming data) is ideal for minimizing pressure on the GC.\n *\n * @param input - Input value\n * @param fn - Visitor function\n */\nexport function extract(\n  input: string, fn: VisitorFn\n): void {\n\n  let block = 0                        /* Current block */\n  let start = 0                        /* Current start offset */\n  let end = 0                          /* Current end offset */\n\n  /* Split string into sections */\n  for (let stack = 0; end < input.length; end++) {\n\n    /* Opening tag after non-empty section */\n    if (input.charAt(end) === \"<\" && end > start) {\n      fn(block, Extract.TEXT, start, start = end)\n\n    /* Closing tag */\n    } else if (input.charAt(end) === \">\") {\n      if (input.charAt(start + 1) === \"/\") {\n        if (--stack === 0)\n          fn(block++, Extract.TAG_CLOSE, start, end + 1)\n\n      /* Tag is not self-closing */\n      } else if (input.charAt(end - 1) !== \"/\") {\n        if (stack++ === 0)\n          fn(block, Extract.TAG_OPEN, start, end + 1)\n      }\n\n      /* New section */\n      start = end + 1\n    }\n  }\n\n  /* Add trailing section */\n  if (end > start)\n    fn(block, Extract.TEXT, start, end)\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Position table\n */\nexport type PositionTable = number[][]\n\n/**\n * Position\n */\nexport type Position = number\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Highlight all occurrences in a string\n *\n * This function receives a field's value (e.g. like `title` or `text`), it's\n * position table that was generated during indexing, and the positions found\n * when executing the query. It then highlights all occurrences, and returns\n * their concatenation. In case of multiple blocks, two are returned.\n *\n * @param input - Input value\n * @param table - Table for indexing\n * @param positions - Occurrences\n * @param full - Full results\n *\n * @returns Highlighted string value\n */\nexport function highlight(\n  input: string, table: PositionTable, positions: Position[], full = false\n): string {\n  return highlightAll([input], table, positions, full).pop()!\n}\n\n/**\n * Highlight all occurrences in a set of strings\n *\n * @param inputs - Input values\n * @param table - Table for indexing\n * @param positions - Occurrences\n * @param full - Full results\n *\n * @returns Highlighted string values\n */\nexport function highlightAll(\n  inputs: string[], table: PositionTable, positions: Position[], full = false\n): string[] {\n\n  /* Map blocks to input values */\n  const mapping = [0]\n  for (let t = 1; t < table.length; t++) {\n    const prev = table[t - 1]\n    const next = table[t]\n\n    /* Check if table points to new block */\n    const p = prev[prev.length - 1] >>> 2 & 0x3FF\n    const q = next[0]               >>> 12\n\n    /* Add block to mapping */\n    mapping.push(+(p > q) + mapping[mapping.length - 1])\n  }\n\n  /* Highlight strings one after another */\n  return inputs.map((input, i) => {\n    let cursor = 0\n\n    /* Map occurrences to blocks */\n    const blocks = new Map<number, number[]>()\n    for (const p of positions.sort((a, b) => a - b)) {\n      const index = p & 0xFFFFF\n      const block = p >>> 20\n      if (mapping[block] !== i)\n        continue\n\n      /* Ensure presence of block group */\n      let group = blocks.get(block)\n      if (typeof group === \"undefined\")\n        blocks.set(block, group = [])\n\n      /* Add index to group */\n      group.push(index)\n    }\n\n    /* Just return string, if no occurrences */\n    if (blocks.size === 0)\n      return input\n\n    /* Compute slices */\n    const slices: string[] = []\n    for (const [block, indexes] of blocks) {\n      const t = table[block]\n\n      /* Extract positions and length */\n      const start  = t[0]            >>> 12\n      const end    = t[t.length - 1] >>> 12\n      const length = t[t.length - 1] >>> 2 & 0x3FF\n\n      /* Add prefix, if full results are desired */\n      if (full && start > cursor)\n        slices.push(input.slice(cursor, start))\n\n      /* Extract and highlight slice */\n      let slice = input.slice(start, end + length)\n      for (const j of indexes.sort((a, b) => b - a)) {\n\n        /* Retrieve offset and length of match */\n        const p = (t[j] >>> 12) - start\n        const q = (t[j] >>> 2 & 0x3FF) + p\n\n        /* Wrap occurrence */\n        slice = [\n          slice.slice(0, p),\n          \"<mark>\",\n          slice.slice(p, q),\n          \"</mark>\",\n          slice.slice(q)\n        ].join(\"\")\n      }\n\n      /* Update cursor */\n      cursor = end + length\n\n      /* Append slice and abort if we have two */\n      if (slices.push(slice) === 2)\n        break\n    }\n\n    /* Add suffix, if full results are desired */\n    if (full && cursor < input.length)\n      slices.push(input.slice(cursor))\n\n    /* Return highlighted slices */\n    return slices.join(\"\")\n  })\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { split } from \"../_\"\nimport {\n  Extract,\n  extract\n} from \"../extract\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Split a string or set of strings into tokens\n *\n * This tokenizer supersedes the default tokenizer that is provided by Lunr.js,\n * as it is aware of HTML tags and allows for multi-character splitting.\n *\n * It takes the given inputs, splits each of them into markup and text sections,\n * tokenizes and segments (if necessary) each of them, and then indexes them in\n * a table by using a compact bit representation. Bitwise techniques are used\n * to write and read from the table during indexing and querying.\n *\n * @see https://bit.ly/3W3Xw4J - Search: better, faster, smaller\n *\n * @param input - Input value(s)\n *\n * @returns Tokens\n */\nexport function tokenize(\n  input?: string | string[]\n): lunr.Token[] {\n  const tokens: lunr.Token[] = []\n  if (typeof input === \"undefined\")\n    return tokens\n\n  /* Tokenize strings one after another */\n  const inputs = Array.isArray(input) ? input : [input]\n  for (let i = 0; i < inputs.length; i++) {\n    const table = lunr.tokenizer.table\n    const total = table.length\n\n    /* Split string into sections and tokenize content blocks */\n    extract(inputs[i], (block, type, start, end) => {\n      table[block += total] ||= []\n      switch (type) {\n\n        /* Handle markup */\n        case Extract.TAG_OPEN:\n        case Extract.TAG_CLOSE:\n          table[block].push(\n            start       << 12 |\n            end - start <<  2 |\n            type\n          )\n          break\n\n        /* Handle text content */\n        case Extract.TEXT:\n          const section = inputs[i].slice(start, end)\n          split(section, lunr.tokenizer.separator, (index, until) => {\n\n            /**\n             * Apply segmenter after tokenization. Note that the segmenter will\n             * also split words at word boundaries, which is not what we want,\n             * so we need to check if we can somehow mitigate this behavior.\n             */\n            if (typeof lunr.segmenter !== \"undefined\") {\n              const subsection = section.slice(index, until)\n              if (/^[MHIK]$/.test(lunr.segmenter.ctype_(subsection))) {\n                const segments = lunr.segmenter.segment(subsection)\n                for (let s = 0, l = 0; s < segments.length; s++) {\n\n                  /* Add block to section */\n                  table[block] ||= []\n                  table[block].push(\n                    start + index + l  << 12 |\n                    segments[s].length <<  2 |\n                    type\n                  )\n\n                  /* Add token with position */\n                  tokens.push(new lunr.Token(\n                    segments[s].toLowerCase(), {\n                      position: block << 20 | table[block].length - 1\n                    }\n                  ))\n\n                  /* Keep track of length */\n                  l += segments[s].length\n                }\n                return\n              }\n            }\n\n            /* Add block to section */\n            table[block].push(\n              start + index << 12 |\n              until - index <<  2 |\n              type\n            )\n\n            /* Add token with position */\n            tokens.push(new lunr.Token(\n              section.slice(index, until).toLowerCase(), {\n                position: block << 20 | table[block].length - 1\n              }\n            ))\n          })\n      }\n    })\n  }\n\n  /* Return tokens */\n  return tokens\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Visitor function\n *\n * @param value - String value\n *\n * @returns String term(s)\n */\ntype VisitorFn = (\n  value: string\n) => string | string[]\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Default transformation function\n *\n * 1. Trim excess whitespace from left and right.\n *\n * 2. Search for parts in quotation marks and prepend a `+` modifier to denote\n *    that the resulting document must contain all parts, converting the query\n *    to an `AND` query (as opposed to the default `OR` behavior). While users\n *    may expect parts enclosed in quotation marks to map to span queries, i.e.\n *    for which order is important, Lunr.js doesn't support them, so the best\n *    we can do is to convert the parts to an `AND` query.\n *\n * 3. Replace control characters which are not located at the beginning of the\n *    query or preceded by white space, or are not followed by a non-whitespace\n *    character or are at the end of the query string. Furthermore, filter\n *    unmatched quotation marks.\n *\n * 4. Split the query string at whitespace, then pass each part to the visitor\n *    function for tokenization, and append a wildcard to every resulting term\n *    that is not explicitly marked with a `+`, `-`, `~` or `^` modifier, since\n *    it ensures consistent and stable ranking when multiple terms are entered.\n *    Also, if a fuzzy or boost modifier are given, but no numeric value has\n *    been entered, default to 1 to not induce a query error.\n *\n * @param query - Query value\n * @param fn - Visitor function\n *\n * @returns Transformed query value\n */\nexport function transform(\n  query: string, fn: VisitorFn = term => term\n): string {\n  return query\n\n    /* => 1 */\n    .trim()\n\n    /* => 2 */\n    .split(/\"([^\"]+)\"/g)\n      .map((parts, index) => index & 1\n        ? parts.replace(/^\\b|^(?![^\\x00-\\x7F]|$)|\\s+/g, \" +\")\n        : parts\n      )\n      .join(\"\")\n\n    /* => 3 */\n    .replace(/\"|(?:^|\\s+)[*+\\-:^~]+(?=\\s+|$)/g, \"\")\n\n    /* => 4 */\n    .split(/\\s+/g)\n      .reduce((prev, term) => {\n        const next = fn(term)\n        return [...prev, ...Array.isArray(next) ? next : [next]]\n      }, [] as string[])\n      .map(term => /([~^]$)/.test(term) ? `${term}1` : term)\n      .map(term => /(^[+-]|[~^]\\d+$)/.test(term) ? term : `${term}*`)\n      .join(\" \")\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { split } from \"../../internal\"\nimport { transform } from \"../transform\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search query clause\n */\nexport interface SearchQueryClause {\n  presence: lunr.Query.presence        /* Clause presence */\n  term: string                         /* Clause term */\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Search query terms\n */\nexport type SearchQueryTerms = Record<string, boolean>\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Transform search query\n *\n * This function lexes the given search query and applies the transformation\n * function to each term, preserving markup like `+` and `-` modifiers.\n *\n * @param query - Search query\n *\n * @returns Search query\n */\nexport function transformSearchQuery(\n  query: string\n): string {\n\n  /* Split query terms with tokenizer */\n  return transform(query, part => {\n    const terms: string[] = []\n\n    /* Initialize lexer and analyze part */\n    const lexer = new lunr.QueryLexer(part)\n    lexer.run()\n\n    /* Extract and tokenize term from lexeme */\n    for (const { type, str: term, start, end } of lexer.lexemes)\n      switch (type) {\n\n        /* Hack: remove colon - see https://bit.ly/3wD3T3I */\n        case \"FIELD\":\n          if (![\"title\", \"text\", \"tags\"].includes(term))\n            part = [\n              part.slice(0, end),\n              \" \",\n              part.slice(end + 1)\n            ].join(\"\")\n          break\n\n        /* Tokenize term */\n        case \"TERM\":\n          split(term, lunr.tokenizer.separator, (...range) => {\n            terms.push([\n              part.slice(0, start),\n              term.slice(...range),\n              part.slice(end)\n            ].join(\"\"))\n          })\n      }\n\n    /* Return terms */\n    return terms\n  })\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Parse a search query for analysis\n *\n * Lunr.js itself has a bug where it doesn't detect or remove wildcards for\n * query clauses, so we must do this here.\n *\n * @see https://bit.ly/3DpTGtz - GitHub issue\n *\n * @param value - Query value\n *\n * @returns Search query clauses\n */\nexport function parseSearchQuery(\n  value: string\n): SearchQueryClause[] {\n  const query  = new lunr.Query([\"title\", \"text\", \"tags\"])\n  const parser = new lunr.QueryParser(value, query)\n\n  /* Parse Search query */\n  parser.parse()\n  for (const clause of query.clauses) {\n    clause.usePipeline = true\n\n    /* Handle leading wildcard */\n    if (clause.term.startsWith(\"*\")) {\n      clause.wildcard = lunr.Query.wildcard.LEADING\n      clause.term = clause.term.slice(1)\n    }\n\n    /* Handle trailing wildcard */\n    if (clause.term.endsWith(\"*\")) {\n      clause.wildcard = lunr.Query.wildcard.TRAILING\n      clause.term = clause.term.slice(0, -1)\n    }\n  }\n\n  /* Return query clauses */\n  return query.clauses\n}\n\n/**\n * Analyze the search query clauses in regard to the search terms found\n *\n * @param query - Search query clauses\n * @param terms - Search terms\n *\n * @returns Search query terms\n */\nexport function getSearchQueryTerms(\n  query: SearchQueryClause[], terms: string[]\n): SearchQueryTerms {\n  const clauses = new Set<SearchQueryClause>(query)\n\n  /* Match query clauses against terms */\n  const result: SearchQueryTerms = {}\n  for (let t = 0; t < terms.length; t++)\n    for (const clause of clauses)\n      if (terms[t].startsWith(clause.term)) {\n        result[clause.term] = true\n        clauses.delete(clause)\n      }\n\n  /* Annotate unmatched non-stopword query clauses */\n  for (const clause of clauses)\n    if (lunr.stopWordFilter?.(clause.term))\n      result[clause.term] = false\n\n  /* Return query terms */\n  return result\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Segment a search query using the inverted index\n *\n * This function implements a clever approach to text segmentation for Asian\n * languages, as it used the information already available in the search index.\n * The idea is to greedily segment the search query based on the tokens that are\n * already part of the index, as described in the linked issue.\n *\n * @see https://bit.ly/3lwjrk7 - GitHub issue\n *\n * @param query - Query value\n * @param index - Inverted index\n *\n * @returns Segmented query value\n */\nexport function segment(\n  query: string, index: object\n): Iterable<string> {\n  const segments = new Set<string>()\n\n  /* Segment search query */\n  const wordcuts = new Uint16Array(query.length)\n  for (let i = 0; i < query.length; i++)\n    for (let j = i + 1; j < query.length; j++) {\n      const value = query.slice(i, j)\n      if (value in index)\n        wordcuts[i] = j - i\n    }\n\n  /* Compute longest matches with minimum overlap */\n  const stack = [0]\n  for (let s = stack.length; s > 0;) {\n    const p = stack[--s]\n    for (let q = 1; q < wordcuts[p]; q++)\n      if (wordcuts[p + q] > wordcuts[p] - q) {\n        segments.add(query.slice(p, p + q))\n        stack[s++] = p + q\n      }\n\n    /* Continue at end of query string */\n    const q = p + wordcuts[p]\n    if (wordcuts[q] && q < query.length - 1)\n      stack[s++] = q\n\n    /* Add current segment */\n    segments.add(query.slice(p, q))\n  }\n\n  // @todo fix this case in the code block above, this is a hotfix\n  if (segments.has(\"\"))\n    return new Set([query])\n\n  /* Return segmented query value */\n  return segments\n}\n", "/*\n * Copyright (c) 2016-2023 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  SearchDocument,\n  SearchIndex,\n  SearchOptions,\n  setupSearchDocumentMap\n} from \"../config\"\nimport {\n  Position,\n  PositionTable,\n  highlight,\n  highlightAll,\n  tokenize\n} from \"../internal\"\nimport {\n  SearchQueryTerms,\n  getSearchQueryTerms,\n  parseSearchQuery,\n  segment,\n  transformSearchQuery\n} from \"../query\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search item\n */\nexport interface SearchItem\n  extends SearchDocument\n{\n  score: number                        /* Score (relevance) */\n  terms: SearchQueryTerms              /* Search query terms */\n}\n\n/**\n * Search result\n */\nexport interface SearchResult {\n  items: SearchItem[][]                /* Search items */\n  suggest?: string[]                   /* Search suggestions */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Create field extractor factory\n *\n * @param table - Position table map\n *\n * @returns Extractor factory\n */\nfunction extractor(table: Map<string, PositionTable>) {\n  return (name: keyof SearchDocument) => {\n    return (doc: SearchDocument) => {\n      if (typeof doc[name] === \"undefined\")\n        return undefined\n\n      /* Compute identifier and initialize table */\n      const id = [doc.location, name].join(\":\")\n      table.set(id, lunr.tokenizer.table = [])\n\n      /* Return field value */\n      return doc[name]\n    }\n  }\n}\n\n/**\n * Compute the difference of two lists of strings\n *\n * @param a - 1st list of strings\n * @param b - 2nd list of strings\n *\n * @returns Difference\n */\nfunction difference(a: string[], b: string[]): string[] {\n  const [x, y] = [new Set(a), new Set(b)]\n  return [\n    ...new Set([...x].filter(value => !y.has(value)))\n  ]\n}\n\n/* ----------------------------------------------------------------------------\n * Class\n * ------------------------------------------------------------------------- */\n\n/**\n * Search index\n */\nexport class Search {\n\n  /**\n   * Search document map\n   */\n  protected map: Map<string, SearchDocument>\n\n  /**\n   * Search options\n   */\n  protected options: SearchOptions\n\n  /**\n   * The underlying Lunr.js search index\n   */\n  protected index: lunr.Index\n\n  /**\n   * Internal position table map\n   */\n  protected table: Map<string, PositionTable>\n\n  /**\n   * Create the search integration\n   *\n   * @param data - Search index\n   */\n  public constructor({ config, docs, options }: SearchIndex) {\n    const field = extractor(this.table = new Map())\n\n    /* Set up document map and options */\n    this.map = setupSearchDocumentMap(docs)\n    this.options = options\n\n    /* Set up document index */\n    this.index = lunr(function () {\n      this.metadataWhitelist = [\"position\"]\n      this.b(0)\n\n      /* Set up (multi-)language support */\n      if (config.lang.length === 1 && config.lang[0] !== \"en\") {\n        // @ts-expect-error - namespace indexing not supported\n        this.use(lunr[config.lang[0]])\n      } else if (config.lang.length > 1) {\n        this.use(lunr.multiLanguage(...config.lang))\n      }\n\n      /* Set up custom tokenizer (must be after language setup) */\n      this.tokenizer = tokenize as typeof lunr.tokenizer\n      lunr.tokenizer.separator = new RegExp(config.separator)\n\n      /* Set up custom segmenter, if loaded */\n      lunr.segmenter = \"TinySegmenter\" in lunr\n        ? new lunr.TinySegmenter()\n        : undefined\n\n      /* Compute functions to be removed from the pipeline */\n      const fns = difference([\n        \"trimmer\", \"stopWordFilter\", \"stemmer\"\n      ], config.pipeline)\n\n      /* Remove functions from the pipeline for registered languages */\n      for (const lang of config.lang.map(language => (\n        // @ts-expect-error - namespace indexing not supported\n        language === \"en\" ? lunr : lunr[language]\n      )))\n        for (const fn of fns) {\n          this.pipeline.remove(lang[fn])\n          this.searchPipeline.remove(lang[fn])\n        }\n\n      /* Set up index reference */\n      this.ref(\"location\")\n\n      /* Set up index fields */\n      this.field(\"title\", { boost: 1e3, extractor: field(\"title\") })\n      this.field(\"text\",  { boost: 1e0, extractor: field(\"text\") })\n      this.field(\"tags\",  { boost: 1e6, extractor: field(\"tags\") })\n\n      /* Add documents to index */\n      for (const doc of docs)\n        this.add(doc, { boost: doc.boost })\n    })\n  }\n\n  /**\n   * Search for matching documents\n   *\n   * @param query - Search query\n   *\n   * @returns Search result\n   */\n  public search(query: string): SearchResult {\n\n    // Experimental Chinese segmentation\n    query = query.replace(/\\p{sc=Han}+/gu, value => {\n      return [...segment(value, this.index.invertedIndex)]\n        .join(\"* \")\n    })\n\n    // @todo: move segmenter (above) into transformSearchQuery\n    query = transformSearchQuery(query)\n    if (!query)\n      return { items: [] }\n\n    /* Parse query to extract clauses for analysis */\n    const clauses = parseSearchQuery(query)\n      .filter(clause => (\n        clause.presence !== lunr.Query.presence.PROHIBITED\n      ))\n\n    /* Perform search and post-process results */\n    const groups = this.index.search(query)\n\n      /* Apply post-query boosts based on title and search query terms */\n      .reduce<SearchItem[]>((item, { ref, score, matchData }) => {\n        let doc = this.map.get(ref)\n        if (typeof doc !== \"undefined\") {\n\n          /* Shallow copy document */\n          doc = { ...doc }\n          if (doc.tags)\n            doc.tags = [...doc.tags]\n\n          /* Compute and analyze search query terms */\n          const terms = getSearchQueryTerms(\n            clauses,\n            Object.keys(matchData.metadata)\n          )\n\n          /* Highlight matches in fields */\n          for (const field of this.index.fields) {\n            if (typeof doc[field] === \"undefined\")\n              continue\n\n            /* Collect positions from matches */\n            const positions: Position[] = []\n            for (const match of Object.values(matchData.metadata))\n              if (typeof match[field] !== \"undefined\")\n                positions.push(...match[field].position)\n\n            /* Skip highlighting, if no positions were collected */\n            if (!positions.length)\n              continue\n\n            /* Load table and determine highlighting method */\n            const table = this.table.get([doc.location, field].join(\":\"))!\n            const fn = Array.isArray(doc[field])\n              ? highlightAll\n              : highlight\n\n            // @ts-expect-error - stop moaning, TypeScript!\n            doc[field] = fn(doc[field], table, positions, field !== \"text\")\n          }\n\n          /* Highlight title and text and apply post-query boosts */\n          const boost = +!doc.parent +\n            Object.values(terms)\n              .filter(t => t).length /\n            Object.keys(terms).length\n\n          /* Append item */\n          item.push({\n            ...doc,\n            score: score * (1 + boost ** 2),\n            terms\n          })\n        }\n        return item\n      }, [])\n\n      /* Sort search results again after applying boosts */\n      .sort((a, b) => b.score - a.score)\n\n      /* Group search results by article */\n      .reduce((items, result) => {\n        const doc = this.map.get(result.location)\n        if (typeof doc !== \"undefined\") {\n          const ref = doc.parent\n            ? doc.parent.location\n            : doc.location\n          items.set(ref, [...items.get(ref) || [], result])\n        }\n        return items\n      }, new Map<string, SearchItem[]>())\n\n    /* Ensure that every item set has an article */\n    for (const [ref, items] of groups)\n      if (!items.find(item => item.location === ref)) {\n        const doc = this.map.get(ref)!\n        items.push({ ...doc, score: 0, terms: {} })\n      }\n\n    /* Generate search suggestions, if desired */\n    let suggest: string[] | undefined\n    if (this.options.suggest) {\n      const titles = this.index.query(builder => {\n        for (const clause of clauses)\n          builder.term(clause.term, {\n            fields: [\"title\"],\n            presence: lunr.Query.presence.REQUIRED,\n            wildcard: lunr.Query.wildcard.TRAILING\n          })\n      })\n\n      /* Retrieve suggestions for best match */\n      suggest = titles.length\n        ? Object.keys(titles[0].matchData.metadata)\n        : []\n    }\n\n    /* Return search result */\n    return {\n      items: [...groups.values()],\n      ...typeof suggest !== \"undefined\" && { suggest }\n    }\n  }\n}\n"],
+  "sourcesContent": ["/**\n * lunr - http://lunrjs.com - A bit like Solr, but much smaller and not as bright - 2.3.9\n * Copyright (C) 2020 Oliver Nightingale\n * @license MIT\n */\n\n;(function(){\n\n/**\n * A convenience function for configuring and constructing\n * a new lunr Index.\n *\n * A lunr.Builder instance is created and the pipeline setup\n * with a trimmer, stop word filter and stemmer.\n *\n * This builder object is yielded to the configuration function\n * that is passed as a parameter, allowing the list of fields\n * and other builder parameters to be customised.\n *\n * All documents _must_ be added within the passed config function.\n *\n * @example\n * var idx = lunr(function () {\n *   this.field('title')\n *   this.field('body')\n *   this.ref('id')\n *\n *   documents.forEach(function (doc) {\n *     this.add(doc)\n *   }, this)\n * })\n *\n * @see {@link lunr.Builder}\n * @see {@link lunr.Pipeline}\n * @see {@link lunr.trimmer}\n * @see {@link lunr.stopWordFilter}\n * @see {@link lunr.stemmer}\n * @namespace {function} lunr\n */\nvar lunr = function (config) {\n  var builder = new lunr.Builder\n\n  builder.pipeline.add(\n    lunr.trimmer,\n    lunr.stopWordFilter,\n    lunr.stemmer\n  )\n\n  builder.searchPipeline.add(\n    lunr.stemmer\n  )\n\n  config.call(builder, builder)\n  return builder.build()\n}\n\nlunr.version = \"2.3.9\"\n/*!\n * lunr.utils\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * A namespace containing utils for the rest of the lunr library\n * @namespace lunr.utils\n */\nlunr.utils = {}\n\n/**\n * Print a warning message to the console.\n *\n * @param {String} message The message to be printed.\n * @memberOf lunr.utils\n * @function\n */\nlunr.utils.warn = (function (global) {\n  /* eslint-disable no-console */\n  return function (message) {\n    if (global.console && console.warn) {\n      console.warn(message)\n    }\n  }\n  /* eslint-enable no-console */\n})(this)\n\n/**\n * Convert an object to a string.\n *\n * In the case of `null` and `undefined` the function returns\n * the empty string, in all other cases the result of calling\n * `toString` on the passed object is returned.\n *\n * @param {Any} obj The object to convert to a string.\n * @return {String} string representation of the passed object.\n * @memberOf lunr.utils\n */\nlunr.utils.asString = function (obj) {\n  if (obj === void 0 || obj === null) {\n    return \"\"\n  } else {\n    return obj.toString()\n  }\n}\n\n/**\n * Clones an object.\n *\n * Will create a copy of an existing object such that any mutations\n * on the copy cannot affect the original.\n *\n * Only shallow objects are supported, passing a nested object to this\n * function will cause a TypeError.\n *\n * Objects with primitives, and arrays of primitives are supported.\n *\n * @param {Object} obj The object to clone.\n * @return {Object} a clone of the passed object.\n * @throws {TypeError} when a nested object is passed.\n * @memberOf Utils\n */\nlunr.utils.clone = function (obj) {\n  if (obj === null || obj === undefined) {\n    return obj\n  }\n\n  var clone = Object.create(null),\n      keys = Object.keys(obj)\n\n  for (var i = 0; i < keys.length; i++) {\n    var key = keys[i],\n        val = obj[key]\n\n    if (Array.isArray(val)) {\n      clone[key] = val.slice()\n      continue\n    }\n\n    if (typeof val === 'string' ||\n        typeof val === 'number' ||\n        typeof val === 'boolean') {\n      clone[key] = val\n      continue\n    }\n\n    throw new TypeError(\"clone is not deep and does not support nested objects\")\n  }\n\n  return clone\n}\nlunr.FieldRef = function (docRef, fieldName, stringValue) {\n  this.docRef = docRef\n  this.fieldName = fieldName\n  this._stringValue = stringValue\n}\n\nlunr.FieldRef.joiner = \"/\"\n\nlunr.FieldRef.fromString = function (s) {\n  var n = s.indexOf(lunr.FieldRef.joiner)\n\n  if (n === -1) {\n    throw \"malformed field ref string\"\n  }\n\n  var fieldRef = s.slice(0, n),\n      docRef = s.slice(n + 1)\n\n  return new lunr.FieldRef (docRef, fieldRef, s)\n}\n\nlunr.FieldRef.prototype.toString = function () {\n  if (this._stringValue == undefined) {\n    this._stringValue = this.fieldName + lunr.FieldRef.joiner + this.docRef\n  }\n\n  return this._stringValue\n}\n/*!\n * lunr.Set\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * A lunr set.\n *\n * @constructor\n */\nlunr.Set = function (elements) {\n  this.elements = Object.create(null)\n\n  if (elements) {\n    this.length = elements.length\n\n    for (var i = 0; i < this.length; i++) {\n      this.elements[elements[i]] = true\n    }\n  } else {\n    this.length = 0\n  }\n}\n\n/**\n * A complete set that contains all elements.\n *\n * @static\n * @readonly\n * @type {lunr.Set}\n */\nlunr.Set.complete = {\n  intersect: function (other) {\n    return other\n  },\n\n  union: function () {\n    return this\n  },\n\n  contains: function () {\n    return true\n  }\n}\n\n/**\n * An empty set that contains no elements.\n *\n * @static\n * @readonly\n * @type {lunr.Set}\n */\nlunr.Set.empty = {\n  intersect: function () {\n    return this\n  },\n\n  union: function (other) {\n    return other\n  },\n\n  contains: function () {\n    return false\n  }\n}\n\n/**\n * Returns true if this set contains the specified object.\n *\n * @param {object} object - Object whose presence in this set is to be tested.\n * @returns {boolean} - True if this set contains the specified object.\n */\nlunr.Set.prototype.contains = function (object) {\n  return !!this.elements[object]\n}\n\n/**\n * Returns a new set containing only the elements that are present in both\n * this set and the specified set.\n *\n * @param {lunr.Set} other - set to intersect with this set.\n * @returns {lunr.Set} a new set that is the intersection of this and the specified set.\n */\n\nlunr.Set.prototype.intersect = function (other) {\n  var a, b, elements, intersection = []\n\n  if (other === lunr.Set.complete) {\n    return this\n  }\n\n  if (other === lunr.Set.empty) {\n    return other\n  }\n\n  if (this.length < other.length) {\n    a = this\n    b = other\n  } else {\n    a = other\n    b = this\n  }\n\n  elements = Object.keys(a.elements)\n\n  for (var i = 0; i < elements.length; i++) {\n    var element = elements[i]\n    if (element in b.elements) {\n      intersection.push(element)\n    }\n  }\n\n  return new lunr.Set (intersection)\n}\n\n/**\n * Returns a new set combining the elements of this and the specified set.\n *\n * @param {lunr.Set} other - set to union with this set.\n * @return {lunr.Set} a new set that is the union of this and the specified set.\n */\n\nlunr.Set.prototype.union = function (other) {\n  if (other === lunr.Set.complete) {\n    return lunr.Set.complete\n  }\n\n  if (other === lunr.Set.empty) {\n    return this\n  }\n\n  return new lunr.Set(Object.keys(this.elements).concat(Object.keys(other.elements)))\n}\n/**\n * A function to calculate the inverse document frequency for\n * a posting. This is shared between the builder and the index\n *\n * @private\n * @param {object} posting - The posting for a given term\n * @param {number} documentCount - The total number of documents.\n */\nlunr.idf = function (posting, documentCount) {\n  var documentsWithTerm = 0\n\n  for (var fieldName in posting) {\n    if (fieldName == '_index') continue // Ignore the term index, its not a field\n    documentsWithTerm += Object.keys(posting[fieldName]).length\n  }\n\n  var x = (documentCount - documentsWithTerm + 0.5) / (documentsWithTerm + 0.5)\n\n  return Math.log(1 + Math.abs(x))\n}\n\n/**\n * A token wraps a string representation of a token\n * as it is passed through the text processing pipeline.\n *\n * @constructor\n * @param {string} [str=''] - The string token being wrapped.\n * @param {object} [metadata={}] - Metadata associated with this token.\n */\nlunr.Token = function (str, metadata) {\n  this.str = str || \"\"\n  this.metadata = metadata || {}\n}\n\n/**\n * Returns the token string that is being wrapped by this object.\n *\n * @returns {string}\n */\nlunr.Token.prototype.toString = function () {\n  return this.str\n}\n\n/**\n * A token update function is used when updating or optionally\n * when cloning a token.\n *\n * @callback lunr.Token~updateFunction\n * @param {string} str - The string representation of the token.\n * @param {Object} metadata - All metadata associated with this token.\n */\n\n/**\n * Applies the given function to the wrapped string token.\n *\n * @example\n * token.update(function (str, metadata) {\n *   return str.toUpperCase()\n * })\n *\n * @param {lunr.Token~updateFunction} fn - A function to apply to the token string.\n * @returns {lunr.Token}\n */\nlunr.Token.prototype.update = function (fn) {\n  this.str = fn(this.str, this.metadata)\n  return this\n}\n\n/**\n * Creates a clone of this token. Optionally a function can be\n * applied to the cloned token.\n *\n * @param {lunr.Token~updateFunction} [fn] - An optional function to apply to the cloned token.\n * @returns {lunr.Token}\n */\nlunr.Token.prototype.clone = function (fn) {\n  fn = fn || function (s) { return s }\n  return new lunr.Token (fn(this.str, this.metadata), this.metadata)\n}\n/*!\n * lunr.tokenizer\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * A function for splitting a string into tokens ready to be inserted into\n * the search index. Uses `lunr.tokenizer.separator` to split strings, change\n * the value of this property to change how strings are split into tokens.\n *\n * This tokenizer will convert its parameter to a string by calling `toString` and\n * then will split this string on the character in `lunr.tokenizer.separator`.\n * Arrays will have their elements converted to strings and wrapped in a lunr.Token.\n *\n * Optional metadata can be passed to the tokenizer, this metadata will be cloned and\n * added as metadata to every token that is created from the object to be tokenized.\n *\n * @static\n * @param {?(string|object|object[])} obj - The object to convert into tokens\n * @param {?object} metadata - Optional metadata to associate with every token\n * @returns {lunr.Token[]}\n * @see {@link lunr.Pipeline}\n */\nlunr.tokenizer = function (obj, metadata) {\n  if (obj == null || obj == undefined) {\n    return []\n  }\n\n  if (Array.isArray(obj)) {\n    return obj.map(function (t) {\n      return new lunr.Token(\n        lunr.utils.asString(t).toLowerCase(),\n        lunr.utils.clone(metadata)\n      )\n    })\n  }\n\n  var str = obj.toString().toLowerCase(),\n      len = str.length,\n      tokens = []\n\n  for (var sliceEnd = 0, sliceStart = 0; sliceEnd <= len; sliceEnd++) {\n    var char = str.charAt(sliceEnd),\n        sliceLength = sliceEnd - sliceStart\n\n    if ((char.match(lunr.tokenizer.separator) || sliceEnd == len)) {\n\n      if (sliceLength > 0) {\n        var tokenMetadata = lunr.utils.clone(metadata) || {}\n        tokenMetadata[\"position\"] = [sliceStart, sliceLength]\n        tokenMetadata[\"index\"] = tokens.length\n\n        tokens.push(\n          new lunr.Token (\n            str.slice(sliceStart, sliceEnd),\n            tokenMetadata\n          )\n        )\n      }\n\n      sliceStart = sliceEnd + 1\n    }\n\n  }\n\n  return tokens\n}\n\n/**\n * The separator used to split a string into tokens. Override this property to change the behaviour of\n * `lunr.tokenizer` behaviour when tokenizing strings. By default this splits on whitespace and hyphens.\n *\n * @static\n * @see lunr.tokenizer\n */\nlunr.tokenizer.separator = /[\\s\\-]+/\n/*!\n * lunr.Pipeline\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * lunr.Pipelines maintain an ordered list of functions to be applied to all\n * tokens in documents entering the search index and queries being ran against\n * the index.\n *\n * An instance of lunr.Index created with the lunr shortcut will contain a\n * pipeline with a stop word filter and an English language stemmer. Extra\n * functions can be added before or after either of these functions or these\n * default functions can be removed.\n *\n * When run the pipeline will call each function in turn, passing a token, the\n * index of that token in the original list of all tokens and finally a list of\n * all the original tokens.\n *\n * The output of functions in the pipeline will be passed to the next function\n * in the pipeline. To exclude a token from entering the index the function\n * should return undefined, the rest of the pipeline will not be called with\n * this token.\n *\n * For serialisation of pipelines to work, all functions used in an instance of\n * a pipeline should be registered with lunr.Pipeline. Registered functions can\n * then be loaded. If trying to load a serialised pipeline that uses functions\n * that are not registered an error will be thrown.\n *\n * If not planning on serialising the pipeline then registering pipeline functions\n * is not necessary.\n *\n * @constructor\n */\nlunr.Pipeline = function () {\n  this._stack = []\n}\n\nlunr.Pipeline.registeredFunctions = Object.create(null)\n\n/**\n * A pipeline function maps lunr.Token to lunr.Token. A lunr.Token contains the token\n * string as well as all known metadata. A pipeline function can mutate the token string\n * or mutate (or add) metadata for a given token.\n *\n * A pipeline function can indicate that the passed token should be discarded by returning\n * null, undefined or an empty string. This token will not be passed to any downstream pipeline\n * functions and will not be added to the index.\n *\n * Multiple tokens can be returned by returning an array of tokens. Each token will be passed\n * to any downstream pipeline functions and all will returned tokens will be added to the index.\n *\n * Any number of pipeline functions may be chained together using a lunr.Pipeline.\n *\n * @interface lunr.PipelineFunction\n * @param {lunr.Token} token - A token from the document being processed.\n * @param {number} i - The index of this token in the complete list of tokens for this document/field.\n * @param {lunr.Token[]} tokens - All tokens for this document/field.\n * @returns {(?lunr.Token|lunr.Token[])}\n */\n\n/**\n * Register a function with the pipeline.\n *\n * Functions that are used in the pipeline should be registered if the pipeline\n * needs to be serialised, or a serialised pipeline needs to be loaded.\n *\n * Registering a function does not add it to a pipeline, functions must still be\n * added to instances of the pipeline for them to be used when running a pipeline.\n *\n * @param {lunr.PipelineFunction} fn - The function to check for.\n * @param {String} label - The label to register this function with\n */\nlunr.Pipeline.registerFunction = function (fn, label) {\n  if (label in this.registeredFunctions) {\n    lunr.utils.warn('Overwriting existing registered function: ' + label)\n  }\n\n  fn.label = label\n  lunr.Pipeline.registeredFunctions[fn.label] = fn\n}\n\n/**\n * Warns if the function is not registered as a Pipeline function.\n *\n * @param {lunr.PipelineFunction} fn - The function to check for.\n * @private\n */\nlunr.Pipeline.warnIfFunctionNotRegistered = function (fn) {\n  var isRegistered = fn.label && (fn.label in this.registeredFunctions)\n\n  if (!isRegistered) {\n    lunr.utils.warn('Function is not registered with pipeline. This may cause problems when serialising the index.\\n', fn)\n  }\n}\n\n/**\n * Loads a previously serialised pipeline.\n *\n * All functions to be loaded must already be registered with lunr.Pipeline.\n * If any function from the serialised data has not been registered then an\n * error will be thrown.\n *\n * @param {Object} serialised - The serialised pipeline to load.\n * @returns {lunr.Pipeline}\n */\nlunr.Pipeline.load = function (serialised) {\n  var pipeline = new lunr.Pipeline\n\n  serialised.forEach(function (fnName) {\n    var fn = lunr.Pipeline.registeredFunctions[fnName]\n\n    if (fn) {\n      pipeline.add(fn)\n    } else {\n      throw new Error('Cannot load unregistered function: ' + fnName)\n    }\n  })\n\n  return pipeline\n}\n\n/**\n * Adds new functions to the end of the pipeline.\n *\n * Logs a warning if the function has not been registered.\n *\n * @param {lunr.PipelineFunction[]} functions - Any number of functions to add to the pipeline.\n */\nlunr.Pipeline.prototype.add = function () {\n  var fns = Array.prototype.slice.call(arguments)\n\n  fns.forEach(function (fn) {\n    lunr.Pipeline.warnIfFunctionNotRegistered(fn)\n    this._stack.push(fn)\n  }, this)\n}\n\n/**\n * Adds a single function after a function that already exists in the\n * pipeline.\n *\n * Logs a warning if the function has not been registered.\n *\n * @param {lunr.PipelineFunction} existingFn - A function that already exists in the pipeline.\n * @param {lunr.PipelineFunction} newFn - The new function to add to the pipeline.\n */\nlunr.Pipeline.prototype.after = function (existingFn, newFn) {\n  lunr.Pipeline.warnIfFunctionNotRegistered(newFn)\n\n  var pos = this._stack.indexOf(existingFn)\n  if (pos == -1) {\n    throw new Error('Cannot find existingFn')\n  }\n\n  pos = pos + 1\n  this._stack.splice(pos, 0, newFn)\n}\n\n/**\n * Adds a single function before a function that already exists in the\n * pipeline.\n *\n * Logs a warning if the function has not been registered.\n *\n * @param {lunr.PipelineFunction} existingFn - A function that already exists in the pipeline.\n * @param {lunr.PipelineFunction} newFn - The new function to add to the pipeline.\n */\nlunr.Pipeline.prototype.before = function (existingFn, newFn) {\n  lunr.Pipeline.warnIfFunctionNotRegistered(newFn)\n\n  var pos = this._stack.indexOf(existingFn)\n  if (pos == -1) {\n    throw new Error('Cannot find existingFn')\n  }\n\n  this._stack.splice(pos, 0, newFn)\n}\n\n/**\n * Removes a function from the pipeline.\n *\n * @param {lunr.PipelineFunction} fn The function to remove from the pipeline.\n */\nlunr.Pipeline.prototype.remove = function (fn) {\n  var pos = this._stack.indexOf(fn)\n  if (pos == -1) {\n    return\n  }\n\n  this._stack.splice(pos, 1)\n}\n\n/**\n * Runs the current list of functions that make up the pipeline against the\n * passed tokens.\n *\n * @param {Array} tokens The tokens to run through the pipeline.\n * @returns {Array}\n */\nlunr.Pipeline.prototype.run = function (tokens) {\n  var stackLength = this._stack.length\n\n  for (var i = 0; i < stackLength; i++) {\n    var fn = this._stack[i]\n    var memo = []\n\n    for (var j = 0; j < tokens.length; j++) {\n      var result = fn(tokens[j], j, tokens)\n\n      if (result === null || result === void 0 || result === '') continue\n\n      if (Array.isArray(result)) {\n        for (var k = 0; k < result.length; k++) {\n          memo.push(result[k])\n        }\n      } else {\n        memo.push(result)\n      }\n    }\n\n    tokens = memo\n  }\n\n  return tokens\n}\n\n/**\n * Convenience method for passing a string through a pipeline and getting\n * strings out. This method takes care of wrapping the passed string in a\n * token and mapping the resulting tokens back to strings.\n *\n * @param {string} str - The string to pass through the pipeline.\n * @param {?object} metadata - Optional metadata to associate with the token\n * passed to the pipeline.\n * @returns {string[]}\n */\nlunr.Pipeline.prototype.runString = function (str, metadata) {\n  var token = new lunr.Token (str, metadata)\n\n  return this.run([token]).map(function (t) {\n    return t.toString()\n  })\n}\n\n/**\n * Resets the pipeline by removing any existing processors.\n *\n */\nlunr.Pipeline.prototype.reset = function () {\n  this._stack = []\n}\n\n/**\n * Returns a representation of the pipeline ready for serialisation.\n *\n * Logs a warning if the function has not been registered.\n *\n * @returns {Array}\n */\nlunr.Pipeline.prototype.toJSON = function () {\n  return this._stack.map(function (fn) {\n    lunr.Pipeline.warnIfFunctionNotRegistered(fn)\n\n    return fn.label\n  })\n}\n/*!\n * lunr.Vector\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * A vector is used to construct the vector space of documents and queries. These\n * vectors support operations to determine the similarity between two documents or\n * a document and a query.\n *\n * Normally no parameters are required for initializing a vector, but in the case of\n * loading a previously dumped vector the raw elements can be provided to the constructor.\n *\n * For performance reasons vectors are implemented with a flat array, where an elements\n * index is immediately followed by its value. E.g. [index, value, index, value]. This\n * allows the underlying array to be as sparse as possible and still offer decent\n * performance when being used for vector calculations.\n *\n * @constructor\n * @param {Number[]} [elements] - The flat list of element index and element value pairs.\n */\nlunr.Vector = function (elements) {\n  this._magnitude = 0\n  this.elements = elements || []\n}\n\n\n/**\n * Calculates the position within the vector to insert a given index.\n *\n * This is used internally by insert and upsert. If there are duplicate indexes then\n * the position is returned as if the value for that index were to be updated, but it\n * is the callers responsibility to check whether there is a duplicate at that index\n *\n * @param {Number} insertIdx - The index at which the element should be inserted.\n * @returns {Number}\n */\nlunr.Vector.prototype.positionForIndex = function (index) {\n  // For an empty vector the tuple can be inserted at the beginning\n  if (this.elements.length == 0) {\n    return 0\n  }\n\n  var start = 0,\n      end = this.elements.length / 2,\n      sliceLength = end - start,\n      pivotPoint = Math.floor(sliceLength / 2),\n      pivotIndex = this.elements[pivotPoint * 2]\n\n  while (sliceLength > 1) {\n    if (pivotIndex < index) {\n      start = pivotPoint\n    }\n\n    if (pivotIndex > index) {\n      end = pivotPoint\n    }\n\n    if (pivotIndex == index) {\n      break\n    }\n\n    sliceLength = end - start\n    pivotPoint = start + Math.floor(sliceLength / 2)\n    pivotIndex = this.elements[pivotPoint * 2]\n  }\n\n  if (pivotIndex == index) {\n    return pivotPoint * 2\n  }\n\n  if (pivotIndex > index) {\n    return pivotPoint * 2\n  }\n\n  if (pivotIndex < index) {\n    return (pivotPoint + 1) * 2\n  }\n}\n\n/**\n * Inserts an element at an index within the vector.\n *\n * Does not allow duplicates, will throw an error if there is already an entry\n * for this index.\n *\n * @param {Number} insertIdx - The index at which the element should be inserted.\n * @param {Number} val - The value to be inserted into the vector.\n */\nlunr.Vector.prototype.insert = function (insertIdx, val) {\n  this.upsert(insertIdx, val, function () {\n    throw \"duplicate index\"\n  })\n}\n\n/**\n * Inserts or updates an existing index within the vector.\n *\n * @param {Number} insertIdx - The index at which the element should be inserted.\n * @param {Number} val - The value to be inserted into the vector.\n * @param {function} fn - A function that is called for updates, the existing value and the\n * requested value are passed as arguments\n */\nlunr.Vector.prototype.upsert = function (insertIdx, val, fn) {\n  this._magnitude = 0\n  var position = this.positionForIndex(insertIdx)\n\n  if (this.elements[position] == insertIdx) {\n    this.elements[position + 1] = fn(this.elements[position + 1], val)\n  } else {\n    this.elements.splice(position, 0, insertIdx, val)\n  }\n}\n\n/**\n * Calculates the magnitude of this vector.\n *\n * @returns {Number}\n */\nlunr.Vector.prototype.magnitude = function () {\n  if (this._magnitude) return this._magnitude\n\n  var sumOfSquares = 0,\n      elementsLength = this.elements.length\n\n  for (var i = 1; i < elementsLength; i += 2) {\n    var val = this.elements[i]\n    sumOfSquares += val * val\n  }\n\n  return this._magnitude = Math.sqrt(sumOfSquares)\n}\n\n/**\n * Calculates the dot product of this vector and another vector.\n *\n * @param {lunr.Vector} otherVector - The vector to compute the dot product with.\n * @returns {Number}\n */\nlunr.Vector.prototype.dot = function (otherVector) {\n  var dotProduct = 0,\n      a = this.elements, b = otherVector.elements,\n      aLen = a.length, bLen = b.length,\n      aVal = 0, bVal = 0,\n      i = 0, j = 0\n\n  while (i < aLen && j < bLen) {\n    aVal = a[i], bVal = b[j]\n    if (aVal < bVal) {\n      i += 2\n    } else if (aVal > bVal) {\n      j += 2\n    } else if (aVal == bVal) {\n      dotProduct += a[i + 1] * b[j + 1]\n      i += 2\n      j += 2\n    }\n  }\n\n  return dotProduct\n}\n\n/**\n * Calculates the similarity between this vector and another vector.\n *\n * @param {lunr.Vector} otherVector - The other vector to calculate the\n * similarity with.\n * @returns {Number}\n */\nlunr.Vector.prototype.similarity = function (otherVector) {\n  return this.dot(otherVector) / this.magnitude() || 0\n}\n\n/**\n * Converts the vector to an array of the elements within the vector.\n *\n * @returns {Number[]}\n */\nlunr.Vector.prototype.toArray = function () {\n  var output = new Array (this.elements.length / 2)\n\n  for (var i = 1, j = 0; i < this.elements.length; i += 2, j++) {\n    output[j] = this.elements[i]\n  }\n\n  return output\n}\n\n/**\n * A JSON serializable representation of the vector.\n *\n * @returns {Number[]}\n */\nlunr.Vector.prototype.toJSON = function () {\n  return this.elements\n}\n/* eslint-disable */\n/*!\n * lunr.stemmer\n * Copyright (C) 2020 Oliver Nightingale\n * Includes code from - http://tartarus.org/~martin/PorterStemmer/js.txt\n */\n\n/**\n * lunr.stemmer is an english language stemmer, this is a JavaScript\n * implementation of the PorterStemmer taken from http://tartarus.org/~martin\n *\n * @static\n * @implements {lunr.PipelineFunction}\n * @param {lunr.Token} token - The string to stem\n * @returns {lunr.Token}\n * @see {@link lunr.Pipeline}\n * @function\n */\nlunr.stemmer = (function(){\n  var step2list = {\n      \"ational\" : \"ate\",\n      \"tional\" : \"tion\",\n      \"enci\" : \"ence\",\n      \"anci\" : \"ance\",\n      \"izer\" : \"ize\",\n      \"bli\" : \"ble\",\n      \"alli\" : \"al\",\n      \"entli\" : \"ent\",\n      \"eli\" : \"e\",\n      \"ousli\" : \"ous\",\n      \"ization\" : \"ize\",\n      \"ation\" : \"ate\",\n      \"ator\" : \"ate\",\n      \"alism\" : \"al\",\n      \"iveness\" : \"ive\",\n      \"fulness\" : \"ful\",\n      \"ousness\" : \"ous\",\n      \"aliti\" : \"al\",\n      \"iviti\" : \"ive\",\n      \"biliti\" : \"ble\",\n      \"logi\" : \"log\"\n    },\n\n    step3list = {\n      \"icate\" : \"ic\",\n      \"ative\" : \"\",\n      \"alize\" : \"al\",\n      \"iciti\" : \"ic\",\n      \"ical\" : \"ic\",\n      \"ful\" : \"\",\n      \"ness\" : \"\"\n    },\n\n    c = \"[^aeiou]\",          // consonant\n    v = \"[aeiouy]\",          // vowel\n    C = c + \"[^aeiouy]*\",    // consonant sequence\n    V = v + \"[aeiou]*\",      // vowel sequence\n\n    mgr0 = \"^(\" + C + \")?\" + V + C,               // [C]VC... is m>0\n    meq1 = \"^(\" + C + \")?\" + V + C + \"(\" + V + \")?$\",  // [C]VC[V] is m=1\n    mgr1 = \"^(\" + C + \")?\" + V + C + V + C,       // [C]VCVC... is m>1\n    s_v = \"^(\" + C + \")?\" + v;                   // vowel in stem\n\n  var re_mgr0 = new RegExp(mgr0);\n  var re_mgr1 = new RegExp(mgr1);\n  var re_meq1 = new RegExp(meq1);\n  var re_s_v = new RegExp(s_v);\n\n  var re_1a = /^(.+?)(ss|i)es$/;\n  var re2_1a = /^(.+?)([^s])s$/;\n  var re_1b = /^(.+?)eed$/;\n  var re2_1b = /^(.+?)(ed|ing)$/;\n  var re_1b_2 = /.$/;\n  var re2_1b_2 = /(at|bl|iz)$/;\n  var re3_1b_2 = new RegExp(\"([^aeiouylsz])\\\\1$\");\n  var re4_1b_2 = new RegExp(\"^\" + C + v + \"[^aeiouwxy]$\");\n\n  var re_1c = /^(.+?[^aeiou])y$/;\n  var re_2 = /^(.+?)(ational|tional|enci|anci|izer|bli|alli|entli|eli|ousli|ization|ation|ator|alism|iveness|fulness|ousness|aliti|iviti|biliti|logi)$/;\n\n  var re_3 = /^(.+?)(icate|ative|alize|iciti|ical|ful|ness)$/;\n\n  var re_4 = /^(.+?)(al|ance|ence|er|ic|able|ible|ant|ement|ment|ent|ou|ism|ate|iti|ous|ive|ize)$/;\n  var re2_4 = /^(.+?)(s|t)(ion)$/;\n\n  var re_5 = /^(.+?)e$/;\n  var re_5_1 = /ll$/;\n  var re3_5 = new RegExp(\"^\" + C + v + \"[^aeiouwxy]$\");\n\n  var porterStemmer = function porterStemmer(w) {\n    var stem,\n      suffix,\n      firstch,\n      re,\n      re2,\n      re3,\n      re4;\n\n    if (w.length < 3) { return w; }\n\n    firstch = w.substr(0,1);\n    if (firstch == \"y\") {\n      w = firstch.toUpperCase() + w.substr(1);\n    }\n\n    // Step 1a\n    re = re_1a\n    re2 = re2_1a;\n\n    if (re.test(w)) { w = w.replace(re,\"$1$2\"); }\n    else if (re2.test(w)) { w = w.replace(re2,\"$1$2\"); }\n\n    // Step 1b\n    re = re_1b;\n    re2 = re2_1b;\n    if (re.test(w)) {\n      var fp = re.exec(w);\n      re = re_mgr0;\n      if (re.test(fp[1])) {\n        re = re_1b_2;\n        w = w.replace(re,\"\");\n      }\n    } else if (re2.test(w)) {\n      var fp = re2.exec(w);\n      stem = fp[1];\n      re2 = re_s_v;\n      if (re2.test(stem)) {\n        w = stem;\n        re2 = re2_1b_2;\n        re3 = re3_1b_2;\n        re4 = re4_1b_2;\n        if (re2.test(w)) { w = w + \"e\"; }\n        else if (re3.test(w)) { re = re_1b_2; w = w.replace(re,\"\"); }\n        else if (re4.test(w)) { w = w + \"e\"; }\n      }\n    }\n\n    // Step 1c - replace suffix y or Y by i if preceded by a non-vowel which is not the first letter of the word (so cry -> cri, by -> by, say -> say)\n    re = re_1c;\n    if (re.test(w)) {\n      var fp = re.exec(w);\n      stem = fp[1];\n      w = stem + \"i\";\n    }\n\n    // Step 2\n    re = re_2;\n    if (re.test(w)) {\n      var fp = re.exec(w);\n      stem = fp[1];\n      suffix = fp[2];\n      re = re_mgr0;\n      if (re.test(stem)) {\n        w = stem + step2list[suffix];\n      }\n    }\n\n    // Step 3\n    re = re_3;\n    if (re.test(w)) {\n      var fp = re.exec(w);\n      stem = fp[1];\n      suffix = fp[2];\n      re = re_mgr0;\n      if (re.test(stem)) {\n        w = stem + step3list[suffix];\n      }\n    }\n\n    // Step 4\n    re = re_4;\n    re2 = re2_4;\n    if (re.test(w)) {\n      var fp = re.exec(w);\n      stem = fp[1];\n      re = re_mgr1;\n      if (re.test(stem)) {\n        w = stem;\n      }\n    } else if (re2.test(w)) {\n      var fp = re2.exec(w);\n      stem = fp[1] + fp[2];\n      re2 = re_mgr1;\n      if (re2.test(stem)) {\n        w = stem;\n      }\n    }\n\n    // Step 5\n    re = re_5;\n    if (re.test(w)) {\n      var fp = re.exec(w);\n      stem = fp[1];\n      re = re_mgr1;\n      re2 = re_meq1;\n      re3 = re3_5;\n      if (re.test(stem) || (re2.test(stem) && !(re3.test(stem)))) {\n        w = stem;\n      }\n    }\n\n    re = re_5_1;\n    re2 = re_mgr1;\n    if (re.test(w) && re2.test(w)) {\n      re = re_1b_2;\n      w = w.replace(re,\"\");\n    }\n\n    // and turn initial Y back to y\n\n    if (firstch == \"y\") {\n      w = firstch.toLowerCase() + w.substr(1);\n    }\n\n    return w;\n  };\n\n  return function (token) {\n    return token.update(porterStemmer);\n  }\n})();\n\nlunr.Pipeline.registerFunction(lunr.stemmer, 'stemmer')\n/*!\n * lunr.stopWordFilter\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * lunr.generateStopWordFilter builds a stopWordFilter function from the provided\n * list of stop words.\n *\n * The built in lunr.stopWordFilter is built using this generator and can be used\n * to generate custom stopWordFilters for applications or non English languages.\n *\n * @function\n * @param {Array} token The token to pass through the filter\n * @returns {lunr.PipelineFunction}\n * @see lunr.Pipeline\n * @see lunr.stopWordFilter\n */\nlunr.generateStopWordFilter = function (stopWords) {\n  var words = stopWords.reduce(function (memo, stopWord) {\n    memo[stopWord] = stopWord\n    return memo\n  }, {})\n\n  return function (token) {\n    if (token && words[token.toString()] !== token.toString()) return token\n  }\n}\n\n/**\n * lunr.stopWordFilter is an English language stop word list filter, any words\n * contained in the list will not be passed through the filter.\n *\n * This is intended to be used in the Pipeline. If the token does not pass the\n * filter then undefined will be returned.\n *\n * @function\n * @implements {lunr.PipelineFunction}\n * @params {lunr.Token} token - A token to check for being a stop word.\n * @returns {lunr.Token}\n * @see {@link lunr.Pipeline}\n */\nlunr.stopWordFilter = lunr.generateStopWordFilter([\n  'a',\n  'able',\n  'about',\n  'across',\n  'after',\n  'all',\n  'almost',\n  'also',\n  'am',\n  'among',\n  'an',\n  'and',\n  'any',\n  'are',\n  'as',\n  'at',\n  'be',\n  'because',\n  'been',\n  'but',\n  'by',\n  'can',\n  'cannot',\n  'could',\n  'dear',\n  'did',\n  'do',\n  'does',\n  'either',\n  'else',\n  'ever',\n  'every',\n  'for',\n  'from',\n  'get',\n  'got',\n  'had',\n  'has',\n  'have',\n  'he',\n  'her',\n  'hers',\n  'him',\n  'his',\n  'how',\n  'however',\n  'i',\n  'if',\n  'in',\n  'into',\n  'is',\n  'it',\n  'its',\n  'just',\n  'least',\n  'let',\n  'like',\n  'likely',\n  'may',\n  'me',\n  'might',\n  'most',\n  'must',\n  'my',\n  'neither',\n  'no',\n  'nor',\n  'not',\n  'of',\n  'off',\n  'often',\n  'on',\n  'only',\n  'or',\n  'other',\n  'our',\n  'own',\n  'rather',\n  'said',\n  'say',\n  'says',\n  'she',\n  'should',\n  'since',\n  'so',\n  'some',\n  'than',\n  'that',\n  'the',\n  'their',\n  'them',\n  'then',\n  'there',\n  'these',\n  'they',\n  'this',\n  'tis',\n  'to',\n  'too',\n  'twas',\n  'us',\n  'wants',\n  'was',\n  'we',\n  'were',\n  'what',\n  'when',\n  'where',\n  'which',\n  'while',\n  'who',\n  'whom',\n  'why',\n  'will',\n  'with',\n  'would',\n  'yet',\n  'you',\n  'your'\n])\n\nlunr.Pipeline.registerFunction(lunr.stopWordFilter, 'stopWordFilter')\n/*!\n * lunr.trimmer\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * lunr.trimmer is a pipeline function for trimming non word\n * characters from the beginning and end of tokens before they\n * enter the index.\n *\n * This implementation may not work correctly for non latin\n * characters and should either be removed or adapted for use\n * with languages with non-latin characters.\n *\n * @static\n * @implements {lunr.PipelineFunction}\n * @param {lunr.Token} token The token to pass through the filter\n * @returns {lunr.Token}\n * @see lunr.Pipeline\n */\nlunr.trimmer = function (token) {\n  return token.update(function (s) {\n    return s.replace(/^\\W+/, '').replace(/\\W+$/, '')\n  })\n}\n\nlunr.Pipeline.registerFunction(lunr.trimmer, 'trimmer')\n/*!\n * lunr.TokenSet\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * A token set is used to store the unique list of all tokens\n * within an index. Token sets are also used to represent an\n * incoming query to the index, this query token set and index\n * token set are then intersected to find which tokens to look\n * up in the inverted index.\n *\n * A token set can hold multiple tokens, as in the case of the\n * index token set, or it can hold a single token as in the\n * case of a simple query token set.\n *\n * Additionally token sets are used to perform wildcard matching.\n * Leading, contained and trailing wildcards are supported, and\n * from this edit distance matching can also be provided.\n *\n * Token sets are implemented as a minimal finite state automata,\n * where both common prefixes and suffixes are shared between tokens.\n * This helps to reduce the space used for storing the token set.\n *\n * @constructor\n */\nlunr.TokenSet = function () {\n  this.final = false\n  this.edges = {}\n  this.id = lunr.TokenSet._nextId\n  lunr.TokenSet._nextId += 1\n}\n\n/**\n * Keeps track of the next, auto increment, identifier to assign\n * to a new tokenSet.\n *\n * TokenSets require a unique identifier to be correctly minimised.\n *\n * @private\n */\nlunr.TokenSet._nextId = 1\n\n/**\n * Creates a TokenSet instance from the given sorted array of words.\n *\n * @param {String[]} arr - A sorted array of strings to create the set from.\n * @returns {lunr.TokenSet}\n * @throws Will throw an error if the input array is not sorted.\n */\nlunr.TokenSet.fromArray = function (arr) {\n  var builder = new lunr.TokenSet.Builder\n\n  for (var i = 0, len = arr.length; i < len; i++) {\n    builder.insert(arr[i])\n  }\n\n  builder.finish()\n  return builder.root\n}\n\n/**\n * Creates a token set from a query clause.\n *\n * @private\n * @param {Object} clause - A single clause from lunr.Query.\n * @param {string} clause.term - The query clause term.\n * @param {number} [clause.editDistance] - The optional edit distance for the term.\n * @returns {lunr.TokenSet}\n */\nlunr.TokenSet.fromClause = function (clause) {\n  if ('editDistance' in clause) {\n    return lunr.TokenSet.fromFuzzyString(clause.term, clause.editDistance)\n  } else {\n    return lunr.TokenSet.fromString(clause.term)\n  }\n}\n\n/**\n * Creates a token set representing a single string with a specified\n * edit distance.\n *\n * Insertions, deletions, substitutions and transpositions are each\n * treated as an edit distance of 1.\n *\n * Increasing the allowed edit distance will have a dramatic impact\n * on the performance of both creating and intersecting these TokenSets.\n * It is advised to keep the edit distance less than 3.\n *\n * @param {string} str - The string to create the token set from.\n * @param {number} editDistance - The allowed edit distance to match.\n * @returns {lunr.Vector}\n */\nlunr.TokenSet.fromFuzzyString = function (str, editDistance) {\n  var root = new lunr.TokenSet\n\n  var stack = [{\n    node: root,\n    editsRemaining: editDistance,\n    str: str\n  }]\n\n  while (stack.length) {\n    var frame = stack.pop()\n\n    // no edit\n    if (frame.str.length > 0) {\n      var char = frame.str.charAt(0),\n          noEditNode\n\n      if (char in frame.node.edges) {\n        noEditNode = frame.node.edges[char]\n      } else {\n        noEditNode = new lunr.TokenSet\n        frame.node.edges[char] = noEditNode\n      }\n\n      if (frame.str.length == 1) {\n        noEditNode.final = true\n      }\n\n      stack.push({\n        node: noEditNode,\n        editsRemaining: frame.editsRemaining,\n        str: frame.str.slice(1)\n      })\n    }\n\n    if (frame.editsRemaining == 0) {\n      continue\n    }\n\n    // insertion\n    if (\"*\" in frame.node.edges) {\n      var insertionNode = frame.node.edges[\"*\"]\n    } else {\n      var insertionNode = new lunr.TokenSet\n      frame.node.edges[\"*\"] = insertionNode\n    }\n\n    if (frame.str.length == 0) {\n      insertionNode.final = true\n    }\n\n    stack.push({\n      node: insertionNode,\n      editsRemaining: frame.editsRemaining - 1,\n      str: frame.str\n    })\n\n    // deletion\n    // can only do a deletion if we have enough edits remaining\n    // and if there are characters left to delete in the string\n    if (frame.str.length > 1) {\n      stack.push({\n        node: frame.node,\n        editsRemaining: frame.editsRemaining - 1,\n        str: frame.str.slice(1)\n      })\n    }\n\n    // deletion\n    // just removing the last character from the str\n    if (frame.str.length == 1) {\n      frame.node.final = true\n    }\n\n    // substitution\n    // can only do a substitution if we have enough edits remaining\n    // and if there are characters left to substitute\n    if (frame.str.length >= 1) {\n      if (\"*\" in frame.node.edges) {\n        var substitutionNode = frame.node.edges[\"*\"]\n      } else {\n        var substitutionNode = new lunr.TokenSet\n        frame.node.edges[\"*\"] = substitutionNode\n      }\n\n      if (frame.str.length == 1) {\n        substitutionNode.final = true\n      }\n\n      stack.push({\n        node: substitutionNode,\n        editsRemaining: frame.editsRemaining - 1,\n        str: frame.str.slice(1)\n      })\n    }\n\n    // transposition\n    // can only do a transposition if there are edits remaining\n    // and there are enough characters to transpose\n    if (frame.str.length > 1) {\n      var charA = frame.str.charAt(0),\n          charB = frame.str.charAt(1),\n          transposeNode\n\n      if (charB in frame.node.edges) {\n        transposeNode = frame.node.edges[charB]\n      } else {\n        transposeNode = new lunr.TokenSet\n        frame.node.edges[charB] = transposeNode\n      }\n\n      if (frame.str.length == 1) {\n        transposeNode.final = true\n      }\n\n      stack.push({\n        node: transposeNode,\n        editsRemaining: frame.editsRemaining - 1,\n        str: charA + frame.str.slice(2)\n      })\n    }\n  }\n\n  return root\n}\n\n/**\n * Creates a TokenSet from a string.\n *\n * The string may contain one or more wildcard characters (*)\n * that will allow wildcard matching when intersecting with\n * another TokenSet.\n *\n * @param {string} str - The string to create a TokenSet from.\n * @returns {lunr.TokenSet}\n */\nlunr.TokenSet.fromString = function (str) {\n  var node = new lunr.TokenSet,\n      root = node\n\n  /*\n   * Iterates through all characters within the passed string\n   * appending a node for each character.\n   *\n   * When a wildcard character is found then a self\n   * referencing edge is introduced to continually match\n   * any number of any characters.\n   */\n  for (var i = 0, len = str.length; i < len; i++) {\n    var char = str[i],\n        final = (i == len - 1)\n\n    if (char == \"*\") {\n      node.edges[char] = node\n      node.final = final\n\n    } else {\n      var next = new lunr.TokenSet\n      next.final = final\n\n      node.edges[char] = next\n      node = next\n    }\n  }\n\n  return root\n}\n\n/**\n * Converts this TokenSet into an array of strings\n * contained within the TokenSet.\n *\n * This is not intended to be used on a TokenSet that\n * contains wildcards, in these cases the results are\n * undefined and are likely to cause an infinite loop.\n *\n * @returns {string[]}\n */\nlunr.TokenSet.prototype.toArray = function () {\n  var words = []\n\n  var stack = [{\n    prefix: \"\",\n    node: this\n  }]\n\n  while (stack.length) {\n    var frame = stack.pop(),\n        edges = Object.keys(frame.node.edges),\n        len = edges.length\n\n    if (frame.node.final) {\n      /* In Safari, at this point the prefix is sometimes corrupted, see:\n       * https://github.com/olivernn/lunr.js/issues/279 Calling any\n       * String.prototype method forces Safari to \"cast\" this string to what\n       * it's supposed to be, fixing the bug. */\n      frame.prefix.charAt(0)\n      words.push(frame.prefix)\n    }\n\n    for (var i = 0; i < len; i++) {\n      var edge = edges[i]\n\n      stack.push({\n        prefix: frame.prefix.concat(edge),\n        node: frame.node.edges[edge]\n      })\n    }\n  }\n\n  return words\n}\n\n/**\n * Generates a string representation of a TokenSet.\n *\n * This is intended to allow TokenSets to be used as keys\n * in objects, largely to aid the construction and minimisation\n * of a TokenSet. As such it is not designed to be a human\n * friendly representation of the TokenSet.\n *\n * @returns {string}\n */\nlunr.TokenSet.prototype.toString = function () {\n  // NOTE: Using Object.keys here as this.edges is very likely\n  // to enter 'hash-mode' with many keys being added\n  //\n  // avoiding a for-in loop here as it leads to the function\n  // being de-optimised (at least in V8). From some simple\n  // benchmarks the performance is comparable, but allowing\n  // V8 to optimize may mean easy performance wins in the future.\n\n  if (this._str) {\n    return this._str\n  }\n\n  var str = this.final ? '1' : '0',\n      labels = Object.keys(this.edges).sort(),\n      len = labels.length\n\n  for (var i = 0; i < len; i++) {\n    var label = labels[i],\n        node = this.edges[label]\n\n    str = str + label + node.id\n  }\n\n  return str\n}\n\n/**\n * Returns a new TokenSet that is the intersection of\n * this TokenSet and the passed TokenSet.\n *\n * This intersection will take into account any wildcards\n * contained within the TokenSet.\n *\n * @param {lunr.TokenSet} b - An other TokenSet to intersect with.\n * @returns {lunr.TokenSet}\n */\nlunr.TokenSet.prototype.intersect = function (b) {\n  var output = new lunr.TokenSet,\n      frame = undefined\n\n  var stack = [{\n    qNode: b,\n    output: output,\n    node: this\n  }]\n\n  while (stack.length) {\n    frame = stack.pop()\n\n    // NOTE: As with the #toString method, we are using\n    // Object.keys and a for loop instead of a for-in loop\n    // as both of these objects enter 'hash' mode, causing\n    // the function to be de-optimised in V8\n    var qEdges = Object.keys(frame.qNode.edges),\n        qLen = qEdges.length,\n        nEdges = Object.keys(frame.node.edges),\n        nLen = nEdges.length\n\n    for (var q = 0; q < qLen; q++) {\n      var qEdge = qEdges[q]\n\n      for (var n = 0; n < nLen; n++) {\n        var nEdge = nEdges[n]\n\n        if (nEdge == qEdge || qEdge == '*') {\n          var node = frame.node.edges[nEdge],\n              qNode = frame.qNode.edges[qEdge],\n              final = node.final && qNode.final,\n              next = undefined\n\n          if (nEdge in frame.output.edges) {\n            // an edge already exists for this character\n            // no need to create a new node, just set the finality\n            // bit unless this node is already final\n            next = frame.output.edges[nEdge]\n            next.final = next.final || final\n\n          } else {\n            // no edge exists yet, must create one\n            // set the finality bit and insert it\n            // into the output\n            next = new lunr.TokenSet\n            next.final = final\n            frame.output.edges[nEdge] = next\n          }\n\n          stack.push({\n            qNode: qNode,\n            output: next,\n            node: node\n          })\n        }\n      }\n    }\n  }\n\n  return output\n}\nlunr.TokenSet.Builder = function () {\n  this.previousWord = \"\"\n  this.root = new lunr.TokenSet\n  this.uncheckedNodes = []\n  this.minimizedNodes = {}\n}\n\nlunr.TokenSet.Builder.prototype.insert = function (word) {\n  var node,\n      commonPrefix = 0\n\n  if (word < this.previousWord) {\n    throw new Error (\"Out of order word insertion\")\n  }\n\n  for (var i = 0; i < word.length && i < this.previousWord.length; i++) {\n    if (word[i] != this.previousWord[i]) break\n    commonPrefix++\n  }\n\n  this.minimize(commonPrefix)\n\n  if (this.uncheckedNodes.length == 0) {\n    node = this.root\n  } else {\n    node = this.uncheckedNodes[this.uncheckedNodes.length - 1].child\n  }\n\n  for (var i = commonPrefix; i < word.length; i++) {\n    var nextNode = new lunr.TokenSet,\n        char = word[i]\n\n    node.edges[char] = nextNode\n\n    this.uncheckedNodes.push({\n      parent: node,\n      char: char,\n      child: nextNode\n    })\n\n    node = nextNode\n  }\n\n  node.final = true\n  this.previousWord = word\n}\n\nlunr.TokenSet.Builder.prototype.finish = function () {\n  this.minimize(0)\n}\n\nlunr.TokenSet.Builder.prototype.minimize = function (downTo) {\n  for (var i = this.uncheckedNodes.length - 1; i >= downTo; i--) {\n    var node = this.uncheckedNodes[i],\n        childKey = node.child.toString()\n\n    if (childKey in this.minimizedNodes) {\n      node.parent.edges[node.char] = this.minimizedNodes[childKey]\n    } else {\n      // Cache the key for this node since\n      // we know it can't change anymore\n      node.child._str = childKey\n\n      this.minimizedNodes[childKey] = node.child\n    }\n\n    this.uncheckedNodes.pop()\n  }\n}\n/*!\n * lunr.Index\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * An index contains the built index of all documents and provides a query interface\n * to the index.\n *\n * Usually instances of lunr.Index will not be created using this constructor, instead\n * lunr.Builder should be used to construct new indexes, or lunr.Index.load should be\n * used to load previously built and serialized indexes.\n *\n * @constructor\n * @param {Object} attrs - The attributes of the built search index.\n * @param {Object} attrs.invertedIndex - An index of term/field to document reference.\n * @param {Object<string, lunr.Vector>} attrs.fieldVectors - Field vectors\n * @param {lunr.TokenSet} attrs.tokenSet - An set of all corpus tokens.\n * @param {string[]} attrs.fields - The names of indexed document fields.\n * @param {lunr.Pipeline} attrs.pipeline - The pipeline to use for search terms.\n */\nlunr.Index = function (attrs) {\n  this.invertedIndex = attrs.invertedIndex\n  this.fieldVectors = attrs.fieldVectors\n  this.tokenSet = attrs.tokenSet\n  this.fields = attrs.fields\n  this.pipeline = attrs.pipeline\n}\n\n/**\n * A result contains details of a document matching a search query.\n * @typedef {Object} lunr.Index~Result\n * @property {string} ref - The reference of the document this result represents.\n * @property {number} score - A number between 0 and 1 representing how similar this document is to the query.\n * @property {lunr.MatchData} matchData - Contains metadata about this match including which term(s) caused the match.\n */\n\n/**\n * Although lunr provides the ability to create queries using lunr.Query, it also provides a simple\n * query language which itself is parsed into an instance of lunr.Query.\n *\n * For programmatically building queries it is advised to directly use lunr.Query, the query language\n * is best used for human entered text rather than program generated text.\n *\n * At its simplest queries can just be a single term, e.g. `hello`, multiple terms are also supported\n * and will be combined with OR, e.g `hello world` will match documents that contain either 'hello'\n * or 'world', though those that contain both will rank higher in the results.\n *\n * Wildcards can be included in terms to match one or more unspecified characters, these wildcards can\n * be inserted anywhere within the term, and more than one wildcard can exist in a single term. Adding\n * wildcards will increase the number of documents that will be found but can also have a negative\n * impact on query performance, especially with wildcards at the beginning of a term.\n *\n * Terms can be restricted to specific fields, e.g. `title:hello`, only documents with the term\n * hello in the title field will match this query. Using a field not present in the index will lead\n * to an error being thrown.\n *\n * Modifiers can also be added to terms, lunr supports edit distance and boost modifiers on terms. A term\n * boost will make documents matching that term score higher, e.g. `foo^5`. Edit distance is also supported\n * to provide fuzzy matching, e.g. 'hello~2' will match documents with hello with an edit distance of 2.\n * Avoid large values for edit distance to improve query performance.\n *\n * Each term also supports a presence modifier. By default a term's presence in document is optional, however\n * this can be changed to either required or prohibited. For a term's presence to be required in a document the\n * term should be prefixed with a '+', e.g. `+foo bar` is a search for documents that must contain 'foo' and\n * optionally contain 'bar'. Conversely a leading '-' sets the terms presence to prohibited, i.e. it must not\n * appear in a document, e.g. `-foo bar` is a search for documents that do not contain 'foo' but may contain 'bar'.\n *\n * To escape special characters the backslash character '\\' can be used, this allows searches to include\n * characters that would normally be considered modifiers, e.g. `foo\\~2` will search for a term \"foo~2\" instead\n * of attempting to apply a boost of 2 to the search term \"foo\".\n *\n * @typedef {string} lunr.Index~QueryString\n * @example <caption>Simple single term query</caption>\n * hello\n * @example <caption>Multiple term query</caption>\n * hello world\n * @example <caption>term scoped to a field</caption>\n * title:hello\n * @example <caption>term with a boost of 10</caption>\n * hello^10\n * @example <caption>term with an edit distance of 2</caption>\n * hello~2\n * @example <caption>terms with presence modifiers</caption>\n * -foo +bar baz\n */\n\n/**\n * Performs a search against the index using lunr query syntax.\n *\n * Results will be returned sorted by their score, the most relevant results\n * will be returned first.  For details on how the score is calculated, please see\n * the {@link https://lunrjs.com/guides/searching.html#scoring|guide}.\n *\n * For more programmatic querying use lunr.Index#query.\n *\n * @param {lunr.Index~QueryString} queryString - A string containing a lunr query.\n * @throws {lunr.QueryParseError} If the passed query string cannot be parsed.\n * @returns {lunr.Index~Result[]}\n */\nlunr.Index.prototype.search = function (queryString) {\n  return this.query(function (query) {\n    var parser = new lunr.QueryParser(queryString, query)\n    parser.parse()\n  })\n}\n\n/**\n * A query builder callback provides a query object to be used to express\n * the query to perform on the index.\n *\n * @callback lunr.Index~queryBuilder\n * @param {lunr.Query} query - The query object to build up.\n * @this lunr.Query\n */\n\n/**\n * Performs a query against the index using the yielded lunr.Query object.\n *\n * If performing programmatic queries against the index, this method is preferred\n * over lunr.Index#search so as to avoid the additional query parsing overhead.\n *\n * A query object is yielded to the supplied function which should be used to\n * express the query to be run against the index.\n *\n * Note that although this function takes a callback parameter it is _not_ an\n * asynchronous operation, the callback is just yielded a query object to be\n * customized.\n *\n * @param {lunr.Index~queryBuilder} fn - A function that is used to build the query.\n * @returns {lunr.Index~Result[]}\n */\nlunr.Index.prototype.query = function (fn) {\n  // for each query clause\n  // * process terms\n  // * expand terms from token set\n  // * find matching documents and metadata\n  // * get document vectors\n  // * score documents\n\n  var query = new lunr.Query(this.fields),\n      matchingFields = Object.create(null),\n      queryVectors = Object.create(null),\n      termFieldCache = Object.create(null),\n      requiredMatches = Object.create(null),\n      prohibitedMatches = Object.create(null)\n\n  /*\n   * To support field level boosts a query vector is created per\n   * field. An empty vector is eagerly created to support negated\n   * queries.\n   */\n  for (var i = 0; i < this.fields.length; i++) {\n    queryVectors[this.fields[i]] = new lunr.Vector\n  }\n\n  fn.call(query, query)\n\n  for (var i = 0; i < query.clauses.length; i++) {\n    /*\n     * Unless the pipeline has been disabled for this term, which is\n     * the case for terms with wildcards, we need to pass the clause\n     * term through the search pipeline. A pipeline returns an array\n     * of processed terms. Pipeline functions may expand the passed\n     * term, which means we may end up performing multiple index lookups\n     * for a single query term.\n     */\n    var clause = query.clauses[i],\n        terms = null,\n        clauseMatches = lunr.Set.empty\n\n    if (clause.usePipeline) {\n      terms = this.pipeline.runString(clause.term, {\n        fields: clause.fields\n      })\n    } else {\n      terms = [clause.term]\n    }\n\n    for (var m = 0; m < terms.length; m++) {\n      var term = terms[m]\n\n      /*\n       * Each term returned from the pipeline needs to use the same query\n       * clause object, e.g. the same boost and or edit distance. The\n       * simplest way to do this is to re-use the clause object but mutate\n       * its term property.\n       */\n      clause.term = term\n\n      /*\n       * From the term in the clause we create a token set which will then\n       * be used to intersect the indexes token set to get a list of terms\n       * to lookup in the inverted index\n       */\n      var termTokenSet = lunr.TokenSet.fromClause(clause),\n          expandedTerms = this.tokenSet.intersect(termTokenSet).toArray()\n\n      /*\n       * If a term marked as required does not exist in the tokenSet it is\n       * impossible for the search to return any matches. We set all the field\n       * scoped required matches set to empty and stop examining any further\n       * clauses.\n       */\n      if (expandedTerms.length === 0 && clause.presence === lunr.Query.presence.REQUIRED) {\n        for (var k = 0; k < clause.fields.length; k++) {\n          var field = clause.fields[k]\n          requiredMatches[field] = lunr.Set.empty\n        }\n\n        break\n      }\n\n      for (var j = 0; j < expandedTerms.length; j++) {\n        /*\n         * For each term get the posting and termIndex, this is required for\n         * building the query vector.\n         */\n        var expandedTerm = expandedTerms[j],\n            posting = this.invertedIndex[expandedTerm],\n            termIndex = posting._index\n\n        for (var k = 0; k < clause.fields.length; k++) {\n          /*\n           * For each field that this query term is scoped by (by default\n           * all fields are in scope) we need to get all the document refs\n           * that have this term in that field.\n           *\n           * The posting is the entry in the invertedIndex for the matching\n           * term from above.\n           */\n          var field = clause.fields[k],\n              fieldPosting = posting[field],\n              matchingDocumentRefs = Object.keys(fieldPosting),\n              termField = expandedTerm + \"/\" + field,\n              matchingDocumentsSet = new lunr.Set(matchingDocumentRefs)\n\n          /*\n           * if the presence of this term is required ensure that the matching\n           * documents are added to the set of required matches for this clause.\n           *\n           */\n          if (clause.presence == lunr.Query.presence.REQUIRED) {\n            clauseMatches = clauseMatches.union(matchingDocumentsSet)\n\n            if (requiredMatches[field] === undefined) {\n              requiredMatches[field] = lunr.Set.complete\n            }\n          }\n\n          /*\n           * if the presence of this term is prohibited ensure that the matching\n           * documents are added to the set of prohibited matches for this field,\n           * creating that set if it does not yet exist.\n           */\n          if (clause.presence == lunr.Query.presence.PROHIBITED) {\n            if (prohibitedMatches[field] === undefined) {\n              prohibitedMatches[field] = lunr.Set.empty\n            }\n\n            prohibitedMatches[field] = prohibitedMatches[field].union(matchingDocumentsSet)\n\n            /*\n             * Prohibited matches should not be part of the query vector used for\n             * similarity scoring and no metadata should be extracted so we continue\n             * to the next field\n             */\n            continue\n          }\n\n          /*\n           * The query field vector is populated using the termIndex found for\n           * the term and a unit value with the appropriate boost applied.\n           * Using upsert because there could already be an entry in the vector\n           * for the term we are working with. In that case we just add the scores\n           * together.\n           */\n          queryVectors[field].upsert(termIndex, clause.boost, function (a, b) { return a + b })\n\n          /**\n           * If we've already seen this term, field combo then we've already collected\n           * the matching documents and metadata, no need to go through all that again\n           */\n          if (termFieldCache[termField]) {\n            continue\n          }\n\n          for (var l = 0; l < matchingDocumentRefs.length; l++) {\n            /*\n             * All metadata for this term/field/document triple\n             * are then extracted and collected into an instance\n             * of lunr.MatchData ready to be returned in the query\n             * results\n             */\n            var matchingDocumentRef = matchingDocumentRefs[l],\n                matchingFieldRef = new lunr.FieldRef (matchingDocumentRef, field),\n                metadata = fieldPosting[matchingDocumentRef],\n                fieldMatch\n\n            if ((fieldMatch = matchingFields[matchingFieldRef]) === undefined) {\n              matchingFields[matchingFieldRef] = new lunr.MatchData (expandedTerm, field, metadata)\n            } else {\n              fieldMatch.add(expandedTerm, field, metadata)\n            }\n\n          }\n\n          termFieldCache[termField] = true\n        }\n      }\n    }\n\n    /**\n     * If the presence was required we need to update the requiredMatches field sets.\n     * We do this after all fields for the term have collected their matches because\n     * the clause terms presence is required in _any_ of the fields not _all_ of the\n     * fields.\n     */\n    if (clause.presence === lunr.Query.presence.REQUIRED) {\n      for (var k = 0; k < clause.fields.length; k++) {\n        var field = clause.fields[k]\n        requiredMatches[field] = requiredMatches[field].intersect(clauseMatches)\n      }\n    }\n  }\n\n  /**\n   * Need to combine the field scoped required and prohibited\n   * matching documents into a global set of required and prohibited\n   * matches\n   */\n  var allRequiredMatches = lunr.Set.complete,\n      allProhibitedMatches = lunr.Set.empty\n\n  for (var i = 0; i < this.fields.length; i++) {\n    var field = this.fields[i]\n\n    if (requiredMatches[field]) {\n      allRequiredMatches = allRequiredMatches.intersect(requiredMatches[field])\n    }\n\n    if (prohibitedMatches[field]) {\n      allProhibitedMatches = allProhibitedMatches.union(prohibitedMatches[field])\n    }\n  }\n\n  var matchingFieldRefs = Object.keys(matchingFields),\n      results = [],\n      matches = Object.create(null)\n\n  /*\n   * If the query is negated (contains only prohibited terms)\n   * we need to get _all_ fieldRefs currently existing in the\n   * index. This is only done when we know that the query is\n   * entirely prohibited terms to avoid any cost of getting all\n   * fieldRefs unnecessarily.\n   *\n   * Additionally, blank MatchData must be created to correctly\n   * populate the results.\n   */\n  if (query.isNegated()) {\n    matchingFieldRefs = Object.keys(this.fieldVectors)\n\n    for (var i = 0; i < matchingFieldRefs.length; i++) {\n      var matchingFieldRef = matchingFieldRefs[i]\n      var fieldRef = lunr.FieldRef.fromString(matchingFieldRef)\n      matchingFields[matchingFieldRef] = new lunr.MatchData\n    }\n  }\n\n  for (var i = 0; i < matchingFieldRefs.length; i++) {\n    /*\n     * Currently we have document fields that match the query, but we\n     * need to return documents. The matchData and scores are combined\n     * from multiple fields belonging to the same document.\n     *\n     * Scores are calculated by field, using the query vectors created\n     * above, and combined into a final document score using addition.\n     */\n    var fieldRef = lunr.FieldRef.fromString(matchingFieldRefs[i]),\n        docRef = fieldRef.docRef\n\n    if (!allRequiredMatches.contains(docRef)) {\n      continue\n    }\n\n    if (allProhibitedMatches.contains(docRef)) {\n      continue\n    }\n\n    var fieldVector = this.fieldVectors[fieldRef],\n        score = queryVectors[fieldRef.fieldName].similarity(fieldVector),\n        docMatch\n\n    if ((docMatch = matches[docRef]) !== undefined) {\n      docMatch.score += score\n      docMatch.matchData.combine(matchingFields[fieldRef])\n    } else {\n      var match = {\n        ref: docRef,\n        score: score,\n        matchData: matchingFields[fieldRef]\n      }\n      matches[docRef] = match\n      results.push(match)\n    }\n  }\n\n  /*\n   * Sort the results objects by score, highest first.\n   */\n  return results.sort(function (a, b) {\n    return b.score - a.score\n  })\n}\n\n/**\n * Prepares the index for JSON serialization.\n *\n * The schema for this JSON blob will be described in a\n * separate JSON schema file.\n *\n * @returns {Object}\n */\nlunr.Index.prototype.toJSON = function () {\n  var invertedIndex = Object.keys(this.invertedIndex)\n    .sort()\n    .map(function (term) {\n      return [term, this.invertedIndex[term]]\n    }, this)\n\n  var fieldVectors = Object.keys(this.fieldVectors)\n    .map(function (ref) {\n      return [ref, this.fieldVectors[ref].toJSON()]\n    }, this)\n\n  return {\n    version: lunr.version,\n    fields: this.fields,\n    fieldVectors: fieldVectors,\n    invertedIndex: invertedIndex,\n    pipeline: this.pipeline.toJSON()\n  }\n}\n\n/**\n * Loads a previously serialized lunr.Index\n *\n * @param {Object} serializedIndex - A previously serialized lunr.Index\n * @returns {lunr.Index}\n */\nlunr.Index.load = function (serializedIndex) {\n  var attrs = {},\n      fieldVectors = {},\n      serializedVectors = serializedIndex.fieldVectors,\n      invertedIndex = Object.create(null),\n      serializedInvertedIndex = serializedIndex.invertedIndex,\n      tokenSetBuilder = new lunr.TokenSet.Builder,\n      pipeline = lunr.Pipeline.load(serializedIndex.pipeline)\n\n  if (serializedIndex.version != lunr.version) {\n    lunr.utils.warn(\"Version mismatch when loading serialised index. Current version of lunr '\" + lunr.version + \"' does not match serialized index '\" + serializedIndex.version + \"'\")\n  }\n\n  for (var i = 0; i < serializedVectors.length; i++) {\n    var tuple = serializedVectors[i],\n        ref = tuple[0],\n        elements = tuple[1]\n\n    fieldVectors[ref] = new lunr.Vector(elements)\n  }\n\n  for (var i = 0; i < serializedInvertedIndex.length; i++) {\n    var tuple = serializedInvertedIndex[i],\n        term = tuple[0],\n        posting = tuple[1]\n\n    tokenSetBuilder.insert(term)\n    invertedIndex[term] = posting\n  }\n\n  tokenSetBuilder.finish()\n\n  attrs.fields = serializedIndex.fields\n\n  attrs.fieldVectors = fieldVectors\n  attrs.invertedIndex = invertedIndex\n  attrs.tokenSet = tokenSetBuilder.root\n  attrs.pipeline = pipeline\n\n  return new lunr.Index(attrs)\n}\n/*!\n * lunr.Builder\n * Copyright (C) 2020 Oliver Nightingale\n */\n\n/**\n * lunr.Builder performs indexing on a set of documents and\n * returns instances of lunr.Index ready for querying.\n *\n * All configuration of the index is done via the builder, the\n * fields to index, the document reference, the text processing\n * pipeline and document scoring parameters are all set on the\n * builder before indexing.\n *\n * @constructor\n * @property {string} _ref - Internal reference to the document reference field.\n * @property {string[]} _fields - Internal reference to the document fields to index.\n * @property {object} invertedIndex - The inverted index maps terms to document fields.\n * @property {object} documentTermFrequencies - Keeps track of document term frequencies.\n * @property {object} documentLengths - Keeps track of the length of documents added to the index.\n * @property {lunr.tokenizer} tokenizer - Function for splitting strings into tokens for indexing.\n * @property {lunr.Pipeline} pipeline - The pipeline performs text processing on tokens before indexing.\n * @property {lunr.Pipeline} searchPipeline - A pipeline for processing search terms before querying the index.\n * @property {number} documentCount - Keeps track of the total number of documents indexed.\n * @property {number} _b - A parameter to control field length normalization, setting this to 0 disabled normalization, 1 fully normalizes field lengths, the default value is 0.75.\n * @property {number} _k1 - A parameter to control how quickly an increase in term frequency results in term frequency saturation, the default value is 1.2.\n * @property {number} termIndex - A counter incremented for each unique term, used to identify a terms position in the vector space.\n * @property {array} metadataWhitelist - A list of metadata keys that have been whitelisted for entry in the index.\n */\nlunr.Builder = function () {\n  this._ref = \"id\"\n  this._fields = Object.create(null)\n  this._documents = Object.create(null)\n  this.invertedIndex = Object.create(null)\n  this.fieldTermFrequencies = {}\n  this.fieldLengths = {}\n  this.tokenizer = lunr.tokenizer\n  this.pipeline = new lunr.Pipeline\n  this.searchPipeline = new lunr.Pipeline\n  this.documentCount = 0\n  this._b = 0.75\n  this._k1 = 1.2\n  this.termIndex = 0\n  this.metadataWhitelist = []\n}\n\n/**\n * Sets the document field used as the document reference. Every document must have this field.\n * The type of this field in the document should be a string, if it is not a string it will be\n * coerced into a string by calling toString.\n *\n * The default ref is 'id'.\n *\n * The ref should _not_ be changed during indexing, it should be set before any documents are\n * added to the index. Changing it during indexing can lead to inconsistent results.\n *\n * @param {string} ref - The name of the reference field in the document.\n */\nlunr.Builder.prototype.ref = function (ref) {\n  this._ref = ref\n}\n\n/**\n * A function that is used to extract a field from a document.\n *\n * Lunr expects a field to be at the top level of a document, if however the field\n * is deeply nested within a document an extractor function can be used to extract\n * the right field for indexing.\n *\n * @callback fieldExtractor\n * @param {object} doc - The document being added to the index.\n * @returns {?(string|object|object[])} obj - The object that will be indexed for this field.\n * @example <caption>Extracting a nested field</caption>\n * function (doc) { return doc.nested.field }\n */\n\n/**\n * Adds a field to the list of document fields that will be indexed. Every document being\n * indexed should have this field. Null values for this field in indexed documents will\n * not cause errors but will limit the chance of that document being retrieved by searches.\n *\n * All fields should be added before adding documents to the index. Adding fields after\n * a document has been indexed will have no effect on already indexed documents.\n *\n * Fields can be boosted at build time. This allows terms within that field to have more\n * importance when ranking search results. Use a field boost to specify that matches within\n * one field are more important than other fields.\n *\n * @param {string} fieldName - The name of a field to index in all documents.\n * @param {object} attributes - Optional attributes associated with this field.\n * @param {number} [attributes.boost=1] - Boost applied to all terms within this field.\n * @param {fieldExtractor} [attributes.extractor] - Function to extract a field from a document.\n * @throws {RangeError} fieldName cannot contain unsupported characters '/'\n */\nlunr.Builder.prototype.field = function (fieldName, attributes) {\n  if (/\\//.test(fieldName)) {\n    throw new RangeError (\"Field '\" + fieldName + \"' contains illegal character '/'\")\n  }\n\n  this._fields[fieldName] = attributes || {}\n}\n\n/**\n * A parameter to tune the amount of field length normalisation that is applied when\n * calculating relevance scores. A value of 0 will completely disable any normalisation\n * and a value of 1 will fully normalise field lengths. The default is 0.75. Values of b\n * will be clamped to the range 0 - 1.\n *\n * @param {number} number - The value to set for this tuning parameter.\n */\nlunr.Builder.prototype.b = function (number) {\n  if (number < 0) {\n    this._b = 0\n  } else if (number > 1) {\n    this._b = 1\n  } else {\n    this._b = number\n  }\n}\n\n/**\n * A parameter that controls the speed at which a rise in term frequency results in term\n * frequency saturation. The default value is 1.2. Setting this to a higher value will give\n * slower saturation levels, a lower value will result in quicker saturation.\n *\n * @param {number} number - The value to set for this tuning parameter.\n */\nlunr.Builder.prototype.k1 = function (number) {\n  this._k1 = number\n}\n\n/**\n * Adds a document to the index.\n *\n * Before adding fields to the index the index should have been fully setup, with the document\n * ref and all fields to index already having been specified.\n *\n * The document must have a field name as specified by the ref (by default this is 'id') and\n * it should have all fields defined for indexing, though null or undefined values will not\n * cause errors.\n *\n * Entire documents can be boosted at build time. Applying a boost to a document indicates that\n * this document should rank higher in search results than other documents.\n *\n * @param {object} doc - The document to add to the index.\n * @param {object} attributes - Optional attributes associated with this document.\n * @param {number} [attributes.boost=1] - Boost applied to all terms within this document.\n */\nlunr.Builder.prototype.add = function (doc, attributes) {\n  var docRef = doc[this._ref],\n      fields = Object.keys(this._fields)\n\n  this._documents[docRef] = attributes || {}\n  this.documentCount += 1\n\n  for (var i = 0; i < fields.length; i++) {\n    var fieldName = fields[i],\n        extractor = this._fields[fieldName].extractor,\n        field = extractor ? extractor(doc) : doc[fieldName],\n        tokens = this.tokenizer(field, {\n          fields: [fieldName]\n        }),\n        terms = this.pipeline.run(tokens),\n        fieldRef = new lunr.FieldRef (docRef, fieldName),\n        fieldTerms = Object.create(null)\n\n    this.fieldTermFrequencies[fieldRef] = fieldTerms\n    this.fieldLengths[fieldRef] = 0\n\n    // store the length of this field for this document\n    this.fieldLengths[fieldRef] += terms.length\n\n    // calculate term frequencies for this field\n    for (var j = 0; j < terms.length; j++) {\n      var term = terms[j]\n\n      if (fieldTerms[term] == undefined) {\n        fieldTerms[term] = 0\n      }\n\n      fieldTerms[term] += 1\n\n      // add to inverted index\n      // create an initial posting if one doesn't exist\n      if (this.invertedIndex[term] == undefined) {\n        var posting = Object.create(null)\n        posting[\"_index\"] = this.termIndex\n        this.termIndex += 1\n\n        for (var k = 0; k < fields.length; k++) {\n          posting[fields[k]] = Object.create(null)\n        }\n\n        this.invertedIndex[term] = posting\n      }\n\n      // add an entry for this term/fieldName/docRef to the invertedIndex\n      if (this.invertedIndex[term][fieldName][docRef] == undefined) {\n        this.invertedIndex[term][fieldName][docRef] = Object.create(null)\n      }\n\n      // store all whitelisted metadata about this token in the\n      // inverted index\n      for (var l = 0; l < this.metadataWhitelist.length; l++) {\n        var metadataKey = this.metadataWhitelist[l],\n            metadata = term.metadata[metadataKey]\n\n        if (this.invertedIndex[term][fieldName][docRef][metadataKey] == undefined) {\n          this.invertedIndex[term][fieldName][docRef][metadataKey] = []\n        }\n\n        this.invertedIndex[term][fieldName][docRef][metadataKey].push(metadata)\n      }\n    }\n\n  }\n}\n\n/**\n * Calculates the average document length for this index\n *\n * @private\n */\nlunr.Builder.prototype.calculateAverageFieldLengths = function () {\n\n  var fieldRefs = Object.keys(this.fieldLengths),\n      numberOfFields = fieldRefs.length,\n      accumulator = {},\n      documentsWithField = {}\n\n  for (var i = 0; i < numberOfFields; i++) {\n    var fieldRef = lunr.FieldRef.fromString(fieldRefs[i]),\n        field = fieldRef.fieldName\n\n    documentsWithField[field] || (documentsWithField[field] = 0)\n    documentsWithField[field] += 1\n\n    accumulator[field] || (accumulator[field] = 0)\n    accumulator[field] += this.fieldLengths[fieldRef]\n  }\n\n  var fields = Object.keys(this._fields)\n\n  for (var i = 0; i < fields.length; i++) {\n    var fieldName = fields[i]\n    accumulator[fieldName] = accumulator[fieldName] / documentsWithField[fieldName]\n  }\n\n  this.averageFieldLength = accumulator\n}\n\n/**\n * Builds a vector space model of every document using lunr.Vector\n *\n * @private\n */\nlunr.Builder.prototype.createFieldVectors = function () {\n  var fieldVectors = {},\n      fieldRefs = Object.keys(this.fieldTermFrequencies),\n      fieldRefsLength = fieldRefs.length,\n      termIdfCache = Object.create(null)\n\n  for (var i = 0; i < fieldRefsLength; i++) {\n    var fieldRef = lunr.FieldRef.fromString(fieldRefs[i]),\n        fieldName = fieldRef.fieldName,\n        fieldLength = this.fieldLengths[fieldRef],\n        fieldVector = new lunr.Vector,\n        termFrequencies = this.fieldTermFrequencies[fieldRef],\n        terms = Object.keys(termFrequencies),\n        termsLength = terms.length\n\n\n    var fieldBoost = this._fields[fieldName].boost || 1,\n        docBoost = this._documents[fieldRef.docRef].boost || 1\n\n    for (var j = 0; j < termsLength; j++) {\n      var term = terms[j],\n          tf = termFrequencies[term],\n          termIndex = this.invertedIndex[term]._index,\n          idf, score, scoreWithPrecision\n\n      if (termIdfCache[term] === undefined) {\n        idf = lunr.idf(this.invertedIndex[term], this.documentCount)\n        termIdfCache[term] = idf\n      } else {\n        idf = termIdfCache[term]\n      }\n\n      score = idf * ((this._k1 + 1) * tf) / (this._k1 * (1 - this._b + this._b * (fieldLength / this.averageFieldLength[fieldName])) + tf)\n      score *= fieldBoost\n      score *= docBoost\n      scoreWithPrecision = Math.round(score * 1000) / 1000\n      // Converts 1.23456789 to 1.234.\n      // Reducing the precision so that the vectors take up less\n      // space when serialised. Doing it now so that they behave\n      // the same before and after serialisation. Also, this is\n      // the fastest approach to reducing a number's precision in\n      // JavaScript.\n\n      fieldVector.insert(termIndex, scoreWithPrecision)\n    }\n\n    fieldVectors[fieldRef] = fieldVector\n  }\n\n  this.fieldVectors = fieldVectors\n}\n\n/**\n * Creates a token set of all tokens in the index using lunr.TokenSet\n *\n * @private\n */\nlunr.Builder.prototype.createTokenSet = function () {\n  this.tokenSet = lunr.TokenSet.fromArray(\n    Object.keys(this.invertedIndex).sort()\n  )\n}\n\n/**\n * Builds the index, creating an instance of lunr.Index.\n *\n * This completes the indexing process and should only be called\n * once all documents have been added to the index.\n *\n * @returns {lunr.Index}\n */\nlunr.Builder.prototype.build = function () {\n  this.calculateAverageFieldLengths()\n  this.createFieldVectors()\n  this.createTokenSet()\n\n  return new lunr.Index({\n    invertedIndex: this.invertedIndex,\n    fieldVectors: this.fieldVectors,\n    tokenSet: this.tokenSet,\n    fields: Object.keys(this._fields),\n    pipeline: this.searchPipeline\n  })\n}\n\n/**\n * Applies a plugin to the index builder.\n *\n * A plugin is a function that is called with the index builder as its context.\n * Plugins can be used to customise or extend the behaviour of the index\n * in some way. A plugin is just a function, that encapsulated the custom\n * behaviour that should be applied when building the index.\n *\n * The plugin function will be called with the index builder as its argument, additional\n * arguments can also be passed when calling use. The function will be called\n * with the index builder as its context.\n *\n * @param {Function} plugin The plugin to apply.\n */\nlunr.Builder.prototype.use = function (fn) {\n  var args = Array.prototype.slice.call(arguments, 1)\n  args.unshift(this)\n  fn.apply(this, args)\n}\n/**\n * Contains and collects metadata about a matching document.\n * A single instance of lunr.MatchData is returned as part of every\n * lunr.Index~Result.\n *\n * @constructor\n * @param {string} term - The term this match data is associated with\n * @param {string} field - The field in which the term was found\n * @param {object} metadata - The metadata recorded about this term in this field\n * @property {object} metadata - A cloned collection of metadata associated with this document.\n * @see {@link lunr.Index~Result}\n */\nlunr.MatchData = function (term, field, metadata) {\n  var clonedMetadata = Object.create(null),\n      metadataKeys = Object.keys(metadata || {})\n\n  // Cloning the metadata to prevent the original\n  // being mutated during match data combination.\n  // Metadata is kept in an array within the inverted\n  // index so cloning the data can be done with\n  // Array#slice\n  for (var i = 0; i < metadataKeys.length; i++) {\n    var key = metadataKeys[i]\n    clonedMetadata[key] = metadata[key].slice()\n  }\n\n  this.metadata = Object.create(null)\n\n  if (term !== undefined) {\n    this.metadata[term] = Object.create(null)\n    this.metadata[term][field] = clonedMetadata\n  }\n}\n\n/**\n * An instance of lunr.MatchData will be created for every term that matches a\n * document. However only one instance is required in a lunr.Index~Result. This\n * method combines metadata from another instance of lunr.MatchData with this\n * objects metadata.\n *\n * @param {lunr.MatchData} otherMatchData - Another instance of match data to merge with this one.\n * @see {@link lunr.Index~Result}\n */\nlunr.MatchData.prototype.combine = function (otherMatchData) {\n  var terms = Object.keys(otherMatchData.metadata)\n\n  for (var i = 0; i < terms.length; i++) {\n    var term = terms[i],\n        fields = Object.keys(otherMatchData.metadata[term])\n\n    if (this.metadata[term] == undefined) {\n      this.metadata[term] = Object.create(null)\n    }\n\n    for (var j = 0; j < fields.length; j++) {\n      var field = fields[j],\n          keys = Object.keys(otherMatchData.metadata[term][field])\n\n      if (this.metadata[term][field] == undefined) {\n        this.metadata[term][field] = Object.create(null)\n      }\n\n      for (var k = 0; k < keys.length; k++) {\n        var key = keys[k]\n\n        if (this.metadata[term][field][key] == undefined) {\n          this.metadata[term][field][key] = otherMatchData.metadata[term][field][key]\n        } else {\n          this.metadata[term][field][key] = this.metadata[term][field][key].concat(otherMatchData.metadata[term][field][key])\n        }\n\n      }\n    }\n  }\n}\n\n/**\n * Add metadata for a term/field pair to this instance of match data.\n *\n * @param {string} term - The term this match data is associated with\n * @param {string} field - The field in which the term was found\n * @param {object} metadata - The metadata recorded about this term in this field\n */\nlunr.MatchData.prototype.add = function (term, field, metadata) {\n  if (!(term in this.metadata)) {\n    this.metadata[term] = Object.create(null)\n    this.metadata[term][field] = metadata\n    return\n  }\n\n  if (!(field in this.metadata[term])) {\n    this.metadata[term][field] = metadata\n    return\n  }\n\n  var metadataKeys = Object.keys(metadata)\n\n  for (var i = 0; i < metadataKeys.length; i++) {\n    var key = metadataKeys[i]\n\n    if (key in this.metadata[term][field]) {\n      this.metadata[term][field][key] = this.metadata[term][field][key].concat(metadata[key])\n    } else {\n      this.metadata[term][field][key] = metadata[key]\n    }\n  }\n}\n/**\n * A lunr.Query provides a programmatic way of defining queries to be performed\n * against a {@link lunr.Index}.\n *\n * Prefer constructing a lunr.Query using the {@link lunr.Index#query} method\n * so the query object is pre-initialized with the right index fields.\n *\n * @constructor\n * @property {lunr.Query~Clause[]} clauses - An array of query clauses.\n * @property {string[]} allFields - An array of all available fields in a lunr.Index.\n */\nlunr.Query = function (allFields) {\n  this.clauses = []\n  this.allFields = allFields\n}\n\n/**\n * Constants for indicating what kind of automatic wildcard insertion will be used when constructing a query clause.\n *\n * This allows wildcards to be added to the beginning and end of a term without having to manually do any string\n * concatenation.\n *\n * The wildcard constants can be bitwise combined to select both leading and trailing wildcards.\n *\n * @constant\n * @default\n * @property {number} wildcard.NONE - The term will have no wildcards inserted, this is the default behaviour\n * @property {number} wildcard.LEADING - Prepend the term with a wildcard, unless a leading wildcard already exists\n * @property {number} wildcard.TRAILING - Append a wildcard to the term, unless a trailing wildcard already exists\n * @see lunr.Query~Clause\n * @see lunr.Query#clause\n * @see lunr.Query#term\n * @example <caption>query term with trailing wildcard</caption>\n * query.term('foo', { wildcard: lunr.Query.wildcard.TRAILING })\n * @example <caption>query term with leading and trailing wildcard</caption>\n * query.term('foo', {\n *   wildcard: lunr.Query.wildcard.LEADING | lunr.Query.wildcard.TRAILING\n * })\n */\n\nlunr.Query.wildcard = new String (\"*\")\nlunr.Query.wildcard.NONE = 0\nlunr.Query.wildcard.LEADING = 1\nlunr.Query.wildcard.TRAILING = 2\n\n/**\n * Constants for indicating what kind of presence a term must have in matching documents.\n *\n * @constant\n * @enum {number}\n * @see lunr.Query~Clause\n * @see lunr.Query#clause\n * @see lunr.Query#term\n * @example <caption>query term with required presence</caption>\n * query.term('foo', { presence: lunr.Query.presence.REQUIRED })\n */\nlunr.Query.presence = {\n  /**\n   * Term's presence in a document is optional, this is the default value.\n   */\n  OPTIONAL: 1,\n\n  /**\n   * Term's presence in a document is required, documents that do not contain\n   * this term will not be returned.\n   */\n  REQUIRED: 2,\n\n  /**\n   * Term's presence in a document is prohibited, documents that do contain\n   * this term will not be returned.\n   */\n  PROHIBITED: 3\n}\n\n/**\n * A single clause in a {@link lunr.Query} contains a term and details on how to\n * match that term against a {@link lunr.Index}.\n *\n * @typedef {Object} lunr.Query~Clause\n * @property {string[]} fields - The fields in an index this clause should be matched against.\n * @property {number} [boost=1] - Any boost that should be applied when matching this clause.\n * @property {number} [editDistance] - Whether the term should have fuzzy matching applied, and how fuzzy the match should be.\n * @property {boolean} [usePipeline] - Whether the term should be passed through the search pipeline.\n * @property {number} [wildcard=lunr.Query.wildcard.NONE] - Whether the term should have wildcards appended or prepended.\n * @property {number} [presence=lunr.Query.presence.OPTIONAL] - The terms presence in any matching documents.\n */\n\n/**\n * Adds a {@link lunr.Query~Clause} to this query.\n *\n * Unless the clause contains the fields to be matched all fields will be matched. In addition\n * a default boost of 1 is applied to the clause.\n *\n * @param {lunr.Query~Clause} clause - The clause to add to this query.\n * @see lunr.Query~Clause\n * @returns {lunr.Query}\n */\nlunr.Query.prototype.clause = function (clause) {\n  if (!('fields' in clause)) {\n    clause.fields = this.allFields\n  }\n\n  if (!('boost' in clause)) {\n    clause.boost = 1\n  }\n\n  if (!('usePipeline' in clause)) {\n    clause.usePipeline = true\n  }\n\n  if (!('wildcard' in clause)) {\n    clause.wildcard = lunr.Query.wildcard.NONE\n  }\n\n  if ((clause.wildcard & lunr.Query.wildcard.LEADING) && (clause.term.charAt(0) != lunr.Query.wildcard)) {\n    clause.term = \"*\" + clause.term\n  }\n\n  if ((clause.wildcard & lunr.Query.wildcard.TRAILING) && (clause.term.slice(-1) != lunr.Query.wildcard)) {\n    clause.term = \"\" + clause.term + \"*\"\n  }\n\n  if (!('presence' in clause)) {\n    clause.presence = lunr.Query.presence.OPTIONAL\n  }\n\n  this.clauses.push(clause)\n\n  return this\n}\n\n/**\n * A negated query is one in which every clause has a presence of\n * prohibited. These queries require some special processing to return\n * the expected results.\n *\n * @returns boolean\n */\nlunr.Query.prototype.isNegated = function () {\n  for (var i = 0; i < this.clauses.length; i++) {\n    if (this.clauses[i].presence != lunr.Query.presence.PROHIBITED) {\n      return false\n    }\n  }\n\n  return true\n}\n\n/**\n * Adds a term to the current query, under the covers this will create a {@link lunr.Query~Clause}\n * to the list of clauses that make up this query.\n *\n * The term is used as is, i.e. no tokenization will be performed by this method. Instead conversion\n * to a token or token-like string should be done before calling this method.\n *\n * The term will be converted to a string by calling `toString`. Multiple terms can be passed as an\n * array, each term in the array will share the same options.\n *\n * @param {object|object[]} term - The term(s) to add to the query.\n * @param {object} [options] - Any additional properties to add to the query clause.\n * @returns {lunr.Query}\n * @see lunr.Query#clause\n * @see lunr.Query~Clause\n * @example <caption>adding a single term to a query</caption>\n * query.term(\"foo\")\n * @example <caption>adding a single term to a query and specifying search fields, term boost and automatic trailing wildcard</caption>\n * query.term(\"foo\", {\n *   fields: [\"title\"],\n *   boost: 10,\n *   wildcard: lunr.Query.wildcard.TRAILING\n * })\n * @example <caption>using lunr.tokenizer to convert a string to tokens before using them as terms</caption>\n * query.term(lunr.tokenizer(\"foo bar\"))\n */\nlunr.Query.prototype.term = function (term, options) {\n  if (Array.isArray(term)) {\n    term.forEach(function (t) { this.term(t, lunr.utils.clone(options)) }, this)\n    return this\n  }\n\n  var clause = options || {}\n  clause.term = term.toString()\n\n  this.clause(clause)\n\n  return this\n}\nlunr.QueryParseError = function (message, start, end) {\n  this.name = \"QueryParseError\"\n  this.message = message\n  this.start = start\n  this.end = end\n}\n\nlunr.QueryParseError.prototype = new Error\nlunr.QueryLexer = function (str) {\n  this.lexemes = []\n  this.str = str\n  this.length = str.length\n  this.pos = 0\n  this.start = 0\n  this.escapeCharPositions = []\n}\n\nlunr.QueryLexer.prototype.run = function () {\n  var state = lunr.QueryLexer.lexText\n\n  while (state) {\n    state = state(this)\n  }\n}\n\nlunr.QueryLexer.prototype.sliceString = function () {\n  var subSlices = [],\n      sliceStart = this.start,\n      sliceEnd = this.pos\n\n  for (var i = 0; i < this.escapeCharPositions.length; i++) {\n    sliceEnd = this.escapeCharPositions[i]\n    subSlices.push(this.str.slice(sliceStart, sliceEnd))\n    sliceStart = sliceEnd + 1\n  }\n\n  subSlices.push(this.str.slice(sliceStart, this.pos))\n  this.escapeCharPositions.length = 0\n\n  return subSlices.join('')\n}\n\nlunr.QueryLexer.prototype.emit = function (type) {\n  this.lexemes.push({\n    type: type,\n    str: this.sliceString(),\n    start: this.start,\n    end: this.pos\n  })\n\n  this.start = this.pos\n}\n\nlunr.QueryLexer.prototype.escapeCharacter = function () {\n  this.escapeCharPositions.push(this.pos - 1)\n  this.pos += 1\n}\n\nlunr.QueryLexer.prototype.next = function () {\n  if (this.pos >= this.length) {\n    return lunr.QueryLexer.EOS\n  }\n\n  var char = this.str.charAt(this.pos)\n  this.pos += 1\n  return char\n}\n\nlunr.QueryLexer.prototype.width = function () {\n  return this.pos - this.start\n}\n\nlunr.QueryLexer.prototype.ignore = function () {\n  if (this.start == this.pos) {\n    this.pos += 1\n  }\n\n  this.start = this.pos\n}\n\nlunr.QueryLexer.prototype.backup = function () {\n  this.pos -= 1\n}\n\nlunr.QueryLexer.prototype.acceptDigitRun = function () {\n  var char, charCode\n\n  do {\n    char = this.next()\n    charCode = char.charCodeAt(0)\n  } while (charCode > 47 && charCode < 58)\n\n  if (char != lunr.QueryLexer.EOS) {\n    this.backup()\n  }\n}\n\nlunr.QueryLexer.prototype.more = function () {\n  return this.pos < this.length\n}\n\nlunr.QueryLexer.EOS = 'EOS'\nlunr.QueryLexer.FIELD = 'FIELD'\nlunr.QueryLexer.TERM = 'TERM'\nlunr.QueryLexer.EDIT_DISTANCE = 'EDIT_DISTANCE'\nlunr.QueryLexer.BOOST = 'BOOST'\nlunr.QueryLexer.PRESENCE = 'PRESENCE'\n\nlunr.QueryLexer.lexField = function (lexer) {\n  lexer.backup()\n  lexer.emit(lunr.QueryLexer.FIELD)\n  lexer.ignore()\n  return lunr.QueryLexer.lexText\n}\n\nlunr.QueryLexer.lexTerm = function (lexer) {\n  if (lexer.width() > 1) {\n    lexer.backup()\n    lexer.emit(lunr.QueryLexer.TERM)\n  }\n\n  lexer.ignore()\n\n  if (lexer.more()) {\n    return lunr.QueryLexer.lexText\n  }\n}\n\nlunr.QueryLexer.lexEditDistance = function (lexer) {\n  lexer.ignore()\n  lexer.acceptDigitRun()\n  lexer.emit(lunr.QueryLexer.EDIT_DISTANCE)\n  return lunr.QueryLexer.lexText\n}\n\nlunr.QueryLexer.lexBoost = function (lexer) {\n  lexer.ignore()\n  lexer.acceptDigitRun()\n  lexer.emit(lunr.QueryLexer.BOOST)\n  return lunr.QueryLexer.lexText\n}\n\nlunr.QueryLexer.lexEOS = function (lexer) {\n  if (lexer.width() > 0) {\n    lexer.emit(lunr.QueryLexer.TERM)\n  }\n}\n\n// This matches the separator used when tokenising fields\n// within a document. These should match otherwise it is\n// not possible to search for some tokens within a document.\n//\n// It is possible for the user to change the separator on the\n// tokenizer so it _might_ clash with any other of the special\n// characters already used within the search string, e.g. :.\n//\n// This means that it is possible to change the separator in\n// such a way that makes some words unsearchable using a search\n// string.\nlunr.QueryLexer.termSeparator = lunr.tokenizer.separator\n\nlunr.QueryLexer.lexText = function (lexer) {\n  while (true) {\n    var char = lexer.next()\n\n    if (char == lunr.QueryLexer.EOS) {\n      return lunr.QueryLexer.lexEOS\n    }\n\n    // Escape character is '\\'\n    if (char.charCodeAt(0) == 92) {\n      lexer.escapeCharacter()\n      continue\n    }\n\n    if (char == \":\") {\n      return lunr.QueryLexer.lexField\n    }\n\n    if (char == \"~\") {\n      lexer.backup()\n      if (lexer.width() > 0) {\n        lexer.emit(lunr.QueryLexer.TERM)\n      }\n      return lunr.QueryLexer.lexEditDistance\n    }\n\n    if (char == \"^\") {\n      lexer.backup()\n      if (lexer.width() > 0) {\n        lexer.emit(lunr.QueryLexer.TERM)\n      }\n      return lunr.QueryLexer.lexBoost\n    }\n\n    // \"+\" indicates term presence is required\n    // checking for length to ensure that only\n    // leading \"+\" are considered\n    if (char == \"+\" && lexer.width() === 1) {\n      lexer.emit(lunr.QueryLexer.PRESENCE)\n      return lunr.QueryLexer.lexText\n    }\n\n    // \"-\" indicates term presence is prohibited\n    // checking for length to ensure that only\n    // leading \"-\" are considered\n    if (char == \"-\" && lexer.width() === 1) {\n      lexer.emit(lunr.QueryLexer.PRESENCE)\n      return lunr.QueryLexer.lexText\n    }\n\n    if (char.match(lunr.QueryLexer.termSeparator)) {\n      return lunr.QueryLexer.lexTerm\n    }\n  }\n}\n\nlunr.QueryParser = function (str, query) {\n  this.lexer = new lunr.QueryLexer (str)\n  this.query = query\n  this.currentClause = {}\n  this.lexemeIdx = 0\n}\n\nlunr.QueryParser.prototype.parse = function () {\n  this.lexer.run()\n  this.lexemes = this.lexer.lexemes\n\n  var state = lunr.QueryParser.parseClause\n\n  while (state) {\n    state = state(this)\n  }\n\n  return this.query\n}\n\nlunr.QueryParser.prototype.peekLexeme = function () {\n  return this.lexemes[this.lexemeIdx]\n}\n\nlunr.QueryParser.prototype.consumeLexeme = function () {\n  var lexeme = this.peekLexeme()\n  this.lexemeIdx += 1\n  return lexeme\n}\n\nlunr.QueryParser.prototype.nextClause = function () {\n  var completedClause = this.currentClause\n  this.query.clause(completedClause)\n  this.currentClause = {}\n}\n\nlunr.QueryParser.parseClause = function (parser) {\n  var lexeme = parser.peekLexeme()\n\n  if (lexeme == undefined) {\n    return\n  }\n\n  switch (lexeme.type) {\n    case lunr.QueryLexer.PRESENCE:\n      return lunr.QueryParser.parsePresence\n    case lunr.QueryLexer.FIELD:\n      return lunr.QueryParser.parseField\n    case lunr.QueryLexer.TERM:\n      return lunr.QueryParser.parseTerm\n    default:\n      var errorMessage = \"expected either a field or a term, found \" + lexeme.type\n\n      if (lexeme.str.length >= 1) {\n        errorMessage += \" with value '\" + lexeme.str + \"'\"\n      }\n\n      throw new lunr.QueryParseError (errorMessage, lexeme.start, lexeme.end)\n  }\n}\n\nlunr.QueryParser.parsePresence = function (parser) {\n  var lexeme = parser.consumeLexeme()\n\n  if (lexeme == undefined) {\n    return\n  }\n\n  switch (lexeme.str) {\n    case \"-\":\n      parser.currentClause.presence = lunr.Query.presence.PROHIBITED\n      break\n    case \"+\":\n      parser.currentClause.presence = lunr.Query.presence.REQUIRED\n      break\n    default:\n      var errorMessage = \"unrecognised presence operator'\" + lexeme.str + \"'\"\n      throw new lunr.QueryParseError (errorMessage, lexeme.start, lexeme.end)\n  }\n\n  var nextLexeme = parser.peekLexeme()\n\n  if (nextLexeme == undefined) {\n    var errorMessage = \"expecting term or field, found nothing\"\n    throw new lunr.QueryParseError (errorMessage, lexeme.start, lexeme.end)\n  }\n\n  switch (nextLexeme.type) {\n    case lunr.QueryLexer.FIELD:\n      return lunr.QueryParser.parseField\n    case lunr.QueryLexer.TERM:\n      return lunr.QueryParser.parseTerm\n    default:\n      var errorMessage = \"expecting term or field, found '\" + nextLexeme.type + \"'\"\n      throw new lunr.QueryParseError (errorMessage, nextLexeme.start, nextLexeme.end)\n  }\n}\n\nlunr.QueryParser.parseField = function (parser) {\n  var lexeme = parser.consumeLexeme()\n\n  if (lexeme == undefined) {\n    return\n  }\n\n  if (parser.query.allFields.indexOf(lexeme.str) == -1) {\n    var possibleFields = parser.query.allFields.map(function (f) { return \"'\" + f + \"'\" }).join(', '),\n        errorMessage = \"unrecognised field '\" + lexeme.str + \"', possible fields: \" + possibleFields\n\n    throw new lunr.QueryParseError (errorMessage, lexeme.start, lexeme.end)\n  }\n\n  parser.currentClause.fields = [lexeme.str]\n\n  var nextLexeme = parser.peekLexeme()\n\n  if (nextLexeme == undefined) {\n    var errorMessage = \"expecting term, found nothing\"\n    throw new lunr.QueryParseError (errorMessage, lexeme.start, lexeme.end)\n  }\n\n  switch (nextLexeme.type) {\n    case lunr.QueryLexer.TERM:\n      return lunr.QueryParser.parseTerm\n    default:\n      var errorMessage = \"expecting term, found '\" + nextLexeme.type + \"'\"\n      throw new lunr.QueryParseError (errorMessage, nextLexeme.start, nextLexeme.end)\n  }\n}\n\nlunr.QueryParser.parseTerm = function (parser) {\n  var lexeme = parser.consumeLexeme()\n\n  if (lexeme == undefined) {\n    return\n  }\n\n  parser.currentClause.term = lexeme.str.toLowerCase()\n\n  if (lexeme.str.indexOf(\"*\") != -1) {\n    parser.currentClause.usePipeline = false\n  }\n\n  var nextLexeme = parser.peekLexeme()\n\n  if (nextLexeme == undefined) {\n    parser.nextClause()\n    return\n  }\n\n  switch (nextLexeme.type) {\n    case lunr.QueryLexer.TERM:\n      parser.nextClause()\n      return lunr.QueryParser.parseTerm\n    case lunr.QueryLexer.FIELD:\n      parser.nextClause()\n      return lunr.QueryParser.parseField\n    case lunr.QueryLexer.EDIT_DISTANCE:\n      return lunr.QueryParser.parseEditDistance\n    case lunr.QueryLexer.BOOST:\n      return lunr.QueryParser.parseBoost\n    case lunr.QueryLexer.PRESENCE:\n      parser.nextClause()\n      return lunr.QueryParser.parsePresence\n    default:\n      var errorMessage = \"Unexpected lexeme type '\" + nextLexeme.type + \"'\"\n      throw new lunr.QueryParseError (errorMessage, nextLexeme.start, nextLexeme.end)\n  }\n}\n\nlunr.QueryParser.parseEditDistance = function (parser) {\n  var lexeme = parser.consumeLexeme()\n\n  if (lexeme == undefined) {\n    return\n  }\n\n  var editDistance = parseInt(lexeme.str, 10)\n\n  if (isNaN(editDistance)) {\n    var errorMessage = \"edit distance must be numeric\"\n    throw new lunr.QueryParseError (errorMessage, lexeme.start, lexeme.end)\n  }\n\n  parser.currentClause.editDistance = editDistance\n\n  var nextLexeme = parser.peekLexeme()\n\n  if (nextLexeme == undefined) {\n    parser.nextClause()\n    return\n  }\n\n  switch (nextLexeme.type) {\n    case lunr.QueryLexer.TERM:\n      parser.nextClause()\n      return lunr.QueryParser.parseTerm\n    case lunr.QueryLexer.FIELD:\n      parser.nextClause()\n      return lunr.QueryParser.parseField\n    case lunr.QueryLexer.EDIT_DISTANCE:\n      return lunr.QueryParser.parseEditDistance\n    case lunr.QueryLexer.BOOST:\n      return lunr.QueryParser.parseBoost\n    case lunr.QueryLexer.PRESENCE:\n      parser.nextClause()\n      return lunr.QueryParser.parsePresence\n    default:\n      var errorMessage = \"Unexpected lexeme type '\" + nextLexeme.type + \"'\"\n      throw new lunr.QueryParseError (errorMessage, nextLexeme.start, nextLexeme.end)\n  }\n}\n\nlunr.QueryParser.parseBoost = function (parser) {\n  var lexeme = parser.consumeLexeme()\n\n  if (lexeme == undefined) {\n    return\n  }\n\n  var boost = parseInt(lexeme.str, 10)\n\n  if (isNaN(boost)) {\n    var errorMessage = \"boost must be numeric\"\n    throw new lunr.QueryParseError (errorMessage, lexeme.start, lexeme.end)\n  }\n\n  parser.currentClause.boost = boost\n\n  var nextLexeme = parser.peekLexeme()\n\n  if (nextLexeme == undefined) {\n    parser.nextClause()\n    return\n  }\n\n  switch (nextLexeme.type) {\n    case lunr.QueryLexer.TERM:\n      parser.nextClause()\n      return lunr.QueryParser.parseTerm\n    case lunr.QueryLexer.FIELD:\n      parser.nextClause()\n      return lunr.QueryParser.parseField\n    case lunr.QueryLexer.EDIT_DISTANCE:\n      return lunr.QueryParser.parseEditDistance\n    case lunr.QueryLexer.BOOST:\n      return lunr.QueryParser.parseBoost\n    case lunr.QueryLexer.PRESENCE:\n      parser.nextClause()\n      return lunr.QueryParser.parsePresence\n    default:\n      var errorMessage = \"Unexpected lexeme type '\" + nextLexeme.type + \"'\"\n      throw new lunr.QueryParseError (errorMessage, nextLexeme.start, nextLexeme.end)\n  }\n}\n\n  /**\n   * export the module via AMD, CommonJS or as a browser global\n   * Export code from https://github.com/umdjs/umd/blob/master/returnExports.js\n   */\n  ;(function (root, factory) {\n    if (typeof define === 'function' && define.amd) {\n      // AMD. Register as an anonymous module.\n      define(factory)\n    } else if (typeof exports === 'object') {\n      /**\n       * Node. Does not work with strict CommonJS, but\n       * only CommonJS-like enviroments that support module.exports,\n       * like Node.\n       */\n      module.exports = factory()\n    } else {\n      // Browser globals (root is window)\n      root.lunr = factory()\n    }\n  }(this, function () {\n    /**\n     * Just return a value to define the module export.\n     * This example returns an object, but the module\n     * can return a function as the exported value.\n     */\n    return lunr\n  }))\n})();\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A RTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport lunr from \"lunr\"\n\nimport { getElement } from \"~/browser/element/_\"\nimport \"~/polyfills\"\n\nimport { Search } from \"../../_\"\nimport { SearchConfig } from \"../../config\"\nimport {\n  SearchMessage,\n  SearchMessageType\n} from \"../message\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Add support for `iframe-worker` shim\n *\n * While `importScripts` is synchronous when executed inside of a web worker,\n * it's not possible to provide a synchronous shim implementation. The cool\n * thing is that awaiting a non-Promise will convert it into a Promise, so\n * extending the type definition to return a `Promise` shouldn't break anything.\n *\n * @see https://bit.ly/2PjDnXi - GitHub comment\n *\n * @param urls - Scripts to load\n *\n * @returns Promise resolving with no result\n */\ndeclare global {\n  function importScripts(...urls: string[]): Promise<void> | void\n}\n\n/* ----------------------------------------------------------------------------\n * Data\n * ------------------------------------------------------------------------- */\n\n/**\n * Search index\n */\nlet index: Search\n\n/* ----------------------------------------------------------------------------\n * Helper functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Fetch (= import) multi-language support through `lunr-languages`\n *\n * This function automatically imports the stemmers necessary to process the\n * languages which are defined as part of the search configuration.\n *\n * If the worker runs inside of an `iframe` (when using `iframe-worker` as\n * a shim), the base URL for the stemmers to be loaded must be determined by\n * searching for the first `script` element with a `src` attribute, which will\n * contain the contents of this script.\n *\n * @param config - Search configuration\n *\n * @returns Promise resolving with no result\n */\nasync function setupSearchLanguages(\n  config: SearchConfig\n): Promise<void> {\n  let base = \"../lunr\"\n\n  /* Detect `iframe-worker` and fix base URL */\n  if (typeof parent !== \"undefined\" && \"IFrameWorker\" in parent) {\n    const worker = getElement<HTMLScriptElement>(\"script[src]\")!\n    const [path] = worker.src.split(\"/worker\")\n\n    /* Prefix base with path */\n    base = base.replace(\"..\", path)\n  }\n\n  /* Add scripts for languages */\n  const scripts = []\n  for (const lang of config.lang) {\n    switch (lang) {\n\n      /* Add segmenter for Japanese */\n      case \"ja\":\n        scripts.push(`${base}/tinyseg.js`)\n        break\n\n      /* Add segmenter for Hindi and Thai */\n      case \"hi\":\n      case \"th\":\n        scripts.push(`${base}/wordcut.js`)\n        break\n    }\n\n    /* Add language support */\n    if (lang !== \"en\")\n      scripts.push(`${base}/min/lunr.${lang}.min.js`)\n  }\n\n  /* Add multi-language support */\n  if (config.lang.length > 1)\n    scripts.push(`${base}/min/lunr.multi.min.js`)\n\n  /* Load scripts synchronously */\n  if (scripts.length)\n    await importScripts(\n      `${base}/min/lunr.stemmer.support.min.js`,\n      ...scripts\n    )\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Message handler\n *\n * @param message - Source message\n *\n * @returns Target message\n */\nexport async function handler(\n  message: SearchMessage\n): Promise<SearchMessage> {\n  switch (message.type) {\n\n    /* Search setup message */\n    case SearchMessageType.SETUP:\n      await setupSearchLanguages(message.data.config)\n      index = new Search(message.data)\n      return {\n        type: SearchMessageType.READY\n      }\n\n    /* Search query message */\n    case SearchMessageType.QUERY:\n      const query = message.data\n      try {\n        return {\n          type: SearchMessageType.RESULT,\n          data: index.search(query)\n        }\n\n      /* Return empty result in case of error */\n      } catch (err) {\n        console.warn(`Invalid query: ${query} \u2013 see https://bit.ly/2s3ChXG`)\n        console.warn(err)\n        return {\n          type: SearchMessageType.RESULT,\n          data: { items: [] }\n        }\n      }\n\n    /* All other messages */\n    default:\n      throw new TypeError(\"Invalid message type\")\n  }\n}\n\n/* ----------------------------------------------------------------------------\n * Worker\n * ------------------------------------------------------------------------- */\n\n/* Expose Lunr.js in global scope, or stemmers won't work */\nself.lunr = lunr\n\n/* Handle messages */\naddEventListener(\"message\", async ev => {\n  postMessage(await handler(ev.data))\n})\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Retrieve all elements matching the query selector\n *\n * @template T - Element type\n *\n * @param selector - Query selector\n * @param node - Node of reference\n *\n * @returns Elements\n */\nexport function getElements<T extends keyof HTMLElementTagNameMap>(\n  selector: T, node?: ParentNode\n): HTMLElementTagNameMap[T][]\n\nexport function getElements<T extends HTMLElement>(\n  selector: string, node?: ParentNode\n): T[]\n\nexport function getElements<T extends HTMLElement>(\n  selector: string, node: ParentNode = document\n): T[] {\n  return Array.from(node.querySelectorAll<T>(selector))\n}\n\n/**\n * Retrieve an element matching a query selector or throw a reference error\n *\n * Note that this function assumes that the element is present. If unsure if an\n * element is existent, use the `getOptionalElement` function instead.\n *\n * @template T - Element type\n *\n * @param selector - Query selector\n * @param node - Node of reference\n *\n * @returns Element\n */\nexport function getElement<T extends keyof HTMLElementTagNameMap>(\n  selector: T, node?: ParentNode\n): HTMLElementTagNameMap[T]\n\nexport function getElement<T extends HTMLElement>(\n  selector: string, node?: ParentNode\n): T\n\nexport function getElement<T extends HTMLElement>(\n  selector: string, node: ParentNode = document\n): T {\n  const el = getOptionalElement<T>(selector, node)\n  if (typeof el === \"undefined\")\n    throw new ReferenceError(\n      `Missing element: expected \"${selector}\" to be present`\n    )\n\n  /* Return element */\n  return el\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Retrieve an optional element matching the query selector\n *\n * @template T - Element type\n *\n * @param selector - Query selector\n * @param node - Node of reference\n *\n * @returns Element or nothing\n */\nexport function getOptionalElement<T extends keyof HTMLElementTagNameMap>(\n  selector: T, node?: ParentNode\n): HTMLElementTagNameMap[T] | undefined\n\nexport function getOptionalElement<T extends HTMLElement>(\n  selector: string, node?: ParentNode\n): T | undefined\n\nexport function getOptionalElement<T extends HTMLElement>(\n  selector: string, node: ParentNode = document\n): T | undefined {\n  return node.querySelector<T>(selector) || undefined\n}\n\n/**\n * Retrieve the currently active element\n *\n * @returns Element or nothing\n */\nexport function getActiveElement(): HTMLElement | undefined {\n  return (\n    document.activeElement?.shadowRoot?.activeElement as HTMLElement ??\n    document.activeElement as HTMLElement ??\n    undefined\n  )\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Polyfills\n * ------------------------------------------------------------------------- */\n\n/* Polyfill `Object.entries` */\nif (!Object.entries)\n  Object.entries = function (obj: object) {\n    const data: [string, string][] = []\n    for (const key of Object.keys(obj))\n      // @ts-expect-error - ignore property access warning\n      data.push([key, obj[key]])\n\n    /* Return entries */\n    return data\n  }\n\n/* Polyfill `Object.values` */\nif (!Object.values)\n  Object.values = function (obj: object) {\n    const data: string[] = []\n    for (const key of Object.keys(obj))\n      // @ts-expect-error - ignore property access warning\n      data.push(obj[key])\n\n    /* Return values */\n    return data\n  }\n\n/* ------------------------------------------------------------------------- */\n\n/* Polyfills for `Element` */\nif (typeof Element !== \"undefined\") {\n\n  /* Polyfill `Element.scrollTo` */\n  if (!Element.prototype.scrollTo)\n    Element.prototype.scrollTo = function (\n      x?: ScrollToOptions | number, y?: number\n    ): void {\n      if (typeof x === \"object\") {\n        this.scrollLeft = x.left!\n        this.scrollTop = x.top!\n      } else {\n        this.scrollLeft = x!\n        this.scrollTop = y!\n      }\n    }\n\n  /* Polyfill `Element.replaceWith` */\n  if (!Element.prototype.replaceWith)\n    Element.prototype.replaceWith = function (\n      ...nodes: Array<string | Node>\n    ): void {\n      const parent = this.parentNode\n      if (parent) {\n        if (nodes.length === 0)\n          parent.removeChild(this)\n\n        /* Replace children and create text nodes */\n        for (let i = nodes.length - 1; i >= 0; i--) {\n          let node = nodes[i]\n          if (typeof node === \"string\")\n            node = document.createTextNode(node)\n          else if (node.parentNode)\n            node.parentNode.removeChild(node)\n\n          /* Replace child or insert before previous sibling */\n          if (!i)\n            parent.replaceChild(node, this)\n          else\n            parent.insertBefore(this.previousSibling!, node)\n        }\n      }\n    }\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search configuration\n */\nexport interface SearchConfig {\n  lang: string[]                       /* Search languages */\n  separator: string                    /* Search separator */\n  pipeline: SearchPipelineFn[]         /* Search pipeline */\n}\n\n/**\n * Search document\n */\nexport interface SearchDocument {\n  location: string                     /* Document location */\n  title: string                        /* Document title */\n  text: string                         /* Document text */\n  tags?: string[]                      /* Document tags */\n  boost?: number                       /* Document boost */\n  parent?: SearchDocument              /* Document parent */\n}\n\n/**\n * Search options\n */\nexport interface SearchOptions {\n  suggest: boolean                     /* Search suggestions */\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Search index\n */\nexport interface SearchIndex {\n  config: SearchConfig                 /* Search configuration */\n  docs: SearchDocument[]               /* Search documents */\n  options: SearchOptions               /* Search options */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search pipeline function\n */\ntype SearchPipelineFn =\n  | \"trimmer\"                          /* Trimmer */\n  | \"stopWordFilter\"                   /* Stop word filter */\n  | \"stemmer\"                          /* Stemmer */\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Create a search document map\n *\n * This function creates a mapping of URLs (including anchors) to the actual\n * articles and sections. It relies on the invariant that the search index is\n * ordered with the main article appearing before all sections with anchors.\n * If this is not the case, the logic music be changed.\n *\n * @param docs - Search documents\n *\n * @returns Search document map\n */\nexport function setupSearchDocumentMap(\n  docs: SearchDocument[]\n): Map<string, SearchDocument> {\n  const map = new Map<string, SearchDocument>()\n  for (const doc of docs) {\n    const [path] = doc.location.split(\"#\")\n\n    /* Add document article */\n    const article = map.get(path)\n    if (typeof article === \"undefined\") {\n      map.set(path, doc)\n\n      /* Add document section */\n    } else {\n      map.set(doc.location, doc)\n      doc.parent = article\n    }\n  }\n\n  /* Return search document map */\n  return map\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Visitor function\n *\n * @param start - Start offset\n * @param end - End offset\n */\ntype VisitorFn = (\n  start: number, end: number\n) => void\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Split a string using the given separator\n *\n * @param input - Input value\n * @param separator - Separator\n * @param fn - Visitor function\n */\nexport function split(\n  input: string, separator: RegExp, fn: VisitorFn\n): void {\n  separator = new RegExp(separator, \"g\")\n\n  /* Split string using separator */\n  let match: RegExpExecArray | null\n  let index = 0\n  do {\n    match = separator.exec(input)\n\n    /* Emit non-empty range */\n    const until = match?.index ?? input.length\n    if (index < until)\n      fn(index, until)\n\n    /* Update last index */\n    if (match) {\n      const [term] = match\n      index = match.index + term.length\n\n      /* Support zero-length lookaheads */\n      if (term.length === 0)\n        separator.lastIndex = match.index + 1\n    }\n  } while (match)\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Extraction type\n *\n * This type defines the possible values that are encoded into the first two\n * bits of a section that is part of the blocks of a tokenization table. There\n * are three types of interest: HTML opening and closing tags, as well as the\n * actual text content we need to extract for indexing.\n */\nexport const enum Extract {\n  TAG_OPEN  = 0,                       /* HTML opening tag */\n  TEXT      = 1,                       /* Text content */\n  TAG_CLOSE = 2                        /* HTML closing tag */\n}\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Visitor function\n *\n * @param block - Block index\n * @param type - Extraction type\n * @param start - Start offset\n * @param end - End offset\n */\ntype VisitorFn = (\n  block: number, type: Extract, start: number, end: number\n) => void\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Split a string into markup and text sections\n *\n * This function scans a string and divides it up into sections of markup and\n * text. For each section, it invokes the given visitor function with the block\n * index, extraction type, as well as start and end offsets. Using a visitor\n * function (= streaming data) is ideal for minimizing pressure on the GC.\n *\n * @param input - Input value\n * @param fn - Visitor function\n */\nexport function extract(\n  input: string, fn: VisitorFn\n): void {\n\n  let block = 0                        /* Current block */\n  let start = 0                        /* Current start offset */\n  let end = 0                          /* Current end offset */\n\n  /* Split string into sections */\n  for (let stack = 0; end < input.length; end++) {\n\n    /* Opening tag after non-empty section */\n    if (input.charAt(end) === \"<\" && end > start) {\n      fn(block, Extract.TEXT, start, start = end)\n\n    /* Closing tag */\n    } else if (input.charAt(end) === \">\") {\n      if (input.charAt(start + 1) === \"/\") {\n        if (--stack === 0)\n          fn(block++, Extract.TAG_CLOSE, start, end + 1)\n\n      /* Tag is not self-closing */\n      } else if (input.charAt(end - 1) !== \"/\") {\n        if (stack++ === 0)\n          fn(block, Extract.TAG_OPEN, start, end + 1)\n      }\n\n      /* New section */\n      start = end + 1\n    }\n  }\n\n  /* Add trailing section */\n  if (end > start)\n    fn(block, Extract.TEXT, start, end)\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Position table\n */\nexport type PositionTable = number[][]\n\n/**\n * Position\n */\nexport type Position = number\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Highlight all occurrences in a string\n *\n * This function receives a field's value (e.g. like `title` or `text`), it's\n * position table that was generated during indexing, and the positions found\n * when executing the query. It then highlights all occurrences, and returns\n * their concatenation. In case of multiple blocks, two are returned.\n *\n * @param input - Input value\n * @param table - Table for indexing\n * @param positions - Occurrences\n * @param full - Full results\n *\n * @returns Highlighted string value\n */\nexport function highlight(\n  input: string, table: PositionTable, positions: Position[], full = false\n): string {\n  return highlightAll([input], table, positions, full).pop()!\n}\n\n/**\n * Highlight all occurrences in a set of strings\n *\n * @param inputs - Input values\n * @param table - Table for indexing\n * @param positions - Occurrences\n * @param full - Full results\n *\n * @returns Highlighted string values\n */\nexport function highlightAll(\n  inputs: string[], table: PositionTable, positions: Position[], full = false\n): string[] {\n\n  /* Map blocks to input values */\n  const mapping = [0]\n  for (let t = 1; t < table.length; t++) {\n    const prev = table[t - 1]\n    const next = table[t]\n\n    /* Check if table points to new block */\n    const p = prev[prev.length - 1] >>> 2 & 0x3FF\n    const q = next[0]               >>> 12\n\n    /* Add block to mapping */\n    mapping.push(+(p > q) + mapping[mapping.length - 1])\n  }\n\n  /* Highlight strings one after another */\n  return inputs.map((input, i) => {\n    let cursor = 0\n\n    /* Map occurrences to blocks */\n    const blocks = new Map<number, number[]>()\n    for (const p of positions.sort((a, b) => a - b)) {\n      const index = p & 0xFFFFF\n      const block = p >>> 20\n      if (mapping[block] !== i)\n        continue\n\n      /* Ensure presence of block group */\n      let group = blocks.get(block)\n      if (typeof group === \"undefined\")\n        blocks.set(block, group = [])\n\n      /* Add index to group */\n      group.push(index)\n    }\n\n    /* Just return string, if no occurrences */\n    if (blocks.size === 0)\n      return input\n\n    /* Compute slices */\n    const slices: string[] = []\n    for (const [block, indexes] of blocks) {\n      const t = table[block]\n\n      /* Extract positions and length */\n      const start  = t[0]            >>> 12\n      const end    = t[t.length - 1] >>> 12\n      const length = t[t.length - 1] >>> 2 & 0x3FF\n\n      /* Add prefix, if full results are desired */\n      if (full && start > cursor)\n        slices.push(input.slice(cursor, start))\n\n      /* Extract and highlight slice */\n      let slice = input.slice(start, end + length)\n      for (const j of indexes.sort((a, b) => b - a)) {\n\n        /* Retrieve offset and length of match */\n        const p = (t[j] >>> 12) - start\n        const q = (t[j] >>> 2 & 0x3FF) + p\n\n        /* Wrap occurrence */\n        slice = [\n          slice.slice(0, p),\n          \"<mark>\",\n          slice.slice(p, q),\n          \"</mark>\",\n          slice.slice(q)\n        ].join(\"\")\n      }\n\n      /* Update cursor */\n      cursor = end + length\n\n      /* Append slice and abort if we have two */\n      if (slices.push(slice) === 2)\n        break\n    }\n\n    /* Add suffix, if full results are desired */\n    if (full && cursor < input.length)\n      slices.push(input.slice(cursor))\n\n    /* Return highlighted slices */\n    return slices.join(\"\")\n  })\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { split } from \"../_\"\nimport {\n  Extract,\n  extract\n} from \"../extract\"\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Split a string or set of strings into tokens\n *\n * This tokenizer supersedes the default tokenizer that is provided by Lunr.js,\n * as it is aware of HTML tags and allows for multi-character splitting.\n *\n * It takes the given inputs, splits each of them into markup and text sections,\n * tokenizes and segments (if necessary) each of them, and then indexes them in\n * a table by using a compact bit representation. Bitwise techniques are used\n * to write and read from the table during indexing and querying.\n *\n * @see https://bit.ly/3W3Xw4J - Search: better, faster, smaller\n *\n * @param input - Input value(s)\n *\n * @returns Tokens\n */\nexport function tokenize(\n  input?: string | string[]\n): lunr.Token[] {\n  const tokens: lunr.Token[] = []\n  if (typeof input === \"undefined\")\n    return tokens\n\n  /* Tokenize strings one after another */\n  const inputs = Array.isArray(input) ? input : [input]\n  for (let i = 0; i < inputs.length; i++) {\n    const table = lunr.tokenizer.table\n    const total = table.length\n\n    /* Split string into sections and tokenize content blocks */\n    extract(inputs[i], (block, type, start, end) => {\n      table[block += total] ||= []\n      switch (type) {\n\n        /* Handle markup */\n        case Extract.TAG_OPEN:\n        case Extract.TAG_CLOSE:\n          table[block].push(\n            start       << 12 |\n            end - start <<  2 |\n            type\n          )\n          break\n\n        /* Handle text content */\n        case Extract.TEXT:\n          const section = inputs[i].slice(start, end)\n          split(section, lunr.tokenizer.separator, (index, until) => {\n\n            /**\n             * Apply segmenter after tokenization. Note that the segmenter will\n             * also split words at word boundaries, which is not what we want,\n             * so we need to check if we can somehow mitigate this behavior.\n             */\n            if (typeof lunr.segmenter !== \"undefined\") {\n              const subsection = section.slice(index, until)\n              if (/^[MHIK]$/.test(lunr.segmenter.ctype_(subsection))) {\n                const segments = lunr.segmenter.segment(subsection)\n                for (let s = 0, l = 0; s < segments.length; s++) {\n\n                  /* Add block to section */\n                  table[block] ||= []\n                  table[block].push(\n                    start + index + l  << 12 |\n                    segments[s].length <<  2 |\n                    type\n                  )\n\n                  /* Add token with position */\n                  tokens.push(new lunr.Token(\n                    segments[s].toLowerCase(), {\n                      position: block << 20 | table[block].length - 1\n                    }\n                  ))\n\n                  /* Keep track of length */\n                  l += segments[s].length\n                }\n                return\n              }\n            }\n\n            /* Add block to section */\n            table[block].push(\n              start + index << 12 |\n              until - index <<  2 |\n              type\n            )\n\n            /* Add token with position */\n            tokens.push(new lunr.Token(\n              section.slice(index, until).toLowerCase(), {\n                position: block << 20 | table[block].length - 1\n              }\n            ))\n          })\n      }\n    })\n  }\n\n  /* Return tokens */\n  return tokens\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Helper types\n * ------------------------------------------------------------------------- */\n\n/**\n * Visitor function\n *\n * @param value - String value\n *\n * @returns String term(s)\n */\ntype VisitorFn = (\n  value: string\n) => string | string[]\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Default transformation function\n *\n * 1. Trim excess whitespace from left and right.\n *\n * 2. Search for parts in quotation marks and prepend a `+` modifier to denote\n *    that the resulting document must contain all parts, converting the query\n *    to an `AND` query (as opposed to the default `OR` behavior). While users\n *    may expect parts enclosed in quotation marks to map to span queries, i.e.\n *    for which order is important, Lunr.js doesn't support them, so the best\n *    we can do is to convert the parts to an `AND` query.\n *\n * 3. Replace control characters which are not located at the beginning of the\n *    query or preceded by white space, or are not followed by a non-whitespace\n *    character or are at the end of the query string. Furthermore, filter\n *    unmatched quotation marks.\n *\n * 4. Split the query string at whitespace, then pass each part to the visitor\n *    function for tokenization, and append a wildcard to every resulting term\n *    that is not explicitly marked with a `+`, `-`, `~` or `^` modifier, since\n *    it ensures consistent and stable ranking when multiple terms are entered.\n *    Also, if a fuzzy or boost modifier are given, but no numeric value has\n *    been entered, default to 1 to not induce a query error.\n *\n * @param query - Query value\n * @param fn - Visitor function\n *\n * @returns Transformed query value\n */\nexport function transform(\n  query: string, fn: VisitorFn = term => term\n): string {\n  return query\n\n    /* => 1 */\n    .trim()\n\n    /* => 2 */\n    .split(/\"([^\"]+)\"/g)\n      .map((parts, index) => index & 1\n        ? parts.replace(/^\\b|^(?![^\\x00-\\x7F]|$)|\\s+/g, \" +\")\n        : parts\n      )\n      .join(\"\")\n\n    /* => 3 */\n    .replace(/\"|(?:^|\\s+)[*+\\-:^~]+(?=\\s+|$)/g, \"\")\n\n    /* => 4 */\n    .split(/\\s+/g)\n      .reduce((prev, term) => {\n        const next = fn(term)\n        return [...prev, ...Array.isArray(next) ? next : [next]]\n      }, [] as string[])\n      .map(term => /([~^]$)/.test(term) ? `${term}1` : term)\n      .map(term => /(^[+-]|[~^]\\d+$)/.test(term) ? term : `${term}*`)\n      .join(\" \")\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport { split } from \"../../internal\"\nimport { transform } from \"../transform\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search query clause\n */\nexport interface SearchQueryClause {\n  presence: lunr.Query.presence        /* Clause presence */\n  term: string                         /* Clause term */\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Search query terms\n */\nexport type SearchQueryTerms = Record<string, boolean>\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Transform search query\n *\n * This function lexes the given search query and applies the transformation\n * function to each term, preserving markup like `+` and `-` modifiers.\n *\n * @param query - Search query\n *\n * @returns Search query\n */\nexport function transformSearchQuery(\n  query: string\n): string {\n\n  /* Split query terms with tokenizer */\n  return transform(query, part => {\n    const terms: string[] = []\n\n    /* Initialize lexer and analyze part */\n    const lexer = new lunr.QueryLexer(part)\n    lexer.run()\n\n    /* Extract and tokenize term from lexeme */\n    for (const { type, str: term, start, end } of lexer.lexemes)\n      switch (type) {\n\n        /* Hack: remove colon - see https://bit.ly/3wD3T3I */\n        case \"FIELD\":\n          if (![\"title\", \"text\", \"tags\"].includes(term))\n            part = [\n              part.slice(0, end),\n              \" \",\n              part.slice(end + 1)\n            ].join(\"\")\n          break\n\n        /* Tokenize term */\n        case \"TERM\":\n          split(term, lunr.tokenizer.separator, (...range) => {\n            terms.push([\n              part.slice(0, start),\n              term.slice(...range),\n              part.slice(end)\n            ].join(\"\"))\n          })\n      }\n\n    /* Return terms */\n    return terms\n  })\n}\n\n/* ------------------------------------------------------------------------- */\n\n/**\n * Parse a search query for analysis\n *\n * Lunr.js itself has a bug where it doesn't detect or remove wildcards for\n * query clauses, so we must do this here.\n *\n * @see https://bit.ly/3DpTGtz - GitHub issue\n *\n * @param value - Query value\n *\n * @returns Search query clauses\n */\nexport function parseSearchQuery(\n  value: string\n): SearchQueryClause[] {\n  const query  = new lunr.Query([\"title\", \"text\", \"tags\"])\n  const parser = new lunr.QueryParser(value, query)\n\n  /* Parse Search query */\n  parser.parse()\n  for (const clause of query.clauses) {\n    clause.usePipeline = true\n\n    /* Handle leading wildcard */\n    if (clause.term.startsWith(\"*\")) {\n      clause.wildcard = lunr.Query.wildcard.LEADING\n      clause.term = clause.term.slice(1)\n    }\n\n    /* Handle trailing wildcard */\n    if (clause.term.endsWith(\"*\")) {\n      clause.wildcard = lunr.Query.wildcard.TRAILING\n      clause.term = clause.term.slice(0, -1)\n    }\n  }\n\n  /* Return query clauses */\n  return query.clauses\n}\n\n/**\n * Analyze the search query clauses in regard to the search terms found\n *\n * @param query - Search query clauses\n * @param terms - Search terms\n *\n * @returns Search query terms\n */\nexport function getSearchQueryTerms(\n  query: SearchQueryClause[], terms: string[]\n): SearchQueryTerms {\n  const clauses = new Set<SearchQueryClause>(query)\n\n  /* Match query clauses against terms */\n  const result: SearchQueryTerms = {}\n  for (let t = 0; t < terms.length; t++)\n    for (const clause of clauses)\n      if (terms[t].startsWith(clause.term)) {\n        result[clause.term] = true\n        clauses.delete(clause)\n      }\n\n  /* Annotate unmatched non-stopword query clauses */\n  for (const clause of clauses)\n    if (lunr.stopWordFilter?.(clause.term))\n      result[clause.term] = false\n\n  /* Return query terms */\n  return result\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Segment a search query using the inverted index\n *\n * This function implements a clever approach to text segmentation for Asian\n * languages, as it used the information already available in the search index.\n * The idea is to greedily segment the search query based on the tokens that are\n * already part of the index, as described in the linked issue.\n *\n * @see https://bit.ly/3lwjrk7 - GitHub issue\n *\n * @param query - Query value\n * @param index - Inverted index\n *\n * @returns Segmented query value\n */\nexport function segment(\n  query: string, index: object\n): Iterable<string> {\n  const segments = new Set<string>()\n\n  /* Segment search query */\n  const wordcuts = new Uint16Array(query.length)\n  for (let i = 0; i < query.length; i++)\n    for (let j = i + 1; j < query.length; j++) {\n      const value = query.slice(i, j)\n      if (value in index)\n        wordcuts[i] = j - i\n    }\n\n  /* Compute longest matches with minimum overlap */\n  const stack = [0]\n  for (let s = stack.length; s > 0;) {\n    const p = stack[--s]\n    for (let q = 1; q < wordcuts[p]; q++)\n      if (wordcuts[p + q] > wordcuts[p] - q) {\n        segments.add(query.slice(p, p + q))\n        stack[s++] = p + q\n      }\n\n    /* Continue at end of query string */\n    const q = p + wordcuts[p]\n    if (wordcuts[q] && q < query.length - 1)\n      stack[s++] = q\n\n    /* Add current segment */\n    segments.add(query.slice(p, q))\n  }\n\n  // @todo fix this case in the code block above, this is a hotfix\n  if (segments.has(\"\"))\n    return new Set([query])\n\n  /* Return segmented query value */\n  return segments\n}\n", "/*\n * Copyright (c) 2016-2024 Martin Donath <martin.donath@squidfunk.com>\n *\n * Permission is hereby granted, free of charge, to any person obtaining a copy\n * of this software and associated documentation files (the \"Software\"), to\n * deal in the Software without restriction, including without limitation the\n * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or\n * sell copies of the Software, and to permit persons to whom the Software is\n * furnished to do so, subject to the following conditions:\n *\n * The above copyright notice and this permission notice shall be included in\n * all copies or substantial portions of the Software.\n *\n * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE\n * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS\n * IN THE SOFTWARE.\n */\n\nimport {\n  SearchDocument,\n  SearchIndex,\n  SearchOptions,\n  setupSearchDocumentMap\n} from \"../config\"\nimport {\n  Position,\n  PositionTable,\n  highlight,\n  highlightAll,\n  tokenize\n} from \"../internal\"\nimport {\n  SearchQueryTerms,\n  getSearchQueryTerms,\n  parseSearchQuery,\n  segment,\n  transformSearchQuery\n} from \"../query\"\n\n/* ----------------------------------------------------------------------------\n * Types\n * ------------------------------------------------------------------------- */\n\n/**\n * Search item\n */\nexport interface SearchItem\n  extends SearchDocument\n{\n  score: number                        /* Score (relevance) */\n  terms: SearchQueryTerms              /* Search query terms */\n}\n\n/**\n * Search result\n */\nexport interface SearchResult {\n  items: SearchItem[][]                /* Search items */\n  suggest?: string[]                   /* Search suggestions */\n}\n\n/* ----------------------------------------------------------------------------\n * Functions\n * ------------------------------------------------------------------------- */\n\n/**\n * Create field extractor factory\n *\n * @param table - Position table map\n *\n * @returns Extractor factory\n */\nfunction extractor(table: Map<string, PositionTable>) {\n  return (name: keyof SearchDocument) => {\n    return (doc: SearchDocument) => {\n      if (typeof doc[name] === \"undefined\")\n        return undefined\n\n      /* Compute identifier and initialize table */\n      const id = [doc.location, name].join(\":\")\n      table.set(id, lunr.tokenizer.table = [])\n\n      /* Return field value */\n      return doc[name]\n    }\n  }\n}\n\n/**\n * Compute the difference of two lists of strings\n *\n * @param a - 1st list of strings\n * @param b - 2nd list of strings\n *\n * @returns Difference\n */\nfunction difference(a: string[], b: string[]): string[] {\n  const [x, y] = [new Set(a), new Set(b)]\n  return [\n    ...new Set([...x].filter(value => !y.has(value)))\n  ]\n}\n\n/* ----------------------------------------------------------------------------\n * Class\n * ------------------------------------------------------------------------- */\n\n/**\n * Search index\n */\nexport class Search {\n\n  /**\n   * Search document map\n   */\n  protected map: Map<string, SearchDocument>\n\n  /**\n   * Search options\n   */\n  protected options: SearchOptions\n\n  /**\n   * The underlying Lunr.js search index\n   */\n  protected index: lunr.Index\n\n  /**\n   * Internal position table map\n   */\n  protected table: Map<string, PositionTable>\n\n  /**\n   * Create the search integration\n   *\n   * @param data - Search index\n   */\n  public constructor({ config, docs, options }: SearchIndex) {\n    const field = extractor(this.table = new Map())\n\n    /* Set up document map and options */\n    this.map = setupSearchDocumentMap(docs)\n    this.options = options\n\n    /* Set up document index */\n    this.index = lunr(function () {\n      this.metadataWhitelist = [\"position\"]\n      this.b(0)\n\n      /* Set up (multi-)language support */\n      if (config.lang.length === 1 && config.lang[0] !== \"en\") {\n        // @ts-expect-error - namespace indexing not supported\n        this.use(lunr[config.lang[0]])\n      } else if (config.lang.length > 1) {\n        this.use(lunr.multiLanguage(...config.lang))\n      }\n\n      /* Set up custom tokenizer (must be after language setup) */\n      this.tokenizer = tokenize as typeof lunr.tokenizer\n      lunr.tokenizer.separator = new RegExp(config.separator)\n\n      /* Set up custom segmenter, if loaded */\n      lunr.segmenter = \"TinySegmenter\" in lunr\n        ? new lunr.TinySegmenter()\n        : undefined\n\n      /* Compute functions to be removed from the pipeline */\n      const fns = difference([\n        \"trimmer\", \"stopWordFilter\", \"stemmer\"\n      ], config.pipeline)\n\n      /* Remove functions from the pipeline for registered languages */\n      for (const lang of config.lang.map(language => (\n        // @ts-expect-error - namespace indexing not supported\n        language === \"en\" ? lunr : lunr[language]\n      )))\n        for (const fn of fns) {\n          this.pipeline.remove(lang[fn])\n          this.searchPipeline.remove(lang[fn])\n        }\n\n      /* Set up index reference */\n      this.ref(\"location\")\n\n      /* Set up index fields */\n      this.field(\"title\", { boost: 1e3, extractor: field(\"title\") })\n      this.field(\"text\",  { boost: 1e0, extractor: field(\"text\") })\n      this.field(\"tags\",  { boost: 1e6, extractor: field(\"tags\") })\n\n      /* Add documents to index */\n      for (const doc of docs)\n        this.add(doc, { boost: doc.boost })\n    })\n  }\n\n  /**\n   * Search for matching documents\n   *\n   * @param query - Search query\n   *\n   * @returns Search result\n   */\n  public search(query: string): SearchResult {\n\n    // Experimental Chinese segmentation\n    query = query.replace(/\\p{sc=Han}+/gu, value => {\n      return [...segment(value, this.index.invertedIndex)]\n        .join(\"* \")\n    })\n\n    // @todo: move segmenter (above) into transformSearchQuery\n    query = transformSearchQuery(query)\n    if (!query)\n      return { items: [] }\n\n    /* Parse query to extract clauses for analysis */\n    const clauses = parseSearchQuery(query)\n      .filter(clause => (\n        clause.presence !== lunr.Query.presence.PROHIBITED\n      ))\n\n    /* Perform search and post-process results */\n    const groups = this.index.search(query)\n\n      /* Apply post-query boosts based on title and search query terms */\n      .reduce<SearchItem[]>((item, { ref, score, matchData }) => {\n        let doc = this.map.get(ref)\n        if (typeof doc !== \"undefined\") {\n\n          /* Shallow copy document */\n          doc = { ...doc }\n          if (doc.tags)\n            doc.tags = [...doc.tags]\n\n          /* Compute and analyze search query terms */\n          const terms = getSearchQueryTerms(\n            clauses,\n            Object.keys(matchData.metadata)\n          )\n\n          /* Highlight matches in fields */\n          for (const field of this.index.fields) {\n            if (typeof doc[field] === \"undefined\")\n              continue\n\n            /* Collect positions from matches */\n            const positions: Position[] = []\n            for (const match of Object.values(matchData.metadata))\n              if (typeof match[field] !== \"undefined\")\n                positions.push(...match[field].position)\n\n            /* Skip highlighting, if no positions were collected */\n            if (!positions.length)\n              continue\n\n            /* Load table and determine highlighting method */\n            const table = this.table.get([doc.location, field].join(\":\"))!\n            const fn = Array.isArray(doc[field])\n              ? highlightAll\n              : highlight\n\n            // @ts-expect-error - stop moaning, TypeScript!\n            doc[field] = fn(doc[field], table, positions, field !== \"text\")\n          }\n\n          /* Highlight title and text and apply post-query boosts */\n          const boost = +!doc.parent +\n            Object.values(terms)\n              .filter(t => t).length /\n            Object.keys(terms).length\n\n          /* Append item */\n          item.push({\n            ...doc,\n            score: score * (1 + boost ** 2),\n            terms\n          })\n        }\n        return item\n      }, [])\n\n      /* Sort search results again after applying boosts */\n      .sort((a, b) => b.score - a.score)\n\n      /* Group search results by article */\n      .reduce((items, result) => {\n        const doc = this.map.get(result.location)\n        if (typeof doc !== \"undefined\") {\n          const ref = doc.parent\n            ? doc.parent.location\n            : doc.location\n          items.set(ref, [...items.get(ref) || [], result])\n        }\n        return items\n      }, new Map<string, SearchItem[]>())\n\n    /* Ensure that every item set has an article */\n    for (const [ref, items] of groups)\n      if (!items.find(item => item.location === ref)) {\n        const doc = this.map.get(ref)!\n        items.push({ ...doc, score: 0, terms: {} })\n      }\n\n    /* Generate search suggestions, if desired */\n    let suggest: string[] | undefined\n    if (this.options.suggest) {\n      const titles = this.index.query(builder => {\n        for (const clause of clauses)\n          builder.term(clause.term, {\n            fields: [\"title\"],\n            presence: lunr.Query.presence.REQUIRED,\n            wildcard: lunr.Query.wildcard.TRAILING\n          })\n      })\n\n      /* Retrieve suggestions for best match */\n      suggest = titles.length\n        ? Object.keys(titles[0].matchData.metadata)\n        : []\n    }\n\n    /* Return search result */\n    return {\n      items: [...groups.values()],\n      ...typeof suggest !== \"undefined\" && { suggest }\n    }\n  }\n}\n"],
   "mappings": "6lCAAA,IAAAA,GAAAC,GAAA,CAAAC,EAAAC,KAAA;AAAA;AAAA;AAAA;AAAA,IAME,UAAU,CAiCZ,IAAIC,EAAO,SAAUC,EAAQ,CAC3B,IAAIC,EAAU,IAAIF,EAAK,QAEvB,OAAAE,EAAQ,SAAS,IACfF,EAAK,QACLA,EAAK,eACLA,EAAK,OACP,EAEAE,EAAQ,eAAe,IACrBF,EAAK,OACP,EAEAC,EAAO,KAAKC,EAASA,CAAO,EACrBA,EAAQ,MAAM,CACvB,EAEAF,EAAK,QAAU,QACf;AAAA;AAAA;AAAA,GASAA,EAAK,MAAQ,CAAC,EASdA,EAAK,MAAM,KAAQ,SAAUG,EAAQ,CAEnC,OAAO,SAAUC,EAAS,CACpBD,EAAO,SAAW,QAAQ,MAC5B,QAAQ,KAAKC,CAAO,CAExB,CAEF,EAAG,IAAI,EAaPJ,EAAK,MAAM,SAAW,SAAUK,EAAK,CACnC,OAAsBA,GAAQ,KACrB,GAEAA,EAAI,SAAS,CAExB,EAkBAL,EAAK,MAAM,MAAQ,SAAUK,EAAK,CAChC,GAAIA,GAAQ,KACV,OAAOA,EAMT,QAHIC,EAAQ,OAAO,OAAO,IAAI,EAC1BC,EAAO,OAAO,KAAKF,CAAG,EAEjB,EAAI,EAAG,EAAIE,EAAK,OAAQ,IAAK,CACpC,IAAIC,EAAMD,EAAK,CAAC,EACZE,EAAMJ,EAAIG,CAAG,EAEjB,GAAI,MAAM,QAAQC,CAAG,EAAG,CACtBH,EAAME,CAAG,EAAIC,EAAI,MAAM,EACvB,QACF,CAEA,GAAI,OAAOA,GAAQ,UACf,OAAOA,GAAQ,UACf,OAAOA,GAAQ,UAAW,CAC5BH,EAAME,CAAG,EAAIC,EACb,QACF,CAEA,MAAM,IAAI,UAAU,uDAAuD,CAC7E,CAEA,OAAOH,CACT,EACAN,EAAK,SAAW,SAAUU,EAAQC,EAAWC,EAAa,CACxD,KAAK,OAASF,EACd,KAAK,UAAYC,EACjB,KAAK,aAAeC,CACtB,EAEAZ,EAAK,SAAS,OAAS,IAEvBA,EAAK,SAAS,WAAa,SAAUa,EAAG,CACtC,IAAIC,EAAID,EAAE,QAAQb,EAAK,SAAS,MAAM,EAEtC,GAAIc,IAAM,GACR,KAAM,6BAGR,IAAIC,EAAWF,EAAE,MAAM,EAAGC,CAAC,EACvBJ,EAASG,EAAE,MAAMC,EAAI,CAAC,EAE1B,OAAO,IAAId,EAAK,SAAUU,EAAQK,EAAUF,CAAC,CAC/C,EAEAb,EAAK,SAAS,UAAU,SAAW,UAAY,CAC7C,OAAI,KAAK,cAAgB,OACvB,KAAK,aAAe,KAAK,UAAYA,EAAK,SAAS,OAAS,KAAK,QAG5D,KAAK,YACd,EACA;AAAA;AAAA;AAAA,GAUAA,EAAK,IAAM,SAAUgB,EAAU,CAG7B,GAFA,KAAK,SAAW,OAAO,OAAO,IAAI,EAE9BA,EAAU,CACZ,KAAK,OAASA,EAAS,OAEvB,QAASC,EAAI,EAAGA,EAAI,KAAK,OAAQA,IAC/B,KAAK,SAASD,EAASC,CAAC,CAAC,EAAI,EAEjC,MACE,KAAK,OAAS,CAElB,EASAjB,EAAK,IAAI,SAAW,CAClB,UAAW,SAAUkB,EAAO,CAC1B,OAAOA,CACT,EAEA,MAAO,UAAY,CACjB,OAAO,IACT,EAEA,SAAU,UAAY,CACpB,MAAO,EACT,CACF,EASAlB,EAAK,IAAI,MAAQ,CACf,UAAW,UAAY,CACrB,OAAO,IACT,EAEA,MAAO,SAAUkB,EAAO,CACtB,OAAOA,CACT,EAEA,SAAU,UAAY,CACpB,MAAO,EACT,CACF,EAQAlB,EAAK,IAAI,UAAU,SAAW,SAAUmB,EAAQ,CAC9C,MAAO,CAAC,CAAC,KAAK,SAASA,CAAM,CAC/B,EAUAnB,EAAK,IAAI,UAAU,UAAY,SAAUkB,EAAO,CAC9C,IAAIE,EAAGC,EAAGL,EAAUM,EAAe,CAAC,EAEpC,GAAIJ,IAAUlB,EAAK,IAAI,SACrB,OAAO,KAGT,GAAIkB,IAAUlB,EAAK,IAAI,MACrB,OAAOkB,EAGL,KAAK,OAASA,EAAM,QACtBE,EAAI,KACJC,EAAIH,IAEJE,EAAIF,EACJG,EAAI,MAGNL,EAAW,OAAO,KAAKI,EAAE,QAAQ,EAEjC,QAASH,EAAI,EAAGA,EAAID,EAAS,OAAQC,IAAK,CACxC,IAAIM,EAAUP,EAASC,CAAC,EACpBM,KAAWF,EAAE,UACfC,EAAa,KAAKC,CAAO,CAE7B,CAEA,OAAO,IAAIvB,EAAK,IAAKsB,CAAY,CACnC,EASAtB,EAAK,IAAI,UAAU,MAAQ,SAAUkB,EAAO,CAC1C,OAAIA,IAAUlB,EAAK,IAAI,SACdA,EAAK,IAAI,SAGdkB,IAAUlB,EAAK,IAAI,MACd,KAGF,IAAIA,EAAK,IAAI,OAAO,KAAK,KAAK,QAAQ,EAAE,OAAO,OAAO,KAAKkB,EAAM,QAAQ,CAAC,CAAC,CACpF,EASAlB,EAAK,IAAM,SAAUwB,EAASC,EAAe,CAC3C,IAAIC,EAAoB,EAExB,QAASf,KAAaa,EAChBb,GAAa,WACjBe,GAAqB,OAAO,KAAKF,EAAQb,CAAS,CAAC,EAAE,QAGvD,IAAIgB,GAAKF,EAAgBC,EAAoB,KAAQA,EAAoB,IAEzE,OAAO,KAAK,IAAI,EAAI,KAAK,IAAIC,CAAC,CAAC,CACjC,EAUA3B,EAAK,MAAQ,SAAU4B,EAAKC,EAAU,CACpC,KAAK,IAAMD,GAAO,GAClB,KAAK,SAAWC,GAAY,CAAC,CAC/B,EAOA7B,EAAK,MAAM,UAAU,SAAW,UAAY,CAC1C,OAAO,KAAK,GACd,EAsBAA,EAAK,MAAM,UAAU,OAAS,SAAU8B,EAAI,CAC1C,YAAK,IAAMA,EAAG,KAAK,IAAK,KAAK,QAAQ,EAC9B,IACT,EASA9B,EAAK,MAAM,UAAU,MAAQ,SAAU8B,EAAI,CACzC,OAAAA,EAAKA,GAAM,SAAUjB,EAAG,CAAE,OAAOA,CAAE,EAC5B,IAAIb,EAAK,MAAO8B,EAAG,KAAK,IAAK,KAAK,QAAQ,EAAG,KAAK,QAAQ,CACnE,EACA;AAAA;AAAA;AAAA,GAuBA9B,EAAK,UAAY,SAAUK,EAAKwB,EAAU,CACxC,GAAIxB,GAAO,MAAQA,GAAO,KACxB,MAAO,CAAC,EAGV,GAAI,MAAM,QAAQA,CAAG,EACnB,OAAOA,EAAI,IAAI,SAAU0B,EAAG,CAC1B,OAAO,IAAI/B,EAAK,MACdA,EAAK,MAAM,SAAS+B,CAAC,EAAE,YAAY,EACnC/B,EAAK,MAAM,MAAM6B,CAAQ,CAC3B,CACF,CAAC,EAOH,QAJID,EAAMvB,EAAI,SAAS,EAAE,YAAY,EACjC2B,EAAMJ,EAAI,OACVK,EAAS,CAAC,EAELC,EAAW,EAAGC,EAAa,EAAGD,GAAYF,EAAKE,IAAY,CAClE,IAAIE,EAAOR,EAAI,OAAOM,CAAQ,EAC1BG,EAAcH,EAAWC,EAE7B,GAAKC,EAAK,MAAMpC,EAAK,UAAU,SAAS,GAAKkC,GAAYF,EAAM,CAE7D,GAAIK,EAAc,EAAG,CACnB,IAAIC,EAAgBtC,EAAK,MAAM,MAAM6B,CAAQ,GAAK,CAAC,EACnDS,EAAc,SAAc,CAACH,EAAYE,CAAW,EACpDC,EAAc,MAAWL,EAAO,OAEhCA,EAAO,KACL,IAAIjC,EAAK,MACP4B,EAAI,MAAMO,EAAYD,CAAQ,EAC9BI,CACF,CACF,CACF,CAEAH,EAAaD,EAAW,CAC1B,CAEF,CAEA,OAAOD,CACT,EASAjC,EAAK,UAAU,UAAY,UAC3B;AAAA;AAAA;AAAA,GAkCAA,EAAK,SAAW,UAAY,CAC1B,KAAK,OAAS,CAAC,CACjB,EAEAA,EAAK,SAAS,oBAAsB,OAAO,OAAO,IAAI,EAmCtDA,EAAK,SAAS,iBAAmB,SAAU8B,EAAIS,EAAO,CAChDA,KAAS,KAAK,qBAChBvC,EAAK,MAAM,KAAK,6CAA+CuC,CAAK,EAGtET,EAAG,MAAQS,EACXvC,EAAK,SAAS,oBAAoB8B,EAAG,KAAK,EAAIA,CAChD,EAQA9B,EAAK,SAAS,4BAA8B,SAAU8B,EAAI,CACxD,IAAIU,EAAeV,EAAG,OAAUA,EAAG,SAAS,KAAK,oBAE5CU,GACHxC,EAAK,MAAM,KAAK;AAAA,EAAmG8B,CAAE,CAEzH,EAYA9B,EAAK,SAAS,KAAO,SAAUyC,EAAY,CACzC,IAAIC,EAAW,IAAI1C,EAAK,SAExB,OAAAyC,EAAW,QAAQ,SAAUE,EAAQ,CACnC,IAAIb,EAAK9B,EAAK,SAAS,oBAAoB2C,CAAM,EAEjD,GAAIb,EACFY,EAAS,IAAIZ,CAAE,MAEf,OAAM,IAAI,MAAM,sCAAwCa,CAAM,CAElE,CAAC,EAEMD,CACT,EASA1C,EAAK,SAAS,UAAU,IAAM,UAAY,CACxC,IAAI4C,EAAM,MAAM,UAAU,MAAM,KAAK,SAAS,EAE9CA,EAAI,QAAQ,SAAUd,EAAI,CACxB9B,EAAK,SAAS,4BAA4B8B,CAAE,EAC5C,KAAK,OAAO,KAAKA,CAAE,CACrB,EAAG,IAAI,CACT,EAWA9B,EAAK,SAAS,UAAU,MAAQ,SAAU6C,EAAYC,EAAO,CAC3D9C,EAAK,SAAS,4BAA4B8C,CAAK,EAE/C,IAAIC,EAAM,KAAK,OAAO,QAAQF,CAAU,EACxC,GAAIE,GAAO,GACT,MAAM,IAAI,MAAM,wBAAwB,EAG1CA,EAAMA,EAAM,EACZ,KAAK,OAAO,OAAOA,EAAK,EAAGD,CAAK,CAClC,EAWA9C,EAAK,SAAS,UAAU,OAAS,SAAU6C,EAAYC,EAAO,CAC5D9C,EAAK,SAAS,4BAA4B8C,CAAK,EAE/C,IAAIC,EAAM,KAAK,OAAO,QAAQF,CAAU,EACxC,GAAIE,GAAO,GACT,MAAM,IAAI,MAAM,wBAAwB,EAG1C,KAAK,OAAO,OAAOA,EAAK,EAAGD,CAAK,CAClC,EAOA9C,EAAK,SAAS,UAAU,OAAS,SAAU8B,EAAI,CAC7C,IAAIiB,EAAM,KAAK,OAAO,QAAQjB,CAAE,EAC5BiB,GAAO,IAIX,KAAK,OAAO,OAAOA,EAAK,CAAC,CAC3B,EASA/C,EAAK,SAAS,UAAU,IAAM,SAAUiC,EAAQ,CAG9C,QAFIe,EAAc,KAAK,OAAO,OAErB/B,EAAI,EAAGA,EAAI+B,EAAa/B,IAAK,CAIpC,QAHIa,EAAK,KAAK,OAAOb,CAAC,EAClBgC,EAAO,CAAC,EAEHC,EAAI,EAAGA,EAAIjB,EAAO,OAAQiB,IAAK,CACtC,IAAIC,EAASrB,EAAGG,EAAOiB,CAAC,EAAGA,EAAGjB,CAAM,EAEpC,GAAI,EAAAkB,GAAW,MAA6BA,IAAW,IAEvD,GAAI,MAAM,QAAQA,CAAM,EACtB,QAASC,EAAI,EAAGA,EAAID,EAAO,OAAQC,IACjCH,EAAK,KAAKE,EAAOC,CAAC,CAAC,OAGrBH,EAAK,KAAKE,CAAM,CAEpB,CAEAlB,EAASgB,CACX,CAEA,OAAOhB,CACT,EAYAjC,EAAK,SAAS,UAAU,UAAY,SAAU4B,EAAKC,EAAU,CAC3D,IAAIwB,EAAQ,IAAIrD,EAAK,MAAO4B,EAAKC,CAAQ,EAEzC,OAAO,KAAK,IAAI,CAACwB,CAAK,CAAC,EAAE,IAAI,SAAUtB,EAAG,CACxC,OAAOA,EAAE,SAAS,CACpB,CAAC,CACH,EAMA/B,EAAK,SAAS,UAAU,MAAQ,UAAY,CAC1C,KAAK,OAAS,CAAC,CACjB,EASAA,EAAK,SAAS,UAAU,OAAS,UAAY,CAC3C,OAAO,KAAK,OAAO,IAAI,SAAU8B,EAAI,CACnC,OAAA9B,EAAK,SAAS,4BAA4B8B,CAAE,EAErCA,EAAG,KACZ,CAAC,CACH,EACA;AAAA;AAAA;AAAA,GAqBA9B,EAAK,OAAS,SAAUgB,EAAU,CAChC,KAAK,WAAa,EAClB,KAAK,SAAWA,GAAY,CAAC,CAC/B,EAaAhB,EAAK,OAAO,UAAU,iBAAmB,SAAUsD,EAAO,CAExD,GAAI,KAAK,SAAS,QAAU,EAC1B,MAAO,GAST,QANIC,EAAQ,EACRC,EAAM,KAAK,SAAS,OAAS,EAC7BnB,EAAcmB,EAAMD,EACpBE,EAAa,KAAK,MAAMpB,EAAc,CAAC,EACvCqB,EAAa,KAAK,SAASD,EAAa,CAAC,EAEtCpB,EAAc,IACfqB,EAAaJ,IACfC,EAAQE,GAGNC,EAAaJ,IACfE,EAAMC,GAGJC,GAAcJ,IAIlBjB,EAAcmB,EAAMD,EACpBE,EAAaF,EAAQ,KAAK,MAAMlB,EAAc,CAAC,EAC/CqB,EAAa,KAAK,SAASD,EAAa,CAAC,EAO3C,GAJIC,GAAcJ,GAIdI,EAAaJ,EACf,OAAOG,EAAa,EAGtB,GAAIC,EAAaJ,EACf,OAAQG,EAAa,GAAK,CAE9B,EAWAzD,EAAK,OAAO,UAAU,OAAS,SAAU2D,EAAWlD,EAAK,CACvD,KAAK,OAAOkD,EAAWlD,EAAK,UAAY,CACtC,KAAM,iBACR,CAAC,CACH,EAUAT,EAAK,OAAO,UAAU,OAAS,SAAU2D,EAAWlD,EAAKqB,EAAI,CAC3D,KAAK,WAAa,EAClB,IAAI8B,EAAW,KAAK,iBAAiBD,CAAS,EAE1C,KAAK,SAASC,CAAQ,GAAKD,EAC7B,KAAK,SAASC,EAAW,CAAC,EAAI9B,EAAG,KAAK,SAAS8B,EAAW,CAAC,EAAGnD,CAAG,EAEjE,KAAK,SAAS,OAAOmD,EAAU,EAAGD,EAAWlD,CAAG,CAEpD,EAOAT,EAAK,OAAO,UAAU,UAAY,UAAY,CAC5C,GAAI,KAAK,WAAY,OAAO,KAAK,WAKjC,QAHI6D,EAAe,EACfC,EAAiB,KAAK,SAAS,OAE1B7C,EAAI,EAAGA,EAAI6C,EAAgB7C,GAAK,EAAG,CAC1C,IAAIR,EAAM,KAAK,SAASQ,CAAC,EACzB4C,GAAgBpD,EAAMA,CACxB,CAEA,OAAO,KAAK,WAAa,KAAK,KAAKoD,CAAY,CACjD,EAQA7D,EAAK,OAAO,UAAU,IAAM,SAAU+D,EAAa,CAOjD,QANIC,EAAa,EACb5C,EAAI,KAAK,SAAUC,EAAI0C,EAAY,SACnCE,EAAO7C,EAAE,OAAQ8C,EAAO7C,EAAE,OAC1B8C,EAAO,EAAGC,EAAO,EACjBnD,EAAI,EAAGiC,EAAI,EAERjC,EAAIgD,GAAQf,EAAIgB,GACrBC,EAAO/C,EAAEH,CAAC,EAAGmD,EAAO/C,EAAE6B,CAAC,EACnBiB,EAAOC,EACTnD,GAAK,EACIkD,EAAOC,EAChBlB,GAAK,EACIiB,GAAQC,IACjBJ,GAAc5C,EAAEH,EAAI,CAAC,EAAII,EAAE6B,EAAI,CAAC,EAChCjC,GAAK,EACLiC,GAAK,GAIT,OAAOc,CACT,EASAhE,EAAK,OAAO,UAAU,WAAa,SAAU+D,EAAa,CACxD,OAAO,KAAK,IAAIA,CAAW,EAAI,KAAK,UAAU,GAAK,CACrD,EAOA/D,EAAK,OAAO,UAAU,QAAU,UAAY,CAG1C,QAFIqE,EAAS,IAAI,MAAO,KAAK,SAAS,OAAS,CAAC,EAEvCpD,EAAI,EAAGiC,EAAI,EAAGjC,EAAI,KAAK,SAAS,OAAQA,GAAK,EAAGiC,IACvDmB,EAAOnB,CAAC,EAAI,KAAK,SAASjC,CAAC,EAG7B,OAAOoD,CACT,EAOArE,EAAK,OAAO,UAAU,OAAS,UAAY,CACzC,OAAO,KAAK,QACd,EAEA;AAAA;AAAA;AAAA;AAAA,GAiBAA,EAAK,QAAW,UAAU,CACxB,IAAIsE,EAAY,CACZ,QAAY,MACZ,OAAW,OACX,KAAS,OACT,KAAS,OACT,KAAS,MACT,IAAQ,MACR,KAAS,KACT,MAAU,MACV,IAAQ,IACR,MAAU,MACV,QAAY,MACZ,MAAU,MACV,KAAS,MACT,MAAU,KACV,QAAY,MACZ,QAAY,MACZ,QAAY,MACZ,MAAU,KACV,MAAU,MACV,OAAW,MACX,KAAS,KACX,EAEAC,EAAY,CACV,MAAU,KACV,MAAU,GACV,MAAU,KACV,MAAU,KACV,KAAS,KACT,IAAQ,GACR,KAAS,EACX,EAEAC,EAAI,WACJC,EAAI,WACJC,EAAIF,EAAI,aACRG,EAAIF,EAAI,WAERG,EAAO,KAAOF,EAAI,KAAOC,EAAID,EAC7BG,EAAO,KAAOH,EAAI,KAAOC,EAAID,EAAI,IAAMC,EAAI,MAC3CG,EAAO,KAAOJ,EAAI,KAAOC,EAAID,EAAIC,EAAID,EACrCK,EAAM,KAAOL,EAAI,KAAOD,EAEtBO,EAAU,IAAI,OAAOJ,CAAI,EACzBK,EAAU,IAAI,OAAOH,CAAI,EACzBI,EAAU,IAAI,OAAOL,CAAI,EACzBM,EAAS,IAAI,OAAOJ,CAAG,EAEvBK,EAAQ,kBACRC,EAAS,iBACTC,EAAQ,aACRC,EAAS,kBACTC,EAAU,KACVC,EAAW,cACXC,EAAW,IAAI,OAAO,oBAAoB,EAC1CC,EAAW,IAAI,OAAO,IAAMjB,EAAID,EAAI,cAAc,EAElDmB,EAAQ,mBACRC,EAAO,2IAEPC,EAAO,iDAEPC,EAAO,sFACPC,EAAQ,oBAERC,EAAO,WACPC,EAAS,MACTC,EAAQ,IAAI,OAAO,IAAMzB,EAAID,EAAI,cAAc,EAE/C2B,EAAgB,SAAuBC,EAAG,CAC5C,IAAIC,EACFC,EACAC,EACAC,EACAC,EACAC,EACAC,EAEF,GAAIP,EAAE,OAAS,EAAK,OAAOA,EAiB3B,GAfAG,EAAUH,EAAE,OAAO,EAAE,CAAC,EAClBG,GAAW,MACbH,EAAIG,EAAQ,YAAY,EAAIH,EAAE,OAAO,CAAC,GAIxCI,EAAKrB,EACLsB,EAAMrB,EAEFoB,EAAG,KAAKJ,CAAC,EAAKA,EAAIA,EAAE,QAAQI,EAAG,MAAM,EAChCC,EAAI,KAAKL,CAAC,IAAKA,EAAIA,EAAE,QAAQK,EAAI,MAAM,GAGhDD,EAAKnB,EACLoB,EAAMnB,EACFkB,EAAG,KAAKJ,CAAC,EAAG,CACd,IAAIQ,EAAKJ,EAAG,KAAKJ,CAAC,EAClBI,EAAKzB,EACDyB,EAAG,KAAKI,EAAG,CAAC,CAAC,IACfJ,EAAKjB,EACLa,EAAIA,EAAE,QAAQI,EAAG,EAAE,EAEvB,SAAWC,EAAI,KAAKL,CAAC,EAAG,CACtB,IAAIQ,EAAKH,EAAI,KAAKL,CAAC,EACnBC,EAAOO,EAAG,CAAC,EACXH,EAAMvB,EACFuB,EAAI,KAAKJ,CAAI,IACfD,EAAIC,EACJI,EAAMjB,EACNkB,EAAMjB,EACNkB,EAAMjB,EACFe,EAAI,KAAKL,CAAC,EAAKA,EAAIA,EAAI,IAClBM,EAAI,KAAKN,CAAC,GAAKI,EAAKjB,EAASa,EAAIA,EAAE,QAAQI,EAAG,EAAE,GAChDG,EAAI,KAAKP,CAAC,IAAKA,EAAIA,EAAI,KAEpC,CAIA,GADAI,EAAKb,EACDa,EAAG,KAAKJ,CAAC,EAAG,CACd,IAAIQ,EAAKJ,EAAG,KAAKJ,CAAC,EAClBC,EAAOO,EAAG,CAAC,EACXR,EAAIC,EAAO,GACb,CAIA,GADAG,EAAKZ,EACDY,EAAG,KAAKJ,CAAC,EAAG,CACd,IAAIQ,EAAKJ,EAAG,KAAKJ,CAAC,EAClBC,EAAOO,EAAG,CAAC,EACXN,EAASM,EAAG,CAAC,EACbJ,EAAKzB,EACDyB,EAAG,KAAKH,CAAI,IACdD,EAAIC,EAAOhC,EAAUiC,CAAM,EAE/B,CAIA,GADAE,EAAKX,EACDW,EAAG,KAAKJ,CAAC,EAAG,CACd,IAAIQ,EAAKJ,EAAG,KAAKJ,CAAC,EAClBC,EAAOO,EAAG,CAAC,EACXN,EAASM,EAAG,CAAC,EACbJ,EAAKzB,EACDyB,EAAG,KAAKH,CAAI,IACdD,EAAIC,EAAO/B,EAAUgC,CAAM,EAE/B,CAKA,GAFAE,EAAKV,EACLW,EAAMV,EACFS,EAAG,KAAKJ,CAAC,EAAG,CACd,IAAIQ,EAAKJ,EAAG,KAAKJ,CAAC,EAClBC,EAAOO,EAAG,CAAC,EACXJ,EAAKxB,EACDwB,EAAG,KAAKH,CAAI,IACdD,EAAIC,EAER,SAAWI,EAAI,KAAKL,CAAC,EAAG,CACtB,IAAIQ,EAAKH,EAAI,KAAKL,CAAC,EACnBC,EAAOO,EAAG,CAAC,EAAIA,EAAG,CAAC,EACnBH,EAAMzB,EACFyB,EAAI,KAAKJ,CAAI,IACfD,EAAIC,EAER,CAIA,GADAG,EAAKR,EACDQ,EAAG,KAAKJ,CAAC,EAAG,CACd,IAAIQ,EAAKJ,EAAG,KAAKJ,CAAC,EAClBC,EAAOO,EAAG,CAAC,EACXJ,EAAKxB,EACLyB,EAAMxB,EACNyB,EAAMR,GACFM,EAAG,KAAKH,CAAI,GAAMI,EAAI,KAAKJ,CAAI,GAAK,CAAEK,EAAI,KAAKL,CAAI,KACrDD,EAAIC,EAER,CAEA,OAAAG,EAAKP,EACLQ,EAAMzB,EACFwB,EAAG,KAAKJ,CAAC,GAAKK,EAAI,KAAKL,CAAC,IAC1BI,EAAKjB,EACLa,EAAIA,EAAE,QAAQI,EAAG,EAAE,GAKjBD,GAAW,MACbH,EAAIG,EAAQ,YAAY,EAAIH,EAAE,OAAO,CAAC,GAGjCA,CACT,EAEA,OAAO,SAAUhD,EAAO,CACtB,OAAOA,EAAM,OAAO+C,CAAa,CACnC,CACF,EAAG,EAEHpG,EAAK,SAAS,iBAAiBA,EAAK,QAAS,SAAS,EACtD;AAAA;AAAA;AAAA,GAkBAA,EAAK,uBAAyB,SAAU8G,EAAW,CACjD,IAAIC,EAAQD,EAAU,OAAO,SAAU7D,EAAM+D,EAAU,CACrD,OAAA/D,EAAK+D,CAAQ,EAAIA,EACV/D,CACT,EAAG,CAAC,CAAC,EAEL,OAAO,SAAUI,EAAO,CACtB,GAAIA,GAAS0D,EAAM1D,EAAM,SAAS,CAAC,IAAMA,EAAM,SAAS,EAAG,OAAOA,CACpE,CACF,EAeArD,EAAK,eAAiBA,EAAK,uBAAuB,CAChD,IACA,OACA,QACA,SACA,QACA,MACA,SACA,OACA,KACA,QACA,KACA,MACA,MACA,MACA,KACA,KACA,KACA,UACA,OACA,MACA,KACA,MACA,SACA,QACA,OACA,MACA,KACA,OACA,SACA,OACA,OACA,QACA,MACA,OACA,MACA,MACA,MACA,MACA,OACA,KACA,MACA,OACA,MACA,MACA,MACA,UACA,IACA,KACA,KACA,OACA,KACA,KACA,MACA,OACA,QACA,MACA,OACA,SACA,MACA,KACA,QACA,OACA,OACA,KACA,UACA,KACA,MACA,MACA,KACA,MACA,QACA,KACA,OACA,KACA,QACA,MACA,MACA,SACA,OACA,MACA,OACA,MACA,SACA,QACA,KACA,OACA,OACA,OACA,MACA,QACA,OACA,OACA,QACA,QACA,OACA,OACA,MACA,KACA,MACA,OACA,KACA,QACA,MACA,KACA,OACA,OACA,OACA,QACA,QACA,QACA,MACA,OACA,MACA,OACA,OACA,QACA,MACA,MACA,MACF,CAAC,EAEDA,EAAK,SAAS,iBAAiBA,EAAK,eAAgB,gBAAgB,EACpE;AAAA;AAAA;AAAA,GAoBAA,EAAK,QAAU,SAAUqD,EAAO,CAC9B,OAAOA,EAAM,OAAO,SAAUxC,EAAG,CAC/B,OAAOA,EAAE,QAAQ,OAAQ,EAAE,EAAE,QAAQ,OAAQ,EAAE,CACjD,CAAC,CACH,EAEAb,EAAK,SAAS,iBAAiBA,EAAK,QAAS,SAAS,EACtD;AAAA;AAAA;AAAA,GA0BAA,EAAK,SAAW,UAAY,CAC1B,KAAK,MAAQ,GACb,KAAK,MAAQ,CAAC,EACd,KAAK,GAAKA,EAAK,SAAS,QACxBA,EAAK,SAAS,SAAW,CAC3B,EAUAA,EAAK,SAAS,QAAU,EASxBA,EAAK,SAAS,UAAY,SAAUiH,EAAK,CAGvC,QAFI/G,EAAU,IAAIF,EAAK,SAAS,QAEvBiB,EAAI,EAAGe,EAAMiF,EAAI,OAAQhG,EAAIe,EAAKf,IACzCf,EAAQ,OAAO+G,EAAIhG,CAAC,CAAC,EAGvB,OAAAf,EAAQ,OAAO,EACRA,EAAQ,IACjB,EAWAF,EAAK,SAAS,WAAa,SAAUkH,EAAQ,CAC3C,MAAI,iBAAkBA,EACblH,EAAK,SAAS,gBAAgBkH,EAAO,KAAMA,EAAO,YAAY,EAE9DlH,EAAK,SAAS,WAAWkH,EAAO,IAAI,CAE/C,EAiBAlH,EAAK,SAAS,gBAAkB,SAAU4B,EAAKuF,EAAc,CAS3D,QARIC,EAAO,IAAIpH,EAAK,SAEhBqH,EAAQ,CAAC,CACX,KAAMD,EACN,eAAgBD,EAChB,IAAKvF,CACP,CAAC,EAEMyF,EAAM,QAAQ,CACnB,IAAIC,EAAQD,EAAM,IAAI,EAGtB,GAAIC,EAAM,IAAI,OAAS,EAAG,CACxB,IAAIlF,EAAOkF,EAAM,IAAI,OAAO,CAAC,EACzBC,EAEAnF,KAAQkF,EAAM,KAAK,MACrBC,EAAaD,EAAM,KAAK,MAAMlF,CAAI,GAElCmF,EAAa,IAAIvH,EAAK,SACtBsH,EAAM,KAAK,MAAMlF,CAAI,EAAImF,GAGvBD,EAAM,IAAI,QAAU,IACtBC,EAAW,MAAQ,IAGrBF,EAAM,KAAK,CACT,KAAME,EACN,eAAgBD,EAAM,eACtB,IAAKA,EAAM,IAAI,MAAM,CAAC,CACxB,CAAC,CACH,CAEA,GAAIA,EAAM,gBAAkB,EAK5B,IAAI,MAAOA,EAAM,KAAK,MACpB,IAAIE,EAAgBF,EAAM,KAAK,MAAM,GAAG,MACnC,CACL,IAAIE,EAAgB,IAAIxH,EAAK,SAC7BsH,EAAM,KAAK,MAAM,GAAG,EAAIE,CAC1B,CAgCA,GA9BIF,EAAM,IAAI,QAAU,IACtBE,EAAc,MAAQ,IAGxBH,EAAM,KAAK,CACT,KAAMG,EACN,eAAgBF,EAAM,eAAiB,EACvC,IAAKA,EAAM,GACb,CAAC,EAKGA,EAAM,IAAI,OAAS,GACrBD,EAAM,KAAK,CACT,KAAMC,EAAM,KACZ,eAAgBA,EAAM,eAAiB,EACvC,IAAKA,EAAM,IAAI,MAAM,CAAC,CACxB,CAAC,EAKCA,EAAM,IAAI,QAAU,IACtBA,EAAM,KAAK,MAAQ,IAMjBA,EAAM,IAAI,QAAU,EAAG,CACzB,GAAI,MAAOA,EAAM,KAAK,MACpB,IAAIG,EAAmBH,EAAM,KAAK,MAAM,GAAG,MACtC,CACL,IAAIG,EAAmB,IAAIzH,EAAK,SAChCsH,EAAM,KAAK,MAAM,GAAG,EAAIG,CAC1B,CAEIH,EAAM,IAAI,QAAU,IACtBG,EAAiB,MAAQ,IAG3BJ,EAAM,KAAK,CACT,KAAMI,EACN,eAAgBH,EAAM,eAAiB,EACvC,IAAKA,EAAM,IAAI,MAAM,CAAC,CACxB,CAAC,CACH,CAKA,GAAIA,EAAM,IAAI,OAAS,EAAG,CACxB,IAAII,EAAQJ,EAAM,IAAI,OAAO,CAAC,EAC1BK,EAAQL,EAAM,IAAI,OAAO,CAAC,EAC1BM,EAEAD,KAASL,EAAM,KAAK,MACtBM,EAAgBN,EAAM,KAAK,MAAMK,CAAK,GAEtCC,EAAgB,IAAI5H,EAAK,SACzBsH,EAAM,KAAK,MAAMK,CAAK,EAAIC,GAGxBN,EAAM,IAAI,QAAU,IACtBM,EAAc,MAAQ,IAGxBP,EAAM,KAAK,CACT,KAAMO,EACN,eAAgBN,EAAM,eAAiB,EACvC,IAAKI,EAAQJ,EAAM,IAAI,MAAM,CAAC,CAChC,CAAC,CACH,EACF,CAEA,OAAOF,CACT,EAYApH,EAAK,SAAS,WAAa,SAAU4B,EAAK,CAYxC,QAXIiG,EAAO,IAAI7H,EAAK,SAChBoH,EAAOS,EAUF,EAAI,EAAG7F,EAAMJ,EAAI,OAAQ,EAAII,EAAK,IAAK,CAC9C,IAAII,EAAOR,EAAI,CAAC,EACZkG,EAAS,GAAK9F,EAAM,EAExB,GAAII,GAAQ,IACVyF,EAAK,MAAMzF,CAAI,EAAIyF,EACnBA,EAAK,MAAQC,MAER,CACL,IAAIC,EAAO,IAAI/H,EAAK,SACpB+H,EAAK,MAAQD,EAEbD,EAAK,MAAMzF,CAAI,EAAI2F,EACnBF,EAAOE,CACT,CACF,CAEA,OAAOX,CACT,EAYApH,EAAK,SAAS,UAAU,QAAU,UAAY,CAQ5C,QAPI+G,EAAQ,CAAC,EAETM,EAAQ,CAAC,CACX,OAAQ,GACR,KAAM,IACR,CAAC,EAEMA,EAAM,QAAQ,CACnB,IAAIC,EAAQD,EAAM,IAAI,EAClBW,EAAQ,OAAO,KAAKV,EAAM,KAAK,KAAK,EACpCtF,EAAMgG,EAAM,OAEZV,EAAM,KAAK,QAKbA,EAAM,OAAO,OAAO,CAAC,EACrBP,EAAM,KAAKO,EAAM,MAAM,GAGzB,QAASrG,EAAI,EAAGA,EAAIe,EAAKf,IAAK,CAC5B,IAAIgH,EAAOD,EAAM/G,CAAC,EAElBoG,EAAM,KAAK,CACT,OAAQC,EAAM,OAAO,OAAOW,CAAI,EAChC,KAAMX,EAAM,KAAK,MAAMW,CAAI,CAC7B,CAAC,CACH,CACF,CAEA,OAAOlB,CACT,EAYA/G,EAAK,SAAS,UAAU,SAAW,UAAY,CAS7C,GAAI,KAAK,KACP,OAAO,KAAK,KAOd,QAJI4B,EAAM,KAAK,MAAQ,IAAM,IACzBsG,EAAS,OAAO,KAAK,KAAK,KAAK,EAAE,KAAK,EACtClG,EAAMkG,EAAO,OAER,EAAI,EAAG,EAAIlG,EAAK,IAAK,CAC5B,IAAIO,EAAQ2F,EAAO,CAAC,EAChBL,EAAO,KAAK,MAAMtF,CAAK,EAE3BX,EAAMA,EAAMW,EAAQsF,EAAK,EAC3B,CAEA,OAAOjG,CACT,EAYA5B,EAAK,SAAS,UAAU,UAAY,SAAUqB,EAAG,CAU/C,QATIgD,EAAS,IAAIrE,EAAK,SAClBsH,EAAQ,OAERD,EAAQ,CAAC,CACX,MAAOhG,EACP,OAAQgD,EACR,KAAM,IACR,CAAC,EAEMgD,EAAM,QAAQ,CACnBC,EAAQD,EAAM,IAAI,EAWlB,QALIc,EAAS,OAAO,KAAKb,EAAM,MAAM,KAAK,EACtCc,EAAOD,EAAO,OACdE,EAAS,OAAO,KAAKf,EAAM,KAAK,KAAK,EACrCgB,EAAOD,EAAO,OAETE,EAAI,EAAGA,EAAIH,EAAMG,IAGxB,QAFIC,EAAQL,EAAOI,CAAC,EAEXzH,EAAI,EAAGA,EAAIwH,EAAMxH,IAAK,CAC7B,IAAI2H,EAAQJ,EAAOvH,CAAC,EAEpB,GAAI2H,GAASD,GAASA,GAAS,IAAK,CAClC,IAAIX,EAAOP,EAAM,KAAK,MAAMmB,CAAK,EAC7BC,EAAQpB,EAAM,MAAM,MAAMkB,CAAK,EAC/BV,EAAQD,EAAK,OAASa,EAAM,MAC5BX,EAAO,OAEPU,KAASnB,EAAM,OAAO,OAIxBS,EAAOT,EAAM,OAAO,MAAMmB,CAAK,EAC/BV,EAAK,MAAQA,EAAK,OAASD,IAM3BC,EAAO,IAAI/H,EAAK,SAChB+H,EAAK,MAAQD,EACbR,EAAM,OAAO,MAAMmB,CAAK,EAAIV,GAG9BV,EAAM,KAAK,CACT,MAAOqB,EACP,OAAQX,EACR,KAAMF,CACR,CAAC,CACH,CACF,CAEJ,CAEA,OAAOxD,CACT,EACArE,EAAK,SAAS,QAAU,UAAY,CAClC,KAAK,aAAe,GACpB,KAAK,KAAO,IAAIA,EAAK,SACrB,KAAK,eAAiB,CAAC,EACvB,KAAK,eAAiB,CAAC,CACzB,EAEAA,EAAK,SAAS,QAAQ,UAAU,OAAS,SAAU2I,EAAM,CACvD,IAAId,EACAe,EAAe,EAEnB,GAAID,EAAO,KAAK,aACd,MAAM,IAAI,MAAO,6BAA6B,EAGhD,QAAS,EAAI,EAAG,EAAIA,EAAK,QAAU,EAAI,KAAK,aAAa,QACnDA,EAAK,CAAC,GAAK,KAAK,aAAa,CAAC,EAD6B,IAE/DC,IAGF,KAAK,SAASA,CAAY,EAEtB,KAAK,eAAe,QAAU,EAChCf,EAAO,KAAK,KAEZA,EAAO,KAAK,eAAe,KAAK,eAAe,OAAS,CAAC,EAAE,MAG7D,QAAS,EAAIe,EAAc,EAAID,EAAK,OAAQ,IAAK,CAC/C,IAAIE,EAAW,IAAI7I,EAAK,SACpBoC,EAAOuG,EAAK,CAAC,EAEjBd,EAAK,MAAMzF,CAAI,EAAIyG,EAEnB,KAAK,eAAe,KAAK,CACvB,OAAQhB,EACR,KAAMzF,EACN,MAAOyG,CACT,CAAC,EAEDhB,EAAOgB,CACT,CAEAhB,EAAK,MAAQ,GACb,KAAK,aAAec,CACtB,EAEA3I,EAAK,SAAS,QAAQ,UAAU,OAAS,UAAY,CACnD,KAAK,SAAS,CAAC,CACjB,EAEAA,EAAK,SAAS,QAAQ,UAAU,SAAW,SAAU8I,EAAQ,CAC3D,QAAS7H,EAAI,KAAK,eAAe,OAAS,EAAGA,GAAK6H,EAAQ7H,IAAK,CAC7D,IAAI4G,EAAO,KAAK,eAAe5G,CAAC,EAC5B8H,EAAWlB,EAAK,MAAM,SAAS,EAE/BkB,KAAY,KAAK,eACnBlB,EAAK,OAAO,MAAMA,EAAK,IAAI,EAAI,KAAK,eAAekB,CAAQ,GAI3DlB,EAAK,MAAM,KAAOkB,EAElB,KAAK,eAAeA,CAAQ,EAAIlB,EAAK,OAGvC,KAAK,eAAe,IAAI,CAC1B,CACF,EACA;AAAA;AAAA;AAAA,GAqBA7H,EAAK,MAAQ,SAAUgJ,EAAO,CAC5B,KAAK,cAAgBA,EAAM,cAC3B,KAAK,aAAeA,EAAM,aAC1B,KAAK,SAAWA,EAAM,SACtB,KAAK,OAASA,EAAM,OACpB,KAAK,SAAWA,EAAM,QACxB,EAyEAhJ,EAAK,MAAM,UAAU,OAAS,SAAUiJ,EAAa,CACnD,OAAO,KAAK,MAAM,SAAUC,EAAO,CACjC,IAAIC,EAAS,IAAInJ,EAAK,YAAYiJ,EAAaC,CAAK,EACpDC,EAAO,MAAM,CACf,CAAC,CACH,EA2BAnJ,EAAK,MAAM,UAAU,MAAQ,SAAU8B,EAAI,CAoBzC,QAZIoH,EAAQ,IAAIlJ,EAAK,MAAM,KAAK,MAAM,EAClCoJ,EAAiB,OAAO,OAAO,IAAI,EACnCC,EAAe,OAAO,OAAO,IAAI,EACjCC,EAAiB,OAAO,OAAO,IAAI,EACnCC,EAAkB,OAAO,OAAO,IAAI,EACpCC,EAAoB,OAAO,OAAO,IAAI,EAOjCvI,EAAI,EAAGA,EAAI,KAAK,OAAO,OAAQA,IACtCoI,EAAa,KAAK,OAAOpI,CAAC,CAAC,EAAI,IAAIjB,EAAK,OAG1C8B,EAAG,KAAKoH,EAAOA,CAAK,EAEpB,QAASjI,EAAI,EAAGA,EAAIiI,EAAM,QAAQ,OAAQjI,IAAK,CAS7C,IAAIiG,EAASgC,EAAM,QAAQjI,CAAC,EACxBwI,EAAQ,KACRC,EAAgB1J,EAAK,IAAI,MAEzBkH,EAAO,YACTuC,EAAQ,KAAK,SAAS,UAAUvC,EAAO,KAAM,CAC3C,OAAQA,EAAO,MACjB,CAAC,EAEDuC,EAAQ,CAACvC,EAAO,IAAI,EAGtB,QAASyC,EAAI,EAAGA,EAAIF,EAAM,OAAQE,IAAK,CACrC,IAAIC,EAAOH,EAAME,CAAC,EAQlBzC,EAAO,KAAO0C,EAOd,IAAIC,EAAe7J,EAAK,SAAS,WAAWkH,CAAM,EAC9C4C,EAAgB,KAAK,SAAS,UAAUD,CAAY,EAAE,QAAQ,EAQlE,GAAIC,EAAc,SAAW,GAAK5C,EAAO,WAAalH,EAAK,MAAM,SAAS,SAAU,CAClF,QAASoD,EAAI,EAAGA,EAAI8D,EAAO,OAAO,OAAQ9D,IAAK,CAC7C,IAAI2G,EAAQ7C,EAAO,OAAO9D,CAAC,EAC3BmG,EAAgBQ,CAAK,EAAI/J,EAAK,IAAI,KACpC,CAEA,KACF,CAEA,QAASkD,EAAI,EAAGA,EAAI4G,EAAc,OAAQ5G,IASxC,QAJI8G,EAAeF,EAAc5G,CAAC,EAC9B1B,EAAU,KAAK,cAAcwI,CAAY,EACzCC,EAAYzI,EAAQ,OAEf4B,EAAI,EAAGA,EAAI8D,EAAO,OAAO,OAAQ9D,IAAK,CAS7C,IAAI2G,EAAQ7C,EAAO,OAAO9D,CAAC,EACvB8G,EAAe1I,EAAQuI,CAAK,EAC5BI,EAAuB,OAAO,KAAKD,CAAY,EAC/CE,EAAYJ,EAAe,IAAMD,EACjCM,EAAuB,IAAIrK,EAAK,IAAImK,CAAoB,EAoB5D,GAbIjD,EAAO,UAAYlH,EAAK,MAAM,SAAS,WACzC0J,EAAgBA,EAAc,MAAMW,CAAoB,EAEpDd,EAAgBQ,CAAK,IAAM,SAC7BR,EAAgBQ,CAAK,EAAI/J,EAAK,IAAI,WASlCkH,EAAO,UAAYlH,EAAK,MAAM,SAAS,WAAY,CACjDwJ,EAAkBO,CAAK,IAAM,SAC/BP,EAAkBO,CAAK,EAAI/J,EAAK,IAAI,OAGtCwJ,EAAkBO,CAAK,EAAIP,EAAkBO,CAAK,EAAE,MAAMM,CAAoB,EAO9E,QACF,CAeA,GANAhB,EAAaU,CAAK,EAAE,OAAOE,EAAW/C,EAAO,MAAO,SAAU9F,GAAGC,GAAG,CAAE,OAAOD,GAAIC,EAAE,CAAC,EAMhF,CAAAiI,EAAec,CAAS,EAI5B,SAASE,EAAI,EAAGA,EAAIH,EAAqB,OAAQG,IAAK,CAOpD,IAAIC,EAAsBJ,EAAqBG,CAAC,EAC5CE,EAAmB,IAAIxK,EAAK,SAAUuK,EAAqBR,CAAK,EAChElI,EAAWqI,EAAaK,CAAmB,EAC3CE,GAECA,EAAarB,EAAeoB,CAAgB,KAAO,OACtDpB,EAAeoB,CAAgB,EAAI,IAAIxK,EAAK,UAAWgK,EAAcD,EAAOlI,CAAQ,EAEpF4I,EAAW,IAAIT,EAAcD,EAAOlI,CAAQ,CAGhD,CAEAyH,EAAec,CAAS,EAAI,GAC9B,CAEJ,CAQA,GAAIlD,EAAO,WAAalH,EAAK,MAAM,SAAS,SAC1C,QAASoD,EAAI,EAAGA,EAAI8D,EAAO,OAAO,OAAQ9D,IAAK,CAC7C,IAAI2G,EAAQ7C,EAAO,OAAO9D,CAAC,EAC3BmG,EAAgBQ,CAAK,EAAIR,EAAgBQ,CAAK,EAAE,UAAUL,CAAa,CACzE,CAEJ,CAUA,QAHIgB,EAAqB1K,EAAK,IAAI,SAC9B2K,EAAuB3K,EAAK,IAAI,MAE3BiB,EAAI,EAAGA,EAAI,KAAK,OAAO,OAAQA,IAAK,CAC3C,IAAI8I,EAAQ,KAAK,OAAO9I,CAAC,EAErBsI,EAAgBQ,CAAK,IACvBW,EAAqBA,EAAmB,UAAUnB,EAAgBQ,CAAK,CAAC,GAGtEP,EAAkBO,CAAK,IACzBY,EAAuBA,EAAqB,MAAMnB,EAAkBO,CAAK,CAAC,EAE9E,CAEA,IAAIa,EAAoB,OAAO,KAAKxB,CAAc,EAC9CyB,EAAU,CAAC,EACXC,EAAU,OAAO,OAAO,IAAI,EAYhC,GAAI5B,EAAM,UAAU,EAAG,CACrB0B,EAAoB,OAAO,KAAK,KAAK,YAAY,EAEjD,QAAS3J,EAAI,EAAGA,EAAI2J,EAAkB,OAAQ3J,IAAK,CACjD,IAAIuJ,EAAmBI,EAAkB3J,CAAC,EACtCF,EAAWf,EAAK,SAAS,WAAWwK,CAAgB,EACxDpB,EAAeoB,CAAgB,EAAI,IAAIxK,EAAK,SAC9C,CACF,CAEA,QAASiB,EAAI,EAAGA,EAAI2J,EAAkB,OAAQ3J,IAAK,CASjD,IAAIF,EAAWf,EAAK,SAAS,WAAW4K,EAAkB3J,CAAC,CAAC,EACxDP,EAASK,EAAS,OAEtB,GAAK2J,EAAmB,SAAShK,CAAM,GAInC,CAAAiK,EAAqB,SAASjK,CAAM,EAIxC,KAAIqK,EAAc,KAAK,aAAahK,CAAQ,EACxCiK,EAAQ3B,EAAatI,EAAS,SAAS,EAAE,WAAWgK,CAAW,EAC/DE,EAEJ,IAAKA,EAAWH,EAAQpK,CAAM,KAAO,OACnCuK,EAAS,OAASD,EAClBC,EAAS,UAAU,QAAQ7B,EAAerI,CAAQ,CAAC,MAC9C,CACL,IAAImK,EAAQ,CACV,IAAKxK,EACL,MAAOsK,EACP,UAAW5B,EAAerI,CAAQ,CACpC,EACA+J,EAAQpK,CAAM,EAAIwK,EAClBL,EAAQ,KAAKK,CAAK,CACpB,EACF,CAKA,OAAOL,EAAQ,KAAK,SAAUzJ,GAAGC,GAAG,CAClC,OAAOA,GAAE,MAAQD,GAAE,KACrB,CAAC,CACH,EAUApB,EAAK,MAAM,UAAU,OAAS,UAAY,CACxC,IAAImL,EAAgB,OAAO,KAAK,KAAK,aAAa,EAC/C,KAAK,EACL,IAAI,SAAUvB,EAAM,CACnB,MAAO,CAACA,EAAM,KAAK,cAAcA,CAAI,CAAC,CACxC,EAAG,IAAI,EAELwB,EAAe,OAAO,KAAK,KAAK,YAAY,EAC7C,IAAI,SAAUC,EAAK,CAClB,MAAO,CAACA,EAAK,KAAK,aAAaA,CAAG,EAAE,OAAO,CAAC,CAC9C,EAAG,IAAI,EAET,MAAO,CACL,QAASrL,EAAK,QACd,OAAQ,KAAK,OACb,aAAcoL,EACd,cAAeD,EACf,SAAU,KAAK,SAAS,OAAO,CACjC,CACF,EAQAnL,EAAK,MAAM,KAAO,SAAUsL,EAAiB,CAC3C,IAAItC,EAAQ,CAAC,EACToC,EAAe,CAAC,EAChBG,EAAoBD,EAAgB,aACpCH,EAAgB,OAAO,OAAO,IAAI,EAClCK,EAA0BF,EAAgB,cAC1CG,EAAkB,IAAIzL,EAAK,SAAS,QACpC0C,EAAW1C,EAAK,SAAS,KAAKsL,EAAgB,QAAQ,EAEtDA,EAAgB,SAAWtL,EAAK,SAClCA,EAAK,MAAM,KAAK,4EAA8EA,EAAK,QAAU,sCAAwCsL,EAAgB,QAAU,GAAG,EAGpL,QAASrK,EAAI,EAAGA,EAAIsK,EAAkB,OAAQtK,IAAK,CACjD,IAAIyK,EAAQH,EAAkBtK,CAAC,EAC3BoK,EAAMK,EAAM,CAAC,EACb1K,EAAW0K,EAAM,CAAC,EAEtBN,EAAaC,CAAG,EAAI,IAAIrL,EAAK,OAAOgB,CAAQ,CAC9C,CAEA,QAASC,EAAI,EAAGA,EAAIuK,EAAwB,OAAQvK,IAAK,CACvD,IAAIyK,EAAQF,EAAwBvK,CAAC,EACjC2I,EAAO8B,EAAM,CAAC,EACdlK,EAAUkK,EAAM,CAAC,EAErBD,EAAgB,OAAO7B,CAAI,EAC3BuB,EAAcvB,CAAI,EAAIpI,CACxB,CAEA,OAAAiK,EAAgB,OAAO,EAEvBzC,EAAM,OAASsC,EAAgB,OAE/BtC,EAAM,aAAeoC,EACrBpC,EAAM,cAAgBmC,EACtBnC,EAAM,SAAWyC,EAAgB,KACjCzC,EAAM,SAAWtG,EAEV,IAAI1C,EAAK,MAAMgJ,CAAK,CAC7B,EACA;AAAA;AAAA;AAAA,GA6BAhJ,EAAK,QAAU,UAAY,CACzB,KAAK,KAAO,KACZ,KAAK,QAAU,OAAO,OAAO,IAAI,EACjC,KAAK,WAAa,OAAO,OAAO,IAAI,EACpC,KAAK,cAAgB,OAAO,OAAO,IAAI,EACvC,KAAK,qBAAuB,CAAC,EAC7B,KAAK,aAAe,CAAC,EACrB,KAAK,UAAYA,EAAK,UACtB,KAAK,SAAW,IAAIA,EAAK,SACzB,KAAK,eAAiB,IAAIA,EAAK,SAC/B,KAAK,cAAgB,EACrB,KAAK,GAAK,IACV,KAAK,IAAM,IACX,KAAK,UAAY,EACjB,KAAK,kBAAoB,CAAC,CAC5B,EAcAA,EAAK,QAAQ,UAAU,IAAM,SAAUqL,EAAK,CAC1C,KAAK,KAAOA,CACd,EAkCArL,EAAK,QAAQ,UAAU,MAAQ,SAAUW,EAAWgL,EAAY,CAC9D,GAAI,KAAK,KAAKhL,CAAS,EACrB,MAAM,IAAI,WAAY,UAAYA,EAAY,kCAAkC,EAGlF,KAAK,QAAQA,CAAS,EAAIgL,GAAc,CAAC,CAC3C,EAUA3L,EAAK,QAAQ,UAAU,EAAI,SAAU4L,EAAQ,CACvCA,EAAS,EACX,KAAK,GAAK,EACDA,EAAS,EAClB,KAAK,GAAK,EAEV,KAAK,GAAKA,CAEd,EASA5L,EAAK,QAAQ,UAAU,GAAK,SAAU4L,EAAQ,CAC5C,KAAK,IAAMA,CACb,EAmBA5L,EAAK,QAAQ,UAAU,IAAM,SAAU6L,EAAKF,EAAY,CACtD,IAAIjL,EAASmL,EAAI,KAAK,IAAI,EACtBC,EAAS,OAAO,KAAK,KAAK,OAAO,EAErC,KAAK,WAAWpL,CAAM,EAAIiL,GAAc,CAAC,EACzC,KAAK,eAAiB,EAEtB,QAAS1K,EAAI,EAAGA,EAAI6K,EAAO,OAAQ7K,IAAK,CACtC,IAAIN,EAAYmL,EAAO7K,CAAC,EACpB8K,EAAY,KAAK,QAAQpL,CAAS,EAAE,UACpCoJ,EAAQgC,EAAYA,EAAUF,CAAG,EAAIA,EAAIlL,CAAS,EAClDsB,EAAS,KAAK,UAAU8H,EAAO,CAC7B,OAAQ,CAACpJ,CAAS,CACpB,CAAC,EACD8I,EAAQ,KAAK,SAAS,IAAIxH,CAAM,EAChClB,EAAW,IAAIf,EAAK,SAAUU,EAAQC,CAAS,EAC/CqL,EAAa,OAAO,OAAO,IAAI,EAEnC,KAAK,qBAAqBjL,CAAQ,EAAIiL,EACtC,KAAK,aAAajL,CAAQ,EAAI,EAG9B,KAAK,aAAaA,CAAQ,GAAK0I,EAAM,OAGrC,QAASvG,EAAI,EAAGA,EAAIuG,EAAM,OAAQvG,IAAK,CACrC,IAAI0G,EAAOH,EAAMvG,CAAC,EAUlB,GARI8I,EAAWpC,CAAI,GAAK,OACtBoC,EAAWpC,CAAI,EAAI,GAGrBoC,EAAWpC,CAAI,GAAK,EAIhB,KAAK,cAAcA,CAAI,GAAK,KAAW,CACzC,IAAIpI,EAAU,OAAO,OAAO,IAAI,EAChCA,EAAQ,OAAY,KAAK,UACzB,KAAK,WAAa,EAElB,QAAS4B,EAAI,EAAGA,EAAI0I,EAAO,OAAQ1I,IACjC5B,EAAQsK,EAAO1I,CAAC,CAAC,EAAI,OAAO,OAAO,IAAI,EAGzC,KAAK,cAAcwG,CAAI,EAAIpI,CAC7B,CAGI,KAAK,cAAcoI,CAAI,EAAEjJ,CAAS,EAAED,CAAM,GAAK,OACjD,KAAK,cAAckJ,CAAI,EAAEjJ,CAAS,EAAED,CAAM,EAAI,OAAO,OAAO,IAAI,GAKlE,QAAS4J,EAAI,EAAGA,EAAI,KAAK,kBAAkB,OAAQA,IAAK,CACtD,IAAI2B,EAAc,KAAK,kBAAkB3B,CAAC,EACtCzI,EAAW+H,EAAK,SAASqC,CAAW,EAEpC,KAAK,cAAcrC,CAAI,EAAEjJ,CAAS,EAAED,CAAM,EAAEuL,CAAW,GAAK,OAC9D,KAAK,cAAcrC,CAAI,EAAEjJ,CAAS,EAAED,CAAM,EAAEuL,CAAW,EAAI,CAAC,GAG9D,KAAK,cAAcrC,CAAI,EAAEjJ,CAAS,EAAED,CAAM,EAAEuL,CAAW,EAAE,KAAKpK,CAAQ,CACxE,CACF,CAEF,CACF,EAOA7B,EAAK,QAAQ,UAAU,6BAA+B,UAAY,CAOhE,QALIkM,EAAY,OAAO,KAAK,KAAK,YAAY,EACzCC,EAAiBD,EAAU,OAC3BE,EAAc,CAAC,EACfC,EAAqB,CAAC,EAEjBpL,EAAI,EAAGA,EAAIkL,EAAgBlL,IAAK,CACvC,IAAIF,EAAWf,EAAK,SAAS,WAAWkM,EAAUjL,CAAC,CAAC,EAChD8I,EAAQhJ,EAAS,UAErBsL,EAAmBtC,CAAK,IAAMsC,EAAmBtC,CAAK,EAAI,GAC1DsC,EAAmBtC,CAAK,GAAK,EAE7BqC,EAAYrC,CAAK,IAAMqC,EAAYrC,CAAK,EAAI,GAC5CqC,EAAYrC,CAAK,GAAK,KAAK,aAAahJ,CAAQ,CAClD,CAIA,QAFI+K,EAAS,OAAO,KAAK,KAAK,OAAO,EAE5B7K,EAAI,EAAGA,EAAI6K,EAAO,OAAQ7K,IAAK,CACtC,IAAIN,EAAYmL,EAAO7K,CAAC,EACxBmL,EAAYzL,CAAS,EAAIyL,EAAYzL,CAAS,EAAI0L,EAAmB1L,CAAS,CAChF,CAEA,KAAK,mBAAqByL,CAC5B,EAOApM,EAAK,QAAQ,UAAU,mBAAqB,UAAY,CAMtD,QALIoL,EAAe,CAAC,EAChBc,EAAY,OAAO,KAAK,KAAK,oBAAoB,EACjDI,EAAkBJ,EAAU,OAC5BK,EAAe,OAAO,OAAO,IAAI,EAE5BtL,EAAI,EAAGA,EAAIqL,EAAiBrL,IAAK,CAaxC,QAZIF,EAAWf,EAAK,SAAS,WAAWkM,EAAUjL,CAAC,CAAC,EAChDN,EAAYI,EAAS,UACrByL,EAAc,KAAK,aAAazL,CAAQ,EACxCgK,EAAc,IAAI/K,EAAK,OACvByM,EAAkB,KAAK,qBAAqB1L,CAAQ,EACpD0I,EAAQ,OAAO,KAAKgD,CAAe,EACnCC,EAAcjD,EAAM,OAGpBkD,EAAa,KAAK,QAAQhM,CAAS,EAAE,OAAS,EAC9CiM,EAAW,KAAK,WAAW7L,EAAS,MAAM,EAAE,OAAS,EAEhDmC,EAAI,EAAGA,EAAIwJ,EAAaxJ,IAAK,CACpC,IAAI0G,EAAOH,EAAMvG,CAAC,EACd2J,EAAKJ,EAAgB7C,CAAI,EACzBK,EAAY,KAAK,cAAcL,CAAI,EAAE,OACrCkD,EAAK9B,EAAO+B,EAEZR,EAAa3C,CAAI,IAAM,QACzBkD,EAAM9M,EAAK,IAAI,KAAK,cAAc4J,CAAI,EAAG,KAAK,aAAa,EAC3D2C,EAAa3C,CAAI,EAAIkD,GAErBA,EAAMP,EAAa3C,CAAI,EAGzBoB,EAAQ8B,IAAQ,KAAK,IAAM,GAAKD,IAAO,KAAK,KAAO,EAAI,KAAK,GAAK,KAAK,IAAML,EAAc,KAAK,mBAAmB7L,CAAS,IAAMkM,GACjI7B,GAAS2B,EACT3B,GAAS4B,EACTG,EAAqB,KAAK,MAAM/B,EAAQ,GAAI,EAAI,IAQhDD,EAAY,OAAOd,EAAW8C,CAAkB,CAClD,CAEA3B,EAAarK,CAAQ,EAAIgK,CAC3B,CAEA,KAAK,aAAeK,CACtB,EAOApL,EAAK,QAAQ,UAAU,eAAiB,UAAY,CAClD,KAAK,SAAWA,EAAK,SAAS,UAC5B,OAAO,KAAK,KAAK,aAAa,EAAE,KAAK,CACvC,CACF,EAUAA,EAAK,QAAQ,UAAU,MAAQ,UAAY,CACzC,YAAK,6BAA6B,EAClC,KAAK,mBAAmB,EACxB,KAAK,eAAe,EAEb,IAAIA,EAAK,MAAM,CACpB,cAAe,KAAK,cACpB,aAAc,KAAK,aACnB,SAAU,KAAK,SACf,OAAQ,OAAO,KAAK,KAAK,OAAO,EAChC,SAAU,KAAK,cACjB,CAAC,CACH,EAgBAA,EAAK,QAAQ,UAAU,IAAM,SAAU8B,EAAI,CACzC,IAAIkL,EAAO,MAAM,UAAU,MAAM,KAAK,UAAW,CAAC,EAClDA,EAAK,QAAQ,IAAI,EACjBlL,EAAG,MAAM,KAAMkL,CAAI,CACrB,EAaAhN,EAAK,UAAY,SAAU4J,EAAMG,EAAOlI,EAAU,CAShD,QARIoL,EAAiB,OAAO,OAAO,IAAI,EACnCC,EAAe,OAAO,KAAKrL,GAAY,CAAC,CAAC,EAOpCZ,EAAI,EAAGA,EAAIiM,EAAa,OAAQjM,IAAK,CAC5C,IAAIT,EAAM0M,EAAajM,CAAC,EACxBgM,EAAezM,CAAG,EAAIqB,EAASrB,CAAG,EAAE,MAAM,CAC5C,CAEA,KAAK,SAAW,OAAO,OAAO,IAAI,EAE9BoJ,IAAS,SACX,KAAK,SAASA,CAAI,EAAI,OAAO,OAAO,IAAI,EACxC,KAAK,SAASA,CAAI,EAAEG,CAAK,EAAIkD,EAEjC,EAWAjN,EAAK,UAAU,UAAU,QAAU,SAAUmN,EAAgB,CAG3D,QAFI1D,EAAQ,OAAO,KAAK0D,EAAe,QAAQ,EAEtClM,EAAI,EAAGA,EAAIwI,EAAM,OAAQxI,IAAK,CACrC,IAAI2I,EAAOH,EAAMxI,CAAC,EACd6K,EAAS,OAAO,KAAKqB,EAAe,SAASvD,CAAI,CAAC,EAElD,KAAK,SAASA,CAAI,GAAK,OACzB,KAAK,SAASA,CAAI,EAAI,OAAO,OAAO,IAAI,GAG1C,QAAS1G,EAAI,EAAGA,EAAI4I,EAAO,OAAQ5I,IAAK,CACtC,IAAI6G,EAAQ+B,EAAO5I,CAAC,EAChB3C,EAAO,OAAO,KAAK4M,EAAe,SAASvD,CAAI,EAAEG,CAAK,CAAC,EAEvD,KAAK,SAASH,CAAI,EAAEG,CAAK,GAAK,OAChC,KAAK,SAASH,CAAI,EAAEG,CAAK,EAAI,OAAO,OAAO,IAAI,GAGjD,QAAS3G,EAAI,EAAGA,EAAI7C,EAAK,OAAQ6C,IAAK,CACpC,IAAI5C,EAAMD,EAAK6C,CAAC,EAEZ,KAAK,SAASwG,CAAI,EAAEG,CAAK,EAAEvJ,CAAG,GAAK,KACrC,KAAK,SAASoJ,CAAI,EAAEG,CAAK,EAAEvJ,CAAG,EAAI2M,EAAe,SAASvD,CAAI,EAAEG,CAAK,EAAEvJ,CAAG,EAE1E,KAAK,SAASoJ,CAAI,EAAEG,CAAK,EAAEvJ,CAAG,EAAI,KAAK,SAASoJ,CAAI,EAAEG,CAAK,EAAEvJ,CAAG,EAAE,OAAO2M,EAAe,SAASvD,CAAI,EAAEG,CAAK,EAAEvJ,CAAG,CAAC,CAGtH,CACF,CACF,CACF,EASAR,EAAK,UAAU,UAAU,IAAM,SAAU4J,EAAMG,EAAOlI,EAAU,CAC9D,GAAI,EAAE+H,KAAQ,KAAK,UAAW,CAC5B,KAAK,SAASA,CAAI,EAAI,OAAO,OAAO,IAAI,EACxC,KAAK,SAASA,CAAI,EAAEG,CAAK,EAAIlI,EAC7B,MACF,CAEA,GAAI,EAAEkI,KAAS,KAAK,SAASH,CAAI,GAAI,CACnC,KAAK,SAASA,CAAI,EAAEG,CAAK,EAAIlI,EAC7B,MACF,CAIA,QAFIqL,EAAe,OAAO,KAAKrL,CAAQ,EAE9BZ,EAAI,EAAGA,EAAIiM,EAAa,OAAQjM,IAAK,CAC5C,IAAIT,EAAM0M,EAAajM,CAAC,EAEpBT,KAAO,KAAK,SAASoJ,CAAI,EAAEG,CAAK,EAClC,KAAK,SAASH,CAAI,EAAEG,CAAK,EAAEvJ,CAAG,EAAI,KAAK,SAASoJ,CAAI,EAAEG,CAAK,EAAEvJ,CAAG,EAAE,OAAOqB,EAASrB,CAAG,CAAC,EAEtF,KAAK,SAASoJ,CAAI,EAAEG,CAAK,EAAEvJ,CAAG,EAAIqB,EAASrB,CAAG,CAElD,CACF,EAYAR,EAAK,MAAQ,SAAUoN,EAAW,CAChC,KAAK,QAAU,CAAC,EAChB,KAAK,UAAYA,CACnB,EA0BApN,EAAK,MAAM,SAAW,IAAI,OAAQ,GAAG,EACrCA,EAAK,MAAM,SAAS,KAAO,EAC3BA,EAAK,MAAM,SAAS,QAAU,EAC9BA,EAAK,MAAM,SAAS,SAAW,EAa/BA,EAAK,MAAM,SAAW,CAIpB,SAAU,EAMV,SAAU,EAMV,WAAY,CACd,EAyBAA,EAAK,MAAM,UAAU,OAAS,SAAUkH,EAAQ,CAC9C,MAAM,WAAYA,IAChBA,EAAO,OAAS,KAAK,WAGjB,UAAWA,IACfA,EAAO,MAAQ,GAGX,gBAAiBA,IACrBA,EAAO,YAAc,IAGjB,aAAcA,IAClBA,EAAO,SAAWlH,EAAK,MAAM,SAAS,MAGnCkH,EAAO,SAAWlH,EAAK,MAAM,SAAS,SAAakH,EAAO,KAAK,OAAO,CAAC,GAAKlH,EAAK,MAAM,WAC1FkH,EAAO,KAAO,IAAMA,EAAO,MAGxBA,EAAO,SAAWlH,EAAK,MAAM,SAAS,UAAckH,EAAO,KAAK,MAAM,EAAE,GAAKlH,EAAK,MAAM,WAC3FkH,EAAO,KAAO,GAAKA,EAAO,KAAO,KAG7B,aAAcA,IAClBA,EAAO,SAAWlH,EAAK,MAAM,SAAS,UAGxC,KAAK,QAAQ,KAAKkH,CAAM,EAEjB,IACT,EASAlH,EAAK,MAAM,UAAU,UAAY,UAAY,CAC3C,QAASiB,EAAI,EAAGA,EAAI,KAAK,QAAQ,OAAQA,IACvC,GAAI,KAAK,QAAQA,CAAC,EAAE,UAAYjB,EAAK,MAAM,SAAS,WAClD,MAAO,GAIX,MAAO,EACT,EA4BAA,EAAK,MAAM,UAAU,KAAO,SAAU4J,EAAMyD,EAAS,CACnD,GAAI,MAAM,QAAQzD,CAAI,EACpB,OAAAA,EAAK,QAAQ,SAAU7H,EAAG,CAAE,KAAK,KAAKA,EAAG/B,EAAK,MAAM,MAAMqN,CAAO,CAAC,CAAE,EAAG,IAAI,EACpE,KAGT,IAAInG,EAASmG,GAAW,CAAC,EACzB,OAAAnG,EAAO,KAAO0C,EAAK,SAAS,EAE5B,KAAK,OAAO1C,CAAM,EAEX,IACT,EACAlH,EAAK,gBAAkB,SAAUI,EAASmD,EAAOC,EAAK,CACpD,KAAK,KAAO,kBACZ,KAAK,QAAUpD,EACf,KAAK,MAAQmD,EACb,KAAK,IAAMC,CACb,EAEAxD,EAAK,gBAAgB,UAAY,IAAI,MACrCA,EAAK,WAAa,SAAU4B,EAAK,CAC/B,KAAK,QAAU,CAAC,EAChB,KAAK,IAAMA,EACX,KAAK,OAASA,EAAI,OAClB,KAAK,IAAM,EACX,KAAK,MAAQ,EACb,KAAK,oBAAsB,CAAC,CAC9B,EAEA5B,EAAK,WAAW,UAAU,IAAM,UAAY,CAG1C,QAFIsN,EAAQtN,EAAK,WAAW,QAErBsN,GACLA,EAAQA,EAAM,IAAI,CAEtB,EAEAtN,EAAK,WAAW,UAAU,YAAc,UAAY,CAKlD,QAJIuN,EAAY,CAAC,EACbpL,EAAa,KAAK,MAClBD,EAAW,KAAK,IAEX,EAAI,EAAG,EAAI,KAAK,oBAAoB,OAAQ,IACnDA,EAAW,KAAK,oBAAoB,CAAC,EACrCqL,EAAU,KAAK,KAAK,IAAI,MAAMpL,EAAYD,CAAQ,CAAC,EACnDC,EAAaD,EAAW,EAG1B,OAAAqL,EAAU,KAAK,KAAK,IAAI,MAAMpL,EAAY,KAAK,GAAG,CAAC,EACnD,KAAK,oBAAoB,OAAS,EAE3BoL,EAAU,KAAK,EAAE,CAC1B,EAEAvN,EAAK,WAAW,UAAU,KAAO,SAAUwN,EAAM,CAC/C,KAAK,QAAQ,KAAK,CAChB,KAAMA,EACN,IAAK,KAAK,YAAY,EACtB,MAAO,KAAK,MACZ,IAAK,KAAK,GACZ,CAAC,EAED,KAAK,MAAQ,KAAK,GACpB,EAEAxN,EAAK,WAAW,UAAU,gBAAkB,UAAY,CACtD,KAAK,oBAAoB,KAAK,KAAK,IAAM,CAAC,EAC1C,KAAK,KAAO,CACd,EAEAA,EAAK,WAAW,UAAU,KAAO,UAAY,CAC3C,GAAI,KAAK,KAAO,KAAK,OACnB,OAAOA,EAAK,WAAW,IAGzB,IAAIoC,EAAO,KAAK,IAAI,OAAO,KAAK,GAAG,EACnC,YAAK,KAAO,EACLA,CACT,EAEApC,EAAK,WAAW,UAAU,MAAQ,UAAY,CAC5C,OAAO,KAAK,IAAM,KAAK,KACzB,EAEAA,EAAK,WAAW,UAAU,OAAS,UAAY,CACzC,KAAK,OAAS,KAAK,MACrB,KAAK,KAAO,GAGd,KAAK,MAAQ,KAAK,GACpB,EAEAA,EAAK,WAAW,UAAU,OAAS,UAAY,CAC7C,KAAK,KAAO,CACd,EAEAA,EAAK,WAAW,UAAU,eAAiB,UAAY,CACrD,IAAIoC,EAAMqL,EAEV,GACErL,EAAO,KAAK,KAAK,EACjBqL,EAAWrL,EAAK,WAAW,CAAC,QACrBqL,EAAW,IAAMA,EAAW,IAEjCrL,GAAQpC,EAAK,WAAW,KAC1B,KAAK,OAAO,CAEhB,EAEAA,EAAK,WAAW,UAAU,KAAO,UAAY,CAC3C,OAAO,KAAK,IAAM,KAAK,MACzB,EAEAA,EAAK,WAAW,IAAM,MACtBA,EAAK,WAAW,MAAQ,QACxBA,EAAK,WAAW,KAAO,OACvBA,EAAK,WAAW,cAAgB,gBAChCA,EAAK,WAAW,MAAQ,QACxBA,EAAK,WAAW,SAAW,WAE3BA,EAAK,WAAW,SAAW,SAAU0N,EAAO,CAC1C,OAAAA,EAAM,OAAO,EACbA,EAAM,KAAK1N,EAAK,WAAW,KAAK,EAChC0N,EAAM,OAAO,EACN1N,EAAK,WAAW,OACzB,EAEAA,EAAK,WAAW,QAAU,SAAU0N,EAAO,CAQzC,GAPIA,EAAM,MAAM,EAAI,IAClBA,EAAM,OAAO,EACbA,EAAM,KAAK1N,EAAK,WAAW,IAAI,GAGjC0N,EAAM,OAAO,EAETA,EAAM,KAAK,EACb,OAAO1N,EAAK,WAAW,OAE3B,EAEAA,EAAK,WAAW,gBAAkB,SAAU0N,EAAO,CACjD,OAAAA,EAAM,OAAO,EACbA,EAAM,eAAe,EACrBA,EAAM,KAAK1N,EAAK,WAAW,aAAa,EACjCA,EAAK,WAAW,OACzB,EAEAA,EAAK,WAAW,SAAW,SAAU0N,EAAO,CAC1C,OAAAA,EAAM,OAAO,EACbA,EAAM,eAAe,EACrBA,EAAM,KAAK1N,EAAK,WAAW,KAAK,EACzBA,EAAK,WAAW,OACzB,EAEAA,EAAK,WAAW,OAAS,SAAU0N,EAAO,CACpCA,EAAM,MAAM,EAAI,GAClBA,EAAM,KAAK1N,EAAK,WAAW,IAAI,CAEnC,EAaAA,EAAK,WAAW,cAAgBA,EAAK,UAAU,UAE/CA,EAAK,WAAW,QAAU,SAAU0N,EAAO,CACzC,OAAa,CACX,IAAItL,EAAOsL,EAAM,KAAK,EAEtB,GAAItL,GAAQpC,EAAK,WAAW,IAC1B,OAAOA,EAAK,WAAW,OAIzB,GAAIoC,EAAK,WAAW,CAAC,GAAK,GAAI,CAC5BsL,EAAM,gBAAgB,EACtB,QACF,CAEA,GAAItL,GAAQ,IACV,OAAOpC,EAAK,WAAW,SAGzB,GAAIoC,GAAQ,IACV,OAAAsL,EAAM,OAAO,EACTA,EAAM,MAAM,EAAI,GAClBA,EAAM,KAAK1N,EAAK,WAAW,IAAI,EAE1BA,EAAK,WAAW,gBAGzB,GAAIoC,GAAQ,IACV,OAAAsL,EAAM,OAAO,EACTA,EAAM,MAAM,EAAI,GAClBA,EAAM,KAAK1N,EAAK,WAAW,IAAI,EAE1BA,EAAK,WAAW,SAczB,GARIoC,GAAQ,KAAOsL,EAAM,MAAM,IAAM,GAQjCtL,GAAQ,KAAOsL,EAAM,MAAM,IAAM,EACnC,OAAAA,EAAM,KAAK1N,EAAK,WAAW,QAAQ,EAC5BA,EAAK,WAAW,QAGzB,GAAIoC,EAAK,MAAMpC,EAAK,WAAW,aAAa,EAC1C,OAAOA,EAAK,WAAW,OAE3B,CACF,EAEAA,EAAK,YAAc,SAAU4B,EAAKsH,EAAO,CACvC,KAAK,MAAQ,IAAIlJ,EAAK,WAAY4B,CAAG,EACrC,KAAK,MAAQsH,EACb,KAAK,cAAgB,CAAC,EACtB,KAAK,UAAY,CACnB,EAEAlJ,EAAK,YAAY,UAAU,MAAQ,UAAY,CAC7C,KAAK,MAAM,IAAI,EACf,KAAK,QAAU,KAAK,MAAM,QAI1B,QAFIsN,EAAQtN,EAAK,YAAY,YAEtBsN,GACLA,EAAQA,EAAM,IAAI,EAGpB,OAAO,KAAK,KACd,EAEAtN,EAAK,YAAY,UAAU,WAAa,UAAY,CAClD,OAAO,KAAK,QAAQ,KAAK,SAAS,CACpC,EAEAA,EAAK,YAAY,UAAU,cAAgB,UAAY,CACrD,IAAI2N,EAAS,KAAK,WAAW,EAC7B,YAAK,WAAa,EACXA,CACT,EAEA3N,EAAK,YAAY,UAAU,WAAa,UAAY,CAClD,IAAI4N,EAAkB,KAAK,cAC3B,KAAK,MAAM,OAAOA,CAAe,EACjC,KAAK,cAAgB,CAAC,CACxB,EAEA5N,EAAK,YAAY,YAAc,SAAUmJ,EAAQ,CAC/C,IAAIwE,EAASxE,EAAO,WAAW,EAE/B,GAAIwE,GAAU,KAId,OAAQA,EAAO,KAAM,CACnB,KAAK3N,EAAK,WAAW,SACnB,OAAOA,EAAK,YAAY,cAC1B,KAAKA,EAAK,WAAW,MACnB,OAAOA,EAAK,YAAY,WAC1B,KAAKA,EAAK,WAAW,KACnB,OAAOA,EAAK,YAAY,UAC1B,QACE,IAAI6N,EAAe,4CAA8CF,EAAO,KAExE,MAAIA,EAAO,IAAI,QAAU,IACvBE,GAAgB,gBAAkBF,EAAO,IAAM,KAG3C,IAAI3N,EAAK,gBAAiB6N,EAAcF,EAAO,MAAOA,EAAO,GAAG,CAC1E,CACF,EAEA3N,EAAK,YAAY,cAAgB,SAAUmJ,EAAQ,CACjD,IAAIwE,EAASxE,EAAO,cAAc,EAElC,GAAIwE,GAAU,KAId,QAAQA,EAAO,IAAK,CAClB,IAAK,IACHxE,EAAO,cAAc,SAAWnJ,EAAK,MAAM,SAAS,WACpD,MACF,IAAK,IACHmJ,EAAO,cAAc,SAAWnJ,EAAK,MAAM,SAAS,SACpD,MACF,QACE,IAAI6N,EAAe,kCAAoCF,EAAO,IAAM,IACpE,MAAM,IAAI3N,EAAK,gBAAiB6N,EAAcF,EAAO,MAAOA,EAAO,GAAG,CAC1E,CAEA,IAAIG,EAAa3E,EAAO,WAAW,EAEnC,GAAI2E,GAAc,KAAW,CAC3B,IAAID,EAAe,yCACnB,MAAM,IAAI7N,EAAK,gBAAiB6N,EAAcF,EAAO,MAAOA,EAAO,GAAG,CACxE,CAEA,OAAQG,EAAW,KAAM,CACvB,KAAK9N,EAAK,WAAW,MACnB,OAAOA,EAAK,YAAY,WAC1B,KAAKA,EAAK,WAAW,KACnB,OAAOA,EAAK,YAAY,UAC1B,QACE,IAAI6N,EAAe,mCAAqCC,EAAW,KAAO,IAC1E,MAAM,IAAI9N,EAAK,gBAAiB6N,EAAcC,EAAW,MAAOA,EAAW,GAAG,CAClF,EACF,EAEA9N,EAAK,YAAY,WAAa,SAAUmJ,EAAQ,CAC9C,IAAIwE,EAASxE,EAAO,cAAc,EAElC,GAAIwE,GAAU,KAId,IAAIxE,EAAO,MAAM,UAAU,QAAQwE,EAAO,GAAG,GAAK,GAAI,CACpD,IAAII,EAAiB5E,EAAO,MAAM,UAAU,IAAI,SAAU6E,EAAG,CAAE,MAAO,IAAMA,EAAI,GAAI,CAAC,EAAE,KAAK,IAAI,EAC5FH,EAAe,uBAAyBF,EAAO,IAAM,uBAAyBI,EAElF,MAAM,IAAI/N,EAAK,gBAAiB6N,EAAcF,EAAO,MAAOA,EAAO,GAAG,CACxE,CAEAxE,EAAO,cAAc,OAAS,CAACwE,EAAO,GAAG,EAEzC,IAAIG,EAAa3E,EAAO,WAAW,EAEnC,GAAI2E,GAAc,KAAW,CAC3B,IAAID,EAAe,gCACnB,MAAM,IAAI7N,EAAK,gBAAiB6N,EAAcF,EAAO,MAAOA,EAAO,GAAG,CACxE,CAEA,OAAQG,EAAW,KAAM,CACvB,KAAK9N,EAAK,WAAW,KACnB,OAAOA,EAAK,YAAY,UAC1B,QACE,IAAI6N,EAAe,0BAA4BC,EAAW,KAAO,IACjE,MAAM,IAAI9N,EAAK,gBAAiB6N,EAAcC,EAAW,MAAOA,EAAW,GAAG,CAClF,EACF,EAEA9N,EAAK,YAAY,UAAY,SAAUmJ,EAAQ,CAC7C,IAAIwE,EAASxE,EAAO,cAAc,EAElC,GAAIwE,GAAU,KAId,CAAAxE,EAAO,cAAc,KAAOwE,EAAO,IAAI,YAAY,EAE/CA,EAAO,IAAI,QAAQ,GAAG,GAAK,KAC7BxE,EAAO,cAAc,YAAc,IAGrC,IAAI2E,EAAa3E,EAAO,WAAW,EAEnC,GAAI2E,GAAc,KAAW,CAC3B3E,EAAO,WAAW,EAClB,MACF,CAEA,OAAQ2E,EAAW,KAAM,CACvB,KAAK9N,EAAK,WAAW,KACnB,OAAAmJ,EAAO,WAAW,EACXnJ,EAAK,YAAY,UAC1B,KAAKA,EAAK,WAAW,MACnB,OAAAmJ,EAAO,WAAW,EACXnJ,EAAK,YAAY,WAC1B,KAAKA,EAAK,WAAW,cACnB,OAAOA,EAAK,YAAY,kBAC1B,KAAKA,EAAK,WAAW,MACnB,OAAOA,EAAK,YAAY,WAC1B,KAAKA,EAAK,WAAW,SACnB,OAAAmJ,EAAO,WAAW,EACXnJ,EAAK,YAAY,cAC1B,QACE,IAAI6N,EAAe,2BAA6BC,EAAW,KAAO,IAClE,MAAM,IAAI9N,EAAK,gBAAiB6N,EAAcC,EAAW,MAAOA,EAAW,GAAG,CAClF,EACF,EAEA9N,EAAK,YAAY,kBAAoB,SAAUmJ,EAAQ,CACrD,IAAIwE,EAASxE,EAAO,cAAc,EAElC,GAAIwE,GAAU,KAId,KAAIxG,EAAe,SAASwG,EAAO,IAAK,EAAE,EAE1C,GAAI,MAAMxG,CAAY,EAAG,CACvB,IAAI0G,EAAe,gCACnB,MAAM,IAAI7N,EAAK,gBAAiB6N,EAAcF,EAAO,MAAOA,EAAO,GAAG,CACxE,CAEAxE,EAAO,cAAc,aAAehC,EAEpC,IAAI2G,EAAa3E,EAAO,WAAW,EAEnC,GAAI2E,GAAc,KAAW,CAC3B3E,EAAO,WAAW,EAClB,MACF,CAEA,OAAQ2E,EAAW,KAAM,CACvB,KAAK9N,EAAK,WAAW,KACnB,OAAAmJ,EAAO,WAAW,EACXnJ,EAAK,YAAY,UAC1B,KAAKA,EAAK,WAAW,MACnB,OAAAmJ,EAAO,WAAW,EACXnJ,EAAK,YAAY,WAC1B,KAAKA,EAAK,WAAW,cACnB,OAAOA,EAAK,YAAY,kBAC1B,KAAKA,EAAK,WAAW,MACnB,OAAOA,EAAK,YAAY,WAC1B,KAAKA,EAAK,WAAW,SACnB,OAAAmJ,EAAO,WAAW,EACXnJ,EAAK,YAAY,cAC1B,QACE,IAAI6N,EAAe,2BAA6BC,EAAW,KAAO,IAClE,MAAM,IAAI9N,EAAK,gBAAiB6N,EAAcC,EAAW,MAAOA,EAAW,GAAG,CAClF,EACF,EAEA9N,EAAK,YAAY,WAAa,SAAUmJ,EAAQ,CAC9C,IAAIwE,EAASxE,EAAO,cAAc,EAElC,GAAIwE,GAAU,KAId,KAAIM,EAAQ,SAASN,EAAO,IAAK,EAAE,EAEnC,GAAI,MAAMM,CAAK,EAAG,CAChB,IAAIJ,EAAe,wBACnB,MAAM,IAAI7N,EAAK,gBAAiB6N,EAAcF,EAAO,MAAOA,EAAO,GAAG,CACxE,CAEAxE,EAAO,cAAc,MAAQ8E,EAE7B,IAAIH,EAAa3E,EAAO,WAAW,EAEnC,GAAI2E,GAAc,KAAW,CAC3B3E,EAAO,WAAW,EAClB,MACF,CAEA,OAAQ2E,EAAW,KAAM,CACvB,KAAK9N,EAAK,WAAW,KACnB,OAAAmJ,EAAO,WAAW,EACXnJ,EAAK,YAAY,UAC1B,KAAKA,EAAK,WAAW,MACnB,OAAAmJ,EAAO,WAAW,EACXnJ,EAAK,YAAY,WAC1B,KAAKA,EAAK,WAAW,cACnB,OAAOA,EAAK,YAAY,kBAC1B,KAAKA,EAAK,WAAW,MACnB,OAAOA,EAAK,YAAY,WAC1B,KAAKA,EAAK,WAAW,SACnB,OAAAmJ,EAAO,WAAW,EACXnJ,EAAK,YAAY,cAC1B,QACE,IAAI6N,EAAe,2BAA6BC,EAAW,KAAO,IAClE,MAAM,IAAI9N,EAAK,gBAAiB6N,EAAcC,EAAW,MAAOA,EAAW,GAAG,CAClF,EACF,EAMI,SAAU1G,EAAM8G,EAAS,CACrB,OAAO,QAAW,YAAc,OAAO,IAEzC,OAAOA,CAAO,EACL,OAAOpO,GAAY,SAM5BC,GAAO,QAAUmO,EAAQ,EAGzB9G,EAAK,KAAO8G,EAAQ,CAExB,EAAE,KAAM,UAAY,CAMlB,OAAOlO,CACT,CAAC,CACH,GAAG,IC53GH,IAAAmO,GAAiB,SCiDV,SAASC,GACdC,EAAkBC,EAAmB,SAClC,CACH,IAAMC,EAAKC,GAAsBH,EAAUC,CAAI,EAC/C,GAAI,OAAOC,GAAO,YAChB,MAAM,IAAI,eACR,8BAA8BF,CAAQ,iBACxC,EAGF,OAAOE,CACT,CAsBO,SAASC,GACdH,EAAkBC,EAAmB,SACtB,CACf,OAAOA,EAAK,cAAiBD,CAAQ,GAAK,MAC5C,CCjFK,OAAO,UACV,OAAO,QAAU,SAAUI,EAAa,CACtC,IAAMC,EAA2B,CAAC,EAClC,QAAWC,KAAO,OAAO,KAAKF,CAAG,EAE/BC,EAAK,KAAK,CAACC,EAAKF,EAAIE,CAAG,CAAC,CAAC,EAG3B,OAAOD,CACT,GAGG,OAAO,SACV,OAAO,OAAS,SAAUD,EAAa,CACrC,IAAMC,EAAiB,CAAC,EACxB,QAAWC,KAAO,OAAO,KAAKF,CAAG,EAE/BC,EAAK,KAAKD,EAAIE,CAAG,CAAC,EAGpB,OAAOD,CACT,GAKE,OAAO,SAAY,cAGhB,QAAQ,UAAU,WACrB,QAAQ,UAAU,SAAW,SAC3BE,EAA8BC,EACxB,CACF,OAAOD,GAAM,UACf,KAAK,WAAaA,EAAE,KACpB,KAAK,UAAYA,EAAE,MAEnB,KAAK,WAAaA,EAClB,KAAK,UAAYC,EAErB,GAGG,QAAQ,UAAU,cACrB,QAAQ,UAAU,YAAc,YAC3BC,EACG,CACN,IAAMC,EAAS,KAAK,WACpB,GAAIA,EAAQ,CACND,EAAM,SAAW,GACnBC,EAAO,YAAY,IAAI,EAGzB,QAASC,EAAIF,EAAM,OAAS,EAAGE,GAAK,EAAGA,IAAK,CAC1C,IAAIC,EAAOH,EAAME,CAAC,EACd,OAAOC,GAAS,SAClBA,EAAO,SAAS,eAAeA,CAAI,EAC5BA,EAAK,YACZA,EAAK,WAAW,YAAYA,CAAI,EAG7BD,EAGHD,EAAO,aAAa,KAAK,gBAAkBE,CAAI,EAF/CF,EAAO,aAAaE,EAAM,IAAI,CAGlC,CACF,CACF,ICDG,SAASC,GACdC,EAC6B,CAC7B,IAAMC,EAAM,IAAI,IAChB,QAAWC,KAAOF,EAAM,CACtB,GAAM,CAACG,CAAI,EAAID,EAAI,SAAS,MAAM,GAAG,EAG/BE,EAAUH,EAAI,IAAIE,CAAI,EACxB,OAAOC,GAAY,YACrBH,EAAI,IAAIE,EAAMD,CAAG,GAIjBD,EAAI,IAAIC,EAAI,SAAUA,CAAG,EACzBA,EAAI,OAASE,EAEjB,CAGA,OAAOH,CACT,CCnEO,SAASI,EACdC,EAAeC,EAAmBC,EAC5B,CAjDR,IAAAC,EAkDEF,EAAY,IAAI,OAAOA,EAAW,GAAG,EAGrC,IAAIG,EACAC,EAAQ,EACZ,EAAG,CACDD,EAAQH,EAAU,KAAKD,CAAK,EAG5B,IAAMM,GAAQH,EAAAC,GAAA,YAAAA,EAAO,QAAP,KAAAD,EAAgBH,EAAM,OAKpC,GAJIK,EAAQC,GACVJ,EAAGG,EAAOC,CAAK,EAGbF,EAAO,CACT,GAAM,CAACG,CAAI,EAAIH,EACfC,EAAQD,EAAM,MAAQG,EAAK,OAGvBA,EAAK,SAAW,IAClBN,EAAU,UAAYG,EAAM,MAAQ,EACxC,CACF,OAASA,EACX,CCFO,SAASI,GACdC,EAAeC,EACT,CAEN,IAAIC,EAAQ,EACRC,EAAQ,EACRC,EAAM,EAGV,QAASC,EAAQ,EAAGD,EAAMJ,EAAM,OAAQI,IAGlCJ,EAAM,OAAOI,CAAG,IAAM,KAAOA,EAAMD,EACrCF,EAAGC,EAAO,EAAcC,EAAOA,EAAQC,CAAG,EAGjCJ,EAAM,OAAOI,CAAG,IAAM,MAC3BJ,EAAM,OAAOG,EAAQ,CAAC,IAAM,IAC1B,EAAEE,IAAU,GACdJ,EAAGC,IAAS,EAAmBC,EAAOC,EAAM,CAAC,EAGtCJ,EAAM,OAAOI,EAAM,CAAC,IAAM,KAC/BC,MAAY,GACdJ,EAAGC,EAAO,EAAkBC,EAAOC,EAAM,CAAC,EAI9CD,EAAQC,EAAM,GAKdA,EAAMD,GACRF,EAAGC,EAAO,EAAcC,EAAOC,CAAG,CACtC,CCnDO,SAASE,GACdC,EAAeC,EAAsBC,EAAuBC,EAAO,GAC3D,CACR,OAAOC,EAAa,CAACJ,CAAK,EAAGC,EAAOC,EAAWC,CAAI,EAAE,IAAI,CAC3D,CAYO,SAASC,EACdC,EAAkBJ,EAAsBC,EAAuBC,EAAO,GAC5D,CAGV,IAAMG,EAAU,CAAC,CAAC,EAClB,QAASC,EAAI,EAAGA,EAAIN,EAAM,OAAQM,IAAK,CACrC,IAAMC,EAAOP,EAAMM,EAAI,CAAC,EAClBE,EAAOR,EAAMM,CAAC,EAGdG,EAAIF,EAAKA,EAAK,OAAS,CAAC,IAAM,EAAI,KAClCG,EAAIF,EAAK,CAAC,IAAoB,GAGpCH,EAAQ,KAAK,EAAEI,EAAIC,GAAKL,EAAQA,EAAQ,OAAS,CAAC,CAAC,CACrD,CAGA,OAAOD,EAAO,IAAI,CAACL,EAAOY,IAAM,CAC9B,IAAIC,EAAS,EAGPC,EAAS,IAAI,IACnB,QAAWJ,KAAKR,EAAU,KAAK,CAACa,EAAGC,IAAMD,EAAIC,CAAC,EAAG,CAC/C,IAAMC,EAAQP,EAAI,QACZQ,EAAQR,IAAM,GACpB,GAAIJ,EAAQY,CAAK,IAAMN,EACrB,SAGF,IAAIO,EAAQL,EAAO,IAAII,CAAK,EACxB,OAAOC,GAAU,aACnBL,EAAO,IAAII,EAAOC,EAAQ,CAAC,CAAC,EAG9BA,EAAM,KAAKF,CAAK,CAClB,CAGA,GAAIH,EAAO,OAAS,EAClB,OAAOd,EAGT,IAAMoB,EAAmB,CAAC,EAC1B,OAAW,CAACF,EAAOG,CAAO,IAAKP,EAAQ,CACrC,IAAMP,EAAIN,EAAMiB,CAAK,EAGfI,EAASf,EAAE,CAAC,IAAiB,GAC7BgB,EAAShB,EAAEA,EAAE,OAAS,CAAC,IAAM,GAC7BiB,EAASjB,EAAEA,EAAE,OAAS,CAAC,IAAM,EAAI,KAGnCJ,GAAQmB,EAAQT,GAClBO,EAAO,KAAKpB,EAAM,MAAMa,EAAQS,CAAK,CAAC,EAGxC,IAAIG,EAAQzB,EAAM,MAAMsB,EAAOC,EAAMC,CAAM,EAC3C,QAAWE,KAAKL,EAAQ,KAAK,CAACN,EAAGC,IAAMA,EAAID,CAAC,EAAG,CAG7C,IAAML,GAAKH,EAAEmB,CAAC,IAAM,IAAMJ,EACpBX,GAAKJ,EAAEmB,CAAC,IAAM,EAAI,MAAShB,EAGjCe,EAAQ,CACNA,EAAM,MAAM,EAAGf,CAAC,EAChB,SACAe,EAAM,MAAMf,EAAGC,CAAC,EAChB,UACAc,EAAM,MAAMd,CAAC,CACf,EAAE,KAAK,EAAE,CACX,CAMA,GAHAE,EAASU,EAAMC,EAGXJ,EAAO,KAAKK,CAAK,IAAM,EACzB,KACJ,CAGA,OAAItB,GAAQU,EAASb,EAAM,QACzBoB,EAAO,KAAKpB,EAAM,MAAMa,CAAM,CAAC,EAG1BO,EAAO,KAAK,EAAE,CACvB,CAAC,CACH,CChHO,SAASO,GACdC,EACc,CACd,IAAMC,EAAuB,CAAC,EAC9B,GAAI,OAAOD,GAAU,YACnB,OAAOC,EAGT,IAAMC,EAAS,MAAM,QAAQF,CAAK,EAAIA,EAAQ,CAACA,CAAK,EACpD,QAASG,EAAI,EAAGA,EAAID,EAAO,OAAQC,IAAK,CACtC,IAAMC,EAAQ,KAAK,UAAU,MACvBC,EAAQD,EAAM,OAGpBE,GAAQJ,EAAOC,CAAC,EAAG,CAACI,EAAOC,EAAMC,EAAOC,IAAQ,CA/DpD,IAAAC,EAiEM,OADAP,EAAAO,EAAMJ,GAASF,KAAfD,EAAAO,GAA0B,CAAC,GACnBH,EAAM,CAGZ,OACA,OACEJ,EAAMG,CAAK,EAAE,KACXE,GAAe,GACfC,EAAMD,GAAU,EAChBD,CACF,EACA,MAGF,OACE,IAAMI,EAAUV,EAAOC,CAAC,EAAE,MAAMM,EAAOC,CAAG,EAC1CG,EAAMD,EAAS,KAAK,UAAU,UAAW,CAACE,EAAOC,IAAU,CAOzD,GAAI,OAAO,KAAK,WAAc,YAAa,CACzC,IAAMC,EAAaJ,EAAQ,MAAME,EAAOC,CAAK,EAC7C,GAAI,WAAW,KAAK,KAAK,UAAU,OAAOC,CAAU,CAAC,EAAG,CACtD,IAAMC,EAAW,KAAK,UAAU,QAAQD,CAAU,EAClD,QAASE,EAAI,EAAGC,EAAI,EAAGD,EAAID,EAAS,OAAQC,IAG1Cd,EAAAG,KAAAH,EAAAG,GAAiB,CAAC,GAClBH,EAAMG,CAAK,EAAE,KACXE,EAAQK,EAAQK,GAAM,GACtBF,EAASC,CAAC,EAAE,QAAW,EACvBV,CACF,EAGAP,EAAO,KAAK,IAAI,KAAK,MACnBgB,EAASC,CAAC,EAAE,YAAY,EAAG,CACzB,SAAUX,GAAS,GAAKH,EAAMG,CAAK,EAAE,OAAS,CAChD,CACF,CAAC,EAGDY,GAAKF,EAASC,CAAC,EAAE,OAEnB,MACF,CACF,CAGAd,EAAMG,CAAK,EAAE,KACXE,EAAQK,GAAS,GACjBC,EAAQD,GAAU,EAClBN,CACF,EAGAP,EAAO,KAAK,IAAI,KAAK,MACnBW,EAAQ,MAAME,EAAOC,CAAK,EAAE,YAAY,EAAG,CACzC,SAAUR,GAAS,GAAKH,EAAMG,CAAK,EAAE,OAAS,CAChD,CACF,CAAC,CACH,CAAC,CACL,CACF,CAAC,CACH,CAGA,OAAON,CACT,CCjEO,SAASmB,GACdC,EAAeC,EAAgBC,GAAQA,EAC/B,CACR,OAAOF,EAGJ,KAAK,EAGL,MAAM,YAAY,EAChB,IAAI,CAACG,EAAOC,IAAUA,EAAQ,EAC3BD,EAAM,QAAQ,+BAAgC,IAAI,EAClDA,CACJ,EACC,KAAK,EAAE,EAGT,QAAQ,kCAAmC,EAAE,EAG7C,MAAM,MAAM,EACV,OAAO,CAACE,EAAMH,IAAS,CACtB,IAAMI,EAAOL,EAAGC,CAAI,EACpB,MAAO,CAAC,GAAGG,EAAM,GAAG,MAAM,QAAQC,CAAI,EAAIA,EAAO,CAACA,CAAI,CAAC,CACzD,EAAG,CAAC,CAAa,EAChB,IAAIJ,GAAQ,UAAU,KAAKA,CAAI,EAAI,GAAGA,CAAI,IAAMA,CAAI,EACpD,IAAIA,GAAQ,mBAAmB,KAAKA,CAAI,EAAIA,EAAO,GAAGA,CAAI,GAAG,EAC7D,KAAK,GAAG,CACf,CCxCO,SAASK,GACdC,EACQ,CAGR,OAAOC,GAAUD,EAAOE,GAAQ,CAC9B,IAAMC,EAAkB,CAAC,EAGnBC,EAAQ,IAAI,KAAK,WAAWF,CAAI,EACtCE,EAAM,IAAI,EAGV,OAAW,CAAE,KAAAC,EAAM,IAAKC,EAAM,MAAAC,EAAO,IAAAC,CAAI,IAAKJ,EAAM,QAClD,OAAQC,EAAM,CAGZ,IAAK,QACE,CAAC,QAAS,OAAQ,MAAM,EAAE,SAASC,CAAI,IAC1CJ,EAAO,CACLA,EAAK,MAAM,EAAGM,CAAG,EACjB,IACAN,EAAK,MAAMM,EAAM,CAAC,CACpB,EAAE,KAAK,EAAE,GACX,MAGF,IAAK,OACHC,EAAMH,EAAM,KAAK,UAAU,UAAW,IAAII,IAAU,CAClDP,EAAM,KAAK,CACTD,EAAK,MAAM,EAAGK,CAAK,EACnBD,EAAK,MAAM,GAAGI,CAAK,EACnBR,EAAK,MAAMM,CAAG,CAChB,EAAE,KAAK,EAAE,CAAC,CACZ,CAAC,CACL,CAGF,OAAOL,CACT,CAAC,CACH,CAgBO,SAASQ,GACdC,EACqB,CACrB,IAAMZ,EAAS,IAAI,KAAK,MAAM,CAAC,QAAS,OAAQ,MAAM,CAAC,EACxC,IAAI,KAAK,YAAYY,EAAOZ,CAAK,EAGzC,MAAM,EACb,QAAWa,KAAUb,EAAM,QACzBa,EAAO,YAAc,GAGjBA,EAAO,KAAK,WAAW,GAAG,IAC5BA,EAAO,SAAW,KAAK,MAAM,SAAS,QACtCA,EAAO,KAAOA,EAAO,KAAK,MAAM,CAAC,GAI/BA,EAAO,KAAK,SAAS,GAAG,IAC1BA,EAAO,SAAW,KAAK,MAAM,SAAS,SACtCA,EAAO,KAAOA,EAAO,KAAK,MAAM,EAAG,EAAE,GAKzC,OAAOb,EAAM,OACf,CAUO,SAASc,GACdd,EAA4BG,EACV,CAxJpB,IAAAY,EAyJE,IAAMC,EAAU,IAAI,IAAuBhB,CAAK,EAG1CiB,EAA2B,CAAC,EAClC,QAASC,EAAI,EAAGA,EAAIf,EAAM,OAAQe,IAChC,QAAWL,KAAUG,EACfb,EAAMe,CAAC,EAAE,WAAWL,EAAO,IAAI,IACjCI,EAAOJ,EAAO,IAAI,EAAI,GACtBG,EAAQ,OAAOH,CAAM,GAI3B,QAAWA,KAAUG,GACfD,EAAA,KAAK,iBAAL,MAAAA,EAAA,UAAsBF,EAAO,QAC/BI,EAAOJ,EAAO,IAAI,EAAI,IAG1B,OAAOI,CACT,CClIO,SAASE,GACdC,EAAeC,EACG,CAClB,IAAMC,EAAW,IAAI,IAGfC,EAAW,IAAI,YAAYH,EAAM,MAAM,EAC7C,QAASI,EAAI,EAAGA,EAAIJ,EAAM,OAAQI,IAChC,QAASC,EAAID,EAAI,EAAGC,EAAIL,EAAM,OAAQK,IACtBL,EAAM,MAAMI,EAAGC,CAAC,IACjBJ,IACXE,EAASC,CAAC,EAAIC,EAAID,GAIxB,IAAME,EAAQ,CAAC,CAAC,EAChB,QAAS,EAAIA,EAAM,OAAQ,EAAI,GAAI,CACjC,IAAMC,EAAID,EAAM,EAAE,CAAC,EACnB,QAASE,EAAI,EAAGA,EAAIL,EAASI,CAAC,EAAGC,IAC3BL,EAASI,EAAIC,CAAC,EAAIL,EAASI,CAAC,EAAIC,IAClCN,EAAS,IAAIF,EAAM,MAAMO,EAAGA,EAAIC,CAAC,CAAC,EAClCF,EAAM,GAAG,EAAIC,EAAIC,GAIrB,IAAMA,EAAID,EAAIJ,EAASI,CAAC,EACpBJ,EAASK,CAAC,GAAKA,EAAIR,EAAM,OAAS,IACpCM,EAAM,GAAG,EAAIE,GAGfN,EAAS,IAAIF,EAAM,MAAMO,EAAGC,CAAC,CAAC,CAChC,CAGA,OAAIN,EAAS,IAAI,EAAE,EACV,IAAI,IAAI,CAACF,CAAK,CAAC,EAGjBE,CACT,CCJA,SAASO,GAAUC,EAAmC,CACpD,OAAQC,GACEC,GAAwB,CAC9B,GAAI,OAAOA,EAAID,CAAI,GAAM,YACvB,OAGF,IAAME,EAAK,CAACD,EAAI,SAAUD,CAAI,EAAE,KAAK,GAAG,EACxC,OAAAD,EAAM,IAAIG,EAAI,KAAK,UAAU,MAAQ,CAAC,CAAC,EAGhCD,EAAID,CAAI,CACjB,CAEJ,CAUA,SAASG,GAAWC,EAAaC,EAAuB,CACtD,GAAM,CAACC,EAAGC,CAAC,EAAI,CAAC,IAAI,IAAIH,CAAC,EAAG,IAAI,IAAIC,CAAC,CAAC,EACtC,MAAO,CACL,GAAG,IAAI,IAAI,CAAC,GAAGC,CAAC,EAAE,OAAOE,GAAS,CAACD,EAAE,IAAIC,CAAK,CAAC,CAAC,CAClD,CACF,CASO,IAAMC,EAAN,KAAa,CA2BX,YAAY,CAAE,OAAAC,EAAQ,KAAAC,EAAM,QAAAC,CAAQ,EAAgB,CACzD,IAAMC,EAAQf,GAAU,KAAK,MAAQ,IAAI,GAAK,EAG9C,KAAK,IAAMgB,GAAuBH,CAAI,EACtC,KAAK,QAAUC,EAGf,KAAK,MAAQ,KAAK,UAAY,CAC5B,KAAK,kBAAoB,CAAC,UAAU,EACpC,KAAK,EAAE,CAAC,EAGJF,EAAO,KAAK,SAAW,GAAKA,EAAO,KAAK,CAAC,IAAM,KAEjD,KAAK,IAAI,KAAKA,EAAO,KAAK,CAAC,CAAC,CAAC,EACpBA,EAAO,KAAK,OAAS,GAC9B,KAAK,IAAI,KAAK,cAAc,GAAGA,EAAO,IAAI,CAAC,EAI7C,KAAK,UAAYK,GACjB,KAAK,UAAU,UAAY,IAAI,OAAOL,EAAO,SAAS,EAGtD,KAAK,UAAY,kBAAmB,KAChC,IAAI,KAAK,cACT,OAGJ,IAAMM,EAAMb,GAAW,CACrB,UAAW,iBAAkB,SAC/B,EAAGO,EAAO,QAAQ,EAGlB,QAAWO,KAAQP,EAAO,KAAK,IAAIQ,GAEjCA,IAAa,KAAO,KAAO,KAAKA,CAAQ,CACzC,EACC,QAAWC,KAAMH,EACf,KAAK,SAAS,OAAOC,EAAKE,CAAE,CAAC,EAC7B,KAAK,eAAe,OAAOF,EAAKE,CAAE,CAAC,EAIvC,KAAK,IAAI,UAAU,EAGnB,KAAK,MAAM,QAAS,CAAE,MAAO,IAAK,UAAWN,EAAM,OAAO,CAAE,CAAC,EAC7D,KAAK,MAAM,OAAS,CAAE,MAAO,EAAK,UAAWA,EAAM,MAAM,CAAE,CAAC,EAC5D,KAAK,MAAM,OAAS,CAAE,MAAO,IAAK,UAAWA,EAAM,MAAM,CAAE,CAAC,EAG5D,QAAWZ,KAAOU,EAChB,KAAK,IAAIV,EAAK,CAAE,MAAOA,EAAI,KAAM,CAAC,CACtC,CAAC,CACH,CASO,OAAOmB,EAA6B,CAUzC,GAPAA,EAAQA,EAAM,QAAQ,WAAC,eAAY,IAAE,EAAEZ,GAC9B,CAAC,GAAGa,GAAQb,EAAO,KAAK,MAAM,aAAa,CAAC,EAChD,KAAK,IAAI,CACb,EAGDY,EAAQE,GAAqBF,CAAK,EAC9B,CAACA,EACH,MAAO,CAAE,MAAO,CAAC,CAAE,EAGrB,IAAMG,EAAUC,GAAiBJ,CAAK,EACnC,OAAOK,GACNA,EAAO,WAAa,KAAK,MAAM,SAAS,UACzC,EAGGC,EAAS,KAAK,MAAM,OAAON,CAAK,EAGnC,OAAqB,CAACO,EAAM,CAAE,IAAAC,EAAK,MAAAC,EAAO,UAAAC,CAAU,IAAM,CACzD,IAAI7B,EAAM,KAAK,IAAI,IAAI2B,CAAG,EAC1B,GAAI,OAAO3B,GAAQ,YAAa,CAG9BA,EAAM8B,EAAA,GAAK9B,GACPA,EAAI,OACNA,EAAI,KAAO,CAAC,GAAGA,EAAI,IAAI,GAGzB,IAAM+B,EAAQC,GACZV,EACA,OAAO,KAAKO,EAAU,QAAQ,CAChC,EAGA,QAAWjB,KAAS,KAAK,MAAM,OAAQ,CACrC,GAAI,OAAOZ,EAAIY,CAAK,GAAM,YACxB,SAGF,IAAMqB,EAAwB,CAAC,EAC/B,QAAWC,KAAS,OAAO,OAAOL,EAAU,QAAQ,EAC9C,OAAOK,EAAMtB,CAAK,GAAM,aAC1BqB,EAAU,KAAK,GAAGC,EAAMtB,CAAK,EAAE,QAAQ,EAG3C,GAAI,CAACqB,EAAU,OACb,SAGF,IAAMnC,EAAQ,KAAK,MAAM,IAAI,CAACE,EAAI,SAAUY,CAAK,EAAE,KAAK,GAAG,CAAC,EACtDM,EAAK,MAAM,QAAQlB,EAAIY,CAAK,CAAC,EAC/BuB,EACAC,GAGJpC,EAAIY,CAAK,EAAIM,EAAGlB,EAAIY,CAAK,EAAGd,EAAOmC,EAAWrB,IAAU,MAAM,CAChE,CAGA,IAAMyB,EAAQ,CAAC,CAACrC,EAAI,OAClB,OAAO,OAAO+B,CAAK,EAChB,OAAOO,GAAKA,CAAC,EAAE,OAClB,OAAO,KAAKP,CAAK,EAAE,OAGrBL,EAAK,KAAKa,EAAAT,EAAA,GACL9B,GADK,CAER,MAAO4B,GAAS,EAAIY,EAAAH,EAAS,IAC7B,MAAAN,CACF,EAAC,CACH,CACA,OAAOL,CACT,EAAG,CAAC,CAAC,EAGJ,KAAK,CAACvB,EAAGC,IAAMA,EAAE,MAAQD,EAAE,KAAK,EAGhC,OAAO,CAACsC,EAAOC,IAAW,CACzB,IAAM1C,EAAM,KAAK,IAAI,IAAI0C,EAAO,QAAQ,EACxC,GAAI,OAAO1C,GAAQ,YAAa,CAC9B,IAAM2B,EAAM3B,EAAI,OACZA,EAAI,OAAO,SACXA,EAAI,SACRyC,EAAM,IAAId,EAAK,CAAC,GAAGc,EAAM,IAAId,CAAG,GAAK,CAAC,EAAGe,CAAM,CAAC,CAClD,CACA,OAAOD,CACT,EAAG,IAAI,GAA2B,EAGpC,OAAW,CAACd,EAAKc,CAAK,IAAKhB,EACzB,GAAI,CAACgB,EAAM,KAAKf,GAAQA,EAAK,WAAaC,CAAG,EAAG,CAC9C,IAAM3B,EAAM,KAAK,IAAI,IAAI2B,CAAG,EAC5Bc,EAAM,KAAKF,EAAAT,EAAA,GAAK9B,GAAL,CAAU,MAAO,EAAG,MAAO,CAAC,CAAE,EAAC,CAC5C,CAGF,IAAI2C,EACJ,GAAI,KAAK,QAAQ,QAAS,CACxB,IAAMC,EAAS,KAAK,MAAM,MAAMC,GAAW,CACzC,QAAWrB,KAAUF,EACnBuB,EAAQ,KAAKrB,EAAO,KAAM,CACxB,OAAQ,CAAC,OAAO,EAChB,SAAU,KAAK,MAAM,SAAS,SAC9B,SAAU,KAAK,MAAM,SAAS,QAChC,CAAC,CACL,CAAC,EAGDmB,EAAUC,EAAO,OACb,OAAO,KAAKA,EAAO,CAAC,EAAE,UAAU,QAAQ,EACxC,CAAC,CACP,CAGA,OAAOd,EAAA,CACL,MAAO,CAAC,GAAGL,EAAO,OAAO,CAAC,GACvB,OAAOkB,GAAY,aAAe,CAAE,QAAAA,CAAQ,EAEnD,CACF,EX5QA,IAAIG,GAqBJ,SAAeC,GACbC,EACe,QAAAC,EAAA,sBACf,IAAIC,EAAO,UAGX,GAAI,OAAO,QAAW,aAAe,iBAAkB,OAAQ,CAC7D,IAAMC,EAASC,GAA8B,aAAa,EACpD,CAACC,CAAI,EAAIF,EAAO,IAAI,MAAM,SAAS,EAGzCD,EAAOA,EAAK,QAAQ,KAAMG,CAAI,CAChC,CAGA,IAAMC,EAAU,CAAC,EACjB,QAAWC,KAAQP,EAAO,KAAM,CAC9B,OAAQO,EAAM,CAGZ,IAAK,KACHD,EAAQ,KAAK,GAAGJ,CAAI,aAAa,EACjC,MAGF,IAAK,KACL,IAAK,KACHI,EAAQ,KAAK,GAAGJ,CAAI,aAAa,EACjC,KACJ,CAGIK,IAAS,MACXD,EAAQ,KAAK,GAAGJ,CAAI,aAAaK,CAAI,SAAS,CAClD,CAGIP,EAAO,KAAK,OAAS,GACvBM,EAAQ,KAAK,GAAGJ,CAAI,wBAAwB,EAG1CI,EAAQ,SACV,MAAM,cACJ,GAAGJ,CAAI,mCACP,GAAGI,CACL,EACJ,GAaA,SAAsBE,GACpBC,EACwB,QAAAR,EAAA,sBACxB,OAAQQ,EAAQ,KAAM,CAGpB,OACE,aAAMV,GAAqBU,EAAQ,KAAK,MAAM,EAC9CX,GAAQ,IAAIY,EAAOD,EAAQ,IAAI,EACxB,CACL,MACF,EAGF,OACE,IAAME,EAAQF,EAAQ,KACtB,GAAI,CACF,MAAO,CACL,OACA,KAAMX,GAAM,OAAOa,CAAK,CAC1B,CAGF,OAASC,EAAK,CACZ,eAAQ,KAAK,kBAAkBD,CAAK,oCAA+B,EACnE,QAAQ,KAAKC,CAAG,EACT,CACL,OACA,KAAM,CAAE,MAAO,CAAC,CAAE,CACpB,CACF,CAGF,QACE,MAAM,IAAI,UAAU,sBAAsB,CAC9C,CACF,GAOA,KAAK,KAAO,GAAAC,QAGZ,iBAAiB,UAAiBC,GAAMb,EAAA,wBACtC,YAAY,MAAMO,GAAQM,EAAG,IAAI,CAAC,CACpC,EAAC",
   "names": ["require_lunr", "__commonJSMin", "exports", "module", "lunr", "config", "builder", "global", "message", "obj", "clone", "keys", "key", "val", "docRef", "fieldName", "stringValue", "s", "n", "fieldRef", "elements", "i", "other", "object", "a", "b", "intersection", "element", "posting", "documentCount", "documentsWithTerm", "x", "str", "metadata", "fn", "t", "len", "tokens", "sliceEnd", "sliceStart", "char", "sliceLength", "tokenMetadata", "label", "isRegistered", "serialised", "pipeline", "fnName", "fns", "existingFn", "newFn", "pos", "stackLength", "memo", "j", "result", "k", "token", "index", "start", "end", "pivotPoint", "pivotIndex", "insertIdx", "position", "sumOfSquares", "elementsLength", "otherVector", "dotProduct", "aLen", "bLen", "aVal", "bVal", "output", "step2list", "step3list", "c", "v", "C", "V", "mgr0", "meq1", "mgr1", "s_v", "re_mgr0", "re_mgr1", "re_meq1", "re_s_v", "re_1a", "re2_1a", "re_1b", "re2_1b", "re_1b_2", "re2_1b_2", "re3_1b_2", "re4_1b_2", "re_1c", "re_2", "re_3", "re_4", "re2_4", "re_5", "re_5_1", "re3_5", "porterStemmer", "w", "stem", "suffix", "firstch", "re", "re2", "re3", "re4", "fp", "stopWords", "words", "stopWord", "arr", "clause", "editDistance", "root", "stack", "frame", "noEditNode", "insertionNode", "substitutionNode", "charA", "charB", "transposeNode", "node", "final", "next", "edges", "edge", "labels", "qEdges", "qLen", "nEdges", "nLen", "q", "qEdge", "nEdge", "qNode", "word", "commonPrefix", "nextNode", "downTo", "childKey", "attrs", "queryString", "query", "parser", "matchingFields", "queryVectors", "termFieldCache", "requiredMatches", "prohibitedMatches", "terms", "clauseMatches", "m", "term", "termTokenSet", "expandedTerms", "field", "expandedTerm", "termIndex", "fieldPosting", "matchingDocumentRefs", "termField", "matchingDocumentsSet", "l", "matchingDocumentRef", "matchingFieldRef", "fieldMatch", "allRequiredMatches", "allProhibitedMatches", "matchingFieldRefs", "results", "matches", "fieldVector", "score", "docMatch", "match", "invertedIndex", "fieldVectors", "ref", "serializedIndex", "serializedVectors", "serializedInvertedIndex", "tokenSetBuilder", "tuple", "attributes", "number", "doc", "fields", "extractor", "fieldTerms", "metadataKey", "fieldRefs", "numberOfFields", "accumulator", "documentsWithField", "fieldRefsLength", "termIdfCache", "fieldLength", "termFrequencies", "termsLength", "fieldBoost", "docBoost", "tf", "idf", "scoreWithPrecision", "args", "clonedMetadata", "metadataKeys", "otherMatchData", "allFields", "options", "state", "subSlices", "type", "charCode", "lexer", "lexeme", "completedClause", "errorMessage", "nextLexeme", "possibleFields", "f", "boost", "factory", "import_lunr", "getElement", "selector", "node", "el", "getOptionalElement", "obj", "data", "key", "x", "y", "nodes", "parent", "i", "node", "setupSearchDocumentMap", "docs", "map", "doc", "path", "article", "split", "input", "separator", "fn", "_a", "match", "index", "until", "term", "extract", "input", "fn", "block", "start", "end", "stack", "highlight", "input", "table", "positions", "full", "highlightAll", "inputs", "mapping", "t", "prev", "next", "p", "q", "i", "cursor", "blocks", "a", "b", "index", "block", "group", "slices", "indexes", "start", "end", "length", "slice", "j", "tokenize", "input", "tokens", "inputs", "i", "table", "total", "extract", "block", "type", "start", "end", "_a", "section", "split", "index", "until", "subsection", "segments", "s", "l", "transform", "query", "fn", "term", "parts", "index", "prev", "next", "transformSearchQuery", "query", "transform", "part", "terms", "lexer", "type", "term", "start", "end", "split", "range", "parseSearchQuery", "value", "clause", "getSearchQueryTerms", "_a", "clauses", "result", "t", "segment", "query", "index", "segments", "wordcuts", "i", "j", "stack", "p", "q", "extractor", "table", "name", "doc", "id", "difference", "a", "b", "x", "y", "value", "Search", "config", "docs", "options", "field", "setupSearchDocumentMap", "tokenize", "fns", "lang", "language", "fn", "query", "segment", "transformSearchQuery", "clauses", "parseSearchQuery", "clause", "groups", "item", "ref", "score", "matchData", "__spreadValues", "terms", "getSearchQueryTerms", "positions", "match", "highlightAll", "highlight", "boost", "t", "__spreadProps", "__pow", "items", "result", "suggest", "titles", "builder", "index", "setupSearchLanguages", "config", "__async", "base", "worker", "getElement", "path", "scripts", "lang", "handler", "message", "Search", "query", "err", "lunr", "ev"]
 }
diff --git a/assets/stylesheets/main.bcfcd587.min.css b/assets/stylesheets/main.bcfcd587.min.css
new file mode 100644
index 0000000000..8102c8c088
--- /dev/null
+++ b/assets/stylesheets/main.bcfcd587.min.css
@@ -0,0 +1 @@
+@charset "UTF-8";html{-webkit-text-size-adjust:none;-moz-text-size-adjust:none;text-size-adjust:none;box-sizing:border-box}*,:after,:before{box-sizing:inherit}@media (prefers-reduced-motion){*,:after,:before{transition:none!important}}body{margin:0}a,button,input,label{-webkit-tap-highlight-color:transparent}a{color:inherit;text-decoration:none}hr{border:0;box-sizing:initial;display:block;height:.05rem;overflow:visible;padding:0}small{font-size:80%}sub,sup{line-height:1em}img{border-style:none}table{border-collapse:initial;border-spacing:0}td,th{font-weight:400;vertical-align:top}button{background:#0000;border:0;font-family:inherit;font-size:inherit;margin:0;padding:0}input{border:0;outline:none}:root{--md-primary-fg-color:#4051b5;--md-primary-fg-color--light:#5d6cc0;--md-primary-fg-color--dark:#303fa1;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3;--md-accent-fg-color:#526cfe;--md-accent-fg-color--transparent:#526cfe1a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-scheme=default]{color-scheme:light}[data-md-color-scheme=default] img[src$="#gh-dark-mode-only"],[data-md-color-scheme=default] img[src$="#only-dark"]{display:none}:root,[data-md-color-scheme=default]{--md-hue:225deg;--md-default-fg-color:#000000de;--md-default-fg-color--light:#0000008a;--md-default-fg-color--lighter:#00000052;--md-default-fg-color--lightest:#00000012;--md-default-bg-color:#fff;--md-default-bg-color--light:#ffffffb3;--md-default-bg-color--lighter:#ffffff4d;--md-default-bg-color--lightest:#ffffff1f;--md-code-fg-color:#36464e;--md-code-bg-color:#f5f5f5;--md-code-hl-color:#4287ff;--md-code-hl-color--light:#4287ff1a;--md-code-hl-number-color:#d52a2a;--md-code-hl-special-color:#db1457;--md-code-hl-function-color:#a846b9;--md-code-hl-constant-color:#6e59d9;--md-code-hl-keyword-color:#3f6ec6;--md-code-hl-string-color:#1c7d4d;--md-code-hl-name-color:var(--md-code-fg-color);--md-code-hl-operator-color:var(--md-default-fg-color--light);--md-code-hl-punctuation-color:var(--md-default-fg-color--light);--md-code-hl-comment-color:var(--md-default-fg-color--light);--md-code-hl-generic-color:var(--md-default-fg-color--light);--md-code-hl-variable-color:var(--md-default-fg-color--light);--md-typeset-color:var(--md-default-fg-color);--md-typeset-a-color:var(--md-primary-fg-color);--md-typeset-del-color:#f5503d26;--md-typeset-ins-color:#0bd57026;--md-typeset-kbd-color:#fafafa;--md-typeset-kbd-accent-color:#fff;--md-typeset-kbd-border-color:#b8b8b8;--md-typeset-mark-color:#ffff0080;--md-typeset-table-color:#0000001f;--md-typeset-table-color--light:rgba(0,0,0,.035);--md-admonition-fg-color:var(--md-default-fg-color);--md-admonition-bg-color:var(--md-default-bg-color);--md-warning-fg-color:#000000de;--md-warning-bg-color:#ff9;--md-footer-fg-color:#fff;--md-footer-fg-color--light:#ffffffb3;--md-footer-fg-color--lighter:#ffffff73;--md-footer-bg-color:#000000de;--md-footer-bg-color--dark:#00000052;--md-shadow-z1:0 0.2rem 0.5rem #0000000d,0 0 0.05rem #0000001a;--md-shadow-z2:0 0.2rem 0.5rem #0000001a,0 0 0.05rem #00000040;--md-shadow-z3:0 0.2rem 0.5rem #0003,0 0 0.05rem #00000059}.md-icon svg{fill:currentcolor;display:block;height:1.2rem;width:1.2rem}body{-webkit-font-smoothing:antialiased;-moz-osx-font-smoothing:grayscale;--md-text-font-family:var(--md-text-font,_),-apple-system,BlinkMacSystemFont,Helvetica,Arial,sans-serif;--md-code-font-family:var(--md-code-font,_),SFMono-Regular,Consolas,Menlo,monospace}aside,body,input{font-feature-settings:"kern","liga";color:var(--md-typeset-color);font-family:var(--md-text-font-family)}code,kbd,pre{font-feature-settings:"kern";font-family:var(--md-code-font-family)}:root{--md-typeset-table-sort-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="m18 21-4-4h3V7h-3l4-4 4 4h-3v10h3M2 19v-2h10v2M2 13v-2h7v2M2 7V5h4v2H2Z"/></svg>');--md-typeset-table-sort-icon--asc:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M19 17h3l-4 4-4-4h3V3h2M2 17h10v2H2M6 5v2H2V5m0 6h7v2H2v-2Z"/></svg>');--md-typeset-table-sort-icon--desc:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M19 7h3l-4-4-4 4h3v14h2M2 17h10v2H2M6 5v2H2V5m0 6h7v2H2v-2Z"/></svg>')}.md-typeset{-webkit-print-color-adjust:exact;color-adjust:exact;font-size:.8rem;line-height:1.6}@media print{.md-typeset{font-size:.68rem}}.md-typeset blockquote,.md-typeset dl,.md-typeset figure,.md-typeset ol,.md-typeset pre,.md-typeset ul{margin-bottom:1em;margin-top:1em}.md-typeset h1{color:var(--md-default-fg-color--light);font-size:2em;line-height:1.3;margin:0 0 1.25em}.md-typeset h1,.md-typeset h2{font-weight:300;letter-spacing:-.01em}.md-typeset h2{font-size:1.5625em;line-height:1.4;margin:1.6em 0 .64em}.md-typeset h3{font-size:1.25em;font-weight:400;letter-spacing:-.01em;line-height:1.5;margin:1.6em 0 .8em}.md-typeset h2+h3{margin-top:.8em}.md-typeset h4{font-weight:700;letter-spacing:-.01em;margin:1em 0}.md-typeset h5,.md-typeset h6{color:var(--md-default-fg-color--light);font-size:.8em;font-weight:700;letter-spacing:-.01em;margin:1.25em 0}.md-typeset h5{text-transform:uppercase}.md-typeset hr{border-bottom:.05rem solid var(--md-default-fg-color--lightest);display:flow-root;margin:1.5em 0}.md-typeset a{color:var(--md-typeset-a-color);word-break:break-word}.md-typeset a,.md-typeset a:before{transition:color 125ms}.md-typeset a:focus,.md-typeset a:hover{color:var(--md-accent-fg-color)}.md-typeset a:focus code,.md-typeset a:hover code{background-color:var(--md-accent-fg-color--transparent)}.md-typeset a code{color:currentcolor;transition:background-color 125ms}.md-typeset a.focus-visible{outline-color:var(--md-accent-fg-color);outline-offset:.2rem}.md-typeset code,.md-typeset kbd,.md-typeset pre{color:var(--md-code-fg-color);direction:ltr;font-variant-ligatures:none}@media print{.md-typeset code,.md-typeset kbd,.md-typeset pre{white-space:pre-wrap}}.md-typeset code{background-color:var(--md-code-bg-color);border-radius:.1rem;-webkit-box-decoration-break:clone;box-decoration-break:clone;font-size:.85em;padding:0 .2941176471em;word-break:break-word}.md-typeset code:not(.focus-visible){-webkit-tap-highlight-color:transparent;outline:none}.md-typeset pre{display:flow-root;line-height:1.4;position:relative}.md-typeset pre>code{-webkit-box-decoration-break:slice;box-decoration-break:slice;box-shadow:none;display:block;margin:0;outline-color:var(--md-accent-fg-color);overflow:auto;padding:.7720588235em 1.1764705882em;scrollbar-color:var(--md-default-fg-color--lighter) #0000;scrollbar-width:thin;touch-action:auto;word-break:normal}.md-typeset pre>code:hover{scrollbar-color:var(--md-accent-fg-color) #0000}.md-typeset pre>code::-webkit-scrollbar{height:.2rem;width:.2rem}.md-typeset pre>code::-webkit-scrollbar-thumb{background-color:var(--md-default-fg-color--lighter)}.md-typeset pre>code::-webkit-scrollbar-thumb:hover{background-color:var(--md-accent-fg-color)}.md-typeset kbd{background-color:var(--md-typeset-kbd-color);border-radius:.1rem;box-shadow:0 .1rem 0 .05rem var(--md-typeset-kbd-border-color),0 .1rem 0 var(--md-typeset-kbd-border-color),0 -.1rem .2rem var(--md-typeset-kbd-accent-color) inset;color:var(--md-default-fg-color);display:inline-block;font-size:.75em;padding:0 .6666666667em;vertical-align:text-top;word-break:break-word}.md-typeset mark{background-color:var(--md-typeset-mark-color);-webkit-box-decoration-break:clone;box-decoration-break:clone;color:inherit;word-break:break-word}.md-typeset abbr{border-bottom:.05rem dotted var(--md-default-fg-color--light);cursor:help;text-decoration:none}.md-typeset small{opacity:.75}[dir=ltr] .md-typeset sub,[dir=ltr] .md-typeset sup{margin-left:.078125em}[dir=rtl] .md-typeset sub,[dir=rtl] .md-typeset sup{margin-right:.078125em}[dir=ltr] .md-typeset blockquote{padding-left:.6rem}[dir=rtl] .md-typeset blockquote{padding-right:.6rem}[dir=ltr] .md-typeset blockquote{border-left:.2rem solid var(--md-default-fg-color--lighter)}[dir=rtl] .md-typeset blockquote{border-right:.2rem solid var(--md-default-fg-color--lighter)}.md-typeset blockquote{color:var(--md-default-fg-color--light);margin-left:0;margin-right:0}.md-typeset ul{list-style-type:disc}[dir=ltr] .md-typeset ol,[dir=ltr] .md-typeset ul{margin-left:.625em}[dir=rtl] .md-typeset ol,[dir=rtl] .md-typeset ul{margin-right:.625em}.md-typeset ol,.md-typeset ul{padding:0}.md-typeset ol:not([hidden]),.md-typeset ul:not([hidden]){display:flow-root}.md-typeset ol ol,.md-typeset ul ol{list-style-type:lower-alpha}.md-typeset ol ol ol,.md-typeset ul ol ol{list-style-type:lower-roman}[dir=ltr] .md-typeset ol li,[dir=ltr] .md-typeset ul li{margin-left:1.25em}[dir=rtl] .md-typeset ol li,[dir=rtl] .md-typeset ul li{margin-right:1.25em}.md-typeset ol li,.md-typeset ul li{margin-bottom:.5em}.md-typeset ol li blockquote,.md-typeset ol li p,.md-typeset ul li blockquote,.md-typeset ul li p{margin:.5em 0}.md-typeset ol li:last-child,.md-typeset ul li:last-child{margin-bottom:0}[dir=ltr] .md-typeset ol li ol,[dir=ltr] .md-typeset ol li ul,[dir=ltr] .md-typeset ul li ol,[dir=ltr] .md-typeset ul li ul{margin-left:.625em}[dir=rtl] .md-typeset ol li ol,[dir=rtl] .md-typeset ol li ul,[dir=rtl] .md-typeset ul li ol,[dir=rtl] .md-typeset ul li ul{margin-right:.625em}.md-typeset ol li ol,.md-typeset ol li ul,.md-typeset ul li ol,.md-typeset ul li ul{margin-bottom:.5em;margin-top:.5em}[dir=ltr] .md-typeset dd{margin-left:1.875em}[dir=rtl] .md-typeset dd{margin-right:1.875em}.md-typeset dd{margin-bottom:1.5em;margin-top:1em}.md-typeset img,.md-typeset svg,.md-typeset video{height:auto;max-width:100%}.md-typeset img[align=left]{margin:1em 1em 1em 0}.md-typeset img[align=right]{margin:1em 0 1em 1em}.md-typeset img[align]:only-child{margin-top:0}.md-typeset figure{display:flow-root;margin:1em auto;max-width:100%;text-align:center;width:-moz-fit-content;width:fit-content}.md-typeset figure img{display:block;margin:0 auto}.md-typeset figcaption{font-style:italic;margin:1em auto;max-width:24rem}.md-typeset iframe{max-width:100%}.md-typeset table:not([class]){background-color:var(--md-default-bg-color);border:.05rem solid var(--md-typeset-table-color);border-radius:.1rem;display:inline-block;font-size:.64rem;max-width:100%;overflow:auto;touch-action:auto}@media print{.md-typeset table:not([class]){display:table}}.md-typeset table:not([class])+*{margin-top:1.5em}.md-typeset table:not([class]) td>:first-child,.md-typeset table:not([class]) th>:first-child{margin-top:0}.md-typeset table:not([class]) td>:last-child,.md-typeset table:not([class]) th>:last-child{margin-bottom:0}.md-typeset table:not([class]) td:not([align]),.md-typeset table:not([class]) th:not([align]){text-align:left}[dir=rtl] .md-typeset table:not([class]) td:not([align]),[dir=rtl] .md-typeset table:not([class]) th:not([align]){text-align:right}.md-typeset table:not([class]) th{font-weight:700;min-width:5rem;padding:.9375em 1.25em;vertical-align:top}.md-typeset table:not([class]) td{border-top:.05rem solid var(--md-typeset-table-color);padding:.9375em 1.25em;vertical-align:top}.md-typeset table:not([class]) tbody tr{transition:background-color 125ms}.md-typeset table:not([class]) tbody tr:hover{background-color:var(--md-typeset-table-color--light);box-shadow:0 .05rem 0 var(--md-default-bg-color) inset}.md-typeset table:not([class]) a{word-break:normal}.md-typeset table th[role=columnheader]{cursor:pointer}[dir=ltr] .md-typeset table th[role=columnheader]:after{margin-left:.5em}[dir=rtl] .md-typeset table th[role=columnheader]:after{margin-right:.5em}.md-typeset table th[role=columnheader]:after{content:"";display:inline-block;height:1.2em;-webkit-mask-image:var(--md-typeset-table-sort-icon);mask-image:var(--md-typeset-table-sort-icon);-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;transition:background-color 125ms;vertical-align:text-bottom;width:1.2em}.md-typeset table th[role=columnheader]:hover:after{background-color:var(--md-default-fg-color--lighter)}.md-typeset table th[role=columnheader][aria-sort=ascending]:after{background-color:var(--md-default-fg-color--light);-webkit-mask-image:var(--md-typeset-table-sort-icon--asc);mask-image:var(--md-typeset-table-sort-icon--asc)}.md-typeset table th[role=columnheader][aria-sort=descending]:after{background-color:var(--md-default-fg-color--light);-webkit-mask-image:var(--md-typeset-table-sort-icon--desc);mask-image:var(--md-typeset-table-sort-icon--desc)}.md-typeset__scrollwrap{margin:1em -.8rem;overflow-x:auto;touch-action:auto}.md-typeset__table{display:inline-block;margin-bottom:.5em;padding:0 .8rem}@media print{.md-typeset__table{display:block}}html .md-typeset__table table{display:table;margin:0;overflow:hidden;width:100%}@media screen and (max-width:44.984375em){.md-content__inner>pre{margin:1em -.8rem}.md-content__inner>pre code{border-radius:0}}.md-typeset .md-author{border-radius:100%;display:block;flex-shrink:0;height:1.6rem;overflow:hidden;position:relative;transition:color 125ms,transform 125ms;width:1.6rem}.md-typeset .md-author img{display:block}.md-typeset .md-author--more{background:var(--md-default-fg-color--lightest);color:var(--md-default-fg-color--lighter);font-size:.6rem;font-weight:700;line-height:1.6rem;text-align:center}.md-typeset .md-author--long{height:2.4rem;width:2.4rem}.md-typeset a.md-author{transform:scale(1)}.md-typeset a.md-author img{filter:grayscale(100%) opacity(75%);transition:filter 125ms}.md-typeset a.md-author:focus,.md-typeset a.md-author:hover{transform:scale(1.1);z-index:1}.md-typeset a.md-author:focus img,.md-typeset a.md-author:hover img{filter:grayscale(0)}.md-banner{background-color:var(--md-footer-bg-color);color:var(--md-footer-fg-color);overflow:auto}@media print{.md-banner{display:none}}.md-banner--warning{background-color:var(--md-warning-bg-color);color:var(--md-warning-fg-color)}.md-banner__inner{font-size:.7rem;margin:.6rem auto;padding:0 .8rem}[dir=ltr] .md-banner__button{float:right}[dir=rtl] .md-banner__button{float:left}.md-banner__button{color:inherit;cursor:pointer;transition:opacity .25s}.no-js .md-banner__button{display:none}.md-banner__button:hover{opacity:.7}html{font-size:125%;height:100%;overflow-x:hidden}@media screen and (min-width:100em){html{font-size:137.5%}}@media screen and (min-width:125em){html{font-size:150%}}body{background-color:var(--md-default-bg-color);display:flex;flex-direction:column;font-size:.5rem;min-height:100%;position:relative;width:100%}@media print{body{display:block}}@media screen and (max-width:59.984375em){body[data-md-scrolllock]{position:fixed}}.md-grid{margin-left:auto;margin-right:auto;max-width:61rem}.md-container{display:flex;flex-direction:column;flex-grow:1}@media print{.md-container{display:block}}.md-main{flex-grow:1}.md-main__inner{display:flex;height:100%;margin-top:1.5rem}.md-ellipsis{overflow:hidden;text-overflow:ellipsis}.md-toggle{display:none}.md-option{height:0;opacity:0;position:absolute;width:0}.md-option:checked+label:not([hidden]){display:block}.md-option.focus-visible+label{outline-color:var(--md-accent-fg-color);outline-style:auto}.md-skip{background-color:var(--md-default-fg-color);border-radius:.1rem;color:var(--md-default-bg-color);font-size:.64rem;margin:.5rem;opacity:0;outline-color:var(--md-accent-fg-color);padding:.3rem .5rem;position:fixed;transform:translateY(.4rem);z-index:-1}.md-skip:focus{opacity:1;transform:translateY(0);transition:transform .25s cubic-bezier(.4,0,.2,1),opacity 175ms 75ms;z-index:10}@page{margin:25mm}:root{--md-clipboard-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M19 21H8V7h11m0-2H8a2 2 0 0 0-2 2v14a2 2 0 0 0 2 2h11a2 2 0 0 0 2-2V7a2 2 0 0 0-2-2m-3-4H4a2 2 0 0 0-2 2v14h2V3h12V1Z"/></svg>')}.md-clipboard{border-radius:.1rem;color:var(--md-default-fg-color--lightest);cursor:pointer;height:1.5em;outline-color:var(--md-accent-fg-color);outline-offset:.1rem;position:absolute;right:.5em;top:.5em;transition:color .25s;width:1.5em;z-index:1}@media print{.md-clipboard{display:none}}.md-clipboard:not(.focus-visible){-webkit-tap-highlight-color:transparent;outline:none}:hover>.md-clipboard{color:var(--md-default-fg-color--light)}.md-clipboard:focus,.md-clipboard:hover{color:var(--md-accent-fg-color)}.md-clipboard:after{background-color:currentcolor;content:"";display:block;height:1.125em;margin:0 auto;-webkit-mask-image:var(--md-clipboard-icon);mask-image:var(--md-clipboard-icon);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;width:1.125em}.md-clipboard--inline{cursor:pointer}.md-clipboard--inline code{transition:color .25s,background-color .25s}.md-clipboard--inline:focus code,.md-clipboard--inline:hover code{background-color:var(--md-accent-fg-color--transparent);color:var(--md-accent-fg-color)}@keyframes consent{0%{opacity:0;transform:translateY(100%)}to{opacity:1;transform:translateY(0)}}@keyframes overlay{0%{opacity:0}to{opacity:1}}.md-consent__overlay{animation:overlay .25s both;-webkit-backdrop-filter:blur(.1rem);backdrop-filter:blur(.1rem);background-color:#0000008a;height:100%;opacity:1;position:fixed;top:0;width:100%;z-index:5}.md-consent__inner{animation:consent .5s cubic-bezier(.1,.7,.1,1) both;background-color:var(--md-default-bg-color);border:0;border-radius:.1rem;bottom:0;box-shadow:0 0 .2rem #0000001a,0 .2rem .4rem #0003;max-height:100%;overflow:auto;padding:0;position:fixed;width:100%;z-index:5}.md-consent__form{padding:.8rem}.md-consent__settings{display:none;margin:1em 0}input:checked+.md-consent__settings{display:block}.md-consent__controls{margin-bottom:.8rem}.md-typeset .md-consent__controls .md-button{display:inline}@media screen and (max-width:44.984375em){.md-typeset .md-consent__controls .md-button{display:block;margin-top:.4rem;text-align:center;width:100%}}.md-consent label{cursor:pointer}.md-content{flex-grow:1;min-width:0}.md-content__inner{margin:0 .8rem 1.2rem;padding-top:.6rem}@media screen and (min-width:76.25em){[dir=ltr] .md-sidebar--primary:not([hidden])~.md-content>.md-content__inner{margin-left:1.2rem}[dir=ltr] .md-sidebar--secondary:not([hidden])~.md-content>.md-content__inner,[dir=rtl] .md-sidebar--primary:not([hidden])~.md-content>.md-content__inner{margin-right:1.2rem}[dir=rtl] .md-sidebar--secondary:not([hidden])~.md-content>.md-content__inner{margin-left:1.2rem}}.md-content__inner:before{content:"";display:block;height:.4rem}.md-content__inner>:last-child{margin-bottom:0}[dir=ltr] .md-content__button{float:right}[dir=rtl] .md-content__button{float:left}[dir=ltr] .md-content__button{margin-left:.4rem}[dir=rtl] .md-content__button{margin-right:.4rem}.md-content__button{margin:.4rem 0;padding:0}@media print{.md-content__button{display:none}}.md-typeset .md-content__button{color:var(--md-default-fg-color--lighter)}.md-content__button svg{display:inline;vertical-align:top}[dir=rtl] .md-content__button svg{transform:scaleX(-1)}[dir=ltr] .md-dialog{right:.8rem}[dir=rtl] .md-dialog{left:.8rem}.md-dialog{background-color:var(--md-default-fg-color);border-radius:.1rem;bottom:.8rem;box-shadow:var(--md-shadow-z3);min-width:11.1rem;opacity:0;padding:.4rem .6rem;pointer-events:none;position:fixed;transform:translateY(100%);transition:transform 0ms .4s,opacity .4s;z-index:4}@media print{.md-dialog{display:none}}.md-dialog--active{opacity:1;pointer-events:auto;transform:translateY(0);transition:transform .4s cubic-bezier(.075,.85,.175,1),opacity .4s}.md-dialog__inner{color:var(--md-default-bg-color);font-size:.7rem}.md-feedback{margin:2em 0 1em;text-align:center}.md-feedback fieldset{border:none;margin:0;padding:0}.md-feedback__title{font-weight:700;margin:1em auto}.md-feedback__inner{position:relative}.md-feedback__list{display:flex;flex-wrap:wrap;place-content:baseline center;position:relative}.md-feedback__list:hover .md-icon:not(:disabled){color:var(--md-default-fg-color--lighter)}:disabled .md-feedback__list{min-height:1.8rem}.md-feedback__icon{color:var(--md-default-fg-color--light);cursor:pointer;flex-shrink:0;margin:0 .1rem;transition:color 125ms}.md-feedback__icon:not(:disabled).md-icon:hover{color:var(--md-accent-fg-color)}.md-feedback__icon:disabled{color:var(--md-default-fg-color--lightest);pointer-events:none}.md-feedback__note{opacity:0;position:relative;transform:translateY(.4rem);transition:transform .4s cubic-bezier(.1,.7,.1,1),opacity .15s}.md-feedback__note>*{margin:0 auto;max-width:16rem}:disabled .md-feedback__note{opacity:1;transform:translateY(0)}.md-footer{background-color:var(--md-footer-bg-color);color:var(--md-footer-fg-color)}@media print{.md-footer{display:none}}.md-footer__inner{justify-content:space-between;overflow:auto;padding:.2rem}.md-footer__inner:not([hidden]){display:flex}.md-footer__link{align-items:end;display:flex;flex-grow:0.01;margin-bottom:.4rem;margin-top:1rem;max-width:100%;outline-color:var(--md-accent-fg-color);overflow:hidden;transition:opacity .25s}.md-footer__link:focus,.md-footer__link:hover{opacity:.7}[dir=rtl] .md-footer__link svg{transform:scaleX(-1)}@media screen and (max-width:44.984375em){.md-footer__link--prev{flex-shrink:0}.md-footer__link--prev .md-footer__title{display:none}}[dir=ltr] .md-footer__link--next{margin-left:auto}[dir=rtl] .md-footer__link--next{margin-right:auto}.md-footer__link--next{text-align:right}[dir=rtl] .md-footer__link--next{text-align:left}.md-footer__title{flex-grow:1;font-size:.9rem;margin-bottom:.7rem;max-width:calc(100% - 2.4rem);padding:0 1rem;white-space:nowrap}.md-footer__button{margin:.2rem;padding:.4rem}.md-footer__direction{font-size:.64rem;opacity:.7}.md-footer-meta{background-color:var(--md-footer-bg-color--dark)}.md-footer-meta__inner{display:flex;flex-wrap:wrap;justify-content:space-between;padding:.2rem}html .md-footer-meta.md-typeset a{color:var(--md-footer-fg-color--light)}html .md-footer-meta.md-typeset a:focus,html .md-footer-meta.md-typeset a:hover{color:var(--md-footer-fg-color)}.md-copyright{color:var(--md-footer-fg-color--lighter);font-size:.64rem;margin:auto .6rem;padding:.4rem 0;width:100%}@media screen and (min-width:45em){.md-copyright{width:auto}}.md-copyright__highlight{color:var(--md-footer-fg-color--light)}.md-social{display:inline-flex;gap:.2rem;margin:0 .4rem;padding:.2rem 0 .6rem}@media screen and (min-width:45em){.md-social{padding:.6rem 0}}.md-social__link{display:inline-block;height:1.6rem;text-align:center;width:1.6rem}.md-social__link:before{line-height:1.9}.md-social__link svg{fill:currentcolor;max-height:.8rem;vertical-align:-25%}.md-typeset .md-button{border:.1rem solid;border-radius:.1rem;color:var(--md-primary-fg-color);cursor:pointer;display:inline-block;font-weight:700;padding:.625em 2em;transition:color 125ms,background-color 125ms,border-color 125ms}.md-typeset .md-button--primary{background-color:var(--md-primary-fg-color);border-color:var(--md-primary-fg-color);color:var(--md-primary-bg-color)}.md-typeset .md-button:focus,.md-typeset .md-button:hover{background-color:var(--md-accent-fg-color);border-color:var(--md-accent-fg-color);color:var(--md-accent-bg-color)}[dir=ltr] .md-typeset .md-input{border-top-left-radius:.1rem}[dir=ltr] .md-typeset .md-input,[dir=rtl] .md-typeset .md-input{border-top-right-radius:.1rem}[dir=rtl] .md-typeset .md-input{border-top-left-radius:.1rem}.md-typeset .md-input{border-bottom:.1rem solid var(--md-default-fg-color--lighter);box-shadow:var(--md-shadow-z1);font-size:.8rem;height:1.8rem;padding:0 .6rem;transition:border .25s,box-shadow .25s}.md-typeset .md-input:focus,.md-typeset .md-input:hover{border-bottom-color:var(--md-accent-fg-color);box-shadow:var(--md-shadow-z2)}.md-typeset .md-input--stretch{width:100%}.md-header{background-color:var(--md-primary-fg-color);box-shadow:0 0 .2rem #0000,0 .2rem .4rem #0000;color:var(--md-primary-bg-color);display:block;left:0;position:sticky;right:0;top:0;z-index:4}@media print{.md-header{display:none}}.md-header[hidden]{transform:translateY(-100%);transition:transform .25s cubic-bezier(.8,0,.6,1),box-shadow .25s}.md-header--shadow{box-shadow:0 0 .2rem #0000001a,0 .2rem .4rem #0003;transition:transform .25s cubic-bezier(.1,.7,.1,1),box-shadow .25s}.md-header__inner{align-items:center;display:flex;padding:0 .2rem}.md-header__button{color:currentcolor;cursor:pointer;margin:.2rem;outline-color:var(--md-accent-fg-color);padding:.4rem;position:relative;transition:opacity .25s;vertical-align:middle;z-index:1}.md-header__button:hover{opacity:.7}.md-header__button:not([hidden]){display:inline-block}.md-header__button:not(.focus-visible){-webkit-tap-highlight-color:transparent;outline:none}.md-header__button.md-logo{margin:.2rem;padding:.4rem}@media screen and (max-width:76.234375em){.md-header__button.md-logo{display:none}}.md-header__button.md-logo img,.md-header__button.md-logo svg{fill:currentcolor;display:block;height:1.2rem;width:auto}@media screen and (min-width:60em){.md-header__button[for=__search]{display:none}}.no-js .md-header__button[for=__search]{display:none}[dir=rtl] .md-header__button[for=__search] svg{transform:scaleX(-1)}@media screen and (min-width:76.25em){.md-header__button[for=__drawer]{display:none}}.md-header__topic{display:flex;max-width:100%;position:absolute;transition:transform .4s cubic-bezier(.1,.7,.1,1),opacity .15s;white-space:nowrap}.md-header__topic+.md-header__topic{opacity:0;pointer-events:none;transform:translateX(1.25rem);transition:transform .4s cubic-bezier(1,.7,.1,.1),opacity .15s;z-index:-1}[dir=rtl] .md-header__topic+.md-header__topic{transform:translateX(-1.25rem)}.md-header__topic:first-child{font-weight:700}[dir=ltr] .md-header__title{margin-left:1rem;margin-right:.4rem}[dir=rtl] .md-header__title{margin-left:.4rem;margin-right:1rem}.md-header__title{flex-grow:1;font-size:.9rem;height:2.4rem;line-height:2.4rem}.md-header__title--active .md-header__topic{opacity:0;pointer-events:none;transform:translateX(-1.25rem);transition:transform .4s cubic-bezier(1,.7,.1,.1),opacity .15s;z-index:-1}[dir=rtl] .md-header__title--active .md-header__topic{transform:translateX(1.25rem)}.md-header__title--active .md-header__topic+.md-header__topic{opacity:1;pointer-events:auto;transform:translateX(0);transition:transform .4s cubic-bezier(.1,.7,.1,1),opacity .15s;z-index:0}.md-header__title>.md-header__ellipsis{height:100%;position:relative;width:100%}.md-header__option{display:flex;flex-shrink:0;max-width:100%;transition:max-width 0ms .25s,opacity .25s .25s;white-space:nowrap}[data-md-toggle=search]:checked~.md-header .md-header__option{max-width:0;opacity:0;transition:max-width 0ms,opacity 0ms}.md-header__option>input{bottom:0}.md-header__source{display:none}@media screen and (min-width:60em){[dir=ltr] .md-header__source{margin-left:1rem}[dir=rtl] .md-header__source{margin-right:1rem}.md-header__source{display:block;max-width:11.7rem;width:11.7rem}}@media screen and (min-width:76.25em){[dir=ltr] .md-header__source{margin-left:1.4rem}[dir=rtl] .md-header__source{margin-right:1.4rem}}.md-meta{color:var(--md-default-fg-color--light);font-size:.7rem;line-height:1.3}.md-meta__list{display:inline-flex;flex-wrap:wrap;list-style:none;margin:0;padding:0}.md-meta__item:not(:last-child):after{content:"·";margin-left:.2rem;margin-right:.2rem}.md-meta__link{color:var(--md-typeset-a-color)}.md-meta__link:focus,.md-meta__link:hover{color:var(--md-accent-fg-color)}.md-draft{background-color:#ff1744;border-radius:.125em;color:#fff;display:inline-block;font-weight:700;padding-left:.5714285714em;padding-right:.5714285714em}:root{--md-nav-icon--prev:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M20 11v2H8l5.5 5.5-1.42 1.42L4.16 12l7.92-7.92L13.5 5.5 8 11h12Z"/></svg>');--md-nav-icon--next:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M8.59 16.58 13.17 12 8.59 7.41 10 6l6 6-6 6-1.41-1.42Z"/></svg>');--md-toc-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M3 9h14V7H3v2m0 4h14v-2H3v2m0 4h14v-2H3v2m16 0h2v-2h-2v2m0-10v2h2V7h-2m0 6h2v-2h-2v2Z"/></svg>')}.md-nav{font-size:.7rem;line-height:1.3}.md-nav__title{color:var(--md-default-fg-color--light);display:block;font-weight:700;overflow:hidden;padding:0 .6rem;text-overflow:ellipsis}.md-nav__title .md-nav__button{display:none}.md-nav__title .md-nav__button img{height:100%;width:auto}.md-nav__title .md-nav__button.md-logo img,.md-nav__title .md-nav__button.md-logo svg{fill:currentcolor;display:block;height:2.4rem;max-width:100%;object-fit:contain;width:auto}.md-nav__list{list-style:none;margin:0;padding:0}.md-nav__link{align-items:flex-start;display:flex;gap:.4rem;margin-top:.625em;scroll-snap-align:start;transition:color 125ms}.md-nav__link--passed{color:var(--md-default-fg-color--light)}.md-nav__item .md-nav__link--active,.md-nav__item .md-nav__link--active code{color:var(--md-typeset-a-color)}.md-nav__link .md-ellipsis{position:relative}[dir=ltr] .md-nav__link .md-icon:last-child{margin-left:auto}[dir=rtl] .md-nav__link .md-icon:last-child{margin-right:auto}.md-nav__link svg{fill:currentcolor;flex-shrink:0;height:1.3em}.md-nav__link[for]:focus,.md-nav__link[for]:hover,.md-nav__link[href]:focus,.md-nav__link[href]:hover{color:var(--md-accent-fg-color);cursor:pointer}.md-nav__link.focus-visible{outline-color:var(--md-accent-fg-color);outline-offset:.2rem}.md-nav--primary .md-nav__link[for=__toc]{display:none}.md-nav--primary .md-nav__link[for=__toc] .md-icon:after{background-color:currentcolor;display:block;height:100%;-webkit-mask-image:var(--md-toc-icon);mask-image:var(--md-toc-icon);width:100%}.md-nav--primary .md-nav__link[for=__toc]~.md-nav{display:none}.md-nav__container>.md-nav__link{margin-top:0}.md-nav__container>.md-nav__link:first-child{flex-grow:1;min-width:0}.md-nav__icon{flex-shrink:0}.md-nav__source{display:none}@media screen and (max-width:76.234375em){.md-nav--primary,.md-nav--primary .md-nav{background-color:var(--md-default-bg-color);display:flex;flex-direction:column;height:100%;left:0;position:absolute;right:0;top:0;z-index:1}.md-nav--primary .md-nav__item,.md-nav--primary .md-nav__title{font-size:.8rem;line-height:1.5}.md-nav--primary .md-nav__title{background-color:var(--md-default-fg-color--lightest);color:var(--md-default-fg-color--light);cursor:pointer;height:5.6rem;line-height:2.4rem;padding:3rem .8rem .2rem;position:relative;white-space:nowrap}[dir=ltr] .md-nav--primary .md-nav__title .md-nav__icon{left:.4rem}[dir=rtl] .md-nav--primary .md-nav__title .md-nav__icon{right:.4rem}.md-nav--primary .md-nav__title .md-nav__icon{display:block;height:1.2rem;margin:.2rem;position:absolute;top:.4rem;width:1.2rem}.md-nav--primary .md-nav__title .md-nav__icon:after{background-color:currentcolor;content:"";display:block;height:100%;-webkit-mask-image:var(--md-nav-icon--prev);mask-image:var(--md-nav-icon--prev);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;width:100%}.md-nav--primary .md-nav__title~.md-nav__list{background-color:var(--md-default-bg-color);box-shadow:0 .05rem 0 var(--md-default-fg-color--lightest) inset;overflow-y:auto;scroll-snap-type:y mandatory;touch-action:pan-y}.md-nav--primary .md-nav__title~.md-nav__list>:first-child{border-top:0}.md-nav--primary .md-nav__title[for=__drawer]{background-color:var(--md-primary-fg-color);color:var(--md-primary-bg-color);font-weight:700}.md-nav--primary .md-nav__title .md-logo{display:block;left:.2rem;margin:.2rem;padding:.4rem;position:absolute;right:.2rem;top:.2rem}.md-nav--primary .md-nav__list{flex:1}.md-nav--primary .md-nav__item{border-top:.05rem solid var(--md-default-fg-color--lightest)}.md-nav--primary .md-nav__item--active>.md-nav__link{color:var(--md-typeset-a-color)}.md-nav--primary .md-nav__item--active>.md-nav__link:focus,.md-nav--primary .md-nav__item--active>.md-nav__link:hover{color:var(--md-accent-fg-color)}.md-nav--primary .md-nav__link{margin-top:0;padding:.6rem .8rem}.md-nav--primary .md-nav__link svg{margin-top:.1em}.md-nav--primary .md-nav__link>.md-nav__link{padding:0}[dir=ltr] .md-nav--primary .md-nav__link .md-nav__icon{margin-right:-.2rem}[dir=rtl] .md-nav--primary .md-nav__link .md-nav__icon{margin-left:-.2rem}.md-nav--primary .md-nav__link .md-nav__icon{font-size:1.2rem;height:1.2rem;width:1.2rem}.md-nav--primary .md-nav__link .md-nav__icon:after{background-color:currentcolor;content:"";display:block;height:100%;-webkit-mask-image:var(--md-nav-icon--next);mask-image:var(--md-nav-icon--next);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;width:100%}[dir=rtl] .md-nav--primary .md-nav__icon:after{transform:scale(-1)}.md-nav--primary .md-nav--secondary .md-nav{background-color:initial;position:static}[dir=ltr] .md-nav--primary .md-nav--secondary .md-nav .md-nav__link{padding-left:1.4rem}[dir=rtl] .md-nav--primary .md-nav--secondary .md-nav .md-nav__link{padding-right:1.4rem}[dir=ltr] .md-nav--primary .md-nav--secondary .md-nav .md-nav .md-nav__link{padding-left:2rem}[dir=rtl] .md-nav--primary .md-nav--secondary .md-nav .md-nav .md-nav__link{padding-right:2rem}[dir=ltr] .md-nav--primary .md-nav--secondary .md-nav .md-nav .md-nav .md-nav__link{padding-left:2.6rem}[dir=rtl] .md-nav--primary .md-nav--secondary .md-nav .md-nav .md-nav .md-nav__link{padding-right:2.6rem}[dir=ltr] .md-nav--primary .md-nav--secondary .md-nav .md-nav .md-nav .md-nav .md-nav__link{padding-left:3.2rem}[dir=rtl] .md-nav--primary .md-nav--secondary .md-nav .md-nav .md-nav .md-nav .md-nav__link{padding-right:3.2rem}.md-nav--secondary{background-color:initial}.md-nav__toggle~.md-nav{display:flex;opacity:0;transform:translateX(100%);transition:transform .25s cubic-bezier(.8,0,.6,1),opacity 125ms 50ms}[dir=rtl] .md-nav__toggle~.md-nav{transform:translateX(-100%)}.md-nav__toggle:checked~.md-nav{opacity:1;transform:translateX(0);transition:transform .25s cubic-bezier(.4,0,.2,1),opacity 125ms 125ms}.md-nav__toggle:checked~.md-nav>.md-nav__list{-webkit-backface-visibility:hidden;backface-visibility:hidden}}@media screen and (max-width:59.984375em){.md-nav--primary .md-nav__link[for=__toc]{display:flex}.md-nav--primary .md-nav__link[for=__toc] .md-icon:after{content:""}.md-nav--primary .md-nav__link[for=__toc]+.md-nav__link{display:none}.md-nav--primary .md-nav__link[for=__toc]~.md-nav{display:flex}.md-nav__source{background-color:var(--md-primary-fg-color--dark);color:var(--md-primary-bg-color);display:block;padding:0 .2rem}}@media screen and (min-width:60em) and (max-width:76.234375em){.md-nav--integrated .md-nav__link[for=__toc]{display:flex}.md-nav--integrated .md-nav__link[for=__toc] .md-icon:after{content:""}.md-nav--integrated .md-nav__link[for=__toc]+.md-nav__link{display:none}.md-nav--integrated .md-nav__link[for=__toc]~.md-nav{display:flex}}@media screen and (min-width:60em){.md-nav{margin-bottom:-.4rem}.md-nav--secondary .md-nav__title{background:var(--md-default-bg-color);box-shadow:0 0 .4rem .4rem var(--md-default-bg-color);position:sticky;top:0;z-index:1}.md-nav--secondary .md-nav__title[for=__toc]{scroll-snap-align:start}.md-nav--secondary .md-nav__title .md-nav__icon{display:none}[dir=ltr] .md-nav--secondary .md-nav__list{padding-left:.6rem}[dir=rtl] .md-nav--secondary .md-nav__list{padding-right:.6rem}.md-nav--secondary .md-nav__list{padding-bottom:.4rem}[dir=ltr] .md-nav--secondary .md-nav__item>.md-nav__link{margin-right:.4rem}[dir=rtl] .md-nav--secondary .md-nav__item>.md-nav__link{margin-left:.4rem}}@media screen and (min-width:76.25em){.md-nav{margin-bottom:-.4rem;transition:max-height .25s cubic-bezier(.86,0,.07,1)}.md-nav--primary .md-nav__title{background:var(--md-default-bg-color);box-shadow:0 0 .4rem .4rem var(--md-default-bg-color);position:sticky;top:0;z-index:1}.md-nav--primary .md-nav__title[for=__drawer]{scroll-snap-align:start}.md-nav--primary .md-nav__title .md-nav__icon{display:none}[dir=ltr] .md-nav--primary .md-nav__list{padding-left:.6rem}[dir=rtl] .md-nav--primary .md-nav__list{padding-right:.6rem}.md-nav--primary .md-nav__list{padding-bottom:.4rem}[dir=ltr] .md-nav--primary .md-nav__item>.md-nav__link{margin-right:.4rem}[dir=rtl] .md-nav--primary .md-nav__item>.md-nav__link{margin-left:.4rem}.md-nav__toggle~.md-nav{display:grid;grid-template-rows:0fr;opacity:0;transition:grid-template-rows .25s cubic-bezier(.86,0,.07,1),opacity .25s,visibility 0ms .25s;visibility:collapse}.md-nav__toggle~.md-nav>.md-nav__list{overflow:hidden}.md-nav__toggle.md-toggle--indeterminate~.md-nav,.md-nav__toggle:checked~.md-nav{grid-template-rows:1fr;opacity:1;transition:grid-template-rows .25s cubic-bezier(.86,0,.07,1),opacity .15s .1s,visibility 0ms;visibility:visible}.md-nav__toggle.md-toggle--indeterminate~.md-nav{transition:none}.md-nav__item--nested>.md-nav>.md-nav__title{display:none}.md-nav__item--section{display:block;margin:1.25em 0}.md-nav__item--section:last-child{margin-bottom:0}.md-nav__item--section>.md-nav__link{font-weight:700}.md-nav__item--section>.md-nav__link[for]{color:var(--md-default-fg-color--light)}.md-nav__item--section>.md-nav__link:not(.md-nav__container){pointer-events:none}.md-nav__item--section>.md-nav__link .md-icon,.md-nav__item--section>.md-nav__link>[for]{display:none}[dir=ltr] .md-nav__item--section>.md-nav{margin-left:-.6rem}[dir=rtl] .md-nav__item--section>.md-nav{margin-right:-.6rem}.md-nav__item--section>.md-nav{display:block;opacity:1;visibility:visible}.md-nav__item--section>.md-nav>.md-nav__list>.md-nav__item{padding:0}.md-nav__icon{border-radius:100%;height:.9rem;transition:background-color .25s;width:.9rem}.md-nav__icon:hover{background-color:var(--md-accent-fg-color--transparent)}.md-nav__icon:after{background-color:currentcolor;border-radius:100%;content:"";display:inline-block;height:100%;-webkit-mask-image:var(--md-nav-icon--next);mask-image:var(--md-nav-icon--next);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;transition:transform .25s;vertical-align:-.1rem;width:100%}[dir=rtl] .md-nav__icon:after{transform:rotate(180deg)}.md-nav__item--nested .md-nav__toggle:checked~.md-nav__link .md-nav__icon:after,.md-nav__item--nested .md-toggle--indeterminate~.md-nav__link .md-nav__icon:after{transform:rotate(90deg)}.md-nav--lifted>.md-nav__list>.md-nav__item,.md-nav--lifted>.md-nav__title{display:none}.md-nav--lifted>.md-nav__list>.md-nav__item--active{display:block}.md-nav--lifted>.md-nav__list>.md-nav__item--active>.md-nav__link{background:var(--md-default-bg-color);box-shadow:0 0 .4rem .4rem var(--md-default-bg-color);margin-top:0;position:sticky;top:0;z-index:1}.md-nav--lifted>.md-nav__list>.md-nav__item--active>.md-nav__link:not(.md-nav__container){pointer-events:none}.md-nav--lifted>.md-nav__list>.md-nav__item--active.md-nav__item--section{margin:0}[dir=ltr] .md-nav--lifted>.md-nav__list>.md-nav__item>.md-nav:not(.md-nav--secondary){margin-left:-.6rem}[dir=rtl] .md-nav--lifted>.md-nav__list>.md-nav__item>.md-nav:not(.md-nav--secondary){margin-right:-.6rem}.md-nav--lifted>.md-nav__list>.md-nav__item>[for]{color:var(--md-default-fg-color--light)}.md-nav--lifted .md-nav[data-md-level="1"]{grid-template-rows:1fr;opacity:1;visibility:visible}[dir=ltr] .md-nav--integrated>.md-nav__list>.md-nav__item--active .md-nav--secondary{border-left:.05rem solid var(--md-primary-fg-color)}[dir=rtl] .md-nav--integrated>.md-nav__list>.md-nav__item--active .md-nav--secondary{border-right:.05rem solid var(--md-primary-fg-color)}.md-nav--integrated>.md-nav__list>.md-nav__item--active .md-nav--secondary{display:block;margin-bottom:1.25em;opacity:1;visibility:visible}.md-nav--integrated>.md-nav__list>.md-nav__item--active .md-nav--secondary>.md-nav__list{overflow:visible;padding-bottom:0}.md-nav--integrated>.md-nav__list>.md-nav__item--active .md-nav--secondary>.md-nav__title{display:none}}.md-pagination{font-size:.8rem;font-weight:700;gap:.4rem}.md-pagination,.md-pagination>*{align-items:center;display:flex;justify-content:center}.md-pagination>*{border-radius:.2rem;height:1.8rem;min-width:1.8rem;text-align:center}.md-pagination__current{background-color:var(--md-default-fg-color--lightest);color:var(--md-default-fg-color--light)}.md-pagination__link{transition:color 125ms,background-color 125ms}.md-pagination__link:focus,.md-pagination__link:hover{background-color:var(--md-accent-fg-color--transparent);color:var(--md-accent-fg-color)}.md-pagination__link:focus svg,.md-pagination__link:hover svg{color:var(--md-accent-fg-color)}.md-pagination__link.focus-visible{outline-color:var(--md-accent-fg-color);outline-offset:.2rem}.md-pagination__link svg{fill:currentcolor;color:var(--md-default-fg-color--lighter);display:block;max-height:100%;width:1.2rem}.md-post__back{border-bottom:.05rem solid var(--md-default-fg-color--lightest);margin-bottom:1.2rem;padding-bottom:1.2rem}@media screen and (max-width:76.234375em){.md-post__back{display:none}}[dir=rtl] .md-post__back svg{transform:scaleX(-1)}.md-post__authors{display:flex;flex-direction:column;gap:.6rem;margin:0 .6rem 1.2rem}.md-post .md-post__meta a{transition:color 125ms}.md-post .md-post__meta a:focus,.md-post .md-post__meta a:hover{color:var(--md-accent-fg-color)}.md-post__title{color:var(--md-default-fg-color--light);font-weight:700}.md-post--excerpt{margin-bottom:3.2rem}.md-post--excerpt .md-post__header{align-items:center;display:flex;gap:.6rem;min-height:1.6rem}.md-post--excerpt .md-post__authors{align-items:center;display:inline-flex;flex-direction:row;gap:.2rem;margin:0;min-height:2.4rem}[dir=ltr] .md-post--excerpt .md-post__meta .md-meta__list{margin-right:.4rem}[dir=rtl] .md-post--excerpt .md-post__meta .md-meta__list{margin-left:.4rem}.md-post--excerpt .md-post__content>:first-child{--md-scroll-margin:6rem;margin-top:0}.md-post>.md-nav--secondary{margin:1em 0}.md-profile{align-items:center;display:flex;font-size:.7rem;gap:.6rem;line-height:1.4;width:100%}.md-profile__description{flex-grow:1}.md-content--post{display:flex}@media screen and (max-width:76.234375em){.md-content--post{flex-flow:column-reverse}}.md-content--post>.md-content__inner{min-width:0}@media screen and (min-width:76.25em){[dir=ltr] .md-content--post>.md-content__inner{margin-left:1.2rem}[dir=rtl] .md-content--post>.md-content__inner{margin-right:1.2rem}}@media screen and (max-width:76.234375em){.md-sidebar.md-sidebar--post{padding:0;position:static;width:100%}.md-sidebar.md-sidebar--post .md-sidebar__scrollwrap{overflow:visible}.md-sidebar.md-sidebar--post .md-sidebar__inner{padding:0}.md-sidebar.md-sidebar--post .md-post__meta{margin-left:.6rem;margin-right:.6rem}.md-sidebar.md-sidebar--post .md-nav__item{border:none;display:inline}.md-sidebar.md-sidebar--post .md-nav__list{display:inline-flex;flex-wrap:wrap;gap:.6rem;padding-bottom:.6rem;padding-top:.6rem}.md-sidebar.md-sidebar--post .md-nav__link{padding:0}.md-sidebar.md-sidebar--post .md-nav{height:auto;margin-bottom:0;position:static}}:root{--md-progress-value:0;--md-progress-delay:400ms}.md-progress{background:var(--md-primary-bg-color);height:.075rem;opacity:min(clamp(0,var(--md-progress-value),1),clamp(0,100 - var(--md-progress-value),1));position:fixed;top:0;transform:scaleX(calc(var(--md-progress-value)*1%));transform-origin:left;transition:transform .5s cubic-bezier(.19,1,.22,1),opacity .25s var(--md-progress-delay);width:100%;z-index:4}:root{--md-search-result-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M14 2H6a2 2 0 0 0-2 2v16a2 2 0 0 0 2 2h7c-.41-.25-.8-.56-1.14-.9-.33-.33-.61-.7-.86-1.1H6V4h7v5h5v1.18c.71.16 1.39.43 2 .82V8l-6-6m6.31 16.9c1.33-2.11.69-4.9-1.4-6.22-2.11-1.33-4.91-.68-6.22 1.4-1.34 2.11-.69 4.89 1.4 6.22 1.46.93 3.32.93 4.79.02L22 23.39 23.39 22l-3.08-3.1m-3.81.1a2.5 2.5 0 0 1-2.5-2.5 2.5 2.5 0 0 1 2.5-2.5 2.5 2.5 0 0 1 2.5 2.5 2.5 2.5 0 0 1-2.5 2.5Z"/></svg>')}.md-search{position:relative}@media screen and (min-width:60em){.md-search{padding:.2rem 0}}.no-js .md-search{display:none}.md-search__overlay{opacity:0;z-index:1}@media screen and (max-width:59.984375em){[dir=ltr] .md-search__overlay{left:-2.2rem}[dir=rtl] .md-search__overlay{right:-2.2rem}.md-search__overlay{background-color:var(--md-default-bg-color);border-radius:1rem;height:2rem;overflow:hidden;pointer-events:none;position:absolute;top:-1rem;transform-origin:center;transition:transform .3s .1s,opacity .2s .2s;width:2rem}[data-md-toggle=search]:checked~.md-header .md-search__overlay{opacity:1;transition:transform .4s,opacity .1s}}@media screen and (min-width:60em){[dir=ltr] .md-search__overlay{left:0}[dir=rtl] .md-search__overlay{right:0}.md-search__overlay{background-color:#0000008a;cursor:pointer;height:0;position:fixed;top:0;transition:width 0ms .25s,height 0ms .25s,opacity .25s;width:0}[data-md-toggle=search]:checked~.md-header .md-search__overlay{height:200vh;opacity:1;transition:width 0ms,height 0ms,opacity .25s;width:100%}}@media screen and (max-width:29.984375em){[data-md-toggle=search]:checked~.md-header .md-search__overlay{transform:scale(45)}}@media screen and (min-width:30em) and (max-width:44.984375em){[data-md-toggle=search]:checked~.md-header .md-search__overlay{transform:scale(60)}}@media screen and (min-width:45em) and (max-width:59.984375em){[data-md-toggle=search]:checked~.md-header .md-search__overlay{transform:scale(75)}}.md-search__inner{-webkit-backface-visibility:hidden;backface-visibility:hidden}@media screen and (max-width:59.984375em){[dir=ltr] .md-search__inner{left:0}[dir=rtl] .md-search__inner{right:0}.md-search__inner{height:0;opacity:0;overflow:hidden;position:fixed;top:0;transform:translateX(5%);transition:width 0ms .3s,height 0ms .3s,transform .15s cubic-bezier(.4,0,.2,1) .15s,opacity .15s .15s;width:0;z-index:2}[dir=rtl] .md-search__inner{transform:translateX(-5%)}[data-md-toggle=search]:checked~.md-header .md-search__inner{height:100%;opacity:1;transform:translateX(0);transition:width 0ms 0ms,height 0ms 0ms,transform .15s cubic-bezier(.1,.7,.1,1) .15s,opacity .15s .15s;width:100%}}@media screen and (min-width:60em){[dir=ltr] .md-search__inner{float:right}[dir=rtl] .md-search__inner{float:left}.md-search__inner{padding:.1rem 0;position:relative;transition:width .25s cubic-bezier(.1,.7,.1,1);width:11.7rem}}@media screen and (min-width:60em) and (max-width:76.234375em){[data-md-toggle=search]:checked~.md-header .md-search__inner{width:23.4rem}}@media screen and (min-width:76.25em){[data-md-toggle=search]:checked~.md-header .md-search__inner{width:34.4rem}}.md-search__form{background-color:var(--md-default-bg-color);box-shadow:0 0 .6rem #0000;height:2.4rem;position:relative;transition:color .25s,background-color .25s;z-index:2}@media screen and (min-width:60em){.md-search__form{background-color:#00000042;border-radius:.1rem;height:1.8rem}.md-search__form:hover{background-color:#ffffff1f}}[data-md-toggle=search]:checked~.md-header .md-search__form{background-color:var(--md-default-bg-color);border-radius:.1rem .1rem 0 0;box-shadow:0 0 .6rem #00000012;color:var(--md-default-fg-color)}[dir=ltr] .md-search__input{padding-left:3.6rem;padding-right:2.2rem}[dir=rtl] .md-search__input{padding-left:2.2rem;padding-right:3.6rem}.md-search__input{background:#0000;font-size:.9rem;height:100%;position:relative;text-overflow:ellipsis;width:100%;z-index:2}.md-search__input::placeholder{transition:color .25s}.md-search__input::placeholder,.md-search__input~.md-search__icon{color:var(--md-default-fg-color--light)}.md-search__input::-ms-clear{display:none}@media screen and (max-width:59.984375em){.md-search__input{font-size:.9rem;height:2.4rem;width:100%}}@media screen and (min-width:60em){[dir=ltr] .md-search__input{padding-left:2.2rem}[dir=rtl] .md-search__input{padding-right:2.2rem}.md-search__input{color:inherit;font-size:.8rem}.md-search__input::placeholder{color:var(--md-primary-bg-color--light)}.md-search__input+.md-search__icon{color:var(--md-primary-bg-color)}[data-md-toggle=search]:checked~.md-header .md-search__input{text-overflow:clip}[data-md-toggle=search]:checked~.md-header .md-search__input+.md-search__icon{color:var(--md-default-fg-color--light)}[data-md-toggle=search]:checked~.md-header .md-search__input::placeholder{color:#0000}}.md-search__icon{cursor:pointer;display:inline-block;height:1.2rem;transition:color .25s,opacity .25s;width:1.2rem}.md-search__icon:hover{opacity:.7}[dir=ltr] .md-search__icon[for=__search]{left:.5rem}[dir=rtl] .md-search__icon[for=__search]{right:.5rem}.md-search__icon[for=__search]{position:absolute;top:.3rem;z-index:2}[dir=rtl] .md-search__icon[for=__search] svg{transform:scaleX(-1)}@media screen and (max-width:59.984375em){[dir=ltr] .md-search__icon[for=__search]{left:.8rem}[dir=rtl] .md-search__icon[for=__search]{right:.8rem}.md-search__icon[for=__search]{top:.6rem}.md-search__icon[for=__search] svg:first-child{display:none}}@media screen and (min-width:60em){.md-search__icon[for=__search]{pointer-events:none}.md-search__icon[for=__search] svg:last-child{display:none}}[dir=ltr] .md-search__options{right:.5rem}[dir=rtl] .md-search__options{left:.5rem}.md-search__options{pointer-events:none;position:absolute;top:.3rem;z-index:2}@media screen and (max-width:59.984375em){[dir=ltr] .md-search__options{right:.8rem}[dir=rtl] .md-search__options{left:.8rem}.md-search__options{top:.6rem}}[dir=ltr] .md-search__options>.md-icon{margin-left:.2rem}[dir=rtl] .md-search__options>.md-icon{margin-right:.2rem}.md-search__options>.md-icon{color:var(--md-default-fg-color--light);opacity:0;transform:scale(.75);transition:transform .15s cubic-bezier(.1,.7,.1,1),opacity .15s}.md-search__options>.md-icon:not(.focus-visible){-webkit-tap-highlight-color:transparent;outline:none}[data-md-toggle=search]:checked~.md-header .md-search__input:valid~.md-search__options>.md-icon{opacity:1;pointer-events:auto;transform:scale(1)}[data-md-toggle=search]:checked~.md-header .md-search__input:valid~.md-search__options>.md-icon:hover{opacity:.7}[dir=ltr] .md-search__suggest{padding-left:3.6rem;padding-right:2.2rem}[dir=rtl] .md-search__suggest{padding-left:2.2rem;padding-right:3.6rem}.md-search__suggest{align-items:center;color:var(--md-default-fg-color--lighter);display:flex;font-size:.9rem;height:100%;opacity:0;position:absolute;top:0;transition:opacity 50ms;white-space:nowrap;width:100%}@media screen and (min-width:60em){[dir=ltr] .md-search__suggest{padding-left:2.2rem}[dir=rtl] .md-search__suggest{padding-right:2.2rem}.md-search__suggest{font-size:.8rem}}[data-md-toggle=search]:checked~.md-header .md-search__suggest{opacity:1;transition:opacity .3s .1s}[dir=ltr] .md-search__output{border-bottom-left-radius:.1rem}[dir=ltr] .md-search__output,[dir=rtl] .md-search__output{border-bottom-right-radius:.1rem}[dir=rtl] .md-search__output{border-bottom-left-radius:.1rem}.md-search__output{overflow:hidden;position:absolute;width:100%;z-index:1}@media screen and (max-width:59.984375em){.md-search__output{bottom:0;top:2.4rem}}@media screen and (min-width:60em){.md-search__output{opacity:0;top:1.9rem;transition:opacity .4s}[data-md-toggle=search]:checked~.md-header .md-search__output{box-shadow:var(--md-shadow-z3);opacity:1}}.md-search__scrollwrap{-webkit-backface-visibility:hidden;backface-visibility:hidden;background-color:var(--md-default-bg-color);height:100%;overflow-y:auto;touch-action:pan-y}@media (-webkit-max-device-pixel-ratio:1),(max-resolution:1dppx){.md-search__scrollwrap{transform:translateZ(0)}}@media screen and (min-width:60em) and (max-width:76.234375em){.md-search__scrollwrap{width:23.4rem}}@media screen and (min-width:76.25em){.md-search__scrollwrap{width:34.4rem}}@media screen and (min-width:60em){.md-search__scrollwrap{max-height:0;scrollbar-color:var(--md-default-fg-color--lighter) #0000;scrollbar-width:thin}[data-md-toggle=search]:checked~.md-header .md-search__scrollwrap{max-height:75vh}.md-search__scrollwrap:hover{scrollbar-color:var(--md-accent-fg-color) #0000}.md-search__scrollwrap::-webkit-scrollbar{height:.2rem;width:.2rem}.md-search__scrollwrap::-webkit-scrollbar-thumb{background-color:var(--md-default-fg-color--lighter)}.md-search__scrollwrap::-webkit-scrollbar-thumb:hover{background-color:var(--md-accent-fg-color)}}.md-search-result{color:var(--md-default-fg-color);word-break:break-word}.md-search-result__meta{background-color:var(--md-default-fg-color--lightest);color:var(--md-default-fg-color--light);font-size:.64rem;line-height:1.8rem;padding:0 .8rem;scroll-snap-align:start}@media screen and (min-width:60em){[dir=ltr] .md-search-result__meta{padding-left:2.2rem}[dir=rtl] .md-search-result__meta{padding-right:2.2rem}}.md-search-result__list{list-style:none;margin:0;padding:0;-webkit-user-select:none;user-select:none}.md-search-result__item{box-shadow:0 -.05rem var(--md-default-fg-color--lightest)}.md-search-result__item:first-child{box-shadow:none}.md-search-result__link{display:block;outline:none;scroll-snap-align:start;transition:background-color .25s}.md-search-result__link:focus,.md-search-result__link:hover{background-color:var(--md-accent-fg-color--transparent)}.md-search-result__link:last-child p:last-child{margin-bottom:.6rem}.md-search-result__more>summary{cursor:pointer;display:block;outline:none;position:sticky;scroll-snap-align:start;top:0;z-index:1}.md-search-result__more>summary::marker{display:none}.md-search-result__more>summary::-webkit-details-marker{display:none}.md-search-result__more>summary>div{color:var(--md-typeset-a-color);font-size:.64rem;padding:.75em .8rem;transition:color .25s,background-color .25s}@media screen and (min-width:60em){[dir=ltr] .md-search-result__more>summary>div{padding-left:2.2rem}[dir=rtl] .md-search-result__more>summary>div{padding-right:2.2rem}}.md-search-result__more>summary:focus>div,.md-search-result__more>summary:hover>div{background-color:var(--md-accent-fg-color--transparent);color:var(--md-accent-fg-color)}.md-search-result__more[open]>summary{background-color:var(--md-default-bg-color)}.md-search-result__article{overflow:hidden;padding:0 .8rem;position:relative}@media screen and (min-width:60em){[dir=ltr] .md-search-result__article{padding-left:2.2rem}[dir=rtl] .md-search-result__article{padding-right:2.2rem}}[dir=ltr] .md-search-result__icon{left:0}[dir=rtl] .md-search-result__icon{right:0}.md-search-result__icon{color:var(--md-default-fg-color--light);height:1.2rem;margin:.5rem;position:absolute;width:1.2rem}@media screen and (max-width:59.984375em){.md-search-result__icon{display:none}}.md-search-result__icon:after{background-color:currentcolor;content:"";display:inline-block;height:100%;-webkit-mask-image:var(--md-search-result-icon);mask-image:var(--md-search-result-icon);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;width:100%}[dir=rtl] .md-search-result__icon:after{transform:scaleX(-1)}.md-search-result .md-typeset{color:var(--md-default-fg-color--light);font-size:.64rem;line-height:1.6}.md-search-result .md-typeset h1{color:var(--md-default-fg-color);font-size:.8rem;font-weight:400;line-height:1.4;margin:.55rem 0}.md-search-result .md-typeset h1 mark{text-decoration:none}.md-search-result .md-typeset h2{color:var(--md-default-fg-color);font-size:.64rem;font-weight:700;line-height:1.6;margin:.5em 0}.md-search-result .md-typeset h2 mark{text-decoration:none}.md-search-result__terms{color:var(--md-default-fg-color);display:block;font-size:.64rem;font-style:italic;margin:.5em 0}.md-search-result mark{background-color:initial;color:var(--md-accent-fg-color);text-decoration:underline}.md-select{position:relative;z-index:1}.md-select__inner{background-color:var(--md-default-bg-color);border-radius:.1rem;box-shadow:var(--md-shadow-z2);color:var(--md-default-fg-color);left:50%;margin-top:.2rem;max-height:0;opacity:0;position:absolute;top:calc(100% - .2rem);transform:translate3d(-50%,.3rem,0);transition:transform .25s 375ms,opacity .25s .25s,max-height 0ms .5s}.md-select:focus-within .md-select__inner,.md-select:hover .md-select__inner{max-height:10rem;opacity:1;transform:translate3d(-50%,0,0);transition:transform .25s cubic-bezier(.1,.7,.1,1),opacity .25s,max-height 0ms}.md-select__inner:after{border-bottom:.2rem solid #0000;border-bottom-color:var(--md-default-bg-color);border-left:.2rem solid #0000;border-right:.2rem solid #0000;border-top:0;content:"";height:0;left:50%;margin-left:-.2rem;margin-top:-.2rem;position:absolute;top:0;width:0}.md-select__list{border-radius:.1rem;font-size:.8rem;list-style-type:none;margin:0;max-height:inherit;overflow:auto;padding:0}.md-select__item{line-height:1.8rem}[dir=ltr] .md-select__link{padding-left:.6rem;padding-right:1.2rem}[dir=rtl] .md-select__link{padding-left:1.2rem;padding-right:.6rem}.md-select__link{cursor:pointer;display:block;outline:none;scroll-snap-align:start;transition:background-color .25s,color .25s;width:100%}.md-select__link:focus,.md-select__link:hover{color:var(--md-accent-fg-color)}.md-select__link:focus{background-color:var(--md-default-fg-color--lightest)}.md-sidebar{align-self:flex-start;flex-shrink:0;padding:1.2rem 0;position:sticky;top:2.4rem;width:12.1rem}@media print{.md-sidebar{display:none}}@media screen and (max-width:76.234375em){[dir=ltr] .md-sidebar--primary{left:-12.1rem}[dir=rtl] .md-sidebar--primary{right:-12.1rem}.md-sidebar--primary{background-color:var(--md-default-bg-color);display:block;height:100%;position:fixed;top:0;transform:translateX(0);transition:transform .25s cubic-bezier(.4,0,.2,1),box-shadow .25s;width:12.1rem;z-index:5}[data-md-toggle=drawer]:checked~.md-container .md-sidebar--primary{box-shadow:var(--md-shadow-z3);transform:translateX(12.1rem)}[dir=rtl] [data-md-toggle=drawer]:checked~.md-container .md-sidebar--primary{transform:translateX(-12.1rem)}.md-sidebar--primary .md-sidebar__scrollwrap{bottom:0;left:0;margin:0;overflow:hidden;position:absolute;right:0;scroll-snap-type:none;top:0}}@media screen and (min-width:76.25em){.md-sidebar{height:0}.no-js .md-sidebar{height:auto}.md-header--lifted~.md-container .md-sidebar{top:4.8rem}}.md-sidebar--secondary{display:none;order:2}@media screen and (min-width:60em){.md-sidebar--secondary{height:0}.no-js .md-sidebar--secondary{height:auto}.md-sidebar--secondary:not([hidden]){display:block}.md-sidebar--secondary .md-sidebar__scrollwrap{touch-action:pan-y}}.md-sidebar__scrollwrap{scrollbar-gutter:stable;-webkit-backface-visibility:hidden;backface-visibility:hidden;margin:0 .2rem;overflow-y:auto;scrollbar-color:var(--md-default-fg-color--lighter) #0000;scrollbar-width:thin}.md-sidebar__scrollwrap::-webkit-scrollbar{height:.2rem;width:.2rem}.md-sidebar__scrollwrap:focus-within,.md-sidebar__scrollwrap:hover{scrollbar-color:var(--md-accent-fg-color) #0000}.md-sidebar__scrollwrap:focus-within::-webkit-scrollbar-thumb,.md-sidebar__scrollwrap:hover::-webkit-scrollbar-thumb{background-color:var(--md-default-fg-color--lighter)}.md-sidebar__scrollwrap:focus-within::-webkit-scrollbar-thumb:hover,.md-sidebar__scrollwrap:hover::-webkit-scrollbar-thumb:hover{background-color:var(--md-accent-fg-color)}@supports selector(::-webkit-scrollbar){.md-sidebar__scrollwrap{scrollbar-gutter:auto}[dir=ltr] .md-sidebar__inner{padding-right:calc(100% - 11.5rem)}[dir=rtl] .md-sidebar__inner{padding-left:calc(100% - 11.5rem)}}@media screen and (max-width:76.234375em){.md-overlay{background-color:#0000008a;height:0;opacity:0;position:fixed;top:0;transition:width 0ms .25s,height 0ms .25s,opacity .25s;width:0;z-index:5}[data-md-toggle=drawer]:checked~.md-overlay{height:100%;opacity:1;transition:width 0ms,height 0ms,opacity .25s;width:100%}}@keyframes facts{0%{height:0}to{height:.65rem}}@keyframes fact{0%{opacity:0;transform:translateY(100%)}50%{opacity:0}to{opacity:1;transform:translateY(0)}}:root{--md-source-forks-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 16 16"><path d="M5 5.372v.878c0 .414.336.75.75.75h4.5a.75.75 0 0 0 .75-.75v-.878a2.25 2.25 0 1 1 1.5 0v.878a2.25 2.25 0 0 1-2.25 2.25h-1.5v2.128a2.251 2.251 0 1 1-1.5 0V8.5h-1.5A2.25 2.25 0 0 1 3.5 6.25v-.878a2.25 2.25 0 1 1 1.5 0ZM5 3.25a.75.75 0 1 0-1.5 0 .75.75 0 0 0 1.5 0Zm6.75.75a.75.75 0 1 0 0-1.5.75.75 0 0 0 0 1.5Zm-3 8.75a.75.75 0 1 0-1.5 0 .75.75 0 0 0 1.5 0Z"/></svg>');--md-source-repositories-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 16 16"><path d="M2 2.5A2.5 2.5 0 0 1 4.5 0h8.75a.75.75 0 0 1 .75.75v12.5a.75.75 0 0 1-.75.75h-2.5a.75.75 0 0 1 0-1.5h1.75v-2h-8a1 1 0 0 0-.714 1.7.75.75 0 1 1-1.072 1.05A2.495 2.495 0 0 1 2 11.5Zm10.5-1h-8a1 1 0 0 0-1 1v6.708A2.486 2.486 0 0 1 4.5 9h8ZM5 12.25a.25.25 0 0 1 .25-.25h3.5a.25.25 0 0 1 .25.25v3.25a.25.25 0 0 1-.4.2l-1.45-1.087a.249.249 0 0 0-.3 0L5.4 15.7a.25.25 0 0 1-.4-.2Z"/></svg>');--md-source-stars-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 16 16"><path d="M8 .25a.75.75 0 0 1 .673.418l1.882 3.815 4.21.612a.75.75 0 0 1 .416 1.279l-3.046 2.97.719 4.192a.751.751 0 0 1-1.088.791L8 12.347l-3.766 1.98a.75.75 0 0 1-1.088-.79l.72-4.194L.818 6.374a.75.75 0 0 1 .416-1.28l4.21-.611L7.327.668A.75.75 0 0 1 8 .25Zm0 2.445L6.615 5.5a.75.75 0 0 1-.564.41l-3.097.45 2.24 2.184a.75.75 0 0 1 .216.664l-.528 3.084 2.769-1.456a.75.75 0 0 1 .698 0l2.77 1.456-.53-3.084a.75.75 0 0 1 .216-.664l2.24-2.183-3.096-.45a.75.75 0 0 1-.564-.41L8 2.694Z"/></svg>');--md-source-version-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 16 16"><path d="M1 7.775V2.75C1 1.784 1.784 1 2.75 1h5.025c.464 0 .91.184 1.238.513l6.25 6.25a1.75 1.75 0 0 1 0 2.474l-5.026 5.026a1.75 1.75 0 0 1-2.474 0l-6.25-6.25A1.752 1.752 0 0 1 1 7.775Zm1.5 0c0 .066.026.13.073.177l6.25 6.25a.25.25 0 0 0 .354 0l5.025-5.025a.25.25 0 0 0 0-.354l-6.25-6.25a.25.25 0 0 0-.177-.073H2.75a.25.25 0 0 0-.25.25ZM6 5a1 1 0 1 1 0 2 1 1 0 0 1 0-2Z"/></svg>')}.md-source{-webkit-backface-visibility:hidden;backface-visibility:hidden;display:block;font-size:.65rem;line-height:1.2;outline-color:var(--md-accent-fg-color);transition:opacity .25s;white-space:nowrap}.md-source:hover{opacity:.7}.md-source__icon{display:inline-block;height:2.4rem;vertical-align:middle;width:2rem}[dir=ltr] .md-source__icon svg{margin-left:.6rem}[dir=rtl] .md-source__icon svg{margin-right:.6rem}.md-source__icon svg{margin-top:.6rem}[dir=ltr] .md-source__icon+.md-source__repository{padding-left:2rem}[dir=rtl] .md-source__icon+.md-source__repository{padding-right:2rem}[dir=ltr] .md-source__icon+.md-source__repository{margin-left:-2rem}[dir=rtl] .md-source__icon+.md-source__repository{margin-right:-2rem}[dir=ltr] .md-source__repository{margin-left:.6rem}[dir=rtl] .md-source__repository{margin-right:.6rem}.md-source__repository{display:inline-block;max-width:calc(100% - 1.2rem);overflow:hidden;text-overflow:ellipsis;vertical-align:middle}.md-source__facts{display:flex;font-size:.55rem;gap:.4rem;list-style-type:none;margin:.1rem 0 0;opacity:.75;overflow:hidden;padding:0;width:100%}.md-source__repository--active .md-source__facts{animation:facts .25s ease-in}.md-source__fact{overflow:hidden;text-overflow:ellipsis}.md-source__repository--active .md-source__fact{animation:fact .4s ease-out}[dir=ltr] .md-source__fact:before{margin-right:.1rem}[dir=rtl] .md-source__fact:before{margin-left:.1rem}.md-source__fact:before{background-color:currentcolor;content:"";display:inline-block;height:.6rem;-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;vertical-align:text-top;width:.6rem}.md-source__fact:nth-child(1n+2){flex-shrink:0}.md-source__fact--version:before{-webkit-mask-image:var(--md-source-version-icon);mask-image:var(--md-source-version-icon)}.md-source__fact--stars:before{-webkit-mask-image:var(--md-source-stars-icon);mask-image:var(--md-source-stars-icon)}.md-source__fact--forks:before{-webkit-mask-image:var(--md-source-forks-icon);mask-image:var(--md-source-forks-icon)}.md-source__fact--repositories:before{-webkit-mask-image:var(--md-source-repositories-icon);mask-image:var(--md-source-repositories-icon)}.md-source-file{margin:1em 0}[dir=ltr] .md-source-file__fact{margin-right:.6rem}[dir=rtl] .md-source-file__fact{margin-left:.6rem}.md-source-file__fact{align-items:center;color:var(--md-default-fg-color--light);display:inline-flex;font-size:.68rem;gap:.3rem}.md-source-file__fact .md-icon{flex-shrink:0;margin-bottom:.05rem}[dir=ltr] .md-source-file__fact .md-author{float:left}[dir=rtl] .md-source-file__fact .md-author{float:right}.md-source-file__fact .md-author{margin-right:.2rem}.md-source-file__fact svg{width:.9rem}:root{--md-status:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M11 9h2V7h-2m1 13c-4.41 0-8-3.59-8-8s3.59-8 8-8 8 3.59 8 8-3.59 8-8 8m0-18A10 10 0 0 0 2 12a10 10 0 0 0 10 10 10 10 0 0 0 10-10A10 10 0 0 0 12 2m-1 15h2v-6h-2v6Z"/></svg>');--md-status--new:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="m23 12-2.44-2.78.34-3.68-3.61-.82-1.89-3.18L12 3 8.6 1.54 6.71 4.72l-3.61.81.34 3.68L1 12l2.44 2.78-.34 3.69 3.61.82 1.89 3.18L12 21l3.4 1.46 1.89-3.18 3.61-.82-.34-3.68L23 12m-10 5h-2v-2h2v2m0-4h-2V7h2v6Z"/></svg>');--md-status--deprecated:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M9 3v1H4v2h1v13a2 2 0 0 0 2 2h10a2 2 0 0 0 2-2V6h1V4h-5V3H9m0 5h2v9H9V8m4 0h2v9h-2V8Z"/></svg>');--md-status--encrypted:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 1 3 5v6c0 5.55 3.84 10.74 9 12 5.16-1.26 9-6.45 9-12V5l-9-4m0 6c1.4 0 2.8 1.1 2.8 2.5V11c.6 0 1.2.6 1.2 1.3v3.5c0 .6-.6 1.2-1.3 1.2H9.2c-.6 0-1.2-.6-1.2-1.3v-3.5c0-.6.6-1.2 1.2-1.2V9.5C9.2 8.1 10.6 7 12 7m0 1.2c-.8 0-1.5.5-1.5 1.3V11h3V9.5c0-.8-.7-1.3-1.5-1.3Z"/></svg>')}.md-status:after{background-color:var(--md-default-fg-color--light);content:"";display:inline-block;height:1.125em;-webkit-mask-image:var(--md-status);mask-image:var(--md-status);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;vertical-align:text-bottom;width:1.125em}.md-status:hover:after{background-color:currentcolor}.md-status--new:after{-webkit-mask-image:var(--md-status--new);mask-image:var(--md-status--new)}.md-status--deprecated:after{-webkit-mask-image:var(--md-status--deprecated);mask-image:var(--md-status--deprecated)}.md-status--encrypted:after{-webkit-mask-image:var(--md-status--encrypted);mask-image:var(--md-status--encrypted)}.md-tabs{background-color:var(--md-primary-fg-color);color:var(--md-primary-bg-color);display:block;line-height:1.3;overflow:auto;width:100%;z-index:3}@media print{.md-tabs{display:none}}@media screen and (max-width:76.234375em){.md-tabs{display:none}}.md-tabs[hidden]{pointer-events:none}[dir=ltr] .md-tabs__list{margin-left:.2rem}[dir=rtl] .md-tabs__list{margin-right:.2rem}.md-tabs__list{contain:content;display:flex;list-style:none;margin:0;overflow:auto;padding:0;scrollbar-width:none;white-space:nowrap}.md-tabs__list::-webkit-scrollbar{display:none}.md-tabs__item{height:2.4rem;padding-left:.6rem;padding-right:.6rem}.md-tabs__item--active .md-tabs__link{color:inherit;opacity:1}.md-tabs__link{-webkit-backface-visibility:hidden;backface-visibility:hidden;display:flex;font-size:.7rem;margin-top:.8rem;opacity:.7;outline-color:var(--md-accent-fg-color);outline-offset:.2rem;transition:transform .4s cubic-bezier(.1,.7,.1,1),opacity .25s}.md-tabs__link:focus,.md-tabs__link:hover{color:inherit;opacity:1}[dir=ltr] .md-tabs__link svg{margin-right:.4rem}[dir=rtl] .md-tabs__link svg{margin-left:.4rem}.md-tabs__link svg{fill:currentcolor;height:1.3em}.md-tabs__item:nth-child(2) .md-tabs__link{transition-delay:20ms}.md-tabs__item:nth-child(3) .md-tabs__link{transition-delay:40ms}.md-tabs__item:nth-child(4) .md-tabs__link{transition-delay:60ms}.md-tabs__item:nth-child(5) .md-tabs__link{transition-delay:80ms}.md-tabs__item:nth-child(6) .md-tabs__link{transition-delay:.1s}.md-tabs__item:nth-child(7) .md-tabs__link{transition-delay:.12s}.md-tabs__item:nth-child(8) .md-tabs__link{transition-delay:.14s}.md-tabs__item:nth-child(9) .md-tabs__link{transition-delay:.16s}.md-tabs__item:nth-child(10) .md-tabs__link{transition-delay:.18s}.md-tabs__item:nth-child(11) .md-tabs__link{transition-delay:.2s}.md-tabs__item:nth-child(12) .md-tabs__link{transition-delay:.22s}.md-tabs__item:nth-child(13) .md-tabs__link{transition-delay:.24s}.md-tabs__item:nth-child(14) .md-tabs__link{transition-delay:.26s}.md-tabs__item:nth-child(15) .md-tabs__link{transition-delay:.28s}.md-tabs__item:nth-child(16) .md-tabs__link{transition-delay:.3s}.md-tabs[hidden] .md-tabs__link{opacity:0;transform:translateY(50%);transition:transform 0ms .1s,opacity .1s}:root{--md-tag-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="m5.41 21 .71-4h-4l.35-2h4l1.06-6h-4l.35-2h4l.71-4h2l-.71 4h6l.71-4h2l-.71 4h4l-.35 2h-4l-1.06 6h4l-.35 2h-4l-.71 4h-2l.71-4h-6l-.71 4h-2M9.53 9l-1.06 6h6l1.06-6h-6Z"/></svg>')}.md-typeset .md-tags:not([hidden]){display:inline-flex;flex-wrap:wrap;gap:.5em;margin-bottom:.75em;margin-top:-.125em}.md-typeset .md-tag{align-items:center;background:var(--md-default-fg-color--lightest);border-radius:2.4rem;display:inline-flex;font-size:.64rem;font-size:min(.8em,.64rem);font-weight:700;gap:.5em;letter-spacing:normal;line-height:1.6;padding:.3125em .78125em}.md-typeset .md-tag[href]{-webkit-tap-highlight-color:transparent;color:inherit;outline:none;transition:color 125ms,background-color 125ms}.md-typeset .md-tag[href]:focus,.md-typeset .md-tag[href]:hover{background-color:var(--md-accent-fg-color);color:var(--md-accent-bg-color)}[id]>.md-typeset .md-tag{vertical-align:text-top}.md-typeset .md-tag-icon:before{background-color:var(--md-default-fg-color--lighter);content:"";display:inline-block;height:1.2em;-webkit-mask-image:var(--md-tag-icon);mask-image:var(--md-tag-icon);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;transition:background-color 125ms;vertical-align:text-bottom;width:1.2em}.md-typeset .md-tag-icon[href]:focus:before,.md-typeset .md-tag-icon[href]:hover:before{background-color:var(--md-accent-bg-color)}@keyframes pulse{0%{transform:scale(.95)}75%{transform:scale(1)}to{transform:scale(.95)}}:root{--md-annotation-bg-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 2A10 10 0 0 0 2 12a10 10 0 0 0 10 10 10 10 0 0 0 10-10A10 10 0 0 0 12 2Z"/></svg>');--md-annotation-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M17 13h-4v4h-2v-4H7v-2h4V7h2v4h4m-5-9A10 10 0 0 0 2 12a10 10 0 0 0 10 10 10 10 0 0 0 10-10A10 10 0 0 0 12 2Z"/></svg>');--md-tooltip-width:20rem}.md-tooltip{-webkit-backface-visibility:hidden;backface-visibility:hidden;background-color:var(--md-default-bg-color);border-radius:.1rem;box-shadow:var(--md-shadow-z2);color:var(--md-default-fg-color);font-family:var(--md-text-font-family);left:clamp(var(--md-tooltip-0,0rem) + .8rem,var(--md-tooltip-x),100vw + var(--md-tooltip-0,0rem) + .8rem - var(--md-tooltip-width) - 2 * .8rem);max-width:calc(100vw - 1.6rem);opacity:0;position:absolute;top:var(--md-tooltip-y);transform:translateY(-.4rem);transition:transform 0ms .25s,opacity .25s,z-index .25s;width:var(--md-tooltip-width);z-index:0}.md-tooltip--active{opacity:1;transform:translateY(0);transition:transform .25s cubic-bezier(.1,.7,.1,1),opacity .25s,z-index 0ms;z-index:2}.md-tooltip--inline{font-weight:700;-webkit-user-select:none;user-select:none;width:auto}.md-tooltip--inline:not(.md-tooltip--active){transform:translateY(.2rem) scale(.9)}.md-tooltip--inline .md-tooltip__inner{font-size:.5rem;padding:.2rem .4rem}[hidden]+.md-tooltip--inline{display:none}.focus-visible>.md-tooltip,.md-tooltip:target{outline:var(--md-accent-fg-color) auto}.md-tooltip__inner{font-size:.64rem;padding:.8rem}.md-tooltip__inner.md-typeset>:first-child{margin-top:0}.md-tooltip__inner.md-typeset>:last-child{margin-bottom:0}.md-annotation{font-weight:400;outline:none;vertical-align:text-bottom;white-space:normal}[dir=rtl] .md-annotation{direction:rtl}code .md-annotation{font-family:var(--md-code-font-family);font-size:inherit}.md-annotation:not([hidden]){display:inline-block;line-height:1.25}.md-annotation__index{border-radius:.01px;cursor:pointer;display:inline-block;margin-left:.4ch;margin-right:.4ch;outline:none;overflow:hidden;position:relative;-webkit-user-select:none;user-select:none;vertical-align:text-top;z-index:0}.md-annotation .md-annotation__index{transition:z-index .25s}@media screen{.md-annotation__index{width:2.2ch}[data-md-visible]>.md-annotation__index{animation:pulse 2s infinite}.md-annotation__index:before{background:var(--md-default-bg-color);-webkit-mask-image:var(--md-annotation-bg-icon);mask-image:var(--md-annotation-bg-icon)}.md-annotation__index:after,.md-annotation__index:before{content:"";height:2.2ch;-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;position:absolute;top:-.1ch;width:2.2ch;z-index:-1}.md-annotation__index:after{background-color:var(--md-default-fg-color--lighter);-webkit-mask-image:var(--md-annotation-icon);mask-image:var(--md-annotation-icon);transform:scale(1.0001);transition:background-color .25s,transform .25s}.md-tooltip--active+.md-annotation__index:after{transform:rotate(45deg)}.md-tooltip--active+.md-annotation__index:after,:hover>.md-annotation__index:after{background-color:var(--md-accent-fg-color)}}.md-tooltip--active+.md-annotation__index{animation-play-state:paused;transition-duration:0ms;z-index:2}.md-annotation__index [data-md-annotation-id]{display:inline-block}@media print{.md-annotation__index [data-md-annotation-id]{background:var(--md-default-fg-color--lighter);border-radius:2ch;color:var(--md-default-bg-color);font-weight:700;padding:0 .6ch;white-space:nowrap}.md-annotation__index [data-md-annotation-id]:after{content:attr(data-md-annotation-id)}}.md-typeset .md-annotation-list{counter-reset:xxx;list-style:none}.md-typeset .md-annotation-list li{position:relative}[dir=ltr] .md-typeset .md-annotation-list li:before{left:-2.125em}[dir=rtl] .md-typeset .md-annotation-list li:before{right:-2.125em}.md-typeset .md-annotation-list li:before{background:var(--md-default-fg-color--lighter);border-radius:2ch;color:var(--md-default-bg-color);content:counter(xxx);counter-increment:xxx;font-size:.8875em;font-weight:700;height:2ch;line-height:1.25;min-width:2ch;padding:0 .6ch;position:absolute;text-align:center;top:.25em}[dir=ltr] .md-top{margin-left:50%}[dir=rtl] .md-top{margin-right:50%}.md-top{background-color:var(--md-default-bg-color);border-radius:1.6rem;box-shadow:var(--md-shadow-z2);color:var(--md-default-fg-color--light);cursor:pointer;display:block;font-size:.7rem;outline:none;padding:.4rem .8rem;position:fixed;top:3.2rem;transform:translate(-50%);transition:color 125ms,background-color 125ms,transform 125ms cubic-bezier(.4,0,.2,1),opacity 125ms;z-index:2}@media print{.md-top{display:none}}[dir=rtl] .md-top{transform:translate(50%)}.md-top[hidden]{opacity:0;pointer-events:none;transform:translate(-50%,.2rem);transition-duration:0ms}[dir=rtl] .md-top[hidden]{transform:translate(50%,.2rem)}.md-top:focus,.md-top:hover{background-color:var(--md-accent-fg-color);color:var(--md-accent-bg-color)}.md-top svg{display:inline-block;vertical-align:-.5em}@keyframes hoverfix{0%{pointer-events:none}}:root{--md-version-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 320 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M137.4 374.6c12.5 12.5 32.8 12.5 45.3 0l128-128c9.2-9.2 11.9-22.9 6.9-34.9S301 191.9 288 191.9L32 192c-12.9 0-24.6 7.8-29.6 19.8s-2.2 25.7 6.9 34.9l128 128z"/></svg>')}.md-version{flex-shrink:0;font-size:.8rem;height:2.4rem}[dir=ltr] .md-version__current{margin-left:1.4rem;margin-right:.4rem}[dir=rtl] .md-version__current{margin-left:.4rem;margin-right:1.4rem}.md-version__current{color:inherit;cursor:pointer;outline:none;position:relative;top:.05rem}[dir=ltr] .md-version__current:after{margin-left:.4rem}[dir=rtl] .md-version__current:after{margin-right:.4rem}.md-version__current:after{background-color:currentcolor;content:"";display:inline-block;height:.6rem;-webkit-mask-image:var(--md-version-icon);mask-image:var(--md-version-icon);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;width:.4rem}.md-version__list{background-color:var(--md-default-bg-color);border-radius:.1rem;box-shadow:var(--md-shadow-z2);color:var(--md-default-fg-color);list-style-type:none;margin:.2rem .8rem;max-height:0;opacity:0;overflow:auto;padding:0;position:absolute;scroll-snap-type:y mandatory;top:.15rem;transition:max-height 0ms .5s,opacity .25s .25s;z-index:3}.md-version:focus-within .md-version__list,.md-version:hover .md-version__list{max-height:10rem;opacity:1;transition:max-height 0ms,opacity .25s}@media (hover:none),(pointer:coarse){.md-version:hover .md-version__list{animation:hoverfix .25s forwards}.md-version:focus-within .md-version__list{animation:none}}.md-version__item{line-height:1.8rem}[dir=ltr] .md-version__link{padding-left:.6rem;padding-right:1.2rem}[dir=rtl] .md-version__link{padding-left:1.2rem;padding-right:.6rem}.md-version__link{cursor:pointer;display:block;outline:none;scroll-snap-align:start;transition:color .25s,background-color .25s;white-space:nowrap;width:100%}.md-version__link:focus,.md-version__link:hover{color:var(--md-accent-fg-color)}.md-version__link:focus{background-color:var(--md-default-fg-color--lightest)}:root{--md-admonition-icon--note:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 2C6.47 2 2 6.47 2 12s4.47 10 10 10 10-4.47 10-10S17.53 2 12 2m3.1 5.07c.14 0 .28.05.4.16l1.27 1.27c.23.22.23.57 0 .78l-1 1-2.05-2.05 1-1c.1-.11.24-.16.38-.16m-1.97 1.74 2.06 2.06-6.06 6.06H7.07v-2.06l6.06-6.06Z"/></svg>');--md-admonition-icon--abstract:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M17 9H7V7h10m0 6H7v-2h10m-3 6H7v-2h7M12 3a1 1 0 0 1 1 1 1 1 0 0 1-1 1 1 1 0 0 1-1-1 1 1 0 0 1 1-1m7 0h-4.18C14.4 1.84 13.3 1 12 1c-1.3 0-2.4.84-2.82 2H5a2 2 0 0 0-2 2v14a2 2 0 0 0 2 2h14a2 2 0 0 0 2-2V5a2 2 0 0 0-2-2Z"/></svg>');--md-admonition-icon--info:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M13 9h-2V7h2m0 10h-2v-6h2m-1-9A10 10 0 0 0 2 12a10 10 0 0 0 10 10 10 10 0 0 0 10-10A10 10 0 0 0 12 2Z"/></svg>');--md-admonition-icon--tip:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M17.66 11.2c-.23-.3-.51-.56-.77-.82-.67-.6-1.43-1.03-2.07-1.66C13.33 7.26 13 4.85 13.95 3c-.95.23-1.78.75-2.49 1.32-2.59 2.08-3.61 5.75-2.39 8.9.04.1.08.2.08.33 0 .22-.15.42-.35.5-.23.1-.47.04-.66-.12a.58.58 0 0 1-.14-.17c-1.13-1.43-1.31-3.48-.55-5.12C5.78 10 4.87 12.3 5 14.47c.06.5.12 1 .29 1.5.14.6.41 1.2.71 1.73 1.08 1.73 2.95 2.97 4.96 3.22 2.14.27 4.43-.12 6.07-1.6 1.83-1.66 2.47-4.32 1.53-6.6l-.13-.26c-.21-.46-.77-1.26-.77-1.26m-3.16 6.3c-.28.24-.74.5-1.1.6-1.12.4-2.24-.16-2.9-.82 1.19-.28 1.9-1.16 2.11-2.05.17-.8-.15-1.46-.28-2.23-.12-.74-.1-1.37.17-2.06.19.38.39.76.63 1.06.77 1 1.98 1.44 2.24 2.8.04.14.06.28.06.43.03.82-.33 1.72-.93 2.27Z"/></svg>');--md-admonition-icon--success:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M21 7 9 19l-5.5-5.5 1.41-1.41L9 16.17 19.59 5.59 21 7Z"/></svg>');--md-admonition-icon--question:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="m15.07 11.25-.9.92C13.45 12.89 13 13.5 13 15h-2v-.5c0-1.11.45-2.11 1.17-2.83l1.24-1.26c.37-.36.59-.86.59-1.41a2 2 0 0 0-2-2 2 2 0 0 0-2 2H8a4 4 0 0 1 4-4 4 4 0 0 1 4 4 3.2 3.2 0 0 1-.93 2.25M13 19h-2v-2h2M12 2A10 10 0 0 0 2 12a10 10 0 0 0 10 10 10 10 0 0 0 10-10c0-5.53-4.5-10-10-10Z"/></svg>');--md-admonition-icon--warning:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M13 14h-2V9h2m0 9h-2v-2h2M1 21h22L12 2 1 21Z"/></svg>');--md-admonition-icon--failure:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M19 6.41 17.59 5 12 10.59 6.41 5 5 6.41 10.59 12 5 17.59 6.41 19 12 13.41 17.59 19 19 17.59 13.41 12 19 6.41Z"/></svg>');--md-admonition-icon--danger:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="m11.5 20 4.86-9.73H13V4l-5 9.73h3.5V20M12 2c2.75 0 5.1 1 7.05 2.95C21 6.9 22 9.25 22 12s-1 5.1-2.95 7.05C17.1 21 14.75 22 12 22s-5.1-1-7.05-2.95C3 17.1 2 14.75 2 12s1-5.1 2.95-7.05C6.9 3 9.25 2 12 2Z"/></svg>');--md-admonition-icon--bug:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M11 13h2v1h-2v-1m10-8v6c0 5.5-3.8 10.7-9 12-5.2-1.3-9-6.5-9-12V5l9-4 9 4m-4 5h-2.2c-.2-.6-.6-1.1-1.1-1.5l1.2-1.2-.7-.7L12.8 8H12c-.2 0-.5 0-.7.1L9.9 6.6l-.8.8 1.2 1.2c-.5.3-.9.8-1.1 1.4H7v1h2v1H7v1h2v1H7v1h2.2c.4 1.2 1.5 2 2.8 2s2.4-.8 2.8-2H17v-1h-2v-1h2v-1h-2v-1h2v-1m-6 2h2v-1h-2v1Z"/></svg>');--md-admonition-icon--example:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M7 2v2h1v14a4 4 0 0 0 4 4 4 4 0 0 0 4-4V4h1V2H7m4 14c-.6 0-1-.4-1-1s.4-1 1-1 1 .4 1 1-.4 1-1 1m2-4c-.6 0-1-.4-1-1s.4-1 1-1 1 .4 1 1-.4 1-1 1m1-5h-4V4h4v3Z"/></svg>');--md-admonition-icon--quote:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M14 17h3l2-4V7h-6v6h3M6 17h3l2-4V7H5v6h3l-2 4Z"/></svg>')}.md-typeset .admonition,.md-typeset details{background-color:var(--md-admonition-bg-color);border:.075rem solid #448aff;border-radius:.2rem;box-shadow:var(--md-shadow-z1);color:var(--md-admonition-fg-color);display:flow-root;font-size:.64rem;margin:1.5625em 0;padding:0 .6rem;page-break-inside:avoid;transition:box-shadow 125ms}@media print{.md-typeset .admonition,.md-typeset details{box-shadow:none}}.md-typeset .admonition:focus-within,.md-typeset details:focus-within{box-shadow:0 0 0 .2rem #448aff1a}.md-typeset .admonition>*,.md-typeset details>*{box-sizing:border-box}.md-typeset .admonition .admonition,.md-typeset .admonition details,.md-typeset details .admonition,.md-typeset details details{margin-bottom:1em;margin-top:1em}.md-typeset .admonition .md-typeset__scrollwrap,.md-typeset details .md-typeset__scrollwrap{margin:1em -.6rem}.md-typeset .admonition .md-typeset__table,.md-typeset details .md-typeset__table{padding:0 .6rem}.md-typeset .admonition>.tabbed-set:only-child,.md-typeset details>.tabbed-set:only-child{margin-top:0}html .md-typeset .admonition>:last-child,html .md-typeset details>:last-child{margin-bottom:.6rem}[dir=ltr] .md-typeset .admonition-title,[dir=ltr] .md-typeset summary{padding-left:2rem;padding-right:.6rem}[dir=rtl] .md-typeset .admonition-title,[dir=rtl] .md-typeset summary{padding-left:.6rem;padding-right:2rem}[dir=ltr] .md-typeset .admonition-title,[dir=ltr] .md-typeset summary{border-left-width:.2rem}[dir=rtl] .md-typeset .admonition-title,[dir=rtl] .md-typeset summary{border-right-width:.2rem}[dir=ltr] .md-typeset .admonition-title,[dir=ltr] .md-typeset summary{border-top-left-radius:.1rem}[dir=ltr] .md-typeset .admonition-title,[dir=ltr] .md-typeset summary,[dir=rtl] .md-typeset .admonition-title,[dir=rtl] .md-typeset summary{border-top-right-radius:.1rem}[dir=rtl] .md-typeset .admonition-title,[dir=rtl] .md-typeset summary{border-top-left-radius:.1rem}.md-typeset .admonition-title,.md-typeset summary{background-color:#448aff1a;border:none;font-weight:700;margin:0 -.6rem;padding-bottom:.4rem;padding-top:.4rem;position:relative}html .md-typeset .admonition-title:last-child,html .md-typeset summary:last-child{margin-bottom:0}[dir=ltr] .md-typeset .admonition-title:before,[dir=ltr] .md-typeset summary:before{left:.6rem}[dir=rtl] .md-typeset .admonition-title:before,[dir=rtl] .md-typeset summary:before{right:.6rem}.md-typeset .admonition-title:before,.md-typeset summary:before{background-color:#448aff;content:"";height:1rem;-webkit-mask-image:var(--md-admonition-icon--note);mask-image:var(--md-admonition-icon--note);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;position:absolute;top:.625em;width:1rem}.md-typeset .admonition-title code,.md-typeset summary code{box-shadow:0 0 0 .05rem var(--md-default-fg-color--lightest)}.md-typeset .admonition.note,.md-typeset details.note{border-color:#448aff}.md-typeset .admonition.note:focus-within,.md-typeset details.note:focus-within{box-shadow:0 0 0 .2rem #448aff1a}.md-typeset .note>.admonition-title,.md-typeset .note>summary{background-color:#448aff1a}.md-typeset .note>.admonition-title:before,.md-typeset .note>summary:before{background-color:#448aff;-webkit-mask-image:var(--md-admonition-icon--note);mask-image:var(--md-admonition-icon--note)}.md-typeset .note>.admonition-title:after,.md-typeset .note>summary:after{color:#448aff}.md-typeset .admonition.abstract,.md-typeset details.abstract{border-color:#00b0ff}.md-typeset .admonition.abstract:focus-within,.md-typeset details.abstract:focus-within{box-shadow:0 0 0 .2rem #00b0ff1a}.md-typeset .abstract>.admonition-title,.md-typeset .abstract>summary{background-color:#00b0ff1a}.md-typeset .abstract>.admonition-title:before,.md-typeset .abstract>summary:before{background-color:#00b0ff;-webkit-mask-image:var(--md-admonition-icon--abstract);mask-image:var(--md-admonition-icon--abstract)}.md-typeset .abstract>.admonition-title:after,.md-typeset .abstract>summary:after{color:#00b0ff}.md-typeset .admonition.info,.md-typeset details.info{border-color:#00b8d4}.md-typeset .admonition.info:focus-within,.md-typeset details.info:focus-within{box-shadow:0 0 0 .2rem #00b8d41a}.md-typeset .info>.admonition-title,.md-typeset .info>summary{background-color:#00b8d41a}.md-typeset .info>.admonition-title:before,.md-typeset .info>summary:before{background-color:#00b8d4;-webkit-mask-image:var(--md-admonition-icon--info);mask-image:var(--md-admonition-icon--info)}.md-typeset .info>.admonition-title:after,.md-typeset .info>summary:after{color:#00b8d4}.md-typeset .admonition.tip,.md-typeset details.tip{border-color:#00bfa5}.md-typeset .admonition.tip:focus-within,.md-typeset details.tip:focus-within{box-shadow:0 0 0 .2rem #00bfa51a}.md-typeset .tip>.admonition-title,.md-typeset .tip>summary{background-color:#00bfa51a}.md-typeset .tip>.admonition-title:before,.md-typeset .tip>summary:before{background-color:#00bfa5;-webkit-mask-image:var(--md-admonition-icon--tip);mask-image:var(--md-admonition-icon--tip)}.md-typeset .tip>.admonition-title:after,.md-typeset .tip>summary:after{color:#00bfa5}.md-typeset .admonition.success,.md-typeset details.success{border-color:#00c853}.md-typeset .admonition.success:focus-within,.md-typeset details.success:focus-within{box-shadow:0 0 0 .2rem #00c8531a}.md-typeset .success>.admonition-title,.md-typeset .success>summary{background-color:#00c8531a}.md-typeset .success>.admonition-title:before,.md-typeset .success>summary:before{background-color:#00c853;-webkit-mask-image:var(--md-admonition-icon--success);mask-image:var(--md-admonition-icon--success)}.md-typeset .success>.admonition-title:after,.md-typeset .success>summary:after{color:#00c853}.md-typeset .admonition.question,.md-typeset details.question{border-color:#64dd17}.md-typeset .admonition.question:focus-within,.md-typeset details.question:focus-within{box-shadow:0 0 0 .2rem #64dd171a}.md-typeset .question>.admonition-title,.md-typeset .question>summary{background-color:#64dd171a}.md-typeset .question>.admonition-title:before,.md-typeset .question>summary:before{background-color:#64dd17;-webkit-mask-image:var(--md-admonition-icon--question);mask-image:var(--md-admonition-icon--question)}.md-typeset .question>.admonition-title:after,.md-typeset .question>summary:after{color:#64dd17}.md-typeset .admonition.warning,.md-typeset details.warning{border-color:#ff9100}.md-typeset .admonition.warning:focus-within,.md-typeset details.warning:focus-within{box-shadow:0 0 0 .2rem #ff91001a}.md-typeset .warning>.admonition-title,.md-typeset .warning>summary{background-color:#ff91001a}.md-typeset .warning>.admonition-title:before,.md-typeset .warning>summary:before{background-color:#ff9100;-webkit-mask-image:var(--md-admonition-icon--warning);mask-image:var(--md-admonition-icon--warning)}.md-typeset .warning>.admonition-title:after,.md-typeset .warning>summary:after{color:#ff9100}.md-typeset .admonition.failure,.md-typeset details.failure{border-color:#ff5252}.md-typeset .admonition.failure:focus-within,.md-typeset details.failure:focus-within{box-shadow:0 0 0 .2rem #ff52521a}.md-typeset .failure>.admonition-title,.md-typeset .failure>summary{background-color:#ff52521a}.md-typeset .failure>.admonition-title:before,.md-typeset .failure>summary:before{background-color:#ff5252;-webkit-mask-image:var(--md-admonition-icon--failure);mask-image:var(--md-admonition-icon--failure)}.md-typeset .failure>.admonition-title:after,.md-typeset .failure>summary:after{color:#ff5252}.md-typeset .admonition.danger,.md-typeset details.danger{border-color:#ff1744}.md-typeset .admonition.danger:focus-within,.md-typeset details.danger:focus-within{box-shadow:0 0 0 .2rem #ff17441a}.md-typeset .danger>.admonition-title,.md-typeset .danger>summary{background-color:#ff17441a}.md-typeset .danger>.admonition-title:before,.md-typeset .danger>summary:before{background-color:#ff1744;-webkit-mask-image:var(--md-admonition-icon--danger);mask-image:var(--md-admonition-icon--danger)}.md-typeset .danger>.admonition-title:after,.md-typeset .danger>summary:after{color:#ff1744}.md-typeset .admonition.bug,.md-typeset details.bug{border-color:#f50057}.md-typeset .admonition.bug:focus-within,.md-typeset details.bug:focus-within{box-shadow:0 0 0 .2rem #f500571a}.md-typeset .bug>.admonition-title,.md-typeset .bug>summary{background-color:#f500571a}.md-typeset .bug>.admonition-title:before,.md-typeset .bug>summary:before{background-color:#f50057;-webkit-mask-image:var(--md-admonition-icon--bug);mask-image:var(--md-admonition-icon--bug)}.md-typeset .bug>.admonition-title:after,.md-typeset .bug>summary:after{color:#f50057}.md-typeset .admonition.example,.md-typeset details.example{border-color:#7c4dff}.md-typeset .admonition.example:focus-within,.md-typeset details.example:focus-within{box-shadow:0 0 0 .2rem #7c4dff1a}.md-typeset .example>.admonition-title,.md-typeset .example>summary{background-color:#7c4dff1a}.md-typeset .example>.admonition-title:before,.md-typeset .example>summary:before{background-color:#7c4dff;-webkit-mask-image:var(--md-admonition-icon--example);mask-image:var(--md-admonition-icon--example)}.md-typeset .example>.admonition-title:after,.md-typeset .example>summary:after{color:#7c4dff}.md-typeset .admonition.quote,.md-typeset details.quote{border-color:#9e9e9e}.md-typeset .admonition.quote:focus-within,.md-typeset details.quote:focus-within{box-shadow:0 0 0 .2rem #9e9e9e1a}.md-typeset .quote>.admonition-title,.md-typeset .quote>summary{background-color:#9e9e9e1a}.md-typeset .quote>.admonition-title:before,.md-typeset .quote>summary:before{background-color:#9e9e9e;-webkit-mask-image:var(--md-admonition-icon--quote);mask-image:var(--md-admonition-icon--quote)}.md-typeset .quote>.admonition-title:after,.md-typeset .quote>summary:after{color:#9e9e9e}:root{--md-footnotes-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M19 7v4H5.83l3.58-3.59L8 6l-6 6 6 6 1.41-1.42L5.83 13H21V7h-2Z"/></svg>')}.md-typeset .footnote{color:var(--md-default-fg-color--light);font-size:.64rem}[dir=ltr] .md-typeset .footnote>ol{margin-left:0}[dir=rtl] .md-typeset .footnote>ol{margin-right:0}.md-typeset .footnote>ol>li{transition:color 125ms}.md-typeset .footnote>ol>li:target{color:var(--md-default-fg-color)}.md-typeset .footnote>ol>li:focus-within .footnote-backref{opacity:1;transform:translateX(0);transition:none}.md-typeset .footnote>ol>li:hover .footnote-backref,.md-typeset .footnote>ol>li:target .footnote-backref{opacity:1;transform:translateX(0)}.md-typeset .footnote>ol>li>:first-child{margin-top:0}.md-typeset .footnote-ref{font-size:.75em;font-weight:700}html .md-typeset .footnote-ref{outline-offset:.1rem}.md-typeset [id^="fnref:"]:target>.footnote-ref{outline:auto}.md-typeset .footnote-backref{color:var(--md-typeset-a-color);display:inline-block;font-size:0;opacity:0;transform:translateX(.25rem);transition:color .25s,transform .25s .25s,opacity 125ms .25s;vertical-align:text-bottom}@media print{.md-typeset .footnote-backref{color:var(--md-typeset-a-color);opacity:1;transform:translateX(0)}}[dir=rtl] .md-typeset .footnote-backref{transform:translateX(-.25rem)}.md-typeset .footnote-backref:hover{color:var(--md-accent-fg-color)}.md-typeset .footnote-backref:before{background-color:currentcolor;content:"";display:inline-block;height:.8rem;-webkit-mask-image:var(--md-footnotes-icon);mask-image:var(--md-footnotes-icon);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;width:.8rem}[dir=rtl] .md-typeset .footnote-backref:before svg{transform:scaleX(-1)}[dir=ltr] .md-typeset .headerlink{margin-left:.5rem}[dir=rtl] .md-typeset .headerlink{margin-right:.5rem}.md-typeset .headerlink{color:var(--md-default-fg-color--lighter);display:inline-block;opacity:0;transition:color .25s,opacity 125ms}@media print{.md-typeset .headerlink{display:none}}.md-typeset .headerlink:focus,.md-typeset :hover>.headerlink,.md-typeset :target>.headerlink{opacity:1;transition:color .25s,opacity 125ms}.md-typeset .headerlink:focus,.md-typeset .headerlink:hover,.md-typeset :target>.headerlink{color:var(--md-accent-fg-color)}.md-typeset :target{--md-scroll-margin:3.6rem;--md-scroll-offset:0rem;scroll-margin-top:calc(var(--md-scroll-margin) - var(--md-scroll-offset))}@media screen and (min-width:76.25em){.md-header--lifted~.md-container .md-typeset :target{--md-scroll-margin:6rem}}.md-typeset h1:target,.md-typeset h2:target,.md-typeset h3:target{--md-scroll-offset:0.2rem}.md-typeset h4:target{--md-scroll-offset:0.15rem}.md-typeset div.arithmatex{overflow:auto}@media screen and (max-width:44.984375em){.md-typeset div.arithmatex{margin:0 -.8rem}.md-typeset div.arithmatex>*{width:min-content}}.md-typeset div.arithmatex>*{margin-left:auto!important;margin-right:auto!important;padding:0 .8rem;touch-action:auto}.md-typeset div.arithmatex>* mjx-container{margin:0!important}.md-typeset div.arithmatex mjx-assistive-mml{height:0}.md-typeset del.critic{background-color:var(--md-typeset-del-color)}.md-typeset del.critic,.md-typeset ins.critic{-webkit-box-decoration-break:clone;box-decoration-break:clone}.md-typeset ins.critic{background-color:var(--md-typeset-ins-color)}.md-typeset .critic.comment{-webkit-box-decoration-break:clone;box-decoration-break:clone;color:var(--md-code-hl-comment-color)}.md-typeset .critic.comment:before{content:"/* "}.md-typeset .critic.comment:after{content:" */"}.md-typeset .critic.block{box-shadow:none;display:block;margin:1em 0;overflow:auto;padding-left:.8rem;padding-right:.8rem}.md-typeset .critic.block>:first-child{margin-top:.5em}.md-typeset .critic.block>:last-child{margin-bottom:.5em}:root{--md-details-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M8.59 16.58 13.17 12 8.59 7.41 10 6l6 6-6 6-1.41-1.42Z"/></svg>')}.md-typeset details{display:flow-root;overflow:visible;padding-top:0}.md-typeset details[open]>summary:after{transform:rotate(90deg)}.md-typeset details:not([open]){box-shadow:none;padding-bottom:0}.md-typeset details:not([open])>summary{border-radius:.1rem}[dir=ltr] .md-typeset summary{padding-right:1.8rem}[dir=rtl] .md-typeset summary{padding-left:1.8rem}[dir=ltr] .md-typeset summary{border-top-left-radius:.1rem}[dir=ltr] .md-typeset summary,[dir=rtl] .md-typeset summary{border-top-right-radius:.1rem}[dir=rtl] .md-typeset summary{border-top-left-radius:.1rem}.md-typeset summary{cursor:pointer;display:block;min-height:1rem;overflow:hidden}.md-typeset summary.focus-visible{outline-color:var(--md-accent-fg-color);outline-offset:.2rem}.md-typeset summary:not(.focus-visible){-webkit-tap-highlight-color:transparent;outline:none}[dir=ltr] .md-typeset summary:after{right:.4rem}[dir=rtl] .md-typeset summary:after{left:.4rem}.md-typeset summary:after{background-color:currentcolor;content:"";height:1rem;-webkit-mask-image:var(--md-details-icon);mask-image:var(--md-details-icon);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;position:absolute;top:.625em;transform:rotate(0deg);transition:transform .25s;width:1rem}[dir=rtl] .md-typeset summary:after{transform:rotate(180deg)}.md-typeset summary::marker{display:none}.md-typeset summary::-webkit-details-marker{display:none}.md-typeset .emojione,.md-typeset .gemoji,.md-typeset .twemoji{--md-icon-size:1.125em;display:inline-flex;height:var(--md-icon-size);vertical-align:text-top}.md-typeset .emojione svg,.md-typeset .gemoji svg,.md-typeset .twemoji svg{fill:currentcolor;max-height:100%;width:var(--md-icon-size)}.md-typeset .lg,.md-typeset .xl,.md-typeset .xxl,.md-typeset .xxxl{vertical-align:text-bottom}.md-typeset .middle{vertical-align:middle}.md-typeset .lg{--md-icon-size:1.5em}.md-typeset .xl{--md-icon-size:2.25em}.md-typeset .xxl{--md-icon-size:3em}.md-typeset .xxxl{--md-icon-size:4em}.highlight .o,.highlight .ow{color:var(--md-code-hl-operator-color)}.highlight .p{color:var(--md-code-hl-punctuation-color)}.highlight .cpf,.highlight .l,.highlight .s,.highlight .s1,.highlight .s2,.highlight .sb,.highlight .sc,.highlight .si,.highlight .ss{color:var(--md-code-hl-string-color)}.highlight .cp,.highlight .se,.highlight .sh,.highlight .sr,.highlight .sx{color:var(--md-code-hl-special-color)}.highlight .il,.highlight .m,.highlight .mb,.highlight .mf,.highlight .mh,.highlight .mi,.highlight .mo{color:var(--md-code-hl-number-color)}.highlight .k,.highlight .kd,.highlight .kn,.highlight .kp,.highlight .kr,.highlight .kt{color:var(--md-code-hl-keyword-color)}.highlight .kc,.highlight .n{color:var(--md-code-hl-name-color)}.highlight .bp,.highlight .nb,.highlight .no{color:var(--md-code-hl-constant-color)}.highlight .nc,.highlight .ne,.highlight .nf,.highlight .nn{color:var(--md-code-hl-function-color)}.highlight .nd,.highlight .ni,.highlight .nl,.highlight .nt{color:var(--md-code-hl-keyword-color)}.highlight .c,.highlight .c1,.highlight .ch,.highlight .cm,.highlight .cs,.highlight .sd{color:var(--md-code-hl-comment-color)}.highlight .na,.highlight .nv,.highlight .vc,.highlight .vg,.highlight .vi{color:var(--md-code-hl-variable-color)}.highlight .ge,.highlight .gh,.highlight .go,.highlight .gp,.highlight .gr,.highlight .gs,.highlight .gt,.highlight .gu{color:var(--md-code-hl-generic-color)}.highlight .gd,.highlight .gi{border-radius:.1rem;margin:0 -.125em;padding:0 .125em}.highlight .gd{background-color:var(--md-typeset-del-color)}.highlight .gi{background-color:var(--md-typeset-ins-color)}.highlight .hll{background-color:var(--md-code-hl-color--light);box-shadow:2px 0 0 0 var(--md-code-hl-color) inset;display:block;margin:0 -1.1764705882em;padding:0 1.1764705882em}.highlight span.filename{background-color:var(--md-code-bg-color);border-bottom:.05rem solid var(--md-default-fg-color--lightest);border-top-left-radius:.1rem;border-top-right-radius:.1rem;display:flow-root;font-size:.85em;font-weight:700;margin-top:1em;padding:.6617647059em 1.1764705882em;position:relative}.highlight span.filename+pre{margin-top:0}.highlight span.filename+pre>code{border-top-left-radius:0;border-top-right-radius:0}.highlight [data-linenos]:before{background-color:var(--md-code-bg-color);box-shadow:-.05rem 0 var(--md-default-fg-color--lightest) inset;color:var(--md-default-fg-color--light);content:attr(data-linenos);float:left;left:-1.1764705882em;margin-left:-1.1764705882em;margin-right:1.1764705882em;padding-left:1.1764705882em;position:sticky;-webkit-user-select:none;user-select:none;z-index:3}.highlight code a[id]{position:absolute;visibility:hidden}.highlight code[data-md-copying] .hll{display:contents}.highlight code[data-md-copying] .md-annotation{display:none}.highlighttable{display:flow-root}.highlighttable tbody,.highlighttable td{display:block;padding:0}.highlighttable tr{display:flex}.highlighttable pre{margin:0}.highlighttable th.filename{flex-grow:1;padding:0;text-align:left}.highlighttable th.filename span.filename{margin-top:0}.highlighttable .linenos{background-color:var(--md-code-bg-color);border-bottom-left-radius:.1rem;border-top-left-radius:.1rem;font-size:.85em;padding:.7720588235em 0 .7720588235em 1.1764705882em;-webkit-user-select:none;user-select:none}.highlighttable .linenodiv{box-shadow:-.05rem 0 var(--md-default-fg-color--lightest) inset;padding-right:.5882352941em}.highlighttable .linenodiv pre{color:var(--md-default-fg-color--light);text-align:right}.highlighttable .code{flex:1;min-width:0}.linenodiv a{color:inherit}.md-typeset .highlighttable{direction:ltr;margin:1em 0}.md-typeset .highlighttable>tbody>tr>.code>div>pre>code{border-bottom-left-radius:0;border-top-left-radius:0}.md-typeset .highlight+.result{border:.05rem solid var(--md-code-bg-color);border-bottom-left-radius:.1rem;border-bottom-right-radius:.1rem;border-top-width:.1rem;margin-top:-1.125em;overflow:visible;padding:0 1em}.md-typeset .highlight+.result:after{clear:both;content:"";display:block}@media screen and (max-width:44.984375em){.md-content__inner>.highlight{margin:1em -.8rem}.md-content__inner>.highlight>.filename,.md-content__inner>.highlight>.highlighttable>tbody>tr>.code>div>pre>code,.md-content__inner>.highlight>.highlighttable>tbody>tr>.filename span.filename,.md-content__inner>.highlight>.highlighttable>tbody>tr>.linenos,.md-content__inner>.highlight>pre>code{border-radius:0}.md-content__inner>.highlight+.result{border-left-width:0;border-radius:0;border-right-width:0;margin-left:-.8rem;margin-right:-.8rem}}.md-typeset .keys kbd:after,.md-typeset .keys kbd:before{-moz-osx-font-smoothing:initial;-webkit-font-smoothing:initial;color:inherit;margin:0;position:relative}.md-typeset .keys span{color:var(--md-default-fg-color--light);padding:0 .2em}.md-typeset .keys .key-alt:before,.md-typeset .keys .key-left-alt:before,.md-typeset .keys .key-right-alt:before{content:"⎇";padding-right:.4em}.md-typeset .keys .key-command:before,.md-typeset .keys .key-left-command:before,.md-typeset .keys .key-right-command:before{content:"⌘";padding-right:.4em}.md-typeset .keys .key-control:before,.md-typeset .keys .key-left-control:before,.md-typeset .keys .key-right-control:before{content:"⌃";padding-right:.4em}.md-typeset .keys .key-left-meta:before,.md-typeset .keys .key-meta:before,.md-typeset .keys .key-right-meta:before{content:"◆";padding-right:.4em}.md-typeset .keys .key-left-option:before,.md-typeset .keys .key-option:before,.md-typeset .keys .key-right-option:before{content:"⌥";padding-right:.4em}.md-typeset .keys .key-left-shift:before,.md-typeset .keys .key-right-shift:before,.md-typeset .keys .key-shift:before{content:"⇧";padding-right:.4em}.md-typeset .keys .key-left-super:before,.md-typeset .keys .key-right-super:before,.md-typeset .keys .key-super:before{content:"❖";padding-right:.4em}.md-typeset .keys .key-left-windows:before,.md-typeset .keys .key-right-windows:before,.md-typeset .keys .key-windows:before{content:"⊞";padding-right:.4em}.md-typeset .keys .key-arrow-down:before{content:"↓";padding-right:.4em}.md-typeset .keys .key-arrow-left:before{content:"←";padding-right:.4em}.md-typeset .keys .key-arrow-right:before{content:"→";padding-right:.4em}.md-typeset .keys .key-arrow-up:before{content:"↑";padding-right:.4em}.md-typeset .keys .key-backspace:before{content:"⌫";padding-right:.4em}.md-typeset .keys .key-backtab:before{content:"⇤";padding-right:.4em}.md-typeset .keys .key-caps-lock:before{content:"⇪";padding-right:.4em}.md-typeset .keys .key-clear:before{content:"⌧";padding-right:.4em}.md-typeset .keys .key-context-menu:before{content:"☰";padding-right:.4em}.md-typeset .keys .key-delete:before{content:"⌦";padding-right:.4em}.md-typeset .keys .key-eject:before{content:"⏏";padding-right:.4em}.md-typeset .keys .key-end:before{content:"⤓";padding-right:.4em}.md-typeset .keys .key-escape:before{content:"⎋";padding-right:.4em}.md-typeset .keys .key-home:before{content:"⤒";padding-right:.4em}.md-typeset .keys .key-insert:before{content:"⎀";padding-right:.4em}.md-typeset .keys .key-page-down:before{content:"⇟";padding-right:.4em}.md-typeset .keys .key-page-up:before{content:"⇞";padding-right:.4em}.md-typeset .keys .key-print-screen:before{content:"⎙";padding-right:.4em}.md-typeset .keys .key-tab:after{content:"⇥";padding-left:.4em}.md-typeset .keys .key-num-enter:after{content:"⌤";padding-left:.4em}.md-typeset .keys .key-enter:after{content:"⏎";padding-left:.4em}:root{--md-tabbed-icon--prev:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M15.41 16.58 10.83 12l4.58-4.59L14 6l-6 6 6 6 1.41-1.42Z"/></svg>');--md-tabbed-icon--next:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M8.59 16.58 13.17 12 8.59 7.41 10 6l6 6-6 6-1.41-1.42Z"/></svg>')}.md-typeset .tabbed-set{border-radius:.1rem;display:flex;flex-flow:column wrap;margin:1em 0;position:relative}.md-typeset .tabbed-set>input{height:0;opacity:0;position:absolute;width:0}.md-typeset .tabbed-set>input:target{--md-scroll-offset:0.625em}.md-typeset .tabbed-set>input.focus-visible~.tabbed-labels:before{background-color:var(--md-accent-fg-color)}.md-typeset .tabbed-labels{-ms-overflow-style:none;box-shadow:0 -.05rem var(--md-default-fg-color--lightest) inset;display:flex;max-width:100%;overflow:auto;scrollbar-width:none}@media print{.md-typeset .tabbed-labels{display:contents}}@media screen{.js .md-typeset .tabbed-labels{position:relative}.js .md-typeset .tabbed-labels:before{background:var(--md-default-fg-color);bottom:0;content:"";display:block;height:2px;left:0;position:absolute;transform:translateX(var(--md-indicator-x));transition:width 225ms,background-color .25s,transform .25s;transition-timing-function:cubic-bezier(.4,0,.2,1);width:var(--md-indicator-width)}}.md-typeset .tabbed-labels::-webkit-scrollbar{display:none}.md-typeset .tabbed-labels>label{border-bottom:.1rem solid #0000;border-radius:.1rem .1rem 0 0;color:var(--md-default-fg-color--light);cursor:pointer;flex-shrink:0;font-size:.64rem;font-weight:700;padding:.78125em 1.25em .625em;scroll-margin-inline-start:1rem;transition:background-color .25s,color .25s;white-space:nowrap;width:auto}@media print{.md-typeset .tabbed-labels>label:first-child{order:1}.md-typeset .tabbed-labels>label:nth-child(2){order:2}.md-typeset .tabbed-labels>label:nth-child(3){order:3}.md-typeset .tabbed-labels>label:nth-child(4){order:4}.md-typeset .tabbed-labels>label:nth-child(5){order:5}.md-typeset .tabbed-labels>label:nth-child(6){order:6}.md-typeset .tabbed-labels>label:nth-child(7){order:7}.md-typeset .tabbed-labels>label:nth-child(8){order:8}.md-typeset .tabbed-labels>label:nth-child(9){order:9}.md-typeset .tabbed-labels>label:nth-child(10){order:10}.md-typeset .tabbed-labels>label:nth-child(11){order:11}.md-typeset .tabbed-labels>label:nth-child(12){order:12}.md-typeset .tabbed-labels>label:nth-child(13){order:13}.md-typeset .tabbed-labels>label:nth-child(14){order:14}.md-typeset .tabbed-labels>label:nth-child(15){order:15}.md-typeset .tabbed-labels>label:nth-child(16){order:16}.md-typeset .tabbed-labels>label:nth-child(17){order:17}.md-typeset .tabbed-labels>label:nth-child(18){order:18}.md-typeset .tabbed-labels>label:nth-child(19){order:19}.md-typeset .tabbed-labels>label:nth-child(20){order:20}}.md-typeset .tabbed-labels>label:hover{color:var(--md-default-fg-color)}.md-typeset .tabbed-labels>label>[href]:first-child{color:inherit}.md-typeset .tabbed-labels--linked>label{padding:0}.md-typeset .tabbed-labels--linked>label>a{display:block;padding:.78125em 1.25em .625em}.md-typeset .tabbed-content{width:100%}@media print{.md-typeset .tabbed-content{display:contents}}.md-typeset .tabbed-block{display:none}@media print{.md-typeset .tabbed-block{display:block}.md-typeset .tabbed-block:first-child{order:1}.md-typeset .tabbed-block:nth-child(2){order:2}.md-typeset .tabbed-block:nth-child(3){order:3}.md-typeset .tabbed-block:nth-child(4){order:4}.md-typeset .tabbed-block:nth-child(5){order:5}.md-typeset .tabbed-block:nth-child(6){order:6}.md-typeset .tabbed-block:nth-child(7){order:7}.md-typeset .tabbed-block:nth-child(8){order:8}.md-typeset .tabbed-block:nth-child(9){order:9}.md-typeset .tabbed-block:nth-child(10){order:10}.md-typeset .tabbed-block:nth-child(11){order:11}.md-typeset .tabbed-block:nth-child(12){order:12}.md-typeset .tabbed-block:nth-child(13){order:13}.md-typeset .tabbed-block:nth-child(14){order:14}.md-typeset .tabbed-block:nth-child(15){order:15}.md-typeset .tabbed-block:nth-child(16){order:16}.md-typeset .tabbed-block:nth-child(17){order:17}.md-typeset .tabbed-block:nth-child(18){order:18}.md-typeset .tabbed-block:nth-child(19){order:19}.md-typeset .tabbed-block:nth-child(20){order:20}}.md-typeset .tabbed-block>.highlight:first-child>pre,.md-typeset .tabbed-block>pre:first-child{margin:0}.md-typeset .tabbed-block>.highlight:first-child>pre>code,.md-typeset .tabbed-block>pre:first-child>code{border-top-left-radius:0;border-top-right-radius:0}.md-typeset .tabbed-block>.highlight:first-child>.filename{border-top-left-radius:0;border-top-right-radius:0;margin:0}.md-typeset .tabbed-block>.highlight:first-child>.highlighttable{margin:0}.md-typeset .tabbed-block>.highlight:first-child>.highlighttable>tbody>tr>.filename span.filename,.md-typeset .tabbed-block>.highlight:first-child>.highlighttable>tbody>tr>.linenos{border-top-left-radius:0;border-top-right-radius:0;margin:0}.md-typeset .tabbed-block>.highlight:first-child>.highlighttable>tbody>tr>.code>div>pre>code{border-top-left-radius:0;border-top-right-radius:0}.md-typeset .tabbed-block>.highlight:first-child+.result{margin-top:-.125em}.md-typeset .tabbed-block>.tabbed-set{margin:0}.md-typeset .tabbed-button{align-self:center;border-radius:100%;color:var(--md-default-fg-color--light);cursor:pointer;display:block;height:.9rem;margin-top:.1rem;pointer-events:auto;transition:background-color .25s;width:.9rem}.md-typeset .tabbed-button:hover{background-color:var(--md-accent-fg-color--transparent);color:var(--md-accent-fg-color)}.md-typeset .tabbed-button:after{background-color:currentcolor;content:"";display:block;height:100%;-webkit-mask-image:var(--md-tabbed-icon--prev);mask-image:var(--md-tabbed-icon--prev);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;transition:background-color .25s,transform .25s;width:100%}.md-typeset .tabbed-control{background:linear-gradient(to right,var(--md-default-bg-color) 60%,#0000);display:flex;height:1.9rem;justify-content:start;pointer-events:none;position:absolute;transition:opacity 125ms;width:1.2rem}[dir=rtl] .md-typeset .tabbed-control{transform:rotate(180deg)}.md-typeset .tabbed-control[hidden]{opacity:0}.md-typeset .tabbed-control--next{background:linear-gradient(to left,var(--md-default-bg-color) 60%,#0000);justify-content:end;right:0}.md-typeset .tabbed-control--next .tabbed-button:after{-webkit-mask-image:var(--md-tabbed-icon--next);mask-image:var(--md-tabbed-icon--next)}@media screen and (max-width:44.984375em){[dir=ltr] .md-content__inner>.tabbed-set .tabbed-labels{padding-left:.8rem}[dir=rtl] .md-content__inner>.tabbed-set .tabbed-labels{padding-right:.8rem}.md-content__inner>.tabbed-set .tabbed-labels{margin:0 -.8rem;max-width:100vw;scroll-padding-inline-start:.8rem}[dir=ltr] .md-content__inner>.tabbed-set .tabbed-labels:after{padding-right:.8rem}[dir=rtl] .md-content__inner>.tabbed-set .tabbed-labels:after{padding-left:.8rem}.md-content__inner>.tabbed-set .tabbed-labels:after{content:""}[dir=ltr] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--prev{padding-left:.8rem}[dir=rtl] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--prev{padding-right:.8rem}[dir=ltr] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--prev{margin-left:-.8rem}[dir=rtl] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--prev{margin-right:-.8rem}.md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--prev{width:2rem}[dir=ltr] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--next{padding-right:.8rem}[dir=rtl] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--next{padding-left:.8rem}[dir=ltr] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--next{margin-right:-.8rem}[dir=rtl] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--next{margin-left:-.8rem}.md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--next{width:2rem}}@media screen{.md-typeset .tabbed-set>input:first-child:checked~.tabbed-labels>:first-child,.md-typeset .tabbed-set>input:nth-child(10):checked~.tabbed-labels>:nth-child(10),.md-typeset .tabbed-set>input:nth-child(11):checked~.tabbed-labels>:nth-child(11),.md-typeset .tabbed-set>input:nth-child(12):checked~.tabbed-labels>:nth-child(12),.md-typeset .tabbed-set>input:nth-child(13):checked~.tabbed-labels>:nth-child(13),.md-typeset .tabbed-set>input:nth-child(14):checked~.tabbed-labels>:nth-child(14),.md-typeset .tabbed-set>input:nth-child(15):checked~.tabbed-labels>:nth-child(15),.md-typeset .tabbed-set>input:nth-child(16):checked~.tabbed-labels>:nth-child(16),.md-typeset .tabbed-set>input:nth-child(17):checked~.tabbed-labels>:nth-child(17),.md-typeset .tabbed-set>input:nth-child(18):checked~.tabbed-labels>:nth-child(18),.md-typeset .tabbed-set>input:nth-child(19):checked~.tabbed-labels>:nth-child(19),.md-typeset .tabbed-set>input:nth-child(2):checked~.tabbed-labels>:nth-child(2),.md-typeset .tabbed-set>input:nth-child(20):checked~.tabbed-labels>:nth-child(20),.md-typeset .tabbed-set>input:nth-child(3):checked~.tabbed-labels>:nth-child(3),.md-typeset .tabbed-set>input:nth-child(4):checked~.tabbed-labels>:nth-child(4),.md-typeset .tabbed-set>input:nth-child(5):checked~.tabbed-labels>:nth-child(5),.md-typeset .tabbed-set>input:nth-child(6):checked~.tabbed-labels>:nth-child(6),.md-typeset .tabbed-set>input:nth-child(7):checked~.tabbed-labels>:nth-child(7),.md-typeset .tabbed-set>input:nth-child(8):checked~.tabbed-labels>:nth-child(8),.md-typeset .tabbed-set>input:nth-child(9):checked~.tabbed-labels>:nth-child(9){color:var(--md-default-fg-color)}.md-typeset .no-js .tabbed-set>input:first-child:checked~.tabbed-labels>:first-child,.md-typeset .no-js .tabbed-set>input:nth-child(10):checked~.tabbed-labels>:nth-child(10),.md-typeset .no-js .tabbed-set>input:nth-child(11):checked~.tabbed-labels>:nth-child(11),.md-typeset .no-js .tabbed-set>input:nth-child(12):checked~.tabbed-labels>:nth-child(12),.md-typeset .no-js .tabbed-set>input:nth-child(13):checked~.tabbed-labels>:nth-child(13),.md-typeset .no-js .tabbed-set>input:nth-child(14):checked~.tabbed-labels>:nth-child(14),.md-typeset .no-js .tabbed-set>input:nth-child(15):checked~.tabbed-labels>:nth-child(15),.md-typeset .no-js .tabbed-set>input:nth-child(16):checked~.tabbed-labels>:nth-child(16),.md-typeset .no-js .tabbed-set>input:nth-child(17):checked~.tabbed-labels>:nth-child(17),.md-typeset .no-js .tabbed-set>input:nth-child(18):checked~.tabbed-labels>:nth-child(18),.md-typeset .no-js .tabbed-set>input:nth-child(19):checked~.tabbed-labels>:nth-child(19),.md-typeset .no-js .tabbed-set>input:nth-child(2):checked~.tabbed-labels>:nth-child(2),.md-typeset .no-js .tabbed-set>input:nth-child(20):checked~.tabbed-labels>:nth-child(20),.md-typeset .no-js .tabbed-set>input:nth-child(3):checked~.tabbed-labels>:nth-child(3),.md-typeset .no-js .tabbed-set>input:nth-child(4):checked~.tabbed-labels>:nth-child(4),.md-typeset .no-js .tabbed-set>input:nth-child(5):checked~.tabbed-labels>:nth-child(5),.md-typeset .no-js .tabbed-set>input:nth-child(6):checked~.tabbed-labels>:nth-child(6),.md-typeset .no-js .tabbed-set>input:nth-child(7):checked~.tabbed-labels>:nth-child(7),.md-typeset .no-js .tabbed-set>input:nth-child(8):checked~.tabbed-labels>:nth-child(8),.md-typeset .no-js .tabbed-set>input:nth-child(9):checked~.tabbed-labels>:nth-child(9),.no-js .md-typeset .tabbed-set>input:first-child:checked~.tabbed-labels>:first-child,.no-js .md-typeset .tabbed-set>input:nth-child(10):checked~.tabbed-labels>:nth-child(10),.no-js .md-typeset .tabbed-set>input:nth-child(11):checked~.tabbed-labels>:nth-child(11),.no-js .md-typeset .tabbed-set>input:nth-child(12):checked~.tabbed-labels>:nth-child(12),.no-js .md-typeset .tabbed-set>input:nth-child(13):checked~.tabbed-labels>:nth-child(13),.no-js .md-typeset .tabbed-set>input:nth-child(14):checked~.tabbed-labels>:nth-child(14),.no-js .md-typeset .tabbed-set>input:nth-child(15):checked~.tabbed-labels>:nth-child(15),.no-js .md-typeset .tabbed-set>input:nth-child(16):checked~.tabbed-labels>:nth-child(16),.no-js .md-typeset .tabbed-set>input:nth-child(17):checked~.tabbed-labels>:nth-child(17),.no-js .md-typeset .tabbed-set>input:nth-child(18):checked~.tabbed-labels>:nth-child(18),.no-js .md-typeset .tabbed-set>input:nth-child(19):checked~.tabbed-labels>:nth-child(19),.no-js .md-typeset .tabbed-set>input:nth-child(2):checked~.tabbed-labels>:nth-child(2),.no-js .md-typeset .tabbed-set>input:nth-child(20):checked~.tabbed-labels>:nth-child(20),.no-js .md-typeset .tabbed-set>input:nth-child(3):checked~.tabbed-labels>:nth-child(3),.no-js .md-typeset .tabbed-set>input:nth-child(4):checked~.tabbed-labels>:nth-child(4),.no-js .md-typeset .tabbed-set>input:nth-child(5):checked~.tabbed-labels>:nth-child(5),.no-js .md-typeset .tabbed-set>input:nth-child(6):checked~.tabbed-labels>:nth-child(6),.no-js .md-typeset .tabbed-set>input:nth-child(7):checked~.tabbed-labels>:nth-child(7),.no-js .md-typeset .tabbed-set>input:nth-child(8):checked~.tabbed-labels>:nth-child(8),.no-js .md-typeset .tabbed-set>input:nth-child(9):checked~.tabbed-labels>:nth-child(9){border-color:var(--md-default-fg-color)}}.md-typeset .tabbed-set>input:first-child.focus-visible~.tabbed-labels>:first-child,.md-typeset .tabbed-set>input:nth-child(10).focus-visible~.tabbed-labels>:nth-child(10),.md-typeset .tabbed-set>input:nth-child(11).focus-visible~.tabbed-labels>:nth-child(11),.md-typeset .tabbed-set>input:nth-child(12).focus-visible~.tabbed-labels>:nth-child(12),.md-typeset .tabbed-set>input:nth-child(13).focus-visible~.tabbed-labels>:nth-child(13),.md-typeset .tabbed-set>input:nth-child(14).focus-visible~.tabbed-labels>:nth-child(14),.md-typeset .tabbed-set>input:nth-child(15).focus-visible~.tabbed-labels>:nth-child(15),.md-typeset .tabbed-set>input:nth-child(16).focus-visible~.tabbed-labels>:nth-child(16),.md-typeset .tabbed-set>input:nth-child(17).focus-visible~.tabbed-labels>:nth-child(17),.md-typeset .tabbed-set>input:nth-child(18).focus-visible~.tabbed-labels>:nth-child(18),.md-typeset .tabbed-set>input:nth-child(19).focus-visible~.tabbed-labels>:nth-child(19),.md-typeset .tabbed-set>input:nth-child(2).focus-visible~.tabbed-labels>:nth-child(2),.md-typeset .tabbed-set>input:nth-child(20).focus-visible~.tabbed-labels>:nth-child(20),.md-typeset .tabbed-set>input:nth-child(3).focus-visible~.tabbed-labels>:nth-child(3),.md-typeset .tabbed-set>input:nth-child(4).focus-visible~.tabbed-labels>:nth-child(4),.md-typeset .tabbed-set>input:nth-child(5).focus-visible~.tabbed-labels>:nth-child(5),.md-typeset .tabbed-set>input:nth-child(6).focus-visible~.tabbed-labels>:nth-child(6),.md-typeset .tabbed-set>input:nth-child(7).focus-visible~.tabbed-labels>:nth-child(7),.md-typeset .tabbed-set>input:nth-child(8).focus-visible~.tabbed-labels>:nth-child(8),.md-typeset .tabbed-set>input:nth-child(9).focus-visible~.tabbed-labels>:nth-child(9){color:var(--md-accent-fg-color)}.md-typeset .tabbed-set>input:first-child:checked~.tabbed-content>:first-child,.md-typeset .tabbed-set>input:nth-child(10):checked~.tabbed-content>:nth-child(10),.md-typeset .tabbed-set>input:nth-child(11):checked~.tabbed-content>:nth-child(11),.md-typeset .tabbed-set>input:nth-child(12):checked~.tabbed-content>:nth-child(12),.md-typeset .tabbed-set>input:nth-child(13):checked~.tabbed-content>:nth-child(13),.md-typeset .tabbed-set>input:nth-child(14):checked~.tabbed-content>:nth-child(14),.md-typeset .tabbed-set>input:nth-child(15):checked~.tabbed-content>:nth-child(15),.md-typeset .tabbed-set>input:nth-child(16):checked~.tabbed-content>:nth-child(16),.md-typeset .tabbed-set>input:nth-child(17):checked~.tabbed-content>:nth-child(17),.md-typeset .tabbed-set>input:nth-child(18):checked~.tabbed-content>:nth-child(18),.md-typeset .tabbed-set>input:nth-child(19):checked~.tabbed-content>:nth-child(19),.md-typeset .tabbed-set>input:nth-child(2):checked~.tabbed-content>:nth-child(2),.md-typeset .tabbed-set>input:nth-child(20):checked~.tabbed-content>:nth-child(20),.md-typeset .tabbed-set>input:nth-child(3):checked~.tabbed-content>:nth-child(3),.md-typeset .tabbed-set>input:nth-child(4):checked~.tabbed-content>:nth-child(4),.md-typeset .tabbed-set>input:nth-child(5):checked~.tabbed-content>:nth-child(5),.md-typeset .tabbed-set>input:nth-child(6):checked~.tabbed-content>:nth-child(6),.md-typeset .tabbed-set>input:nth-child(7):checked~.tabbed-content>:nth-child(7),.md-typeset .tabbed-set>input:nth-child(8):checked~.tabbed-content>:nth-child(8),.md-typeset .tabbed-set>input:nth-child(9):checked~.tabbed-content>:nth-child(9){display:block}:root{--md-tasklist-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M1 12C1 5.925 5.925 1 12 1s11 4.925 11 11-4.925 11-11 11S1 18.075 1 12Zm16.28-2.72a.751.751 0 0 0-.018-1.042.751.751 0 0 0-1.042-.018l-5.97 5.97-2.47-2.47a.751.751 0 0 0-1.042.018.751.751 0 0 0-.018 1.042l3 3a.75.75 0 0 0 1.06 0Z"/></svg>');--md-tasklist-icon--checked:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M1 12C1 5.925 5.925 1 12 1s11 4.925 11 11-4.925 11-11 11S1 18.075 1 12Zm16.28-2.72a.751.751 0 0 0-.018-1.042.751.751 0 0 0-1.042-.018l-5.97 5.97-2.47-2.47a.751.751 0 0 0-1.042.018.751.751 0 0 0-.018 1.042l3 3a.75.75 0 0 0 1.06 0Z"/></svg>')}.md-typeset .task-list-item{list-style-type:none;position:relative}[dir=ltr] .md-typeset .task-list-item [type=checkbox]{left:-2em}[dir=rtl] .md-typeset .task-list-item [type=checkbox]{right:-2em}.md-typeset .task-list-item [type=checkbox]{position:absolute;top:.45em}.md-typeset .task-list-control [type=checkbox]{opacity:0;z-index:-1}[dir=ltr] .md-typeset .task-list-indicator:before{left:-1.5em}[dir=rtl] .md-typeset .task-list-indicator:before{right:-1.5em}.md-typeset .task-list-indicator:before{background-color:var(--md-default-fg-color--lightest);content:"";height:1.25em;-webkit-mask-image:var(--md-tasklist-icon);mask-image:var(--md-tasklist-icon);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;position:absolute;top:.15em;width:1.25em}.md-typeset [type=checkbox]:checked+.task-list-indicator:before{background-color:#00e676;-webkit-mask-image:var(--md-tasklist-icon--checked);mask-image:var(--md-tasklist-icon--checked)}:root>*{--md-mermaid-font-family:var(--md-text-font-family),sans-serif;--md-mermaid-edge-color:var(--md-code-fg-color);--md-mermaid-node-bg-color:var(--md-accent-fg-color--transparent);--md-mermaid-node-fg-color:var(--md-accent-fg-color);--md-mermaid-label-bg-color:var(--md-default-bg-color);--md-mermaid-label-fg-color:var(--md-code-fg-color);--md-mermaid-sequence-actor-bg-color:var(--md-mermaid-label-bg-color);--md-mermaid-sequence-actor-fg-color:var(--md-mermaid-label-fg-color);--md-mermaid-sequence-actor-border-color:var(--md-mermaid-node-fg-color);--md-mermaid-sequence-actor-line-color:var(--md-default-fg-color--lighter);--md-mermaid-sequence-actorman-bg-color:var(--md-mermaid-label-bg-color);--md-mermaid-sequence-actorman-line-color:var(--md-mermaid-node-fg-color);--md-mermaid-sequence-box-bg-color:var(--md-mermaid-node-bg-color);--md-mermaid-sequence-box-fg-color:var(--md-mermaid-edge-color);--md-mermaid-sequence-label-bg-color:var(--md-mermaid-node-bg-color);--md-mermaid-sequence-label-fg-color:var(--md-mermaid-node-fg-color);--md-mermaid-sequence-loop-bg-color:var(--md-mermaid-node-bg-color);--md-mermaid-sequence-loop-fg-color:var(--md-mermaid-edge-color);--md-mermaid-sequence-loop-border-color:var(--md-mermaid-node-fg-color);--md-mermaid-sequence-message-fg-color:var(--md-mermaid-edge-color);--md-mermaid-sequence-message-line-color:var(--md-mermaid-edge-color);--md-mermaid-sequence-note-bg-color:var(--md-mermaid-label-bg-color);--md-mermaid-sequence-note-fg-color:var(--md-mermaid-edge-color);--md-mermaid-sequence-note-border-color:var(--md-mermaid-label-fg-color);--md-mermaid-sequence-number-bg-color:var(--md-mermaid-node-fg-color);--md-mermaid-sequence-number-fg-color:var(--md-accent-bg-color)}.mermaid{line-height:normal;margin:1em 0}.md-typeset .grid{grid-gap:.4rem;display:grid;grid-template-columns:repeat(auto-fit,minmax(min(100%,16rem),1fr));margin:1em 0}.md-typeset .grid.cards>ol,.md-typeset .grid.cards>ul{display:contents}.md-typeset .grid.cards>ol>li,.md-typeset .grid.cards>ul>li,.md-typeset .grid>.card{border:.05rem solid var(--md-default-fg-color--lightest);border-radius:.1rem;display:block;margin:0;padding:.8rem;transition:border .25s,box-shadow .25s}.md-typeset .grid.cards>ol>li:focus-within,.md-typeset .grid.cards>ol>li:hover,.md-typeset .grid.cards>ul>li:focus-within,.md-typeset .grid.cards>ul>li:hover,.md-typeset .grid>.card:focus-within,.md-typeset .grid>.card:hover{border-color:#0000;box-shadow:var(--md-shadow-z2)}.md-typeset .grid.cards>ol>li>hr,.md-typeset .grid.cards>ul>li>hr,.md-typeset .grid>.card>hr{margin-bottom:1em;margin-top:1em}.md-typeset .grid.cards>ol>li>:first-child,.md-typeset .grid.cards>ul>li>:first-child,.md-typeset .grid>.card>:first-child{margin-top:0}.md-typeset .grid.cards>ol>li>:last-child,.md-typeset .grid.cards>ul>li>:last-child,.md-typeset .grid>.card>:last-child{margin-bottom:0}.md-typeset .grid>*,.md-typeset .grid>.admonition,.md-typeset .grid>.highlight>*,.md-typeset .grid>.highlighttable,.md-typeset .grid>.md-typeset details,.md-typeset .grid>details,.md-typeset .grid>pre{margin-bottom:0;margin-top:0}.md-typeset .grid>.highlight>pre:only-child,.md-typeset .grid>.highlight>pre>code,.md-typeset .grid>.highlighttable,.md-typeset .grid>.highlighttable>tbody,.md-typeset .grid>.highlighttable>tbody>tr,.md-typeset .grid>.highlighttable>tbody>tr>.code,.md-typeset .grid>.highlighttable>tbody>tr>.code>.highlight,.md-typeset .grid>.highlighttable>tbody>tr>.code>.highlight>pre,.md-typeset .grid>.highlighttable>tbody>tr>.code>.highlight>pre>code{height:100%}.md-typeset .grid>.tabbed-set{margin-bottom:0;margin-top:0}@media screen and (min-width:45em){[dir=ltr] .md-typeset .inline{float:left}[dir=rtl] .md-typeset .inline{float:right}[dir=ltr] .md-typeset .inline{margin-right:.8rem}[dir=rtl] .md-typeset .inline{margin-left:.8rem}.md-typeset .inline{margin-bottom:.8rem;margin-top:0;width:11.7rem}[dir=ltr] .md-typeset .inline.end{float:right}[dir=rtl] .md-typeset .inline.end{float:left}[dir=ltr] .md-typeset .inline.end{margin-left:.8rem;margin-right:0}[dir=rtl] .md-typeset .inline.end{margin-left:0;margin-right:.8rem}}
\ No newline at end of file
diff --git a/assets/stylesheets/main.bcfcd587.min.css.map b/assets/stylesheets/main.bcfcd587.min.css.map
new file mode 100644
index 0000000000..46829fa5ac
--- /dev/null
+++ b/assets/stylesheets/main.bcfcd587.min.css.map
@@ -0,0 +1 @@
+{"version":3,"sources":["src/templates/assets/stylesheets/main/components/_meta.scss","../../../../src/templates/assets/stylesheets/main.scss","src/templates/assets/stylesheets/main/_resets.scss","src/templates/assets/stylesheets/main/_colors.scss","src/templates/assets/stylesheets/main/_icons.scss","src/templates/assets/stylesheets/main/_typeset.scss","src/templates/assets/stylesheets/utilities/_break.scss","src/templates/assets/stylesheets/main/components/_author.scss","src/templates/assets/stylesheets/main/components/_banner.scss","src/templates/assets/stylesheets/main/components/_base.scss","src/templates/assets/stylesheets/main/components/_clipboard.scss","src/templates/assets/stylesheets/main/components/_consent.scss","src/templates/assets/stylesheets/main/components/_content.scss","src/templates/assets/stylesheets/main/components/_dialog.scss","src/templates/assets/stylesheets/main/components/_feedback.scss","src/templates/assets/stylesheets/main/components/_footer.scss","src/templates/assets/stylesheets/main/components/_form.scss","src/templates/assets/stylesheets/main/components/_header.scss","node_modules/material-design-color/material-color.scss","src/templates/assets/stylesheets/main/components/_nav.scss","src/templates/assets/stylesheets/main/components/_pagination.scss","src/templates/assets/stylesheets/main/components/_post.scss","src/templates/assets/stylesheets/main/components/_progress.scss","src/templates/assets/stylesheets/main/components/_search.scss","src/templates/assets/stylesheets/main/components/_select.scss","src/templates/assets/stylesheets/main/components/_sidebar.scss","src/templates/assets/stylesheets/main/components/_source.scss","src/templates/assets/stylesheets/main/components/_status.scss","src/templates/assets/stylesheets/main/components/_tabs.scss","src/templates/assets/stylesheets/main/components/_tag.scss","src/templates/assets/stylesheets/main/components/_tooltip.scss","src/templates/assets/stylesheets/main/components/_top.scss","src/templates/assets/stylesheets/main/components/_version.scss","src/templates/assets/stylesheets/main/extensions/markdown/_admonition.scss","src/templates/assets/stylesheets/main/extensions/markdown/_footnotes.scss","src/templates/assets/stylesheets/main/extensions/markdown/_toc.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_arithmatex.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_critic.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_details.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_emoji.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_highlight.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_keys.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_tabbed.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_tasklist.scss","src/templates/assets/stylesheets/main/integrations/_mermaid.scss","src/templates/assets/stylesheets/main/modifiers/_grid.scss","src/templates/assets/stylesheets/main/modifiers/_inline.scss"],"names":[],"mappings":"AA0CE,gBCgxCF,CC9xCA,KAEE,6BAAA,CAAA,0BAAA,CAAA,qBAAA,CADA,qBDzBF,CC8BA,iBAGE,kBD3BF,CC8BE,gCANF,iBAOI,yBDzBF,CACF,CC6BA,KACE,QD1BF,CC8BA,qBAIE,uCD3BF,CC+BA,EACE,aAAA,CACA,oBD5BF,CCgCA,GAME,QAAA,CALA,kBAAA,CACA,aAAA,CACA,aAAA,CAEA,gBAAA,CADA,SD3BF,CCiCA,MACE,aD9BF,CCkCA,QAEE,eD/BF,CCmCA,IACE,iBDhCF,CCoCA,MAEE,uBAAA,CADA,gBDhCF,CCqCA,MAEE,eAAA,CACA,kBDlCF,CCsCA,OAKE,gBAAA,CACA,QAAA,CAHA,mBAAA,CACA,iBAAA,CAFA,QAAA,CADA,SD9BF,CCuCA,MACE,QAAA,CACA,YDpCF,CErDA,MAIE,6BAAA,CACA,oCAAA,CACA,mCAAA,CACA,0BAAA,CACA,sCAAA,CAGA,4BAAA,CACA,2CAAA,CACA,yBAAA,CACA,qCFmDF,CE7CA,+BAIE,kBF6CF,CE1CE,oHAEE,YF4CJ,CEnCA,qCAIE,eAAA,CAGA,+BAAA,CACA,sCAAA,CACA,wCAAA,CACA,yCAAA,CACA,0BAAA,CACA,sCAAA,CACA,wCAAA,CACA,yCAAA,CAGA,0BAAA,CACA,0BAAA,CAGA,0BAAA,CACA,mCAAA,CAGA,iCAAA,CACA,kCAAA,CACA,mCAAA,CACA,mCAAA,CACA,kCAAA,CACA,iCAAA,CACA,+CAAA,CACA,6DAAA,CACA,gEAAA,CACA,4DAAA,CACA,4DAAA,CACA,6DAAA,CAGA,6CAAA,CAGA,+CAAA,CAGA,gCAAA,CACA,gCAAA,CAGA,8BAAA,CACA,kCAAA,CACA,qCAAA,CAGA,iCAAA,CAGA,kCAAA,CACA,gDAAA,CAGA,mDAAA,CACA,mDAAA,CAGA,+BAAA,CACA,0BAAA,CAGA,yBAAA,CACA,qCAAA,CACA,uCAAA,CACA,8BAAA,CACA,oCAAA,CAGA,8DAAA,CAKA,8DAAA,CAKA,0DFKF,CG9HE,aAIE,iBAAA,CAHA,aAAA,CAEA,aAAA,CADA,YHmIJ,CIxIA,KACE,kCAAA,CACA,iCAAA,CAGA,uGAAA,CAKA,mFJyIF,CInIA,iBAIE,mCAAA,CACA,6BAAA,CAFA,sCJwIF,CIlIA,aAIE,4BAAA,CADA,sCJsIF,CI7HA,MACE,0NAAA,CACA,mNAAA,CACA,oNJgIF,CIzHA,YAGE,gCAAA,CAAA,kBAAA,CAFA,eAAA,CACA,eJ6HF,CIxHE,aAPF,YAQI,gBJ2HF,CACF,CIxHE,uGAME,iBAAA,CAAA,cJ0HJ,CItHE,eAKE,uCAAA,CAHA,aAAA,CAEA,eAAA,CAHA,iBJ6HJ,CIpHE,8BAPE,eAAA,CAGA,qBJ+HJ,CI3HE,eAEE,kBAAA,CAEA,eAAA,CAHA,oBJ0HJ,CIlHE,eAEE,gBAAA,CACA,eAAA,CAEA,qBAAA,CADA,eAAA,CAHA,mBJwHJ,CIhHE,kBACE,eJkHJ,CI9GE,eAEE,eAAA,CACA,qBAAA,CAFA,YJkHJ,CI5GE,8BAKE,uCAAA,CAFA,cAAA,CACA,eAAA,CAEA,qBAAA,CAJA,eJkHJ,CI1GE,eACE,wBJ4GJ,CIxGE,eAGE,+DAAA,CAFA,iBAAA,CACA,cJ2GJ,CItGE,cACE,+BAAA,CACA,qBJwGJ,CIrGI,mCAEE,sBJsGN,CIlGI,wCACE,+BJoGN,CIjGM,kDACE,uDJmGR,CI9FI,mBACE,kBAAA,CACA,iCJgGN,CI5FI,4BACE,uCAAA,CACA,oBJ8FN,CIzFE,iDAIE,6BAAA,CACA,aAAA,CAFA,2BJ6FJ,CIxFI,aARF,iDASI,oBJ6FJ,CACF,CIzFE,iBAIE,wCAAA,CACA,mBAAA,CACA,kCAAA,CAAA,0BAAA,CAJA,eAAA,CADA,uBAAA,CAEA,qBJ8FJ,CIxFI,qCAEE,uCAAA,CADA,YJ2FN,CIrFE,gBAEE,iBAAA,CACA,eAAA,CAFA,iBJyFJ,CIpFI,qBASE,kCAAA,CAAA,0BAAA,CADA,eAAA,CAPA,aAAA,CAEA,QAAA,CAIA,uCAAA,CAHA,aAAA,CAFA,oCAAA,CASA,yDAAA,CADA,oBAAA,CAJA,iBAAA,CADA,iBJ4FN,CInFM,2BACE,+CJqFR,CIjFM,wCAEE,YAAA,CADA,WJoFR,CI/EM,8CACE,oDJiFR,CI9EQ,oDACE,0CJgFV,CIzEE,gBAOE,4CAAA,CACA,mBAAA,CACA,mKACE,CANF,gCAAA,CAHA,oBAAA,CAEA,eAAA,CADA,uBAAA,CAIA,uBAAA,CADA,qBJ+EJ,CIpEE,iBAGE,6CAAA,CACA,kCAAA,CAAA,0BAAA,CAHA,aAAA,CACA,qBJwEJ,CIlEE,iBAGE,6DAAA,CADA,WAAA,CADA,oBJsEJ,CIhEE,kBACE,WJkEJ,CI9DE,oDAEE,qBJgEJ,CIlEE,oDAEE,sBJgEJ,CI5DE,iCACE,kBJiEJ,CIlEE,iCACE,mBJiEJ,CIlEE,iCAIE,2DJ8DJ,CIlEE,iCAIE,4DJ8DJ,CIlEE,uBAGE,uCAAA,CADA,aAAA,CAAA,cJgEJ,CI1DE,eACE,oBJ4DJ,CIxDE,kDAGE,kBJ0DJ,CI7DE,kDAGE,mBJ0DJ,CI7DE,8BAEE,SJ2DJ,CIvDI,0DACE,iBJ0DN,CItDI,oCACE,2BJyDN,CItDM,0CACE,2BJyDR,CIpDI,wDACE,kBJwDN,CIzDI,wDACE,mBJwDN,CIzDI,oCAEE,kBJuDN,CIpDM,kGAEE,aJwDR,CIpDM,0DACE,eJuDR,CInDM,4HAEE,kBJsDR,CIxDM,4HAEE,mBJsDR,CIxDM,oFACE,kBAAA,CAAA,eJuDR,CIhDE,yBAEE,mBJkDJ,CIpDE,yBAEE,oBJkDJ,CIpDE,eACE,mBAAA,CAAA,cJmDJ,CI9CE,kDAIE,WAAA,CADA,cJiDJ,CIzCI,4BAEE,oBJ2CN,CIvCI,6BAEE,oBJyCN,CIrCI,kCACE,YJuCN,CIlCE,mBACE,iBAAA,CAGA,eAAA,CADA,cAAA,CAEA,iBAAA,CAHA,sBAAA,CAAA,iBJuCJ,CIjCI,uBACE,aAAA,CACA,aJmCN,CI9BE,uBAGE,iBAAA,CADA,eAAA,CADA,eJkCJ,CI5BE,mBACE,cJ8BJ,CI1BE,+BAME,2CAAA,CACA,iDAAA,CACA,mBAAA,CAPA,oBAAA,CAGA,gBAAA,CAFA,cAAA,CACA,aAAA,CAEA,iBJ+BJ,CIzBI,aAXF,+BAYI,aJ4BJ,CACF,CIvBI,iCACE,gBJyBN,CIlBM,8FACE,YJoBR,CIhBM,4FACE,eJkBR,CIbI,8FACE,eJeN,CIZM,kHACE,gBJcR,CITI,kCAGE,eAAA,CAFA,cAAA,CACA,sBAAA,CAEA,kBJWN,CIPI,kCAGE,qDAAA,CAFA,sBAAA,CACA,kBJUN,CILI,wCACE,iCJON,CIJM,8CACE,qDAAA,CACA,sDJMR,CIDI,iCACE,iBJGN,CIEE,wCACE,cJAJ,CIGI,wDAIE,gBJKN,CITI,wDAIE,iBJKN,CITI,8CAME,UAAA,CALA,oBAAA,CAEA,YAAA,CAKA,oDAAA,CAAA,4CAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAHA,iCAAA,CAFA,0BAAA,CAHA,WJON,CIKI,oDACE,oDJHN,CIOI,mEACE,kDAAA,CACA,yDAAA,CAAA,iDJLN,CISI,oEACE,kDAAA,CACA,0DAAA,CAAA,kDJPN,CIYE,wBACE,iBAAA,CACA,eAAA,CACA,iBJVJ,CIcE,mBACE,oBAAA,CAEA,kBAAA,CADA,eJXJ,CIeI,aANF,mBAOI,aJZJ,CACF,CIeI,8BACE,aAAA,CAEA,QAAA,CACA,eAAA,CAFA,UJXN,CKnVI,0CD6WF,uBACE,iBJtBF,CIyBE,4BACE,eJvBJ,CACF,CMlhBE,uBAOE,kBAAA,CALA,aAAA,CACA,aAAA,CAEA,aAAA,CACA,eAAA,CALA,iBAAA,CAOA,sCACE,CALF,YNwhBJ,CM/gBI,2BACE,aNihBN,CM7gBI,6BAME,+CAAA,CAFA,yCAAA,CAHA,eAAA,CACA,eAAA,CACA,kBAAA,CAEA,iBNghBN,CM3gBI,6BAEE,aAAA,CADA,YN8gBN,CMxgBE,wBACE,kBN0gBJ,CMvgBI,4BACE,mCAAA,CACA,uBNygBN,CMrgBI,4DAEE,oBAAA,CADA,SNwgBN,CMpgBM,oEACE,mBNsgBR,CO5jBA,WAGE,0CAAA,CADA,+BAAA,CADA,aPikBF,CO5jBE,aANF,WAOI,YP+jBF,CACF,CO5jBE,oBAEE,2CAAA,CADA,gCP+jBJ,CO1jBE,kBAGE,eAAA,CADA,iBAAA,CADA,eP8jBJ,COxjBE,6BACE,WP6jBJ,CO9jBE,6BACE,UP6jBJ,CO9jBE,mBAEE,aAAA,CACA,cAAA,CACA,uBP0jBJ,COvjBI,0BACE,YPyjBN,COrjBI,yBACE,UPujBN,CQ5lBA,KASE,cAAA,CARA,WAAA,CACA,iBRgmBF,CK5bI,oCGtKJ,KAaI,gBRylBF,CACF,CKjcI,oCGtKJ,KAkBI,cRylBF,CACF,CQplBA,KASE,2CAAA,CAPA,YAAA,CACA,qBAAA,CAKA,eAAA,CAHA,eAAA,CAJA,iBAAA,CAGA,UR0lBF,CQllBE,aAZF,KAaI,aRqlBF,CACF,CKlcI,0CGhJF,yBAII,cRklBJ,CACF,CQzkBA,SAEE,gBAAA,CAAA,iBAAA,CADA,eR6kBF,CQxkBA,cACE,YAAA,CACA,qBAAA,CACA,WR2kBF,CQxkBE,aANF,cAOI,aR2kBF,CACF,CQvkBA,SACE,WR0kBF,CQvkBE,gBACE,YAAA,CACA,WAAA,CACA,iBRykBJ,CQpkBA,aACE,eAAA,CACA,sBRukBF,CQ9jBA,WACE,YRikBF,CQ5jBA,WAGE,QAAA,CACA,SAAA,CAHA,iBAAA,CACA,ORikBF,CQ5jBE,uCACE,aR8jBJ,CQ1jBE,+BAEE,uCAAA,CADA,kBR6jBJ,CQvjBA,SASE,2CAAA,CACA,mBAAA,CAFA,gCAAA,CADA,gBAAA,CADA,YAAA,CAMA,SAAA,CADA,uCAAA,CANA,mBAAA,CAJA,cAAA,CAYA,2BAAA,CATA,URikBF,CQrjBE,eAEE,SAAA,CAIA,uBAAA,CAHA,oEACE,CAHF,UR0jBJ,CQ5iBA,MACE,WR+iBF,CSxsBA,MACE,+PT0sBF,CSpsBA,cASE,mBAAA,CAFA,0CAAA,CACA,cAAA,CAFA,YAAA,CAIA,uCAAA,CACA,oBAAA,CAVA,iBAAA,CAEA,UAAA,CADA,QAAA,CAUA,qBAAA,CAPA,WAAA,CADA,ST+sBF,CSpsBE,aAfF,cAgBI,YTusBF,CACF,CSpsBE,kCAEE,uCAAA,CADA,YTusBJ,CSlsBE,qBACE,uCTosBJ,CShsBE,wCACE,+BTksBJ,CS7rBE,oBAME,6BAAA,CADA,UAAA,CAJA,aAAA,CAEA,cAAA,CACA,aAAA,CAGA,2CAAA,CAAA,mCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CARA,aTusBJ,CS3rBE,sBACE,cT6rBJ,CS1rBI,2BACE,2CT4rBN,CStrBI,kEAEE,uDAAA,CADA,+BTyrBN,CU/vBA,mBACE,GACE,SAAA,CACA,0BVkwBF,CU/vBA,GACE,SAAA,CACA,uBViwBF,CACF,CU7vBA,mBACE,GACE,SV+vBF,CU5vBA,GACE,SV8vBF,CACF,CUnvBE,qBASE,2BAAA,CADA,mCAAA,CAAA,2BAAA,CAFA,0BAAA,CADA,WAAA,CAEA,SAAA,CANA,cAAA,CACA,KAAA,CAEA,UAAA,CADA,SV2vBJ,CUjvBE,mBAcE,mDAAA,CANA,2CAAA,CACA,QAAA,CACA,mBAAA,CARA,QAAA,CASA,kDACE,CAPF,eAAA,CAEA,aAAA,CADA,SAAA,CALA,cAAA,CAGA,UAAA,CADA,SV4vBJ,CU7uBE,kBACE,aV+uBJ,CU3uBE,sBACE,YAAA,CACA,YV6uBJ,CU1uBI,oCACE,aV4uBN,CUvuBE,sBACE,mBVyuBJ,CUtuBI,6CACE,cVwuBN,CKloBI,0CKvGA,6CAKI,aAAA,CAEA,gBAAA,CACA,iBAAA,CAFA,UV0uBN,CACF,CUnuBE,kBACE,cVquBJ,CWt0BA,YACE,WAAA,CAIA,WXs0BF,CWn0BE,mBAEE,qBAAA,CADA,iBXs0BJ,CKzqBI,sCMtJE,4EACE,kBXk0BN,CW9zBI,0JACE,mBXg0BN,CWj0BI,8EACE,kBXg0BN,CACF,CW3zBI,0BAGE,UAAA,CAFA,aAAA,CACA,YX8zBN,CWzzBI,+BACE,eX2zBN,CWrzBE,8BACE,WX0zBJ,CW3zBE,8BACE,UX0zBJ,CW3zBE,8BAIE,iBXuzBJ,CW3zBE,8BAIE,kBXuzBJ,CW3zBE,oBAGE,cAAA,CADA,SXyzBJ,CWpzBI,aAPF,oBAQI,YXuzBJ,CACF,CWpzBI,gCACE,yCXszBN,CWlzBI,wBACE,cAAA,CACA,kBXozBN,CWjzBM,kCACE,oBXmzBR,CYp3BA,qBAeE,WZq3BF,CYp4BA,qBAeE,UZq3BF,CYp4BA,WAOE,2CAAA,CACA,mBAAA,CANA,YAAA,CAOA,8BAAA,CALA,iBAAA,CAMA,SAAA,CALA,mBAAA,CACA,mBAAA,CALA,cAAA,CAaA,0BAAA,CAHA,wCACE,CATF,SZi4BF,CYl3BE,aAlBF,WAmBI,YZq3BF,CACF,CYl3BE,mBAEE,SAAA,CADA,mBAAA,CAKA,uBAAA,CAHA,kEZq3BJ,CY92BE,kBAEE,gCAAA,CADA,eZi3BJ,Can5BA,aACE,gBAAA,CACA,iBbs5BF,Can5BE,sBAGE,WAAA,CADA,QAAA,CADA,Sbu5BJ,Caj5BE,oBAEE,eAAA,CADA,ebo5BJ,Ca/4BE,oBACE,iBbi5BJ,Ca74BE,mBAEE,YAAA,CACA,cAAA,CACA,6BAAA,CAHA,iBbk5BJ,Ca54BI,iDACE,yCb84BN,Ca14BI,6BACE,iBb44BN,Cav4BE,mBAGE,uCAAA,CACA,cAAA,CAHA,aAAA,CACA,cAAA,CAGA,sBby4BJ,Cat4BI,gDACE,+Bbw4BN,Cap4BI,4BACE,0CAAA,CACA,mBbs4BN,Caj4BE,mBAEE,SAAA,CADA,iBAAA,CAKA,2BAAA,CAHA,8Dbo4BJ,Ca93BI,qBAEE,aAAA,CADA,ebi4BN,Ca53BI,6BACE,SAAA,CACA,uBb83BN,Cc58BA,WAEE,0CAAA,CADA,+Bdg9BF,Cc58BE,aALF,WAMI,Yd+8BF,CACF,Cc58BE,kBACE,6BAAA,CAEA,aAAA,CADA,ad+8BJ,Cc38BI,gCACE,Yd68BN,Ccx8BE,iBAOE,eAAA,CANA,YAAA,CAKA,cAAA,CAGA,mBAAA,CAAA,eAAA,CADA,cAAA,CAGA,uCAAA,CADA,eAAA,CAEA,uBds8BJ,Ccn8BI,8CACE,Udq8BN,Ccj8BI,+BACE,oBdm8BN,CKrzBI,0CSvIE,uBACE,ad+7BN,Cc57BM,yCACE,Yd87BR,CACF,Ccz7BI,iCACE,gBd47BN,Cc77BI,iCACE,iBd47BN,Cc77BI,uBAEE,gBd27BN,Ccx7BM,iCACE,ed07BR,Ccp7BE,kBACE,WAAA,CAIA,eAAA,CADA,mBAAA,CAFA,6BAAA,CACA,cAAA,CAGA,kBds7BJ,Ccl7BE,mBAEE,YAAA,CADA,adq7BJ,Cch7BE,sBACE,gBAAA,CACA,Udk7BJ,Cc76BA,gBACE,gDdg7BF,Cc76BE,uBACE,YAAA,CACA,cAAA,CACA,6BAAA,CACA,ad+6BJ,Cc36BE,kCACE,sCd66BJ,Cc16BI,gFACE,+Bd46BN,Ccp6BA,cAKE,wCAAA,CADA,gBAAA,CADA,iBAAA,CADA,eAAA,CADA,Ud26BF,CK/3BI,mCS7CJ,cASI,Udu6BF,CACF,Ccn6BE,yBACE,sCdq6BJ,Cc95BA,WACE,mBAAA,CACA,SAAA,CAEA,cAAA,CADA,qBdk6BF,CK94BI,mCSvBJ,WAQI,edi6BF,CACF,Cc95BE,iBACE,oBAAA,CAEA,aAAA,CACA,iBAAA,CAFA,Ydk6BJ,Cc75BI,wBACE,ed+5BN,Cc35BI,qBAGE,iBAAA,CAFA,gBAAA,CACA,mBd85BN,CepkCE,uBAME,kBAAA,CACA,mBAAA,CAHA,gCAAA,CACA,cAAA,CAJA,oBAAA,CAEA,eAAA,CADA,kBAAA,CAMA,gEfukCJ,CejkCI,gCAEE,2CAAA,CACA,uCAAA,CAFA,gCfqkCN,Ce/jCI,0DAEE,0CAAA,CACA,sCAAA,CAFA,+BfmkCN,Ce5jCE,gCAKE,4BfikCJ,CetkCE,gEAME,6BfgkCJ,CetkCE,gCAME,4BfgkCJ,CetkCE,sBAIE,6DAAA,CAGA,8BAAA,CAJA,eAAA,CAFA,aAAA,CACA,eAAA,CAMA,sCf8jCJ,CezjCI,wDACE,6CAAA,CACA,8Bf2jCN,CevjCI,+BACE,UfyjCN,CgB5mCA,WAOE,2CAAA,CAGA,8CACE,CALF,gCAAA,CADA,aAAA,CAHA,MAAA,CADA,eAAA,CACA,OAAA,CACA,KAAA,CACA,ShBmnCF,CgBxmCE,aAfF,WAgBI,YhB2mCF,CACF,CgBxmCE,mBAIE,2BAAA,CAHA,iEhB2mCJ,CgBpmCE,mBACE,kDACE,CAEF,kEhBomCJ,CgB9lCE,kBAEE,kBAAA,CADA,YAAA,CAEA,ehBgmCJ,CgB5lCE,mBAKE,kBAAA,CAEA,cAAA,CAHA,YAAA,CAIA,uCAAA,CALA,aAAA,CAFA,iBAAA,CAQA,uBAAA,CAHA,qBAAA,CAJA,ShBqmCJ,CgB3lCI,yBACE,UhB6lCN,CgBzlCI,iCACE,oBhB2lCN,CgBvlCI,uCAEE,uCAAA,CADA,YhB0lCN,CgBrlCI,2BAEE,YAAA,CADA,ahBwlCN,CK1+BI,0CW/GA,2BAMI,YhBulCN,CACF,CgBplCM,8DAIE,iBAAA,CAHA,aAAA,CAEA,aAAA,CADA,UhBwlCR,CKxgCI,mCWzEA,iCAII,YhBilCN,CACF,CgB9kCM,wCACE,YhBglCR,CgB5kCM,+CACE,oBhB8kCR,CKnhCI,sCWtDA,iCAII,YhBykCN,CACF,CgBpkCE,kBAEE,YAAA,CACA,cAAA,CAFA,iBAAA,CAIA,8DACE,CAFF,kBhBukCJ,CgBjkCI,oCAGE,SAAA,CADA,mBAAA,CAKA,6BAAA,CAHA,8DACE,CAJF,UhBukCN,CgB9jCM,8CACE,8BhBgkCR,CgB3jCI,8BACE,ehB6jCN,CgBxjCE,4BAGE,gBAAA,CAAA,kBhB4jCJ,CgB/jCE,4BAGE,iBAAA,CAAA,iBhB4jCJ,CgB/jCE,kBACE,WAAA,CAGA,eAAA,CAFA,aAAA,CAGA,kBhB0jCJ,CgBvjCI,4CAGE,SAAA,CADA,mBAAA,CAKA,8BAAA,CAHA,8DACE,CAJF,UhB6jCN,CgBpjCM,sDACE,6BhBsjCR,CgBljCM,8DAGE,SAAA,CADA,mBAAA,CAKA,uBAAA,CAHA,8DACE,CAJF,ShBwjCR,CgB7iCI,uCAGE,WAAA,CAFA,iBAAA,CACA,UhBgjCN,CgB1iCE,mBACE,YAAA,CACA,aAAA,CACA,cAAA,CAEA,+CACE,CAFF,kBhB6iCJ,CgBviCI,8DACE,WAAA,CACA,SAAA,CACA,oChByiCN,CgBhiCI,yBACE,QhBkiCN,CgB7hCE,mBACE,YhB+hCJ,CK3lCI,mCW2DF,6BAQI,gBhB+hCJ,CgBviCA,6BAQI,iBhB+hCJ,CgBviCA,mBAKI,aAAA,CAEA,iBAAA,CADA,ahBiiCJ,CACF,CKnmCI,sCW2DF,6BAaI,kBhB+hCJ,CgB5iCA,6BAaI,mBhB+hCJ,CACF,CD9wCA,SAGE,uCAAA,CAFA,eAAA,CACA,eCkxCF,CD9wCE,eACE,mBAAA,CACA,cAAA,CAGA,eAAA,CADA,QAAA,CADA,SCkxCJ,CD5wCE,sCAEE,WAAA,CADA,iBAAA,CAAA,kBC+wCJ,CD1wCE,eACE,+BC4wCJ,CDzwCI,0CACE,+BC2wCN,CDrwCA,UAKE,wBkBaa,ClBZb,oBAAA,CAFA,UAAA,CAHA,oBAAA,CAEA,eAAA,CADA,0BAAA,CAAA,2BC4wCF,CkB9yCA,MACE,0MAAA,CACA,gMAAA,CACA,yNlBizCF,CkB3yCA,QACE,eAAA,CACA,elB8yCF,CkB3yCE,eAKE,uCAAA,CAJA,aAAA,CAGA,eAAA,CADA,eAAA,CADA,eAAA,CAIA,sBlB6yCJ,CkB1yCI,+BACE,YlB4yCN,CkBzyCM,mCAEE,WAAA,CADA,UlB4yCR,CkBpyCQ,sFAME,iBAAA,CALA,aAAA,CAGA,aAAA,CADA,cAAA,CAEA,kBAAA,CAHA,UlB0yCV,CkB/xCE,cAGE,eAAA,CADA,QAAA,CADA,SlBmyCJ,CkB7xCE,cAGE,sBAAA,CAFA,YAAA,CACA,SAAA,CAEA,iBAAA,CAEA,uBAAA,CADA,sBlBgyCJ,CkB5xCI,sBACE,uClB8xCN,CkBvxCM,6EAEE,+BlByxCR,CkBpxCI,2BAIE,iBlBmxCN,CkB/wCI,4CACE,gBlBixCN,CkBlxCI,4CACE,iBlBixCN,CkB7wCI,kBAGE,iBAAA,CAFA,aAAA,CACA,YlBgxCN,CkB3wCI,sGACE,+BAAA,CACA,clB6wCN,CkBzwCI,4BACE,uCAAA,CACA,oBlB2wCN,CkBvwCI,0CACE,YlBywCN,CkBtwCM,yDAKE,6BAAA,CAJA,aAAA,CAEA,WAAA,CACA,qCAAA,CAAA,6BAAA,CAFA,UlB2wCR,CkBpwCM,kDACE,YlBswCR,CkBhwCE,iCACE,YlBkwCJ,CkB/vCI,6CACE,WAAA,CAGA,WlB+vCN,CkB1vCE,cACE,alB4vCJ,CkBxvCE,gBACE,YlB0vCJ,CKxtCI,0Ca3BA,0CASE,2CAAA,CAHA,YAAA,CACA,qBAAA,CACA,WAAA,CALA,MAAA,CADA,iBAAA,CACA,OAAA,CACA,KAAA,CACA,SlByvCJ,CkB9uCI,+DACE,eAAA,CACA,elBgvCN,CkB5uCI,gCAQE,qDAAA,CAHA,uCAAA,CAEA,cAAA,CALA,aAAA,CAEA,kBAAA,CADA,wBAAA,CAFA,iBAAA,CAKA,kBlBgvCN,CkB3uCM,wDAGE,UlBivCR,CkBpvCM,wDAGE,WlBivCR,CkBpvCM,8CAIE,aAAA,CAEA,aAAA,CACA,YAAA,CANA,iBAAA,CACA,SAAA,CAGA,YlB+uCR,CkB1uCQ,oDAKE,6BAAA,CADA,UAAA,CAHA,aAAA,CAEA,WAAA,CAGA,2CAAA,CAAA,mCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAPA,UlBmvCV,CkBvuCM,8CAGE,2CAAA,CACA,gEACE,CAJF,eAAA,CAKA,4BAAA,CAJA,kBlB4uCR,CkBruCQ,2DACE,YlBuuCV,CkBluCM,8CAGE,2CAAA,CADA,gCAAA,CADA,elBsuCR,CkBhuCM,yCAIE,aAAA,CAFA,UAAA,CAIA,YAAA,CADA,aAAA,CAJA,iBAAA,CACA,WAAA,CACA,SlBquCR,CkB7tCI,+BACE,MlB+tCN,CkB3tCI,+BACE,4DlB6tCN,CkB1tCM,qDACE,+BlB4tCR,CkBztCQ,sHACE,+BlB2tCV,CkBrtCI,+BAEE,YAAA,CADA,mBlBwtCN,CkBptCM,mCACE,elBstCR,CkBltCM,6CACE,SlBotCR,CkBhtCM,uDAGE,mBlBmtCR,CkBttCM,uDAGE,kBlBmtCR,CkBttCM,6CAIE,gBAAA,CAFA,aAAA,CADA,YlBqtCR,CkB/sCQ,mDAKE,6BAAA,CADA,UAAA,CAHA,aAAA,CAEA,WAAA,CAGA,2CAAA,CAAA,mCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAPA,UlBwtCV,CkBxsCM,+CACE,mBlB0sCR,CkBlsCM,4CAEE,wBAAA,CADA,elBqsCR,CkBjsCQ,oEACE,mBlBmsCV,CkBpsCQ,oEACE,oBlBmsCV,CkB/rCQ,4EACE,iBlBisCV,CkBlsCQ,4EACE,kBlBisCV,CkB7rCQ,oFACE,mBlB+rCV,CkBhsCQ,oFACE,oBlB+rCV,CkB3rCQ,4FACE,mBlB6rCV,CkB9rCQ,4FACE,oBlB6rCV,CkBtrCE,mBACE,wBlBwrCJ,CkBprCE,wBACE,YAAA,CACA,SAAA,CAIA,0BAAA,CAHA,oElBurCJ,CkBjrCI,kCACE,2BlBmrCN,CkB9qCE,gCACE,SAAA,CAIA,uBAAA,CAHA,qElBirCJ,CkB3qCI,8CAEE,kCAAA,CAAA,0BlB4qCN,CACF,CK32CI,0CauMA,0CACE,YlBuqCJ,CkBpqCI,yDACE,UlBsqCN,CkBlqCI,wDACE,YlBoqCN,CkBhqCI,kDACE,YlBkqCN,CkB7pCE,gBAIE,iDAAA,CADA,gCAAA,CAFA,aAAA,CACA,elBiqCJ,CACF,CKx6CM,+DagRF,6CACE,YlB2pCJ,CkBxpCI,4DACE,UlB0pCN,CkBtpCI,2DACE,YlBwpCN,CkBppCI,qDACE,YlBspCN,CACF,CKh6CI,mCa7JJ,QA6aI,oBlBopCF,CkB9oCI,kCAME,qCAAA,CACA,qDAAA,CANA,eAAA,CACA,KAAA,CAGA,SlBgpCN,CkB3oCM,6CACE,uBlB6oCR,CkBzoCM,gDACE,YlB2oCR,CkBtoCI,2CACE,kBlByoCN,CkB1oCI,2CACE,mBlByoCN,CkB1oCI,iCAEE,oBlBwoCN,CkBjoCI,yDACE,kBlBmoCN,CkBpoCI,yDACE,iBlBmoCN,CACF,CKz7CI,sCa7JJ,QAydI,oBAAA,CACA,oDlBioCF,CkB3nCI,gCAME,qCAAA,CACA,qDAAA,CANA,eAAA,CACA,KAAA,CAGA,SlB6nCN,CkBxnCM,8CACE,uBlB0nCR,CkBtnCM,8CACE,YlBwnCR,CkBnnCI,yCACE,kBlBsnCN,CkBvnCI,yCACE,mBlBsnCN,CkBvnCI,+BAEE,oBlBqnCN,CkB9mCI,uDACE,kBlBgnCN,CkBjnCI,uDACE,iBlBgnCN,CkB3mCE,wBACE,YAAA,CACA,sBAAA,CAEA,SAAA,CACA,6FACE,CAHF,mBlB+mCJ,CkBvmCI,sCACE,elBymCN,CkBpmCE,iFACE,sBAAA,CAEA,SAAA,CACA,4FACE,CAHF,kBlBwmCJ,CkB/lCE,iDACE,elBimCJ,CkB7lCE,6CACE,YlB+lCJ,CkB3lCE,uBACE,aAAA,CACA,elB6lCJ,CkB1lCI,kCACE,elB4lCN,CkBxlCI,qCACE,elB0lCN,CkBvlCM,0CACE,uClBylCR,CkBrlCM,6DACE,mBlBulCR,CkBnlCM,yFAEE,YlBqlCR,CkBhlCI,yCAEE,kBlBolCN,CkBtlCI,yCAEE,mBlBolCN,CkBtlCI,+BACE,aAAA,CAGA,SAAA,CADA,kBlBmlCN,CkB/kCM,2DACE,SlBilCR,CkB3kCE,cAGE,kBAAA,CADA,YAAA,CAEA,gCAAA,CAHA,WlBglCJ,CkB1kCI,oBACE,uDlB4kCN,CkBxkCI,oBAME,6BAAA,CACA,kBAAA,CAFA,UAAA,CAJA,oBAAA,CAEA,WAAA,CAMA,2CAAA,CAAA,mCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAJA,yBAAA,CAJA,qBAAA,CAFA,UlBolCN,CkBvkCM,8BACE,wBlBykCR,CkBrkCM,kKAEE,uBlBskCR,CkBxjCI,2EACE,YlB6jCN,CkB1jCM,oDACE,alB4jCR,CkBzjCQ,kEAKE,qCAAA,CACA,qDAAA,CAFA,YAAA,CAHA,eAAA,CACA,KAAA,CACA,SlB8jCV,CkBxjCU,0FACE,mBlB0jCZ,CkBrjCQ,0EACE,QlBujCV,CkBljCM,sFACE,kBlBojCR,CkBrjCM,sFACE,mBlBojCR,CkBhjCM,kDACE,uClBkjCR,CkB5iCI,2CACE,sBAAA,CAEA,SAAA,CADA,kBlB+iCN,CkBtiCI,qFAIE,mDlByiCN,CkB7iCI,qFAIE,oDlByiCN,CkB7iCI,2EACE,aAAA,CACA,oBAAA,CAGA,SAAA,CAFA,kBlB0iCN,CkBriCM,yFAEE,gBAAA,CADA,gBlBwiCR,CkBniCM,0FACE,YlBqiCR,CACF,CmBzvDA,eAKE,eAAA,CACA,eAAA,CAJA,SnBgwDF,CmBzvDE,gCANA,kBAAA,CAFA,YAAA,CAGA,sBnBuwDF,CmBlwDE,iBAOE,mBAAA,CAFA,aAAA,CADA,gBAAA,CAEA,iBnB4vDJ,CmBvvDE,wBAEE,qDAAA,CADA,uCnB0vDJ,CmBrvDE,qBACE,6CnBuvDJ,CmBlvDI,sDAEE,uDAAA,CADA,+BnBqvDN,CmBjvDM,8DACE,+BnBmvDR,CmB9uDI,mCACE,uCAAA,CACA,oBnBgvDN,CmB5uDI,yBAKE,iBAAA,CADA,yCAAA,CAHA,aAAA,CAEA,eAAA,CADA,YnBivDN,CoBjyDE,eAGE,+DAAA,CADA,oBAAA,CADA,qBpBsyDJ,CKjnDI,0CetLF,eAOI,YpBoyDJ,CACF,CoB9xDM,6BACE,oBpBgyDR,CoB1xDE,kBACE,YAAA,CACA,qBAAA,CACA,SAAA,CACA,qBpB4xDJ,CoBrxDI,0BACE,sBpBuxDN,CoBpxDM,gEACE,+BpBsxDR,CoBhxDE,gBAEE,uCAAA,CADA,epBmxDJ,CoB9wDE,kBACE,oBpBgxDJ,CoB7wDI,mCAGE,kBAAA,CAFA,YAAA,CACA,SAAA,CAEA,iBpB+wDN,CoB3wDI,oCAIE,kBAAA,CAHA,mBAAA,CACA,kBAAA,CACA,SAAA,CAGA,QAAA,CADA,iBpB8wDN,CoBzwDI,0DACE,kBpB2wDN,CoB5wDI,0DACE,iBpB2wDN,CoBvwDI,iDACE,uBAAA,CAEA,YpBwwDN,CoBnwDE,4BACE,YpBqwDJ,CoB9vDA,YAGE,kBAAA,CAFA,YAAA,CAIA,eAAA,CAHA,SAAA,CAIA,eAAA,CAFA,UpBmwDF,CoB9vDE,yBACE,WpBgwDJ,CoBzvDA,kBACE,YpB4vDF,CKprDI,0CezEJ,kBAKI,wBpB4vDF,CACF,CoBzvDE,qCACE,WpB2vDJ,CK/sDI,sCe7CF,+CAKI,kBpB2vDJ,CoBhwDA,+CAKI,mBpB2vDJ,CACF,CKjsDI,0CerDJ,6BAMI,SAAA,CAFA,eAAA,CACA,UpBwvDF,CoBrvDE,qDACE,gBpBuvDJ,CoBpvDE,gDACE,SpBsvDJ,CoBnvDE,4CACE,iBAAA,CAAA,kBpBqvDJ,CoBlvDE,2CAEE,WAAA,CADA,cpBqvDJ,CoBjvDE,2CACE,mBAAA,CACA,cAAA,CACA,SAAA,CACA,oBAAA,CAAA,iBpBmvDJ,CoBhvDE,2CACE,SpBkvDJ,CoB/uDE,qCAEE,WAAA,CACA,eAAA,CAFA,epBmvDJ,CACF,CqB75DA,MACE,qBAAA,CACA,yBrBg6DF,CqB15DA,aAME,qCAAA,CADA,cAAA,CAEA,0FACE,CAPF,cAAA,CACA,KAAA,CAaA,mDAAA,CACA,qBAAA,CAJA,wFACE,CATF,UAAA,CADA,SrBo6DF,CsB/6DA,MACE,igBtBk7DF,CsB56DA,WACE,iBtB+6DF,CKjxDI,mCiB/JJ,WAKI,etB+6DF,CACF,CsB56DE,kBACE,YtB86DJ,CsB16DE,oBAEE,SAAA,CADA,StB66DJ,CK1wDI,0CiBpKF,8BAkBI,YtB06DJ,CsB57DA,8BAkBI,atB06DJ,CsB57DA,oBAYI,2CAAA,CACA,kBAAA,CAJA,WAAA,CACA,eAAA,CACA,mBAAA,CALA,iBAAA,CACA,SAAA,CAUA,uBAAA,CAHA,4CACE,CAPF,UtBo7DJ,CsBv6DI,+DACE,SAAA,CACA,oCtBy6DN,CACF,CKhzDI,mCiBjJF,8BAyCI,MtBm6DJ,CsB58DA,8BAyCI,OtBm6DJ,CsB58DA,oBAoCI,0BAAA,CADA,cAAA,CADA,QAAA,CAHA,cAAA,CACA,KAAA,CAKA,sDACE,CALF,OtB26DJ,CsBh6DI,+DAME,YAAA,CACA,SAAA,CACA,4CACE,CARF,UtBq6DN,CACF,CK/yDI,0CiBxGA,+DAII,mBtBu5DN,CACF,CK71DM,+DiB/DF,+DASI,mBtBu5DN,CACF,CKl2DM,+DiB/DF,+DAcI,mBtBu5DN,CACF,CsBl5DE,kBAEE,kCAAA,CAAA,0BtBm5DJ,CKj0DI,0CiBpFF,4BAmBI,MtB+4DJ,CsBl6DA,4BAmBI,OtB+4DJ,CsBl6DA,kBAUI,QAAA,CAEA,SAAA,CADA,eAAA,CALA,cAAA,CACA,KAAA,CAWA,wBAAA,CALA,qGACE,CALF,OAAA,CADA,StB05DJ,CsB54DI,4BACE,yBtB84DN,CsB14DI,6DAEE,WAAA,CACA,SAAA,CAMA,uBAAA,CALA,sGACE,CAJF,UtBg5DN,CACF,CK52DI,mCiBjEF,4BA2CI,WtB04DJ,CsBr7DA,4BA2CI,UtB04DJ,CsBr7DA,kBA6CI,eAAA,CAHA,iBAAA,CAIA,8CAAA,CAFA,atBy4DJ,CACF,CK34DM,+DiBOF,6DAII,atBo4DN,CACF,CK13DI,sCiBfA,6DASI,atBo4DN,CACF,CsB/3DE,iBAIE,2CAAA,CACA,0BAAA,CAFA,aAAA,CAFA,iBAAA,CAKA,2CACE,CALF,StBq4DJ,CKv4DI,mCiBAF,iBAaI,0BAAA,CACA,mBAAA,CAFA,atBi4DJ,CsB53DI,uBACE,0BtB83DN,CACF,CsB13DI,4DAEE,2CAAA,CACA,6BAAA,CACA,8BAAA,CAHA,gCtB+3DN,CsBv3DE,4BAKE,mBAAA,CAAA,oBtB43DJ,CsBj4DE,4BAKE,mBAAA,CAAA,oBtB43DJ,CsBj4DE,kBAQE,gBAAA,CAFA,eAAA,CAFA,WAAA,CAHA,iBAAA,CAMA,sBAAA,CAJA,UAAA,CADA,StB+3DJ,CsBt3DI,+BACE,qBtBw3DN,CsBp3DI,kEAEE,uCtBq3DN,CsBj3DI,6BACE,YtBm3DN,CKv5DI,0CiBaF,kBA8BI,eAAA,CADA,aAAA,CADA,UtBo3DJ,CACF,CKj7DI,mCiBgCF,4BAmCI,mBtBo3DJ,CsBv5DA,4BAmCI,oBtBo3DJ,CsBv5DA,kBAqCI,aAAA,CADA,etBm3DJ,CsB/2DI,+BACE,uCtBi3DN,CsB72DI,mCACE,gCtB+2DN,CsB32DI,6DACE,kBtB62DN,CsB12DM,8EACE,uCtB42DR,CsBx2DM,0EACE,WtB02DR,CACF,CsBp2DE,iBAIE,cAAA,CAHA,oBAAA,CAEA,aAAA,CAEA,kCACE,CAJF,YtBy2DJ,CsBj2DI,uBACE,UtBm2DN,CsB/1DI,yCAGE,UtBk2DN,CsBr2DI,yCAGE,WtBk2DN,CsBr2DI,+BACE,iBAAA,CACA,SAAA,CAEA,StBi2DN,CsB91DM,6CACE,oBtBg2DR,CKv8DI,0CiB+FA,yCAcI,UtB+1DN,CsB72DE,yCAcI,WtB+1DN,CsB72DE,+BAaI,StBg2DN,CsB51DM,+CACE,YtB81DR,CACF,CKn+DI,mCiBkHA,+BAwBI,mBtB61DN,CsB11DM,8CACE,YtB41DR,CACF,CsBt1DE,8BAGE,WtB01DJ,CsB71DE,8BAGE,UtB01DJ,CsB71DE,oBAKE,mBAAA,CAJA,iBAAA,CACA,SAAA,CAEA,StBy1DJ,CK/9DI,0CiBkIF,8BAUI,WtBw1DJ,CsBl2DA,8BAUI,UtBw1DJ,CsBl2DA,oBASI,StBy1DJ,CACF,CsBr1DI,uCACE,iBtB21DN,CsB51DI,uCACE,kBtB21DN,CsB51DI,6BAEE,uCAAA,CACA,SAAA,CAIA,oBAAA,CAHA,+DtBw1DN,CsBl1DM,iDAEE,uCAAA,CADA,YtBq1DR,CsBh1DM,gGAGE,SAAA,CADA,mBAAA,CAEA,kBtBi1DR,CsB90DQ,sGACE,UtBg1DV,CsBz0DE,8BAOE,mBAAA,CAAA,oBtBg1DJ,CsBv1DE,8BAOE,mBAAA,CAAA,oBtBg1DJ,CsBv1DE,oBAIE,kBAAA,CAKA,yCAAA,CANA,YAAA,CAKA,eAAA,CAFA,WAAA,CAKA,SAAA,CAVA,iBAAA,CACA,KAAA,CAUA,uBAAA,CAFA,kBAAA,CALA,UtBk1DJ,CKzhEI,mCiBkMF,8BAgBI,mBtB40DJ,CsB51DA,8BAgBI,oBtB40DJ,CsB51DA,oBAiBI,etB20DJ,CACF,CsBx0DI,+DACE,SAAA,CACA,0BtB00DN,CsBr0DE,6BAKE,+BtBw0DJ,CsB70DE,0DAME,gCtBu0DJ,CsB70DE,6BAME,+BtBu0DJ,CsB70DE,mBAIE,eAAA,CAHA,iBAAA,CAEA,UAAA,CADA,StB20DJ,CKxhEI,0CiB2MF,mBAWI,QAAA,CADA,UtBw0DJ,CACF,CKjjEI,mCiB8NF,mBAiBI,SAAA,CADA,UAAA,CAEA,sBtBu0DJ,CsBp0DI,8DACE,8BAAA,CACA,StBs0DN,CACF,CsBj0DE,uBASE,kCAAA,CAAA,0BAAA,CAFA,2CAAA,CANA,WAAA,CACA,eAAA,CAIA,kBtBk0DJ,CsB5zDI,iEAZF,uBAaI,uBtB+zDJ,CACF,CK9lEM,+DiBiRJ,uBAkBI,atB+zDJ,CACF,CK7kEI,sCiB2PF,uBAuBI,atB+zDJ,CACF,CKllEI,mCiB2PF,uBA4BI,YAAA,CAEA,yDAAA,CADA,oBtBg0DJ,CsB5zDI,kEACE,etB8zDN,CsB1zDI,6BACE,+CtB4zDN,CsBxzDI,0CAEE,YAAA,CADA,WtB2zDN,CsBtzDI,gDACE,oDtBwzDN,CsBrzDM,sDACE,0CtBuzDR,CACF,CsBhzDA,kBACE,gCAAA,CACA,qBtBmzDF,CsBhzDE,wBAKE,qDAAA,CADA,uCAAA,CAFA,gBAAA,CACA,kBAAA,CAFA,eAAA,CAKA,uBtBkzDJ,CKtnEI,mCiB8TF,kCAUI,mBtBkzDJ,CsB5zDA,kCAUI,oBtBkzDJ,CACF,CsB9yDE,wBAGE,eAAA,CADA,QAAA,CADA,SAAA,CAIA,wBAAA,CAAA,gBtB+yDJ,CsB3yDE,wBACE,yDtB6yDJ,CsB1yDI,oCACE,etB4yDN,CsBvyDE,wBACE,aAAA,CACA,YAAA,CAEA,uBAAA,CADA,gCtB0yDJ,CsBtyDI,4DACE,uDtBwyDN,CsBpyDI,gDACE,mBtBsyDN,CsBjyDE,gCAKE,cAAA,CADA,aAAA,CAEA,YAAA,CALA,eAAA,CAMA,uBAAA,CALA,KAAA,CACA,StBuyDJ,CsBhyDI,wCACE,YtBkyDN,CsB7xDI,wDACE,YtB+xDN,CsB3xDI,oCAGE,+BAAA,CADA,gBAAA,CADA,mBAAA,CAGA,2CtB6xDN,CKxqEI,mCiBuYA,8CAUI,mBtB2xDN,CsBryDE,8CAUI,oBtB2xDN,CACF,CsBvxDI,oFAEE,uDAAA,CADA,+BtB0xDN,CsBpxDE,sCACE,2CtBsxDJ,CsBjxDE,2BAGE,eAAA,CADA,eAAA,CADA,iBtBqxDJ,CKzrEI,mCiBmaF,qCAOI,mBtBmxDJ,CsB1xDA,qCAOI,oBtBmxDJ,CACF,CsB/wDE,kCAEE,MtBqxDJ,CsBvxDE,kCAEE,OtBqxDJ,CsBvxDE,wBAME,uCAAA,CAFA,aAAA,CACA,YAAA,CAJA,iBAAA,CAEA,YtBoxDJ,CKnrEI,0CiB4ZF,wBAUI,YtBixDJ,CACF,CsB9wDI,8BAKE,6BAAA,CADA,UAAA,CAHA,oBAAA,CAEA,WAAA,CAGA,+CAAA,CAAA,uCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAPA,UtBuxDN,CsB7wDM,wCACE,oBtB+wDR,CsBzwDE,8BAGE,uCAAA,CAFA,gBAAA,CACA,etB4wDJ,CsBxwDI,iCAKE,gCAAA,CAHA,eAAA,CACA,eAAA,CACA,eAAA,CAHA,etB8wDN,CsBvwDM,sCACE,oBtBywDR,CsBpwDI,iCAKE,gCAAA,CAHA,gBAAA,CACA,eAAA,CACA,eAAA,CAHA,atB0wDN,CsBnwDM,sCACE,oBtBqwDR,CsB/vDE,yBAKE,gCAAA,CAJA,aAAA,CAEA,gBAAA,CACA,iBAAA,CAFA,atBowDJ,CsB7vDE,uBAGE,wBAAA,CAFA,+BAAA,CACA,yBtBgwDJ,CuBp6EA,WACE,iBAAA,CACA,SvBu6EF,CuBp6EE,kBAOE,2CAAA,CACA,mBAAA,CACA,8BAAA,CAHA,gCAAA,CAHA,QAAA,CAEA,gBAAA,CADA,YAAA,CAMA,SAAA,CATA,iBAAA,CACA,sBAAA,CAaA,mCAAA,CAJA,oEvBu6EJ,CuBh6EI,6EACE,gBAAA,CACA,SAAA,CAKA,+BAAA,CAJA,8EvBm6EN,CuB35EI,wBAWE,+BAAA,CAAA,8CAAA,CAFA,6BAAA,CAAA,8BAAA,CACA,YAAA,CAFA,UAAA,CAHA,QAAA,CAFA,QAAA,CAIA,kBAAA,CADA,iBAAA,CALA,iBAAA,CACA,KAAA,CAEA,OvBo6EN,CuBx5EE,iBAOE,mBAAA,CAFA,eAAA,CACA,oBAAA,CAHA,QAAA,CAFA,kBAAA,CAGA,aAAA,CAFA,SvB+5EJ,CuBt5EE,iBACE,kBvBw5EJ,CuBp5EE,2BAGE,kBAAA,CAAA,oBvB05EJ,CuB75EE,2BAGE,mBAAA,CAAA,mBvB05EJ,CuB75EE,iBAIE,cAAA,CAHA,aAAA,CAIA,YAAA,CAIA,uBAAA,CAHA,2CACE,CALF,UvB25EJ,CuBj5EI,8CACE,+BvBm5EN,CuB/4EI,uBACE,qDvBi5EN,CwBr+EA,YAIE,qBAAA,CADA,aAAA,CAGA,gBAAA,CALA,eAAA,CACA,UAAA,CAGA,axBy+EF,CwBr+EE,aATF,YAUI,YxBw+EF,CACF,CK1zEI,0CmB3KF,+BAeI,axBm+EJ,CwBl/EA,+BAeI,cxBm+EJ,CwBl/EA,qBAUI,2CAAA,CAHA,aAAA,CAEA,WAAA,CALA,cAAA,CACA,KAAA,CASA,uBAAA,CAHA,iEACE,CAJF,aAAA,CAFA,SxB4+EJ,CwBh+EI,mEACE,8BAAA,CACA,6BxBk+EN,CwB/9EM,6EACE,8BxBi+ER,CwB59EI,6CAEE,QAAA,CAAA,MAAA,CACA,QAAA,CAEA,eAAA,CAJA,iBAAA,CACA,OAAA,CAEA,qBAAA,CAFA,KxBi+EN,CACF,CKz2EI,sCmBtKJ,YAuDI,QxB49EF,CwBz9EE,mBACE,WxB29EJ,CwBv9EE,6CACE,UxBy9EJ,CACF,CwBr9EE,uBACE,YAAA,CACA,OxBu9EJ,CKx3EI,mCmBjGF,uBAMI,QxBu9EJ,CwBp9EI,8BACE,WxBs9EN,CwBl9EI,qCACE,axBo9EN,CwBh9EI,+CACE,kBxBk9EN,CACF,CwB78EE,wBAUE,uBAAA,CANA,kCAAA,CAAA,0BAAA,CAHA,cAAA,CACA,eAAA,CASA,yDAAA,CAFA,oBxB48EJ,CwBv8EI,2CAEE,YAAA,CADA,WxB08EN,CwBr8EI,mEACE,+CxBu8EN,CwBp8EM,qHACE,oDxBs8ER,CwBn8EQ,iIACE,0CxBq8EV,CwBt7EE,wCAGE,wBACE,qBxBs7EJ,CwBl7EE,6BACE,kCxBo7EJ,CwBr7EE,6BACE,iCxBo7EJ,CACF,CKh5EI,0CmB5BF,YAME,0BAAA,CADA,QAAA,CAEA,SAAA,CANA,cAAA,CACA,KAAA,CAMA,sDACE,CALF,OAAA,CADA,SxBq7EF,CwB16EE,4CAEE,WAAA,CACA,SAAA,CACA,4CACE,CAJF,UxB+6EJ,CACF,CyB5lFA,iBACE,GACE,QzB8lFF,CyB3lFA,GACE,azB6lFF,CACF,CyBzlFA,gBACE,GACE,SAAA,CACA,0BzB2lFF,CyBxlFA,IACE,SzB0lFF,CyBvlFA,GACE,SAAA,CACA,uBzBylFF,CACF,CyBjlFA,MACE,+eAAA,CACA,ygBAAA,CACA,mmBAAA,CACA,sfzBmlFF,CyB7kFA,WAOE,kCAAA,CAAA,0BAAA,CANA,aAAA,CACA,gBAAA,CACA,eAAA,CAEA,uCAAA,CAGA,uBAAA,CAJA,kBzBmlFF,CyB5kFE,iBACE,UzB8kFJ,CyB1kFE,iBACE,oBAAA,CAEA,aAAA,CACA,qBAAA,CAFA,UzB8kFJ,CyBzkFI,+BACE,iBzB4kFN,CyB7kFI,+BACE,kBzB4kFN,CyB7kFI,qBAEE,gBzB2kFN,CyBvkFI,kDACE,iBzB0kFN,CyB3kFI,kDACE,kBzB0kFN,CyB3kFI,kDAEE,iBzBykFN,CyB3kFI,kDAEE,kBzBykFN,CyBpkFE,iCAGE,iBzBykFJ,CyB5kFE,iCAGE,kBzBykFJ,CyB5kFE,uBACE,oBAAA,CACA,6BAAA,CAEA,eAAA,CACA,sBAAA,CACA,qBzBskFJ,CyBlkFE,kBACE,YAAA,CAMA,gBAAA,CALA,SAAA,CAMA,oBAAA,CAHA,gBAAA,CAIA,WAAA,CAHA,eAAA,CAFA,SAAA,CADA,UzB0kFJ,CyBjkFI,iDACE,4BzBmkFN,CyB9jFE,iBACE,eAAA,CACA,sBzBgkFJ,CyB7jFI,gDACE,2BzB+jFN,CyB3jFI,kCAIE,kBzBmkFN,CyBvkFI,kCAIE,iBzBmkFN,CyBvkFI,wBAOE,6BAAA,CADA,UAAA,CALA,oBAAA,CAEA,YAAA,CAKA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CALA,uBAAA,CAHA,WzBqkFN,CyBzjFI,iCACE,azB2jFN,CyBvjFI,iCACE,gDAAA,CAAA,wCzByjFN,CyBrjFI,+BACE,8CAAA,CAAA,sCzBujFN,CyBnjFI,+BACE,8CAAA,CAAA,sCzBqjFN,CyBjjFI,sCACE,qDAAA,CAAA,6CzBmjFN,CyB7iFA,gBACE,YzBgjFF,CyB7iFE,gCAIE,kBzBijFJ,CyBrjFE,gCAIE,iBzBijFJ,CyBrjFE,sBAGE,kBAAA,CAGA,uCAAA,CALA,mBAAA,CAIA,gBAAA,CAHA,SzBmjFJ,CyB5iFI,+BACE,aAAA,CACA,oBzB8iFN,CyB1iFI,2CACE,UzB6iFN,CyB9iFI,2CACE,WzB6iFN,CyB9iFI,iCAEE,kBzB4iFN,CyBxiFI,0BACE,WzB0iFN,C0BjuFA,MACE,mSAAA,CACA,oVAAA,CACA,mOAAA,CACA,qZ1BouFF,C0B3tFE,iBAME,kDAAA,CADA,UAAA,CAJA,oBAAA,CAEA,cAAA,CAIA,mCAAA,CAAA,2BAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CANA,0BAAA,CAFA,a1BsuFJ,C0B1tFE,uBACE,6B1B4tFJ,C0BxtFE,sBACE,wCAAA,CAAA,gC1B0tFJ,C0BttFE,6BACE,+CAAA,CAAA,uC1BwtFJ,C0BptFE,4BACE,8CAAA,CAAA,sC1BstFJ,C2BjwFA,SASE,2CAAA,CADA,gCAAA,CAJA,aAAA,CAGA,eAAA,CADA,aAAA,CADA,UAAA,CAFA,S3BwwFF,C2B/vFE,aAZF,SAaI,Y3BkwFF,CACF,CKvlFI,0CsBzLJ,SAkBI,Y3BkwFF,CACF,C2B/vFE,iBACE,mB3BiwFJ,C2B7vFE,yBAIE,iB3BowFJ,C2BxwFE,yBAIE,kB3BowFJ,C2BxwFE,eAQE,eAAA,CAPA,YAAA,CAMA,eAAA,CAJA,QAAA,CAEA,aAAA,CAHA,SAAA,CAWA,oBAAA,CAPA,kB3BkwFJ,C2BxvFI,kCACE,Y3B0vFN,C2BrvFE,eACE,aAAA,CACA,kBAAA,CAAA,mB3BuvFJ,C2BpvFI,sCACE,aAAA,CACA,S3BsvFN,C2BhvFE,eAOE,kCAAA,CAAA,0BAAA,CANA,YAAA,CAEA,eAAA,CADA,gBAAA,CAMA,UAAA,CAJA,uCAAA,CACA,oBAAA,CAIA,8D3BivFJ,C2B5uFI,0CACE,aAAA,CACA,S3B8uFN,C2B1uFI,6BAEE,kB3B6uFN,C2B/uFI,6BAEE,iB3B6uFN,C2B/uFI,mBAGE,iBAAA,CAFA,Y3B8uFN,C2BvuFM,2CACE,qB3ByuFR,C2B1uFM,2CACE,qB3B4uFR,C2B7uFM,2CACE,qB3B+uFR,C2BhvFM,2CACE,qB3BkvFR,C2BnvFM,2CACE,oB3BqvFR,C2BtvFM,2CACE,qB3BwvFR,C2BzvFM,2CACE,qB3B2vFR,C2B5vFM,2CACE,qB3B8vFR,C2B/vFM,4CACE,qB3BiwFR,C2BlwFM,4CACE,oB3BowFR,C2BrwFM,4CACE,qB3BuwFR,C2BxwFM,4CACE,qB3B0wFR,C2B3wFM,4CACE,qB3B6wFR,C2B9wFM,4CACE,qB3BgxFR,C2BjxFM,4CACE,oB3BmxFR,C2B7wFI,gCACE,SAAA,CAIA,yBAAA,CAHA,wC3BgxFN,C4Bn3FA,MACE,wS5Bs3FF,C4B72FE,mCACE,mBAAA,CACA,cAAA,CACA,QAAA,CAEA,mBAAA,CADA,kB5Bi3FJ,C4B52FE,oBAGE,kBAAA,CAOA,+CAAA,CACA,oBAAA,CAVA,mBAAA,CAIA,gBAAA,CACA,0BAAA,CACA,eAAA,CALA,QAAA,CAOA,qBAAA,CADA,eAAA,CAJA,wB5Bq3FJ,C4B32FI,0BAGE,uCAAA,CAFA,aAAA,CACA,YAAA,CAEA,6C5B62FN,C4Bx2FM,gEAEE,0CAAA,CADA,+B5B22FR,C4Br2FI,yBACE,uB5Bu2FN,C4B/1FI,gCAME,oDAAA,CADA,UAAA,CAJA,oBAAA,CAEA,YAAA,CAKA,qCAAA,CAAA,6BAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAJA,iCAAA,CAHA,0BAAA,CAFA,W5B02FN,C4B71FI,wFACE,0C5B+1FN,C6Bz6FA,iBACE,GACE,oB7B46FF,C6Bz6FA,IACE,kB7B26FF,C6Bx6FA,GACE,oB7B06FF,CACF,C6Bl6FA,MACE,0NAAA,CACA,uPAAA,CACA,wB7Bo6FF,C6B95FA,YA6BE,kCAAA,CAAA,0BAAA,CAVA,2CAAA,CACA,mBAAA,CACA,8BAAA,CAHA,gCAAA,CADA,sCAAA,CAdA,+IACE,CAYF,8BAAA,CAMA,SAAA,CArBA,iBAAA,CACA,uBAAA,CAyBA,4BAAA,CAJA,uDACE,CATF,6BAAA,CADA,S7Bk6FF,C6Bh5FE,oBAEE,SAAA,CAKA,uBAAA,CAJA,2EACE,CAHF,S7Bq5FJ,C6B34FE,oBAEE,eAAA,CACA,wBAAA,CAAA,gBAAA,CAFA,U7B+4FJ,C6B14FI,6CACE,qC7B44FN,C6Bx4FI,uCAEE,eAAA,CADA,mB7B24FN,C6Br4FI,6BACE,Y7Bu4FN,C6Bl4FE,8CACE,sC7Bo4FJ,C6Bh4FE,mBAEE,gBAAA,CADA,a7Bm4FJ,C6B/3FI,2CACE,Y7Bi4FN,C6B73FI,0CACE,e7B+3FN,C6Bv3FA,eACE,eAAA,CAGA,YAAA,CADA,0BAAA,CADA,kB7B43FF,C6Bv3FE,yBACE,a7By3FJ,C6Br3FE,oBACE,sCAAA,CACA,iB7Bu3FJ,C6Bn3FE,6BACE,oBAAA,CAGA,gB7Bm3FJ,C6B/2FE,sBAmBE,mBAAA,CAbA,cAAA,CAHA,oBAAA,CACA,gBAAA,CAAA,iBAAA,CAIA,YAAA,CAUA,eAAA,CAjBA,iBAAA,CAMA,wBAAA,CAAA,gBAAA,CAFA,uBAAA,CAHA,S7By3FJ,C6B/2FI,qCACE,uB7Bi3FN,C6Bx2FI,cAtBF,sBAuBI,W7B22FJ,C6Bx2FI,wCACE,2B7B02FN,C6Bt2FI,6BAOE,qCAAA,CACA,+CAAA,CAAA,uC7B22FN,C6Bj2FI,yDAZE,UAAA,CADA,YAAA,CAIA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAVA,iBAAA,CACA,SAAA,CAEA,WAAA,CADA,U7B+3FN,C6Bh3FI,4BAOE,oDAAA,CAMA,4CAAA,CAAA,oCAAA,CADA,uBAAA,CAJA,+C7Bw2FN,C6B71FM,gDACE,uB7B+1FR,C6B31FM,mFACE,0C7B61FR,CACF,C6Bx1FI,0CAGE,2BAAA,CADA,uBAAA,CADA,S7B41FN,C6Bt1FI,8CACE,oB7Bw1FN,C6Br1FM,aAJF,8CASI,8CAAA,CACA,iBAAA,CAHA,gCAAA,CADA,eAAA,CADA,cAAA,CAGA,kB7B01FN,C6Br1FM,oDACE,mC7Bu1FR,CACF,C6B30FE,gCAEE,iBAAA,CADA,e7B+0FJ,C6B30FI,mCACE,iB7B60FN,C6B10FM,oDAGE,a7Bw1FR,C6B31FM,oDAGE,c7Bw1FR,C6B31FM,0CAcE,8CAAA,CACA,iBAAA,CALA,gCAAA,CAEA,oBAAA,CACA,qBAAA,CANA,iBAAA,CACA,eAAA,CAHA,UAAA,CAIA,gBAAA,CALA,aAAA,CAEA,cAAA,CALA,iBAAA,CAUA,iBAAA,CATA,S7By1FR,C8BvmGA,kBAME,e9BmnGF,C8BznGA,kBAME,gB9BmnGF,C8BznGA,QAUE,2CAAA,CACA,oBAAA,CAEA,8BAAA,CALA,uCAAA,CACA,cAAA,CALA,aAAA,CAGA,eAAA,CAKA,YAAA,CAPA,mBAAA,CAJA,cAAA,CACA,UAAA,CAiBA,yBAAA,CALA,mGACE,CAZF,S9BsnGF,C8BnmGE,aAtBF,QAuBI,Y9BsmGF,CACF,C8BnmGE,kBACE,wB9BqmGJ,C8BjmGE,gBAEE,SAAA,CADA,mBAAA,CAGA,+BAAA,CADA,uB9BomGJ,C8BhmGI,0BACE,8B9BkmGN,C8B7lGE,4BAEE,0CAAA,CADA,+B9BgmGJ,C8B3lGE,YACE,oBAAA,CACA,oB9B6lGJ,C+BlpGA,oBACE,GACE,mB/BqpGF,CACF,C+B7oGA,MACE,wf/B+oGF,C+BzoGA,YACE,aAAA,CAEA,eAAA,CADA,a/B6oGF,C+BzoGE,+BAOE,kBAAA,CAAA,kB/B0oGJ,C+BjpGE,+BAOE,iBAAA,CAAA,mB/B0oGJ,C+BjpGE,qBAQE,aAAA,CACA,cAAA,CACA,YAAA,CATA,iBAAA,CAKA,U/B2oGJ,C+BpoGI,qCAIE,iB/B4oGN,C+BhpGI,qCAIE,kB/B4oGN,C+BhpGI,2BAME,6BAAA,CADA,UAAA,CAJA,oBAAA,CAEA,YAAA,CAIA,yCAAA,CAAA,iCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CARA,W/B8oGN,C+BjoGE,kBAUE,2CAAA,CACA,mBAAA,CACA,8BAAA,CAJA,gCAAA,CACA,oBAAA,CAHA,kBAAA,CAFA,YAAA,CASA,SAAA,CANA,aAAA,CAFA,SAAA,CAJA,iBAAA,CAgBA,4BAAA,CAfA,UAAA,CAYA,+CACE,CAZF,S/B+oGJ,C+B9nGI,+EACE,gBAAA,CACA,SAAA,CACA,sC/BgoGN,C+B1nGI,qCAEE,oCACE,gC/B2nGN,C+BvnGI,2CACE,c/BynGN,CACF,C+BpnGE,kBACE,kB/BsnGJ,C+BlnGE,4BAGE,kBAAA,CAAA,oB/BynGJ,C+B5nGE,4BAGE,mBAAA,CAAA,mB/BynGJ,C+B5nGE,kBAKE,cAAA,CAJA,aAAA,CAKA,YAAA,CAIA,uBAAA,CAHA,2CACE,CAJF,kBAAA,CAFA,U/B0nGJ,C+B/mGI,gDACE,+B/BinGN,C+B7mGI,wBACE,qD/B+mGN,CgC/sGA,MAEI,uWAAA,CAAA,8WAAA,CAAA,sPAAA,CAAA,8xBAAA,CAAA,0MAAA,CAAA,gbAAA,CAAA,gMAAA,CAAA,iQAAA,CAAA,0VAAA,CAAA,6aAAA,CAAA,8SAAA,CAAA,gMhCwuGJ,CgC5tGE,4CAME,8CAAA,CACA,4BAAA,CACA,mBAAA,CACA,8BAAA,CAJA,mCAAA,CAJA,iBAAA,CAGA,gBAAA,CADA,iBAAA,CADA,eAAA,CASA,uBAAA,CADA,2BhCguGJ,CgC5tGI,aAdF,4CAeI,ehC+tGJ,CACF,CgC5tGI,sEACE,gChC8tGN,CgCztGI,gDACE,qBhC2tGN,CgCvtGI,gIAEE,iBAAA,CADA,chC0tGN,CgCrtGI,4FACE,iBhCutGN,CgCntGI,kFACE,ehCqtGN,CgCjtGI,0FACE,YhCmtGN,CgC/sGI,8EACE,mBhCitGN,CgC5sGE,sEAGE,iBAAA,CAAA,mBhCstGJ,CgCztGE,sEAGE,kBAAA,CAAA,kBhCstGJ,CgCztGE,sEASE,uBhCgtGJ,CgCztGE,sEASE,wBhCgtGJ,CgCztGE,sEAUE,4BhC+sGJ,CgCztGE,4IAWE,6BhC8sGJ,CgCztGE,sEAWE,4BhC8sGJ,CgCztGE,kDAOE,0BAAA,CACA,WAAA,CAFA,eAAA,CADA,eAAA,CAHA,oBAAA,CAAA,iBAAA,CADA,iBhCwtGJ,CgC3sGI,kFACE,ehC6sGN,CgCzsGI,oFAOE,UhC+sGN,CgCttGI,oFAOE,WhC+sGN,CgCttGI,gEAME,wBfkIU,CenIV,UAAA,CADA,WAAA,CAIA,kDAAA,CAAA,0CAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAVA,iBAAA,CACA,UAAA,CACA,UhCmtGN,CgCvsGI,4DACE,4DhCysGN,CgC3rGE,sDACE,oBhC8rGJ,CgC3rGI,gFACE,gChC6rGN,CgCxrGE,8DACE,0BhC2rGJ,CgCxrGI,4EACE,wBAlBG,CAmBH,kDAAA,CAAA,0ChC0rGN,CgCtrGI,0EACE,ahCwrGN,CgC7sGE,8DACE,oBhCgtGJ,CgC7sGI,wFACE,gChC+sGN,CgC1sGE,sEACE,0BhC6sGJ,CgC1sGI,oFACE,wBAlBG,CAmBH,sDAAA,CAAA,8ChC4sGN,CgCxsGI,kFACE,ahC0sGN,CgC/tGE,sDACE,oBhCkuGJ,CgC/tGI,gFACE,gChCiuGN,CgC5tGE,8DACE,0BhC+tGJ,CgC5tGI,4EACE,wBAlBG,CAmBH,kDAAA,CAAA,0ChC8tGN,CgC1tGI,0EACE,ahC4tGN,CgCjvGE,oDACE,oBhCovGJ,CgCjvGI,8EACE,gChCmvGN,CgC9uGE,4DACE,0BhCivGJ,CgC9uGI,0EACE,wBAlBG,CAmBH,iDAAA,CAAA,yChCgvGN,CgC5uGI,wEACE,ahC8uGN,CgCnwGE,4DACE,oBhCswGJ,CgCnwGI,sFACE,gChCqwGN,CgChwGE,oEACE,0BhCmwGJ,CgChwGI,kFACE,wBAlBG,CAmBH,qDAAA,CAAA,6ChCkwGN,CgC9vGI,gFACE,ahCgwGN,CgCrxGE,8DACE,oBhCwxGJ,CgCrxGI,wFACE,gChCuxGN,CgClxGE,sEACE,0BhCqxGJ,CgClxGI,oFACE,wBAlBG,CAmBH,sDAAA,CAAA,8ChCoxGN,CgChxGI,kFACE,ahCkxGN,CgCvyGE,4DACE,oBhC0yGJ,CgCvyGI,sFACE,gChCyyGN,CgCpyGE,oEACE,0BhCuyGJ,CgCpyGI,kFACE,wBAlBG,CAmBH,qDAAA,CAAA,6ChCsyGN,CgClyGI,gFACE,ahCoyGN,CgCzzGE,4DACE,oBhC4zGJ,CgCzzGI,sFACE,gChC2zGN,CgCtzGE,oEACE,0BhCyzGJ,CgCtzGI,kFACE,wBAlBG,CAmBH,qDAAA,CAAA,6ChCwzGN,CgCpzGI,gFACE,ahCszGN,CgC30GE,0DACE,oBhC80GJ,CgC30GI,oFACE,gChC60GN,CgCx0GE,kEACE,0BhC20GJ,CgCx0GI,gFACE,wBAlBG,CAmBH,oDAAA,CAAA,4ChC00GN,CgCt0GI,8EACE,ahCw0GN,CgC71GE,oDACE,oBhCg2GJ,CgC71GI,8EACE,gChC+1GN,CgC11GE,4DACE,0BhC61GJ,CgC11GI,0EACE,wBAlBG,CAmBH,iDAAA,CAAA,yChC41GN,CgCx1GI,wEACE,ahC01GN,CgC/2GE,4DACE,oBhCk3GJ,CgC/2GI,sFACE,gChCi3GN,CgC52GE,oEACE,0BhC+2GJ,CgC52GI,kFACE,wBAlBG,CAmBH,qDAAA,CAAA,6ChC82GN,CgC12GI,gFACE,ahC42GN,CgCj4GE,wDACE,oBhCo4GJ,CgCj4GI,kFACE,gChCm4GN,CgC93GE,gEACE,0BhCi4GJ,CgC93GI,8EACE,wBAlBG,CAmBH,mDAAA,CAAA,2ChCg4GN,CgC53GI,4EACE,ahC83GN,CiCliHA,MACE,wMjCqiHF,CiC5hHE,sBAEE,uCAAA,CADA,gBjCgiHJ,CiC5hHI,mCACE,ajC8hHN,CiC/hHI,mCACE,cjC8hHN,CiC1hHM,4BACE,sBjC4hHR,CiCzhHQ,mCACE,gCjC2hHV,CiCvhHQ,2DACE,SAAA,CAEA,uBAAA,CADA,ejC0hHV,CiCrhHQ,yGACE,SAAA,CACA,uBjCuhHV,CiCnhHQ,yCACE,YjCqhHV,CiC9gHE,0BACE,eAAA,CACA,ejCghHJ,CiC7gHI,+BACE,oBjC+gHN,CiC1gHE,gDACE,YjC4gHJ,CiCxgHE,8BAIE,+BAAA,CAHA,oBAAA,CAEA,WAAA,CAGA,SAAA,CAKA,4BAAA,CAJA,4DACE,CAHF,0BjC4gHJ,CiCngHI,aAdF,8BAeI,+BAAA,CACA,SAAA,CACA,uBjCsgHJ,CACF,CiCngHI,wCACE,6BjCqgHN,CiCjgHI,oCACE,+BjCmgHN,CiC//GI,qCAKE,6BAAA,CADA,UAAA,CAHA,oBAAA,CAEA,YAAA,CAGA,2CAAA,CAAA,mCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAPA,WjCwgHN,CiC3/GQ,mDACE,oBjC6/GV,CkC3mHE,kCAEE,iBlCinHJ,CkCnnHE,kCAEE,kBlCinHJ,CkCnnHE,wBAGE,yCAAA,CAFA,oBAAA,CAGA,SAAA,CACA,mClC8mHJ,CkCzmHI,aAVF,wBAWI,YlC4mHJ,CACF,CkCxmHE,6FAEE,SAAA,CACA,mClC0mHJ,CkCpmHE,4FAEE,+BlCsmHJ,CkClmHE,oBACE,yBAAA,CACA,uBAAA,CAGA,yElCkmHJ,CKn+GI,sC6BrHE,qDACE,uBlC2lHN,CACF,CkCtlHE,kEACE,yBlCwlHJ,CkCplHE,sBACE,0BlCslHJ,CmCjpHE,2BACE,anCopHJ,CK/9GI,0C8BtLF,2BAKI,enCopHJ,CmCjpHI,6BACE,iBnCmpHN,CACF,CmC/oHI,6BAEE,0BAAA,CAAA,2BAAA,CADA,eAAA,CAEA,iBnCipHN,CmC9oHM,2CACE,kBnCgpHR,CmC1oHI,6CACE,QnC4oHN,CoCxqHE,uBACE,4CpC4qHJ,CoCvqHE,8CAJE,kCAAA,CAAA,0BpC+qHJ,CoC3qHE,uBACE,4CpC0qHJ,CoCrqHE,4BAEE,kCAAA,CAAA,0BAAA,CADA,qCpCwqHJ,CoCpqHI,mCACE,apCsqHN,CoClqHI,kCACE,apCoqHN,CoC/pHE,0BAKE,eAAA,CAJA,aAAA,CAEA,YAAA,CACA,aAAA,CAFA,kBAAA,CAAA,mBpCoqHJ,CoC9pHI,uCACE,epCgqHN,CoC5pHI,sCACE,kBpC8pHN,CqC3sHA,MACE,8LrC8sHF,CqCrsHE,oBAGE,iBAAA,CAEA,gBAAA,CADA,arCusHJ,CqCnsHI,wCACE,uBrCqsHN,CqCjsHI,gCAEE,eAAA,CADA,gBrCosHN,CqC7rHM,wCACE,mBrC+rHR,CqCzrHE,8BAKE,oBrC6rHJ,CqClsHE,8BAKE,mBrC6rHJ,CqClsHE,8BAUE,4BrCwrHJ,CqClsHE,4DAWE,6BrCurHJ,CqClsHE,8BAWE,4BrCurHJ,CqClsHE,oBASE,cAAA,CANA,aAAA,CACA,eAAA,CAIA,erC0rHJ,CqCprHI,kCACE,uCAAA,CACA,oBrCsrHN,CqClrHI,wCAEE,uCAAA,CADA,YrCqrHN,CqChrHI,oCASE,WrCsrHN,CqC/rHI,oCASE,UrCsrHN,CqC/rHI,0BAME,6BAAA,CADA,UAAA,CADA,WAAA,CAMA,yCAAA,CAAA,iCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAZA,iBAAA,CACA,UAAA,CAMA,sBAAA,CADA,yBAAA,CAJA,UrC4rHN,CqC/qHM,oCACE,wBrCirHR,CqC5qHI,4BACE,YrC8qHN,CqCzqHI,4CACE,YrC2qHN,CsCrwHE,+DACE,sBAAA,CAEA,mBAAA,CACA,0BAAA,CACA,uBtCuwHJ,CsCpwHI,2EAGE,iBAAA,CADA,eAAA,CADA,yBtCwwHN,CsCjwHE,mEACE,0BtCmwHJ,CsC/vHE,oBACE,qBtCiwHJ,CsC7vHE,gBACE,oBtC+vHJ,CsC3vHE,gBACE,qBtC6vHJ,CsCzvHE,iBACE,kBtC2vHJ,CsCvvHE,kBACE,kBtCyvHJ,CuClyHE,6BACE,sCvCqyHJ,CuClyHE,cACE,yCvCoyHJ,CuCxxHE,sIACE,oCvC0xHJ,CuClxHE,2EACE,qCvCoxHJ,CuC1wHE,wGACE,oCvC4wHJ,CuCnwHE,yFACE,qCvCqwHJ,CuChwHE,6BACE,kCvCkwHJ,CuC5vHE,6CACE,sCvC8vHJ,CuCvvHE,4DACE,sCvCyvHJ,CuClvHE,4DACE,qCvCovHJ,CuC3uHE,yFACE,qCvC6uHJ,CuCruHE,2EACE,sCvCuuHJ,CuC5tHE,wHACE,qCvC8tHJ,CuCztHE,8BAGE,mBAAA,CADA,gBAAA,CADA,gBvC6tHJ,CuCxtHE,eACE,4CvC0tHJ,CuCvtHE,eACE,4CvCytHJ,CuCrtHE,gBAIE,+CAAA,CACA,kDAAA,CAJA,aAAA,CAEA,wBAAA,CADA,wBvC0tHJ,CuCntHE,yBAOE,wCAAA,CACA,+DAAA,CACA,4BAAA,CACA,6BAAA,CARA,iBAAA,CAGA,eAAA,CACA,eAAA,CAFA,cAAA,CADA,oCAAA,CAFA,iBvC8tHJ,CuCltHI,6BACE,YvCotHN,CuCjtHM,kCACE,wBAAA,CACA,yBvCmtHR,CuC7sHE,iCAaE,wCAAA,CACA,+DAAA,CAJA,uCAAA,CACA,0BAAA,CALA,UAAA,CAJA,oBAAA,CAOA,2BAAA,CADA,2BAAA,CADA,2BAAA,CANA,eAAA,CAWA,wBAAA,CAAA,gBAAA,CAPA,SvCstHJ,CuCpsHE,sBACE,iBAAA,CACA,iBvCssHJ,CuC9rHI,sCACE,gBvCgsHN,CuC5rHI,gDACE,YvC8rHN,CuCprHA,gBACE,iBvCurHF,CuCnrHE,yCACE,aAAA,CACA,SvCqrHJ,CuChrHE,mBACE,YvCkrHJ,CuC7qHE,oBACE,QvC+qHJ,CuC3qHE,4BACE,WAAA,CACA,SAAA,CACA,evC6qHJ,CuC1qHI,0CACE,YvC4qHN,CuCtqHE,yBAKE,wCAAA,CAEA,+BAAA,CADA,4BAAA,CAHA,eAAA,CADA,oDAAA,CAEA,wBAAA,CAAA,gBvC2qHJ,CuCpqHE,2BAEE,+DAAA,CADA,2BvCuqHJ,CuCnqHI,+BACE,uCAAA,CACA,gBvCqqHN,CuChqHE,sBACE,MAAA,CACA,WvCkqHJ,CuC7pHA,aACE,avCgqHF,CuCtpHE,4BAEE,aAAA,CADA,YvC0pHJ,CuCtpHI,wDAEE,2BAAA,CADA,wBvCypHN,CuCnpHE,+BAKE,2CAAA,CAEA,+BAAA,CADA,gCAAA,CADA,sBAAA,CAHA,mBAAA,CACA,gBAAA,CAFA,avC2pHJ,CuClpHI,qCAEE,UAAA,CACA,UAAA,CAFA,avCspHN,CKxxHI,0CkCiJF,8BACE,iBvC2oHF,CuCjoHE,wSAGE,evCuoHJ,CuCnoHE,sCAEE,mBAAA,CACA,eAAA,CADA,oBAAA,CADA,kBAAA,CAAA,mBvCuoHJ,CACF,CwC/9HI,yDAIE,+BAAA,CACA,8BAAA,CAFA,aAAA,CADA,QAAA,CADA,iBxCq+HN,CwC79HI,uBAEE,uCAAA,CADA,cxCg+HN,CwC36HM,iHAEE,WAlDkB,CAiDlB,kBxCs7HR,CwCv7HM,6HAEE,WAlDkB,CAiDlB,kBxCk8HR,CwCn8HM,6HAEE,WAlDkB,CAiDlB,kBxC88HR,CwC/8HM,oHAEE,WAlDkB,CAiDlB,kBxC09HR,CwC39HM,0HAEE,WAlDkB,CAiDlB,kBxCs+HR,CwCv+HM,uHAEE,WAlDkB,CAiDlB,kBxCk/HR,CwCn/HM,uHAEE,WAlDkB,CAiDlB,kBxC8/HR,CwC//HM,6HAEE,WAlDkB,CAiDlB,kBxC0gIR,CwC3gIM,yCAEE,WAlDkB,CAiDlB,kBxC8gIR,CwC/gIM,yCAEE,WAlDkB,CAiDlB,kBxCkhIR,CwCnhIM,0CAEE,WAlDkB,CAiDlB,kBxCshIR,CwCvhIM,uCAEE,WAlDkB,CAiDlB,kBxC0hIR,CwC3hIM,wCAEE,WAlDkB,CAiDlB,kBxC8hIR,CwC/hIM,sCAEE,WAlDkB,CAiDlB,kBxCkiIR,CwCniIM,wCAEE,WAlDkB,CAiDlB,kBxCsiIR,CwCviIM,oCAEE,WAlDkB,CAiDlB,kBxC0iIR,CwC3iIM,2CAEE,WAlDkB,CAiDlB,kBxC8iIR,CwC/iIM,qCAEE,WAlDkB,CAiDlB,kBxCkjIR,CwCnjIM,oCAEE,WAlDkB,CAiDlB,kBxCsjIR,CwCvjIM,kCAEE,WAlDkB,CAiDlB,kBxC0jIR,CwC3jIM,qCAEE,WAlDkB,CAiDlB,kBxC8jIR,CwC/jIM,mCAEE,WAlDkB,CAiDlB,kBxCkkIR,CwCnkIM,qCAEE,WAlDkB,CAiDlB,kBxCskIR,CwCvkIM,wCAEE,WAlDkB,CAiDlB,kBxC0kIR,CwC3kIM,sCAEE,WAlDkB,CAiDlB,kBxC8kIR,CwC/kIM,2CAEE,WAlDkB,CAiDlB,kBxCklIR,CwCvkIM,iCAEE,WAPkB,CAMlB,iBxC0kIR,CwC3kIM,uCAEE,WAPkB,CAMlB,iBxC8kIR,CwC/kIM,mCAEE,WAPkB,CAMlB,iBxCklIR,CyCpqIA,MACE,qMAAA,CACA,mMzCuqIF,CyC9pIE,wBAKE,mBAAA,CAHA,YAAA,CACA,qBAAA,CACA,YAAA,CAHA,iBzCqqIJ,CyC3pII,8BAGE,QAAA,CACA,SAAA,CAHA,iBAAA,CACA,OzC+pIN,CyC1pIM,qCACE,0BzC4pIR,CyC/nIM,kEACE,0CzCioIR,CyC3nIE,2BAKE,uBAAA,CADA,+DAAA,CAHA,YAAA,CACA,cAAA,CACA,aAAA,CAGA,oBzC6nIJ,CyC1nII,aATF,2BAUI,gBzC6nIJ,CACF,CyC1nII,cAGE,+BACE,iBzC0nIN,CyCvnIM,sCAQE,qCAAA,CANA,QAAA,CAKA,UAAA,CAHA,aAAA,CAEA,UAAA,CAHA,MAAA,CAFA,iBAAA,CAaA,2CAAA,CALA,2DACE,CAGF,kDAAA,CARA,+BzC+nIR,CACF,CyCjnII,8CACE,YzCmnIN,CyC/mII,iCASE,+BAAA,CACA,6BAAA,CAJA,uCAAA,CAEA,cAAA,CAPA,aAAA,CAGA,gBAAA,CACA,eAAA,CAFA,8BAAA,CAWA,+BAAA,CAHA,2CACE,CALF,kBAAA,CALA,UzC2nIN,CyC5mIM,aAII,6CACE,OzC2mIV,CyC5mIQ,8CACE,OzC8mIV,CyC/mIQ,8CACE,OzCinIV,CyClnIQ,8CACE,OzConIV,CyCrnIQ,8CACE,OzCunIV,CyCxnIQ,8CACE,OzC0nIV,CyC3nIQ,8CACE,OzC6nIV,CyC9nIQ,8CACE,OzCgoIV,CyCjoIQ,8CACE,OzCmoIV,CyCpoIQ,+CACE,QzCsoIV,CyCvoIQ,+CACE,QzCyoIV,CyC1oIQ,+CACE,QzC4oIV,CyC7oIQ,+CACE,QzC+oIV,CyChpIQ,+CACE,QzCkpIV,CyCnpIQ,+CACE,QzCqpIV,CyCtpIQ,+CACE,QzCwpIV,CyCzpIQ,+CACE,QzC2pIV,CyC5pIQ,+CACE,QzC8pIV,CyC/pIQ,+CACE,QzCiqIV,CyClqIQ,+CACE,QzCoqIV,CACF,CyC/pIM,uCACE,gCzCiqIR,CyC7pIM,oDACE,azC+pIR,CyC1pII,yCACE,SzC4pIN,CyCxpIM,2CACE,aAAA,CACA,8BzC0pIR,CyCppIE,4BACE,UzCspIJ,CyCnpII,aAJF,4BAKI,gBzCspIJ,CACF,CyClpIE,0BACE,YzCopIJ,CyCjpII,aAJF,0BAKI,azCopIJ,CyChpIM,sCACE,OzCkpIR,CyCnpIM,uCACE,OzCqpIR,CyCtpIM,uCACE,OzCwpIR,CyCzpIM,uCACE,OzC2pIR,CyC5pIM,uCACE,OzC8pIR,CyC/pIM,uCACE,OzCiqIR,CyClqIM,uCACE,OzCoqIR,CyCrqIM,uCACE,OzCuqIR,CyCxqIM,uCACE,OzC0qIR,CyC3qIM,wCACE,QzC6qIR,CyC9qIM,wCACE,QzCgrIR,CyCjrIM,wCACE,QzCmrIR,CyCprIM,wCACE,QzCsrIR,CyCvrIM,wCACE,QzCyrIR,CyC1rIM,wCACE,QzC4rIR,CyC7rIM,wCACE,QzC+rIR,CyChsIM,wCACE,QzCksIR,CyCnsIM,wCACE,QzCqsIR,CyCtsIM,wCACE,QzCwsIR,CyCzsIM,wCACE,QzC2sIR,CACF,CyCrsII,+FAEE,QzCusIN,CyCpsIM,yGACE,wBAAA,CACA,yBzCusIR,CyC9rIM,2DAEE,wBAAA,CACA,yBAAA,CAFA,QzCksIR,CyC3rIM,iEACE,QzC6rIR,CyC1rIQ,qLAGE,wBAAA,CACA,yBAAA,CAFA,QzC8rIV,CyCxrIQ,6FACE,wBAAA,CACA,yBzC0rIV,CyCrrIM,yDACE,kBzCurIR,CyClrII,sCACE,QzCorIN,CyC/qIE,2BAEE,iBAAA,CAOA,kBAAA,CAHA,uCAAA,CAEA,cAAA,CAPA,aAAA,CAGA,YAAA,CACA,gBAAA,CAEA,mBAAA,CAGA,gCAAA,CAPA,WzCwrIJ,CyC9qII,iCAEE,uDAAA,CADA,+BzCirIN,CyC5qII,iCAKE,6BAAA,CADA,UAAA,CAHA,aAAA,CAEA,WAAA,CAMA,8CAAA,CAAA,sCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CANA,+CACE,CALF,UzCsrIN,CyCvqIE,4BAOE,yEACE,CANF,YAAA,CAGA,aAAA,CAFA,qBAAA,CAGA,mBAAA,CALA,iBAAA,CAYA,wBAAA,CATA,YzC6qIJ,CyCjqII,sCACE,wBzCmqIN,CyC/pII,oCACE,SzCiqIN,CyC7pII,kCAGE,wEACE,CAFF,mBAAA,CADA,OzCiqIN,CyCvpIM,uDACE,8CAAA,CAAA,sCzCypIR,CKhyII,0CoCqJF,wDAEE,kBzCipIF,CyCnpIA,wDAEE,mBzCipIF,CyCnpIA,8CAGE,eAAA,CAFA,eAAA,CAGA,iCzC+oIF,CyC3oIE,8DACE,mBzC8oIJ,CyC/oIE,8DACE,kBzC8oIJ,CyC/oIE,oDAEE,UzC6oIJ,CyCzoIE,8EAEE,kBzC4oIJ,CyC9oIE,8EAEE,mBzC4oIJ,CyC9oIE,8EAGE,kBzC2oIJ,CyC9oIE,8EAGE,mBzC2oIJ,CyC9oIE,oEACE,UzC6oIJ,CyCvoIE,8EAEE,mBzC0oIJ,CyC5oIE,8EAEE,kBzC0oIJ,CyC5oIE,8EAGE,mBzCyoIJ,CyC5oIE,8EAGE,kBzCyoIJ,CyC5oIE,oEACE,UzC2oIJ,CACF,CyC7nIE,cAHF,olDAII,gCzCgoIF,CyC7nIE,g8GACE,uCzC+nIJ,CACF,CyC1nIA,4sDACE,+BzC6nIF,CyCznIA,wmDACE,azC4nIF,C0ChgJA,MACE,8WAAA,CACA,uX1CmgJF,C0C1/IE,4BAEE,oBAAA,CADA,iB1C8/IJ,C0Cz/II,sDAGE,S1C2/IN,C0C9/II,sDAGE,U1C2/IN,C0C9/II,4CACE,iBAAA,CACA,S1C4/IN,C0Ct/IE,+CAEE,SAAA,CADA,U1Cy/IJ,C0Cp/IE,kDAOE,W1C0/IJ,C0CjgJE,kDAOE,Y1C0/IJ,C0CjgJE,wCAME,qDAAA,CADA,UAAA,CADA,aAAA,CAIA,0CAAA,CAAA,kCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAVA,iBAAA,CACA,SAAA,CACA,Y1C8/IJ,C0Cl/IE,gEACE,wBzB2Wa,CyB1Wb,mDAAA,CAAA,2C1Co/IJ,C2CpiJA,QACE,8DAAA,CAGA,+CAAA,CACA,iEAAA,CACA,oDAAA,CACA,sDAAA,CACA,mDAAA,CAGA,qEAAA,CACA,qEAAA,CACA,wEAAA,CACA,0EAAA,CACA,wEAAA,CACA,yEAAA,CACA,kEAAA,CACA,+DAAA,CACA,oEAAA,CACA,oEAAA,CACA,mEAAA,CACA,gEAAA,CACA,uEAAA,CACA,mEAAA,CACA,qEAAA,CACA,oEAAA,CACA,gEAAA,CACA,wEAAA,CACA,qEAAA,CACA,+D3CmiJF,C2C7hJA,SAEE,kBAAA,CADA,Y3CiiJF,C4CnkJE,kBAUE,cAAA,CATA,YAAA,CACA,kEACE,CAQF,Y5C+jJJ,C4C3jJI,sDACE,gB5C6jJN,C4CvjJI,oFAKE,wDAAA,CACA,mBAAA,CAJA,aAAA,CAEA,QAAA,CADA,aAAA,CAIA,sC5CyjJN,C4CpjJM,iOACE,kBAAA,CACA,8B5CujJR,C4CnjJM,6FACE,iBAAA,CAAA,c5CsjJR,C4CljJM,2HACE,Y5CqjJR,C4CjjJM,wHACE,e5CojJR,C4CriJI,yMAGE,eAAA,CAAA,Y5C6iJN,C4C/hJI,ybAOE,W5CqiJN,C4CjiJI,8BACE,eAAA,CAAA,Y5CmiJN,CK/9II,mCwChKA,8BACE,U7CuoJJ,C6CxoJE,8BACE,W7CuoJJ,C6CxoJE,8BAGE,kB7CqoJJ,C6CxoJE,8BAGE,iB7CqoJJ,C6CxoJE,oBAKE,mBAAA,CADA,YAAA,CAFA,a7CsoJJ,C6ChoJI,kCACE,W7CmoJN,C6CpoJI,kCACE,U7CmoJN,C6CpoJI,kCAEE,iBAAA,CAAA,c7CkoJN,C6CpoJI,kCAEE,aAAA,CAAA,kB7CkoJN,CACF","file":"main.css"}
\ No newline at end of file
diff --git a/assets/stylesheets/main.fad675c6.min.css b/assets/stylesheets/main.fad675c6.min.css
deleted file mode 100644
index 21491eba3b..0000000000
--- a/assets/stylesheets/main.fad675c6.min.css
+++ /dev/null
@@ -1 +0,0 @@
-@charset "UTF-8";html{-webkit-text-size-adjust:none;-moz-text-size-adjust:none;text-size-adjust:none;box-sizing:border-box}*,:after,:before{box-sizing:inherit}@media (prefers-reduced-motion){*,:after,:before{transition:none!important}}body{margin:0}a,button,input,label{-webkit-tap-highlight-color:transparent}a{color:inherit;text-decoration:none}hr{border:0;box-sizing:initial;display:block;height:.05rem;overflow:visible;padding:0}small{font-size:80%}sub,sup{line-height:1em}img{border-style:none}table{border-collapse:initial;border-spacing:0}td,th{font-weight:400;vertical-align:top}button{background:#0000;border:0;font-family:inherit;font-size:inherit;margin:0;padding:0}input{border:0;outline:none}:root{--md-primary-fg-color:#4051b5;--md-primary-fg-color--light:#5d6cc0;--md-primary-fg-color--dark:#303fa1;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3;--md-accent-fg-color:#526cfe;--md-accent-fg-color--transparent:#526cfe1a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-scheme=default]{color-scheme:light}[data-md-color-scheme=default] img[src$="#gh-dark-mode-only"],[data-md-color-scheme=default] img[src$="#only-dark"]{display:none}:root,[data-md-color-scheme=default]{--md-hue:225deg;--md-default-fg-color:#000000de;--md-default-fg-color--light:#0000008a;--md-default-fg-color--lighter:#00000052;--md-default-fg-color--lightest:#00000012;--md-default-bg-color:#fff;--md-default-bg-color--light:#ffffffb3;--md-default-bg-color--lighter:#ffffff4d;--md-default-bg-color--lightest:#ffffff1f;--md-code-fg-color:#36464e;--md-code-bg-color:#f5f5f5;--md-code-hl-color:#4287ff;--md-code-hl-color--light:#4287ff1a;--md-code-hl-number-color:#d52a2a;--md-code-hl-special-color:#db1457;--md-code-hl-function-color:#a846b9;--md-code-hl-constant-color:#6e59d9;--md-code-hl-keyword-color:#3f6ec6;--md-code-hl-string-color:#1c7d4d;--md-code-hl-name-color:var(--md-code-fg-color);--md-code-hl-operator-color:var(--md-default-fg-color--light);--md-code-hl-punctuation-color:var(--md-default-fg-color--light);--md-code-hl-comment-color:var(--md-default-fg-color--light);--md-code-hl-generic-color:var(--md-default-fg-color--light);--md-code-hl-variable-color:var(--md-default-fg-color--light);--md-typeset-color:var(--md-default-fg-color);--md-typeset-a-color:var(--md-primary-fg-color);--md-typeset-del-color:#f5503d26;--md-typeset-ins-color:#0bd57026;--md-typeset-kbd-color:#fafafa;--md-typeset-kbd-accent-color:#fff;--md-typeset-kbd-border-color:#b8b8b8;--md-typeset-mark-color:#ffff0080;--md-typeset-table-color:#0000001f;--md-typeset-table-color--light:rgba(0,0,0,.035);--md-admonition-fg-color:var(--md-default-fg-color);--md-admonition-bg-color:var(--md-default-bg-color);--md-warning-fg-color:#000000de;--md-warning-bg-color:#ff9;--md-footer-fg-color:#fff;--md-footer-fg-color--light:#ffffffb3;--md-footer-fg-color--lighter:#ffffff73;--md-footer-bg-color:#000000de;--md-footer-bg-color--dark:#00000052;--md-shadow-z1:0 0.2rem 0.5rem #0000000d,0 0 0.05rem #0000001a;--md-shadow-z2:0 0.2rem 0.5rem #0000001a,0 0 0.05rem #00000040;--md-shadow-z3:0 0.2rem 0.5rem #0003,0 0 0.05rem #00000059}.md-icon svg{fill:currentcolor;display:block;height:1.2rem;width:1.2rem}body{-webkit-font-smoothing:antialiased;-moz-osx-font-smoothing:grayscale;--md-text-font-family:var(--md-text-font,_),-apple-system,BlinkMacSystemFont,Helvetica,Arial,sans-serif;--md-code-font-family:var(--md-code-font,_),SFMono-Regular,Consolas,Menlo,monospace}aside,body,input{font-feature-settings:"kern","liga";color:var(--md-typeset-color);font-family:var(--md-text-font-family)}code,kbd,pre{font-feature-settings:"kern";font-family:var(--md-code-font-family)}:root{--md-typeset-table-sort-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="m18 21-4-4h3V7h-3l4-4 4 4h-3v10h3M2 19v-2h10v2M2 13v-2h7v2M2 7V5h4v2H2Z"/></svg>');--md-typeset-table-sort-icon--asc:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M19 17h3l-4 4-4-4h3V3h2M2 17h10v2H2M6 5v2H2V5m0 6h7v2H2v-2Z"/></svg>');--md-typeset-table-sort-icon--desc:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M19 7h3l-4-4-4 4h3v14h2M2 17h10v2H2M6 5v2H2V5m0 6h7v2H2v-2Z"/></svg>')}.md-typeset{-webkit-print-color-adjust:exact;color-adjust:exact;font-size:.8rem;line-height:1.6}@media print{.md-typeset{font-size:.68rem}}.md-typeset blockquote,.md-typeset dl,.md-typeset figure,.md-typeset ol,.md-typeset pre,.md-typeset ul{margin-bottom:1em;margin-top:1em}.md-typeset h1{color:var(--md-default-fg-color--light);font-size:2em;line-height:1.3;margin:0 0 1.25em}.md-typeset h1,.md-typeset h2{font-weight:300;letter-spacing:-.01em}.md-typeset h2{font-size:1.5625em;line-height:1.4;margin:1.6em 0 .64em}.md-typeset h3{font-size:1.25em;font-weight:400;letter-spacing:-.01em;line-height:1.5;margin:1.6em 0 .8em}.md-typeset h2+h3{margin-top:.8em}.md-typeset h4{font-weight:700;letter-spacing:-.01em;margin:1em 0}.md-typeset h5,.md-typeset h6{color:var(--md-default-fg-color--light);font-size:.8em;font-weight:700;letter-spacing:-.01em;margin:1.25em 0}.md-typeset h5{text-transform:uppercase}.md-typeset hr{border-bottom:.05rem solid var(--md-default-fg-color--lightest);display:flow-root;margin:1.5em 0}.md-typeset a{color:var(--md-typeset-a-color);word-break:break-word}.md-typeset a,.md-typeset a:before{transition:color 125ms}.md-typeset a:focus,.md-typeset a:hover{color:var(--md-accent-fg-color)}.md-typeset a:focus code,.md-typeset a:hover code{background-color:var(--md-accent-fg-color--transparent)}.md-typeset a code{color:currentcolor;transition:background-color 125ms}.md-typeset a.focus-visible{outline-color:var(--md-accent-fg-color);outline-offset:.2rem}.md-typeset code,.md-typeset kbd,.md-typeset pre{color:var(--md-code-fg-color);direction:ltr;font-variant-ligatures:none}@media print{.md-typeset code,.md-typeset kbd,.md-typeset pre{white-space:pre-wrap}}.md-typeset code{background-color:var(--md-code-bg-color);border-radius:.1rem;-webkit-box-decoration-break:clone;box-decoration-break:clone;font-size:.85em;padding:0 .2941176471em;word-break:break-word}.md-typeset code:not(.focus-visible){-webkit-tap-highlight-color:transparent;outline:none}.md-typeset pre{display:flow-root;line-height:1.4;position:relative}.md-typeset pre>code{-webkit-box-decoration-break:slice;box-decoration-break:slice;box-shadow:none;display:block;margin:0;outline-color:var(--md-accent-fg-color);overflow:auto;padding:.7720588235em 1.1764705882em;scrollbar-color:var(--md-default-fg-color--lighter) #0000;scrollbar-width:thin;touch-action:auto;word-break:normal}.md-typeset pre>code:hover{scrollbar-color:var(--md-accent-fg-color) #0000}.md-typeset pre>code::-webkit-scrollbar{height:.2rem;width:.2rem}.md-typeset pre>code::-webkit-scrollbar-thumb{background-color:var(--md-default-fg-color--lighter)}.md-typeset pre>code::-webkit-scrollbar-thumb:hover{background-color:var(--md-accent-fg-color)}.md-typeset kbd{background-color:var(--md-typeset-kbd-color);border-radius:.1rem;box-shadow:0 .1rem 0 .05rem var(--md-typeset-kbd-border-color),0 .1rem 0 var(--md-typeset-kbd-border-color),0 -.1rem .2rem var(--md-typeset-kbd-accent-color) inset;color:var(--md-default-fg-color);display:inline-block;font-size:.75em;padding:0 .6666666667em;vertical-align:text-top;word-break:break-word}.md-typeset mark{background-color:var(--md-typeset-mark-color);-webkit-box-decoration-break:clone;box-decoration-break:clone;color:inherit;word-break:break-word}.md-typeset abbr{border-bottom:.05rem dotted var(--md-default-fg-color--light);cursor:help;text-decoration:none}@media (hover:none){.md-typeset abbr[title]:focus:after,.md-typeset abbr[title]:hover:after{background-color:var(--md-default-fg-color);border-radius:.1rem;box-shadow:var(--md-shadow-z3);color:var(--md-default-bg-color);content:attr(title);font-size:.7rem;left:.8rem;margin-top:2em;padding:.2rem .3rem;position:absolute;right:.8rem}}.md-typeset small{opacity:.75}[dir=ltr] .md-typeset sub,[dir=ltr] .md-typeset sup{margin-left:.078125em}[dir=rtl] .md-typeset sub,[dir=rtl] .md-typeset sup{margin-right:.078125em}[dir=ltr] .md-typeset blockquote{padding-left:.6rem}[dir=rtl] .md-typeset blockquote{padding-right:.6rem}[dir=ltr] .md-typeset blockquote{border-left:.2rem solid var(--md-default-fg-color--lighter)}[dir=rtl] .md-typeset blockquote{border-right:.2rem solid var(--md-default-fg-color--lighter)}.md-typeset blockquote{color:var(--md-default-fg-color--light);margin-left:0;margin-right:0}.md-typeset ul{list-style-type:disc}[dir=ltr] .md-typeset ol,[dir=ltr] .md-typeset ul{margin-left:.625em}[dir=rtl] .md-typeset ol,[dir=rtl] .md-typeset ul{margin-right:.625em}.md-typeset ol,.md-typeset ul{padding:0}.md-typeset ol:not([hidden]),.md-typeset ul:not([hidden]){display:flow-root}.md-typeset ol ol,.md-typeset ul ol{list-style-type:lower-alpha}.md-typeset ol ol ol,.md-typeset ul ol ol{list-style-type:lower-roman}[dir=ltr] .md-typeset ol li,[dir=ltr] .md-typeset ul li{margin-left:1.25em}[dir=rtl] .md-typeset ol li,[dir=rtl] .md-typeset ul li{margin-right:1.25em}.md-typeset ol li,.md-typeset ul li{margin-bottom:.5em}.md-typeset ol li blockquote,.md-typeset ol li p,.md-typeset ul li blockquote,.md-typeset ul li p{margin:.5em 0}.md-typeset ol li:last-child,.md-typeset ul li:last-child{margin-bottom:0}[dir=ltr] .md-typeset ol li ol,[dir=ltr] .md-typeset ol li ul,[dir=ltr] .md-typeset ul li ol,[dir=ltr] .md-typeset ul li ul{margin-left:.625em}[dir=rtl] .md-typeset ol li ol,[dir=rtl] .md-typeset ol li ul,[dir=rtl] .md-typeset ul li ol,[dir=rtl] .md-typeset ul li ul{margin-right:.625em}.md-typeset ol li ol,.md-typeset ol li ul,.md-typeset ul li ol,.md-typeset ul li ul{margin-bottom:.5em;margin-top:.5em}[dir=ltr] .md-typeset dd{margin-left:1.875em}[dir=rtl] .md-typeset dd{margin-right:1.875em}.md-typeset dd{margin-bottom:1.5em;margin-top:1em}.md-typeset img,.md-typeset svg,.md-typeset video{height:auto;max-width:100%}.md-typeset img[align=left]{margin:1em 1em 1em 0}.md-typeset img[align=right]{margin:1em 0 1em 1em}.md-typeset img[align]:only-child{margin-top:0}.md-typeset figure{display:flow-root;margin:1em auto;max-width:100%;text-align:center;width:-webkit-fit-content;width:-moz-fit-content;width:fit-content}.md-typeset figure img{display:block}.md-typeset figcaption{font-style:italic;margin:1em auto;max-width:24rem}.md-typeset iframe{max-width:100%}.md-typeset table:not([class]){background-color:var(--md-default-bg-color);border:.05rem solid var(--md-typeset-table-color);border-radius:.1rem;display:inline-block;font-size:.64rem;max-width:100%;overflow:auto;touch-action:auto}@media print{.md-typeset table:not([class]){display:table}}.md-typeset table:not([class])+*{margin-top:1.5em}.md-typeset table:not([class]) td>:first-child,.md-typeset table:not([class]) th>:first-child{margin-top:0}.md-typeset table:not([class]) td>:last-child,.md-typeset table:not([class]) th>:last-child{margin-bottom:0}.md-typeset table:not([class]) td:not([align]),.md-typeset table:not([class]) th:not([align]){text-align:left}[dir=rtl] .md-typeset table:not([class]) td:not([align]),[dir=rtl] .md-typeset table:not([class]) th:not([align]){text-align:right}.md-typeset table:not([class]) th{font-weight:700;min-width:5rem;padding:.9375em 1.25em;vertical-align:top}.md-typeset table:not([class]) td{border-top:.05rem solid var(--md-typeset-table-color);padding:.9375em 1.25em;vertical-align:top}.md-typeset table:not([class]) tbody tr{transition:background-color 125ms}.md-typeset table:not([class]) tbody tr:hover{background-color:var(--md-typeset-table-color--light);box-shadow:0 .05rem 0 var(--md-default-bg-color) inset}.md-typeset table:not([class]) a{word-break:normal}.md-typeset table th[role=columnheader]{cursor:pointer}[dir=ltr] .md-typeset table th[role=columnheader]:after{margin-left:.5em}[dir=rtl] .md-typeset table th[role=columnheader]:after{margin-right:.5em}.md-typeset table th[role=columnheader]:after{content:"";display:inline-block;height:1.2em;-webkit-mask-image:var(--md-typeset-table-sort-icon);mask-image:var(--md-typeset-table-sort-icon);-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;transition:background-color 125ms;vertical-align:text-bottom;width:1.2em}.md-typeset table th[role=columnheader]:hover:after{background-color:var(--md-default-fg-color--lighter)}.md-typeset table th[role=columnheader][aria-sort=ascending]:after{background-color:var(--md-default-fg-color--light);-webkit-mask-image:var(--md-typeset-table-sort-icon--asc);mask-image:var(--md-typeset-table-sort-icon--asc)}.md-typeset table th[role=columnheader][aria-sort=descending]:after{background-color:var(--md-default-fg-color--light);-webkit-mask-image:var(--md-typeset-table-sort-icon--desc);mask-image:var(--md-typeset-table-sort-icon--desc)}.md-typeset__scrollwrap{margin:1em -.8rem;overflow-x:auto;touch-action:auto}.md-typeset__table{display:inline-block;margin-bottom:.5em;padding:0 .8rem}@media print{.md-typeset__table{display:block}}html .md-typeset__table table{display:table;margin:0;overflow:hidden;width:100%}@media screen and (max-width:44.984375em){.md-content__inner>pre{margin:1em -.8rem}.md-content__inner>pre code{border-radius:0}}.md-typeset .md-author{display:block;flex-shrink:0;height:1.6rem;overflow:hidden;position:relative;transition:color 125ms,transform 125ms;width:1.6rem}.md-typeset .md-author img{border-radius:100%;display:block}.md-typeset .md-author--more{background:var(--md-default-fg-color--lightest);color:var(--md-default-fg-color--lighter);font-size:.6rem;font-weight:700;line-height:1.6rem;text-align:center}.md-typeset .md-author--long{height:2.4rem;width:2.4rem}.md-typeset a.md-author{transform:scale(1)}.md-typeset a.md-author img{filter:grayscale(100%) opacity(75%);transition:filter 125ms}.md-typeset a.md-author:focus,.md-typeset a.md-author:hover{transform:scale(1.1);z-index:1}.md-typeset a.md-author:focus img,.md-typeset a.md-author:hover img{filter:grayscale(0)}.md-banner{background-color:var(--md-footer-bg-color);color:var(--md-footer-fg-color);overflow:auto}@media print{.md-banner{display:none}}.md-banner--warning{background-color:var(--md-warning-bg-color);color:var(--md-warning-fg-color)}.md-banner__inner{font-size:.7rem;margin:.6rem auto;padding:0 .8rem}[dir=ltr] .md-banner__button{float:right}[dir=rtl] .md-banner__button{float:left}.md-banner__button{color:inherit;cursor:pointer;transition:opacity .25s}.no-js .md-banner__button{display:none}.md-banner__button:hover{opacity:.7}html{font-size:125%;height:100%;overflow-x:hidden}@media screen and (min-width:100em){html{font-size:137.5%}}@media screen and (min-width:125em){html{font-size:150%}}body{background-color:var(--md-default-bg-color);display:flex;flex-direction:column;font-size:.5rem;min-height:100%;position:relative;width:100%}@media print{body{display:block}}@media screen and (max-width:59.984375em){body[data-md-scrolllock]{position:fixed}}.md-grid{margin-left:auto;margin-right:auto;max-width:61rem}.md-container{display:flex;flex-direction:column;flex-grow:1}@media print{.md-container{display:block}}.md-main{flex-grow:1}.md-main__inner{display:flex;height:100%;margin-top:1.5rem}.md-ellipsis{overflow:hidden;text-overflow:ellipsis}.md-toggle{display:none}.md-option{height:0;opacity:0;position:absolute;width:0}.md-option:checked+label:not([hidden]){display:block}.md-option.focus-visible+label{outline-color:var(--md-accent-fg-color);outline-style:auto}.md-skip{background-color:var(--md-default-fg-color);border-radius:.1rem;color:var(--md-default-bg-color);font-size:.64rem;margin:.5rem;opacity:0;outline-color:var(--md-accent-fg-color);padding:.3rem .5rem;position:fixed;transform:translateY(.4rem);z-index:-1}.md-skip:focus{opacity:1;transform:translateY(0);transition:transform .25s cubic-bezier(.4,0,.2,1),opacity 175ms 75ms;z-index:10}@page{margin:25mm}:root{--md-clipboard-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M19 21H8V7h11m0-2H8a2 2 0 0 0-2 2v14a2 2 0 0 0 2 2h11a2 2 0 0 0 2-2V7a2 2 0 0 0-2-2m-3-4H4a2 2 0 0 0-2 2v14h2V3h12V1Z"/></svg>')}.md-clipboard{border-radius:.1rem;color:var(--md-default-fg-color--lightest);cursor:pointer;height:1.5em;outline-color:var(--md-accent-fg-color);outline-offset:.1rem;position:absolute;right:.5em;top:.5em;transition:color .25s;width:1.5em;z-index:1}@media print{.md-clipboard{display:none}}.md-clipboard:not(.focus-visible){-webkit-tap-highlight-color:transparent;outline:none}:hover>.md-clipboard{color:var(--md-default-fg-color--light)}.md-clipboard:focus,.md-clipboard:hover{color:var(--md-accent-fg-color)}.md-clipboard:after{background-color:currentcolor;content:"";display:block;height:1.125em;margin:0 auto;-webkit-mask-image:var(--md-clipboard-icon);mask-image:var(--md-clipboard-icon);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;width:1.125em}.md-clipboard--inline{cursor:pointer}.md-clipboard--inline code{transition:color .25s,background-color .25s}.md-clipboard--inline:focus code,.md-clipboard--inline:hover code{background-color:var(--md-accent-fg-color--transparent);color:var(--md-accent-fg-color)}@keyframes consent{0%{opacity:0;transform:translateY(100%)}to{opacity:1;transform:translateY(0)}}@keyframes overlay{0%{opacity:0}to{opacity:1}}.md-consent__overlay{animation:overlay .25s both;-webkit-backdrop-filter:blur(.1rem);backdrop-filter:blur(.1rem);background-color:#0000008a;height:100%;opacity:1;position:fixed;top:0;width:100%;z-index:5}.md-consent__inner{animation:consent .5s cubic-bezier(.1,.7,.1,1) both;background-color:var(--md-default-bg-color);border:0;border-radius:.1rem;bottom:0;box-shadow:0 0 .2rem #0000001a,0 .2rem .4rem #0003;max-height:100%;overflow:auto;padding:0;position:fixed;width:100%;z-index:5}.md-consent__form{padding:.8rem}.md-consent__settings{display:none;margin:1em 0}input:checked+.md-consent__settings{display:block}.md-consent__controls{margin-bottom:.8rem}.md-typeset .md-consent__controls .md-button{display:inline}@media screen and (max-width:44.984375em){.md-typeset .md-consent__controls .md-button{display:block;margin-top:.4rem;text-align:center;width:100%}}.md-consent label{cursor:pointer}.md-content{flex-grow:1;min-width:0}.md-content__inner{margin:0 .8rem 1.2rem;padding-top:.6rem}@media screen and (min-width:76.25em){[dir=ltr] .md-sidebar--primary:not([hidden])~.md-content>.md-content__inner{margin-left:1.2rem}[dir=ltr] .md-sidebar--secondary:not([hidden])~.md-content>.md-content__inner,[dir=rtl] .md-sidebar--primary:not([hidden])~.md-content>.md-content__inner{margin-right:1.2rem}[dir=rtl] .md-sidebar--secondary:not([hidden])~.md-content>.md-content__inner{margin-left:1.2rem}}.md-content__inner:before{content:"";display:block;height:.4rem}.md-content__inner>:last-child{margin-bottom:0}[dir=ltr] .md-content__button{float:right}[dir=rtl] .md-content__button{float:left}[dir=ltr] .md-content__button{margin-left:.4rem}[dir=rtl] .md-content__button{margin-right:.4rem}.md-content__button{margin:.4rem 0;padding:0}@media print{.md-content__button{display:none}}.md-typeset .md-content__button{color:var(--md-default-fg-color--lighter)}.md-content__button svg{display:inline;vertical-align:top}[dir=rtl] .md-content__button svg{transform:scaleX(-1)}[dir=ltr] .md-dialog{right:.8rem}[dir=rtl] .md-dialog{left:.8rem}.md-dialog{background-color:var(--md-default-fg-color);border-radius:.1rem;bottom:.8rem;box-shadow:var(--md-shadow-z3);min-width:11.1rem;opacity:0;padding:.4rem .6rem;pointer-events:none;position:fixed;transform:translateY(100%);transition:transform 0ms .4s,opacity .4s;z-index:4}@media print{.md-dialog{display:none}}.md-dialog--active{opacity:1;pointer-events:auto;transform:translateY(0);transition:transform .4s cubic-bezier(.075,.85,.175,1),opacity .4s}.md-dialog__inner{color:var(--md-default-bg-color);font-size:.7rem}.md-feedback{margin:2em 0 1em;text-align:center}.md-feedback fieldset{border:none;margin:0;padding:0}.md-feedback__title{font-weight:700;margin:1em auto}.md-feedback__inner{position:relative}.md-feedback__list{display:flex;flex-wrap:wrap;place-content:baseline center;position:relative}.md-feedback__list:hover .md-icon:not(:disabled){color:var(--md-default-fg-color--lighter)}:disabled .md-feedback__list{min-height:1.8rem}.md-feedback__icon{color:var(--md-default-fg-color--light);cursor:pointer;flex-shrink:0;margin:0 .1rem;transition:color 125ms}.md-feedback__icon:not(:disabled).md-icon:hover{color:var(--md-accent-fg-color)}.md-feedback__icon:disabled{color:var(--md-default-fg-color--lightest);pointer-events:none}.md-feedback__note{opacity:0;position:relative;transform:translateY(.4rem);transition:transform .4s cubic-bezier(.1,.7,.1,1),opacity .15s}.md-feedback__note>*{margin:0 auto;max-width:16rem}:disabled .md-feedback__note{opacity:1;transform:translateY(0)}.md-footer{background-color:var(--md-footer-bg-color);color:var(--md-footer-fg-color)}@media print{.md-footer{display:none}}.md-footer__inner{justify-content:space-between;overflow:auto;padding:.2rem}.md-footer__inner:not([hidden]){display:flex}.md-footer__link{align-items:end;display:flex;flex-grow:0.01;margin-bottom:.4rem;margin-top:1rem;max-width:100%;outline-color:var(--md-accent-fg-color);overflow:hidden;transition:opacity .25s}.md-footer__link:focus,.md-footer__link:hover{opacity:.7}[dir=rtl] .md-footer__link svg{transform:scaleX(-1)}@media screen and (max-width:44.984375em){.md-footer__link--prev{flex-shrink:0}.md-footer__link--prev .md-footer__title{display:none}}[dir=ltr] .md-footer__link--next{margin-left:auto}[dir=rtl] .md-footer__link--next{margin-right:auto}.md-footer__link--next{text-align:right}[dir=rtl] .md-footer__link--next{text-align:left}.md-footer__title{flex-grow:1;font-size:.9rem;margin-bottom:.7rem;max-width:calc(100% - 2.4rem);padding:0 1rem;white-space:nowrap}.md-footer__button{margin:.2rem;padding:.4rem}.md-footer__direction{font-size:.64rem;opacity:.7}.md-footer-meta{background-color:var(--md-footer-bg-color--dark)}.md-footer-meta__inner{display:flex;flex-wrap:wrap;justify-content:space-between;padding:.2rem}html .md-footer-meta.md-typeset a{color:var(--md-footer-fg-color--light)}html .md-footer-meta.md-typeset a:focus,html .md-footer-meta.md-typeset a:hover{color:var(--md-footer-fg-color)}.md-copyright{color:var(--md-footer-fg-color--lighter);font-size:.64rem;margin:auto .6rem;padding:.4rem 0;width:100%}@media screen and (min-width:45em){.md-copyright{width:auto}}.md-copyright__highlight{color:var(--md-footer-fg-color--light)}.md-social{display:inline-flex;gap:.2rem;margin:0 .4rem;padding:.2rem 0 .6rem}@media screen and (min-width:45em){.md-social{padding:.6rem 0}}.md-social__link{display:inline-block;height:1.6rem;text-align:center;width:1.6rem}.md-social__link:before{line-height:1.9}.md-social__link svg{fill:currentcolor;max-height:.8rem;vertical-align:-25%}.md-typeset .md-button{border:.1rem solid;border-radius:.1rem;color:var(--md-primary-fg-color);cursor:pointer;display:inline-block;font-weight:700;padding:.625em 2em;transition:color 125ms,background-color 125ms,border-color 125ms}.md-typeset .md-button--primary{background-color:var(--md-primary-fg-color);border-color:var(--md-primary-fg-color);color:var(--md-primary-bg-color)}.md-typeset .md-button:focus,.md-typeset .md-button:hover{background-color:var(--md-accent-fg-color);border-color:var(--md-accent-fg-color);color:var(--md-accent-bg-color)}[dir=ltr] .md-typeset .md-input{border-top-left-radius:.1rem}[dir=ltr] .md-typeset .md-input,[dir=rtl] .md-typeset .md-input{border-top-right-radius:.1rem}[dir=rtl] .md-typeset .md-input{border-top-left-radius:.1rem}.md-typeset .md-input{border-bottom:.1rem solid var(--md-default-fg-color--lighter);box-shadow:var(--md-shadow-z1);font-size:.8rem;height:1.8rem;padding:0 .6rem;transition:border .25s,box-shadow .25s}.md-typeset .md-input:focus,.md-typeset .md-input:hover{border-bottom-color:var(--md-accent-fg-color);box-shadow:var(--md-shadow-z2)}.md-typeset .md-input--stretch{width:100%}.md-header{background-color:var(--md-primary-fg-color);box-shadow:0 0 .2rem #0000,0 .2rem .4rem #0000;color:var(--md-primary-bg-color);display:block;left:0;position:sticky;right:0;top:0;z-index:4}@media print{.md-header{display:none}}.md-header[hidden]{transform:translateY(-100%);transition:transform .25s cubic-bezier(.8,0,.6,1),box-shadow .25s}.md-header--shadow{box-shadow:0 0 .2rem #0000001a,0 .2rem .4rem #0003;transition:transform .25s cubic-bezier(.1,.7,.1,1),box-shadow .25s}.md-header__inner{align-items:center;display:flex;padding:0 .2rem}.md-header__button{color:currentcolor;cursor:pointer;margin:.2rem;outline-color:var(--md-accent-fg-color);padding:.4rem;position:relative;transition:opacity .25s;vertical-align:middle;z-index:1}.md-header__button:hover{opacity:.7}.md-header__button:not([hidden]){display:inline-block}.md-header__button:not(.focus-visible){-webkit-tap-highlight-color:transparent;outline:none}.md-header__button.md-logo{margin:.2rem;padding:.4rem}@media screen and (max-width:76.234375em){.md-header__button.md-logo{display:none}}.md-header__button.md-logo img,.md-header__button.md-logo svg{fill:currentcolor;display:block;height:1.2rem;width:auto}@media screen and (min-width:60em){.md-header__button[for=__search]{display:none}}.no-js .md-header__button[for=__search]{display:none}[dir=rtl] .md-header__button[for=__search] svg{transform:scaleX(-1)}@media screen and (min-width:76.25em){.md-header__button[for=__drawer]{display:none}}.md-header__topic{display:flex;max-width:100%;position:absolute;transition:transform .4s cubic-bezier(.1,.7,.1,1),opacity .15s;white-space:nowrap}.md-header__topic+.md-header__topic{opacity:0;pointer-events:none;transform:translateX(1.25rem);transition:transform .4s cubic-bezier(1,.7,.1,.1),opacity .15s;z-index:-1}[dir=rtl] .md-header__topic+.md-header__topic{transform:translateX(-1.25rem)}.md-header__topic:first-child{font-weight:700}[dir=ltr] .md-header__title{margin-left:1rem;margin-right:.4rem}[dir=rtl] .md-header__title{margin-left:.4rem;margin-right:1rem}.md-header__title{flex-grow:1;font-size:.9rem;height:2.4rem;line-height:2.4rem}.md-header__title--active .md-header__topic{opacity:0;pointer-events:none;transform:translateX(-1.25rem);transition:transform .4s cubic-bezier(1,.7,.1,.1),opacity .15s;z-index:-1}[dir=rtl] .md-header__title--active .md-header__topic{transform:translateX(1.25rem)}.md-header__title--active .md-header__topic+.md-header__topic{opacity:1;pointer-events:auto;transform:translateX(0);transition:transform .4s cubic-bezier(.1,.7,.1,1),opacity .15s;z-index:0}.md-header__title>.md-header__ellipsis{height:100%;position:relative;width:100%}.md-header__option{display:flex;flex-shrink:0;max-width:100%;transition:max-width 0ms .25s,opacity .25s .25s;white-space:nowrap}[data-md-toggle=search]:checked~.md-header .md-header__option{max-width:0;opacity:0;transition:max-width 0ms,opacity 0ms}.md-header__option>input{bottom:0}.md-header__source{display:none}@media screen and (min-width:60em){[dir=ltr] .md-header__source{margin-left:1rem}[dir=rtl] .md-header__source{margin-right:1rem}.md-header__source{display:block;max-width:11.7rem;width:11.7rem}}@media screen and (min-width:76.25em){[dir=ltr] .md-header__source{margin-left:1.4rem}[dir=rtl] .md-header__source{margin-right:1.4rem}}.md-meta{color:var(--md-default-fg-color--light);font-size:.7rem;line-height:1.3}.md-meta__list{display:inline-flex;flex-wrap:wrap;list-style:none;margin:0;padding:0}.md-meta__item:not(:last-child):after{content:"·";margin-left:.2rem;margin-right:.2rem}.md-meta__link{color:var(--md-typeset-a-color)}.md-meta__link:focus,.md-meta__link:hover{color:var(--md-accent-fg-color)}.md-draft{background-color:#ff1744;border-radius:.125em;color:#fff;display:inline-block;font-weight:700;padding-left:.5714285714em;padding-right:.5714285714em}:root{--md-nav-icon--prev:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M20 11v2H8l5.5 5.5-1.42 1.42L4.16 12l7.92-7.92L13.5 5.5 8 11h12Z"/></svg>');--md-nav-icon--next:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M8.59 16.58 13.17 12 8.59 7.41 10 6l6 6-6 6-1.41-1.42Z"/></svg>');--md-toc-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M3 9h14V7H3v2m0 4h14v-2H3v2m0 4h14v-2H3v2m16 0h2v-2h-2v2m0-10v2h2V7h-2m0 6h2v-2h-2v2Z"/></svg>')}.md-nav{font-size:.7rem;line-height:1.3}.md-nav__title{color:var(--md-default-fg-color--light);display:block;font-weight:700;overflow:hidden;padding:0 .6rem;text-overflow:ellipsis}.md-nav__title .md-nav__button{display:none}.md-nav__title .md-nav__button img{height:100%;width:auto}.md-nav__title .md-nav__button.md-logo img,.md-nav__title .md-nav__button.md-logo svg{fill:currentcolor;display:block;height:2.4rem;max-width:100%;object-fit:contain;width:auto}.md-nav__list{list-style:none;margin:0;padding:0}.md-nav__link{align-items:flex-start;display:flex;gap:.4rem;margin-top:.625em;scroll-snap-align:start;transition:color 125ms}.md-nav__link--passed{color:var(--md-default-fg-color--light)}.md-nav__item .md-nav__link--active,.md-nav__item .md-nav__link--active code{color:var(--md-typeset-a-color)}.md-nav__link .md-ellipsis{position:relative}[dir=ltr] .md-nav__link .md-icon:last-child{margin-left:auto}[dir=rtl] .md-nav__link .md-icon:last-child{margin-right:auto}.md-nav__link svg{fill:currentcolor;flex-shrink:0;height:1.3em}.md-nav__link[for]:focus,.md-nav__link[for]:hover,.md-nav__link[href]:focus,.md-nav__link[href]:hover{color:var(--md-accent-fg-color);cursor:pointer}.md-nav__link.focus-visible{outline-color:var(--md-accent-fg-color);outline-offset:.2rem}.md-nav--primary .md-nav__link[for=__toc]{display:none}.md-nav--primary .md-nav__link[for=__toc] .md-icon:after{background-color:currentcolor;display:block;height:100%;-webkit-mask-image:var(--md-toc-icon);mask-image:var(--md-toc-icon);width:100%}.md-nav--primary .md-nav__link[for=__toc]~.md-nav{display:none}.md-nav__container>.md-nav__link{margin-top:0}.md-nav__container>.md-nav__link:first-child{flex-grow:1;min-width:0}.md-nav__icon{flex-shrink:0}.md-nav__source{display:none}@media screen and (max-width:76.234375em){.md-nav--primary,.md-nav--primary .md-nav{background-color:var(--md-default-bg-color);display:flex;flex-direction:column;height:100%;left:0;position:absolute;right:0;top:0;z-index:1}.md-nav--primary .md-nav__item,.md-nav--primary .md-nav__title{font-size:.8rem;line-height:1.5}.md-nav--primary .md-nav__title{background-color:var(--md-default-fg-color--lightest);color:var(--md-default-fg-color--light);cursor:pointer;height:5.6rem;line-height:2.4rem;padding:3rem .8rem .2rem;position:relative;white-space:nowrap}[dir=ltr] .md-nav--primary .md-nav__title .md-nav__icon{left:.4rem}[dir=rtl] .md-nav--primary .md-nav__title .md-nav__icon{right:.4rem}.md-nav--primary .md-nav__title .md-nav__icon{display:block;height:1.2rem;margin:.2rem;position:absolute;top:.4rem;width:1.2rem}.md-nav--primary .md-nav__title .md-nav__icon:after{background-color:currentcolor;content:"";display:block;height:100%;-webkit-mask-image:var(--md-nav-icon--prev);mask-image:var(--md-nav-icon--prev);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;width:100%}.md-nav--primary .md-nav__title~.md-nav__list{background-color:var(--md-default-bg-color);box-shadow:0 .05rem 0 var(--md-default-fg-color--lightest) inset;overflow-y:auto;scroll-snap-type:y mandatory;touch-action:pan-y}.md-nav--primary .md-nav__title~.md-nav__list>:first-child{border-top:0}.md-nav--primary .md-nav__title[for=__drawer]{background-color:var(--md-primary-fg-color);color:var(--md-primary-bg-color);font-weight:700}.md-nav--primary .md-nav__title .md-logo{display:block;left:.2rem;margin:.2rem;padding:.4rem;position:absolute;right:.2rem;top:.2rem}.md-nav--primary .md-nav__list{flex:1}.md-nav--primary .md-nav__item{border-top:.05rem solid var(--md-default-fg-color--lightest)}.md-nav--primary .md-nav__item--active>.md-nav__link{color:var(--md-typeset-a-color)}.md-nav--primary .md-nav__item--active>.md-nav__link:focus,.md-nav--primary .md-nav__item--active>.md-nav__link:hover{color:var(--md-accent-fg-color)}.md-nav--primary .md-nav__link{margin-top:0;padding:.6rem .8rem}.md-nav--primary .md-nav__link svg{margin-top:.1em}.md-nav--primary .md-nav__link>.md-nav__link{padding:0}[dir=ltr] .md-nav--primary .md-nav__link .md-nav__icon{margin-right:-.2rem}[dir=rtl] .md-nav--primary .md-nav__link .md-nav__icon{margin-left:-.2rem}.md-nav--primary .md-nav__link .md-nav__icon{font-size:1.2rem;height:1.2rem;width:1.2rem}.md-nav--primary .md-nav__link .md-nav__icon:after{background-color:currentcolor;content:"";display:block;height:100%;-webkit-mask-image:var(--md-nav-icon--next);mask-image:var(--md-nav-icon--next);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;width:100%}[dir=rtl] .md-nav--primary .md-nav__icon:after{transform:scale(-1)}.md-nav--primary .md-nav--secondary .md-nav{background-color:initial;position:static}[dir=ltr] .md-nav--primary .md-nav--secondary .md-nav .md-nav__link{padding-left:1.4rem}[dir=rtl] .md-nav--primary .md-nav--secondary .md-nav .md-nav__link{padding-right:1.4rem}[dir=ltr] .md-nav--primary .md-nav--secondary .md-nav .md-nav .md-nav__link{padding-left:2rem}[dir=rtl] .md-nav--primary .md-nav--secondary .md-nav .md-nav .md-nav__link{padding-right:2rem}[dir=ltr] .md-nav--primary .md-nav--secondary .md-nav .md-nav .md-nav .md-nav__link{padding-left:2.6rem}[dir=rtl] .md-nav--primary .md-nav--secondary .md-nav .md-nav .md-nav .md-nav__link{padding-right:2.6rem}[dir=ltr] .md-nav--primary .md-nav--secondary .md-nav .md-nav .md-nav .md-nav .md-nav__link{padding-left:3.2rem}[dir=rtl] .md-nav--primary .md-nav--secondary .md-nav .md-nav .md-nav .md-nav .md-nav__link{padding-right:3.2rem}.md-nav--secondary{background-color:initial}.md-nav__toggle~.md-nav{display:flex;opacity:0;transform:translateX(100%);transition:transform .25s cubic-bezier(.8,0,.6,1),opacity 125ms 50ms}[dir=rtl] .md-nav__toggle~.md-nav{transform:translateX(-100%)}.md-nav__toggle:checked~.md-nav{opacity:1;transform:translateX(0);transition:transform .25s cubic-bezier(.4,0,.2,1),opacity 125ms 125ms}.md-nav__toggle:checked~.md-nav>.md-nav__list{-webkit-backface-visibility:hidden;backface-visibility:hidden}}@media screen and (max-width:59.984375em){.md-nav--primary .md-nav__link[for=__toc]{display:flex}.md-nav--primary .md-nav__link[for=__toc] .md-icon:after{content:""}.md-nav--primary .md-nav__link[for=__toc]+.md-nav__link{display:none}.md-nav--primary .md-nav__link[for=__toc]~.md-nav{display:flex}.md-nav__source{background-color:var(--md-primary-fg-color--dark);color:var(--md-primary-bg-color);display:block;padding:0 .2rem}}@media screen and (min-width:60em) and (max-width:76.234375em){.md-nav--integrated .md-nav__link[for=__toc]{display:flex}.md-nav--integrated .md-nav__link[for=__toc] .md-icon:after{content:""}.md-nav--integrated .md-nav__link[for=__toc]+.md-nav__link{display:none}.md-nav--integrated .md-nav__link[for=__toc]~.md-nav{display:flex}}@media screen and (min-width:60em){.md-nav{margin-bottom:-.4rem}.md-nav--secondary .md-nav__title{background:var(--md-default-bg-color);box-shadow:0 0 .4rem .4rem var(--md-default-bg-color);position:sticky;top:0;z-index:1}.md-nav--secondary .md-nav__title[for=__toc]{scroll-snap-align:start}.md-nav--secondary .md-nav__title .md-nav__icon{display:none}[dir=ltr] .md-nav--secondary .md-nav__list{padding-left:.6rem}[dir=rtl] .md-nav--secondary .md-nav__list{padding-right:.6rem}.md-nav--secondary .md-nav__list{padding-bottom:.4rem}[dir=ltr] .md-nav--secondary .md-nav__item>.md-nav__link{margin-right:.4rem}[dir=rtl] .md-nav--secondary .md-nav__item>.md-nav__link{margin-left:.4rem}}@media screen and (min-width:76.25em){.md-nav{margin-bottom:-.4rem;transition:max-height .25s cubic-bezier(.86,0,.07,1)}.md-nav--primary .md-nav__title{background:var(--md-default-bg-color);box-shadow:0 0 .4rem .4rem var(--md-default-bg-color);position:sticky;top:0;z-index:1}.md-nav--primary .md-nav__title[for=__drawer]{scroll-snap-align:start}.md-nav--primary .md-nav__title .md-nav__icon{display:none}[dir=ltr] .md-nav--primary .md-nav__list{padding-left:.6rem}[dir=rtl] .md-nav--primary .md-nav__list{padding-right:.6rem}.md-nav--primary .md-nav__list{padding-bottom:.4rem}[dir=ltr] .md-nav--primary .md-nav__item>.md-nav__link{margin-right:.4rem}[dir=rtl] .md-nav--primary .md-nav__item>.md-nav__link{margin-left:.4rem}.md-nav__toggle~.md-nav{display:grid;grid-template-rows:0fr;opacity:0;transition:grid-template-rows .25s cubic-bezier(.86,0,.07,1),opacity .25s,visibility 0ms .25s;visibility:collapse}.md-nav__toggle~.md-nav>.md-nav__list{overflow:hidden}.md-nav__toggle.md-toggle--indeterminate~.md-nav,.md-nav__toggle:checked~.md-nav{grid-template-rows:1fr;opacity:1;transition:grid-template-rows .25s cubic-bezier(.86,0,.07,1),opacity .15s .1s,visibility 0ms;visibility:visible}.md-nav__toggle.md-toggle--indeterminate~.md-nav{transition:none}.md-nav__item--nested>.md-nav>.md-nav__title{display:none}.md-nav__item--section{display:block;margin:1.25em 0}.md-nav__item--section:last-child{margin-bottom:0}.md-nav__item--section>.md-nav__link{font-weight:700}.md-nav__item--section>.md-nav__link[for]{color:var(--md-default-fg-color--light)}.md-nav__item--section>.md-nav__link:not(.md-nav__container){pointer-events:none}.md-nav__item--section>.md-nav__link .md-icon,.md-nav__item--section>.md-nav__link>[for]{display:none}[dir=ltr] .md-nav__item--section>.md-nav{margin-left:-.6rem}[dir=rtl] .md-nav__item--section>.md-nav{margin-right:-.6rem}.md-nav__item--section>.md-nav{display:block;opacity:1;visibility:visible}.md-nav__item--section>.md-nav>.md-nav__list>.md-nav__item{padding:0}.md-nav__icon{border-radius:100%;height:.9rem;transition:background-color .25s;width:.9rem}.md-nav__icon:hover{background-color:var(--md-accent-fg-color--transparent)}.md-nav__icon:after{background-color:currentcolor;border-radius:100%;content:"";display:inline-block;height:100%;-webkit-mask-image:var(--md-nav-icon--next);mask-image:var(--md-nav-icon--next);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;transition:transform .25s;vertical-align:-.1rem;width:100%}[dir=rtl] .md-nav__icon:after{transform:rotate(180deg)}.md-nav__item--nested .md-nav__toggle:checked~.md-nav__link .md-nav__icon:after,.md-nav__item--nested .md-toggle--indeterminate~.md-nav__link .md-nav__icon:after{transform:rotate(90deg)}.md-nav--lifted>.md-nav__list>.md-nav__item,.md-nav--lifted>.md-nav__title{display:none}.md-nav--lifted>.md-nav__list>.md-nav__item--active{display:block}.md-nav--lifted>.md-nav__list>.md-nav__item--active>.md-nav__link{background:var(--md-default-bg-color);box-shadow:0 0 .4rem .4rem var(--md-default-bg-color);margin-top:0;position:sticky;top:0;z-index:1}.md-nav--lifted>.md-nav__list>.md-nav__item--active>.md-nav__link:not(.md-nav__container){pointer-events:none}.md-nav--lifted>.md-nav__list>.md-nav__item--active.md-nav__item--section{margin:0}[dir=ltr] .md-nav--lifted>.md-nav__list>.md-nav__item>.md-nav{margin-left:-.6rem}[dir=rtl] .md-nav--lifted>.md-nav__list>.md-nav__item>.md-nav{margin-right:-.6rem}.md-nav--lifted>.md-nav__list>.md-nav__item>[for]{color:var(--md-default-fg-color--light)}.md-nav--lifted .md-nav[data-md-level="1"]{grid-template-rows:1fr;opacity:1;visibility:visible}.md-nav--integrated>.md-nav__list>.md-nav__item--active:not(.md-nav__item--nested){padding:0 .6rem}.md-nav--integrated>.md-nav__list>.md-nav__item--active:not(.md-nav__item--nested)>.md-nav__link{padding:0}[dir=ltr] .md-nav--integrated>.md-nav__list>.md-nav__item--active .md-nav--secondary{border-left:.05rem solid var(--md-primary-fg-color)}[dir=rtl] .md-nav--integrated>.md-nav__list>.md-nav__item--active .md-nav--secondary{border-right:.05rem solid var(--md-primary-fg-color)}.md-nav--integrated>.md-nav__list>.md-nav__item--active .md-nav--secondary{display:block;margin-bottom:1.25em;opacity:1;visibility:visible}.md-nav--integrated>.md-nav__list>.md-nav__item--active .md-nav--secondary>.md-nav__list{overflow:visible;padding-bottom:0}.md-nav--integrated>.md-nav__list>.md-nav__item--active .md-nav--secondary>.md-nav__title{display:none}}.md-pagination{font-size:.8rem;font-weight:700;gap:.4rem}.md-pagination,.md-pagination>*{align-items:center;display:flex;justify-content:center}.md-pagination>*{border-radius:.2rem;height:1.8rem;min-width:1.8rem;text-align:center}.md-pagination__current{background-color:var(--md-default-fg-color--lightest);color:var(--md-default-fg-color--light)}.md-pagination__link{transition:color 125ms,background-color 125ms}.md-pagination__link:focus,.md-pagination__link:hover{background-color:var(--md-accent-fg-color--transparent);color:var(--md-accent-fg-color)}.md-pagination__link:focus svg,.md-pagination__link:hover svg{color:var(--md-accent-fg-color)}.md-pagination__link.focus-visible{outline-color:var(--md-accent-fg-color);outline-offset:.2rem}.md-pagination__link svg{fill:currentcolor;color:var(--md-default-fg-color--lighter);display:block;max-height:100%;width:1.2rem}.md-post__back{border-bottom:.05rem solid var(--md-default-fg-color--lightest);margin-bottom:1.2rem;padding-bottom:1.2rem}@media screen and (max-width:76.234375em){.md-post__back{display:none}}[dir=rtl] .md-post__back svg{transform:scaleX(-1)}.md-post__authors{display:flex;flex-direction:column;gap:.6rem;margin:0 .6rem 1.2rem}.md-post .md-post__meta a{transition:color 125ms}.md-post .md-post__meta a:focus,.md-post .md-post__meta a:hover{color:var(--md-accent-fg-color)}.md-post__title{color:var(--md-default-fg-color--light);font-weight:700}.md-post--excerpt{margin-bottom:3.2rem}.md-post--excerpt .md-post__header{align-items:center;display:flex;gap:.6rem;min-height:1.6rem}.md-post--excerpt .md-post__authors{align-items:center;display:inline-flex;flex-direction:row;gap:.2rem;margin:0;min-height:2.4rem}[dir=ltr] .md-post--excerpt .md-post__meta .md-meta__list{margin-right:.4rem}[dir=rtl] .md-post--excerpt .md-post__meta .md-meta__list{margin-left:.4rem}.md-post--excerpt .md-post__content>:first-child{--md-scroll-margin:6rem;margin-top:0}.md-post>.md-nav--secondary{margin:1em 0}.md-profile{align-items:center;display:flex;font-size:.7rem;gap:.6rem;line-height:1.4;width:100%}.md-profile__description{flex-grow:1}.md-content--post{display:flex}@media screen and (max-width:76.234375em){.md-content--post{flex-flow:column-reverse}}.md-content--post>.md-content__inner{min-width:0}@media screen and (min-width:76.25em){[dir=ltr] .md-content--post>.md-content__inner{margin-left:1.2rem}[dir=rtl] .md-content--post>.md-content__inner{margin-right:1.2rem}}@media screen and (max-width:76.234375em){.md-sidebar.md-sidebar--post{padding:0;position:static;width:100%}.md-sidebar.md-sidebar--post .md-sidebar__scrollwrap{overflow:visible}.md-sidebar.md-sidebar--post .md-sidebar__inner{padding:0}.md-sidebar.md-sidebar--post .md-post__meta{margin-left:.6rem;margin-right:.6rem}.md-sidebar.md-sidebar--post .md-nav__item{border:none;display:inline}.md-sidebar.md-sidebar--post .md-nav__list{display:inline-flex;flex-wrap:wrap;gap:.6rem;padding-bottom:.6rem;padding-top:.6rem}.md-sidebar.md-sidebar--post .md-nav__link{padding:0}.md-sidebar.md-sidebar--post .md-nav{height:auto;margin-bottom:0;position:static}}:root{--md-progress-value:0;--md-progress-delay:400ms}.md-progress{background:var(--md-primary-bg-color);height:.075rem;opacity:min(clamp(0,var(--md-progress-value),1),clamp(0,100 - var(--md-progress-value),1));position:fixed;top:0;transform:scaleX(calc(var(--md-progress-value)*1%));transform-origin:left;transition:transform .5s cubic-bezier(.19,1,.22,1),opacity .25s var(--md-progress-delay);width:100%;z-index:4}:root{--md-search-result-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M14 2H6a2 2 0 0 0-2 2v16a2 2 0 0 0 2 2h7c-.41-.25-.8-.56-1.14-.9-.33-.33-.61-.7-.86-1.1H6V4h7v5h5v1.18c.71.16 1.39.43 2 .82V8l-6-6m6.31 16.9c1.33-2.11.69-4.9-1.4-6.22-2.11-1.33-4.91-.68-6.22 1.4-1.34 2.11-.69 4.89 1.4 6.22 1.46.93 3.32.93 4.79.02L22 23.39 23.39 22l-3.08-3.1m-3.81.1a2.5 2.5 0 0 1-2.5-2.5 2.5 2.5 0 0 1 2.5-2.5 2.5 2.5 0 0 1 2.5 2.5 2.5 2.5 0 0 1-2.5 2.5Z"/></svg>')}.md-search{position:relative}@media screen and (min-width:60em){.md-search{padding:.2rem 0}}.no-js .md-search{display:none}.md-search__overlay{opacity:0;z-index:1}@media screen and (max-width:59.984375em){[dir=ltr] .md-search__overlay{left:-2.2rem}[dir=rtl] .md-search__overlay{right:-2.2rem}.md-search__overlay{background-color:var(--md-default-bg-color);border-radius:1rem;height:2rem;overflow:hidden;pointer-events:none;position:absolute;top:-1rem;transform-origin:center;transition:transform .3s .1s,opacity .2s .2s;width:2rem}[data-md-toggle=search]:checked~.md-header .md-search__overlay{opacity:1;transition:transform .4s,opacity .1s}}@media screen and (min-width:60em){[dir=ltr] .md-search__overlay{left:0}[dir=rtl] .md-search__overlay{right:0}.md-search__overlay{background-color:#0000008a;cursor:pointer;height:0;position:fixed;top:0;transition:width 0ms .25s,height 0ms .25s,opacity .25s;width:0}[data-md-toggle=search]:checked~.md-header .md-search__overlay{height:200vh;opacity:1;transition:width 0ms,height 0ms,opacity .25s;width:100%}}@media screen and (max-width:29.984375em){[data-md-toggle=search]:checked~.md-header .md-search__overlay{transform:scale(45)}}@media screen and (min-width:30em) and (max-width:44.984375em){[data-md-toggle=search]:checked~.md-header .md-search__overlay{transform:scale(60)}}@media screen and (min-width:45em) and (max-width:59.984375em){[data-md-toggle=search]:checked~.md-header .md-search__overlay{transform:scale(75)}}.md-search__inner{-webkit-backface-visibility:hidden;backface-visibility:hidden}@media screen and (max-width:59.984375em){[dir=ltr] .md-search__inner{left:0}[dir=rtl] .md-search__inner{right:0}.md-search__inner{height:0;opacity:0;overflow:hidden;position:fixed;top:0;transform:translateX(5%);transition:width 0ms .3s,height 0ms .3s,transform .15s cubic-bezier(.4,0,.2,1) .15s,opacity .15s .15s;width:0;z-index:2}[dir=rtl] .md-search__inner{transform:translateX(-5%)}[data-md-toggle=search]:checked~.md-header .md-search__inner{height:100%;opacity:1;transform:translateX(0);transition:width 0ms 0ms,height 0ms 0ms,transform .15s cubic-bezier(.1,.7,.1,1) .15s,opacity .15s .15s;width:100%}}@media screen and (min-width:60em){[dir=ltr] .md-search__inner{float:right}[dir=rtl] .md-search__inner{float:left}.md-search__inner{padding:.1rem 0;position:relative;transition:width .25s cubic-bezier(.1,.7,.1,1);width:11.7rem}}@media screen and (min-width:60em) and (max-width:76.234375em){[data-md-toggle=search]:checked~.md-header .md-search__inner{width:23.4rem}}@media screen and (min-width:76.25em){[data-md-toggle=search]:checked~.md-header .md-search__inner{width:34.4rem}}.md-search__form{background-color:var(--md-default-bg-color);box-shadow:0 0 .6rem #0000;height:2.4rem;position:relative;transition:color .25s,background-color .25s;z-index:2}@media screen and (min-width:60em){.md-search__form{background-color:#00000042;border-radius:.1rem;height:1.8rem}.md-search__form:hover{background-color:#ffffff1f}}[data-md-toggle=search]:checked~.md-header .md-search__form{background-color:var(--md-default-bg-color);border-radius:.1rem .1rem 0 0;box-shadow:0 0 .6rem #00000012;color:var(--md-default-fg-color)}[dir=ltr] .md-search__input{padding-left:3.6rem;padding-right:2.2rem}[dir=rtl] .md-search__input{padding-left:2.2rem;padding-right:3.6rem}.md-search__input{background:#0000;font-size:.9rem;height:100%;position:relative;text-overflow:ellipsis;width:100%;z-index:2}.md-search__input::placeholder{transition:color .25s}.md-search__input::placeholder,.md-search__input~.md-search__icon{color:var(--md-default-fg-color--light)}.md-search__input::-ms-clear{display:none}@media screen and (max-width:59.984375em){.md-search__input{font-size:.9rem;height:2.4rem;width:100%}}@media screen and (min-width:60em){[dir=ltr] .md-search__input{padding-left:2.2rem}[dir=rtl] .md-search__input{padding-right:2.2rem}.md-search__input{color:inherit;font-size:.8rem}.md-search__input::placeholder{color:var(--md-primary-bg-color--light)}.md-search__input+.md-search__icon{color:var(--md-primary-bg-color)}[data-md-toggle=search]:checked~.md-header .md-search__input{text-overflow:clip}[data-md-toggle=search]:checked~.md-header .md-search__input+.md-search__icon{color:var(--md-default-fg-color--light)}[data-md-toggle=search]:checked~.md-header .md-search__input::placeholder{color:#0000}}.md-search__icon{cursor:pointer;display:inline-block;height:1.2rem;transition:color .25s,opacity .25s;width:1.2rem}.md-search__icon:hover{opacity:.7}[dir=ltr] .md-search__icon[for=__search]{left:.5rem}[dir=rtl] .md-search__icon[for=__search]{right:.5rem}.md-search__icon[for=__search]{position:absolute;top:.3rem;z-index:2}[dir=rtl] .md-search__icon[for=__search] svg{transform:scaleX(-1)}@media screen and (max-width:59.984375em){[dir=ltr] .md-search__icon[for=__search]{left:.8rem}[dir=rtl] .md-search__icon[for=__search]{right:.8rem}.md-search__icon[for=__search]{top:.6rem}.md-search__icon[for=__search] svg:first-child{display:none}}@media screen and (min-width:60em){.md-search__icon[for=__search]{pointer-events:none}.md-search__icon[for=__search] svg:last-child{display:none}}[dir=ltr] .md-search__options{right:.5rem}[dir=rtl] .md-search__options{left:.5rem}.md-search__options{pointer-events:none;position:absolute;top:.3rem;z-index:2}@media screen and (max-width:59.984375em){[dir=ltr] .md-search__options{right:.8rem}[dir=rtl] .md-search__options{left:.8rem}.md-search__options{top:.6rem}}[dir=ltr] .md-search__options>.md-icon{margin-left:.2rem}[dir=rtl] .md-search__options>.md-icon{margin-right:.2rem}.md-search__options>.md-icon{color:var(--md-default-fg-color--light);opacity:0;transform:scale(.75);transition:transform .15s cubic-bezier(.1,.7,.1,1),opacity .15s}.md-search__options>.md-icon:not(.focus-visible){-webkit-tap-highlight-color:transparent;outline:none}[data-md-toggle=search]:checked~.md-header .md-search__input:valid~.md-search__options>.md-icon{opacity:1;pointer-events:auto;transform:scale(1)}[data-md-toggle=search]:checked~.md-header .md-search__input:valid~.md-search__options>.md-icon:hover{opacity:.7}[dir=ltr] .md-search__suggest{padding-left:3.6rem;padding-right:2.2rem}[dir=rtl] .md-search__suggest{padding-left:2.2rem;padding-right:3.6rem}.md-search__suggest{align-items:center;color:var(--md-default-fg-color--lighter);display:flex;font-size:.9rem;height:100%;opacity:0;position:absolute;top:0;transition:opacity 50ms;white-space:nowrap;width:100%}@media screen and (min-width:60em){[dir=ltr] .md-search__suggest{padding-left:2.2rem}[dir=rtl] .md-search__suggest{padding-right:2.2rem}.md-search__suggest{font-size:.8rem}}[data-md-toggle=search]:checked~.md-header .md-search__suggest{opacity:1;transition:opacity .3s .1s}[dir=ltr] .md-search__output{border-bottom-left-radius:.1rem}[dir=ltr] .md-search__output,[dir=rtl] .md-search__output{border-bottom-right-radius:.1rem}[dir=rtl] .md-search__output{border-bottom-left-radius:.1rem}.md-search__output{overflow:hidden;position:absolute;width:100%;z-index:1}@media screen and (max-width:59.984375em){.md-search__output{bottom:0;top:2.4rem}}@media screen and (min-width:60em){.md-search__output{opacity:0;top:1.9rem;transition:opacity .4s}[data-md-toggle=search]:checked~.md-header .md-search__output{box-shadow:var(--md-shadow-z3);opacity:1}}.md-search__scrollwrap{-webkit-backface-visibility:hidden;backface-visibility:hidden;background-color:var(--md-default-bg-color);height:100%;overflow-y:auto;touch-action:pan-y}@media (-webkit-max-device-pixel-ratio:1),(max-resolution:1dppx){.md-search__scrollwrap{transform:translateZ(0)}}@media screen and (min-width:60em) and (max-width:76.234375em){.md-search__scrollwrap{width:23.4rem}}@media screen and (min-width:76.25em){.md-search__scrollwrap{width:34.4rem}}@media screen and (min-width:60em){.md-search__scrollwrap{max-height:0;scrollbar-color:var(--md-default-fg-color--lighter) #0000;scrollbar-width:thin}[data-md-toggle=search]:checked~.md-header .md-search__scrollwrap{max-height:75vh}.md-search__scrollwrap:hover{scrollbar-color:var(--md-accent-fg-color) #0000}.md-search__scrollwrap::-webkit-scrollbar{height:.2rem;width:.2rem}.md-search__scrollwrap::-webkit-scrollbar-thumb{background-color:var(--md-default-fg-color--lighter)}.md-search__scrollwrap::-webkit-scrollbar-thumb:hover{background-color:var(--md-accent-fg-color)}}.md-search-result{color:var(--md-default-fg-color);word-break:break-word}.md-search-result__meta{background-color:var(--md-default-fg-color--lightest);color:var(--md-default-fg-color--light);font-size:.64rem;line-height:1.8rem;padding:0 .8rem;scroll-snap-align:start}@media screen and (min-width:60em){[dir=ltr] .md-search-result__meta{padding-left:2.2rem}[dir=rtl] .md-search-result__meta{padding-right:2.2rem}}.md-search-result__list{list-style:none;margin:0;padding:0;-webkit-user-select:none;user-select:none}.md-search-result__item{box-shadow:0 -.05rem var(--md-default-fg-color--lightest)}.md-search-result__item:first-child{box-shadow:none}.md-search-result__link{display:block;outline:none;scroll-snap-align:start;transition:background-color .25s}.md-search-result__link:focus,.md-search-result__link:hover{background-color:var(--md-accent-fg-color--transparent)}.md-search-result__link:last-child p:last-child{margin-bottom:.6rem}.md-search-result__more>summary{cursor:pointer;display:block;outline:none;position:sticky;scroll-snap-align:start;top:0;z-index:1}.md-search-result__more>summary::marker{display:none}.md-search-result__more>summary::-webkit-details-marker{display:none}.md-search-result__more>summary>div{color:var(--md-typeset-a-color);font-size:.64rem;padding:.75em .8rem;transition:color .25s,background-color .25s}@media screen and (min-width:60em){[dir=ltr] .md-search-result__more>summary>div{padding-left:2.2rem}[dir=rtl] .md-search-result__more>summary>div{padding-right:2.2rem}}.md-search-result__more>summary:focus>div,.md-search-result__more>summary:hover>div{background-color:var(--md-accent-fg-color--transparent);color:var(--md-accent-fg-color)}.md-search-result__more[open]>summary{background-color:var(--md-default-bg-color)}.md-search-result__article{overflow:hidden;padding:0 .8rem;position:relative}@media screen and (min-width:60em){[dir=ltr] .md-search-result__article{padding-left:2.2rem}[dir=rtl] .md-search-result__article{padding-right:2.2rem}}[dir=ltr] .md-search-result__icon{left:0}[dir=rtl] .md-search-result__icon{right:0}.md-search-result__icon{color:var(--md-default-fg-color--light);height:1.2rem;margin:.5rem;position:absolute;width:1.2rem}@media screen and (max-width:59.984375em){.md-search-result__icon{display:none}}.md-search-result__icon:after{background-color:currentcolor;content:"";display:inline-block;height:100%;-webkit-mask-image:var(--md-search-result-icon);mask-image:var(--md-search-result-icon);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;width:100%}[dir=rtl] .md-search-result__icon:after{transform:scaleX(-1)}.md-search-result .md-typeset{color:var(--md-default-fg-color--light);font-size:.64rem;line-height:1.6}.md-search-result .md-typeset h1{color:var(--md-default-fg-color);font-size:.8rem;font-weight:400;line-height:1.4;margin:.55rem 0}.md-search-result .md-typeset h1 mark{text-decoration:none}.md-search-result .md-typeset h2{color:var(--md-default-fg-color);font-size:.64rem;font-weight:700;line-height:1.6;margin:.5em 0}.md-search-result .md-typeset h2 mark{text-decoration:none}.md-search-result__terms{color:var(--md-default-fg-color);display:block;font-size:.64rem;font-style:italic;margin:.5em 0}.md-search-result mark{background-color:initial;color:var(--md-accent-fg-color);text-decoration:underline}.md-select{position:relative;z-index:1}.md-select__inner{background-color:var(--md-default-bg-color);border-radius:.1rem;box-shadow:var(--md-shadow-z2);color:var(--md-default-fg-color);left:50%;margin-top:.2rem;max-height:0;opacity:0;position:absolute;top:calc(100% - .2rem);transform:translate3d(-50%,.3rem,0);transition:transform .25s 375ms,opacity .25s .25s,max-height 0ms .5s}.md-select:focus-within .md-select__inner,.md-select:hover .md-select__inner{max-height:10rem;opacity:1;transform:translate3d(-50%,0,0);transition:transform .25s cubic-bezier(.1,.7,.1,1),opacity .25s,max-height 0ms}.md-select__inner:after{border-bottom:.2rem solid #0000;border-bottom-color:var(--md-default-bg-color);border-left:.2rem solid #0000;border-right:.2rem solid #0000;border-top:0;content:"";height:0;left:50%;margin-left:-.2rem;margin-top:-.2rem;position:absolute;top:0;width:0}.md-select__list{border-radius:.1rem;font-size:.8rem;list-style-type:none;margin:0;max-height:inherit;overflow:auto;padding:0}.md-select__item{line-height:1.8rem}[dir=ltr] .md-select__link{padding-left:.6rem;padding-right:1.2rem}[dir=rtl] .md-select__link{padding-left:1.2rem;padding-right:.6rem}.md-select__link{cursor:pointer;display:block;outline:none;scroll-snap-align:start;transition:background-color .25s,color .25s;width:100%}.md-select__link:focus,.md-select__link:hover{color:var(--md-accent-fg-color)}.md-select__link:focus{background-color:var(--md-default-fg-color--lightest)}.md-sidebar{align-self:flex-start;flex-shrink:0;padding:1.2rem 0;position:sticky;top:2.4rem;width:12.1rem}@media print{.md-sidebar{display:none}}@media screen and (max-width:76.234375em){[dir=ltr] .md-sidebar--primary{left:-12.1rem}[dir=rtl] .md-sidebar--primary{right:-12.1rem}.md-sidebar--primary{background-color:var(--md-default-bg-color);display:block;height:100%;position:fixed;top:0;transform:translateX(0);transition:transform .25s cubic-bezier(.4,0,.2,1),box-shadow .25s;width:12.1rem;z-index:5}[data-md-toggle=drawer]:checked~.md-container .md-sidebar--primary{box-shadow:var(--md-shadow-z3);transform:translateX(12.1rem)}[dir=rtl] [data-md-toggle=drawer]:checked~.md-container .md-sidebar--primary{transform:translateX(-12.1rem)}.md-sidebar--primary .md-sidebar__scrollwrap{bottom:0;left:0;margin:0;overflow:hidden;position:absolute;right:0;scroll-snap-type:none;top:0}}@media screen and (min-width:76.25em){.md-sidebar{height:0}.no-js .md-sidebar{height:auto}.md-header--lifted~.md-container .md-sidebar{top:4.8rem}}.md-sidebar--secondary{display:none;order:2}@media screen and (min-width:60em){.md-sidebar--secondary{height:0}.no-js .md-sidebar--secondary{height:auto}.md-sidebar--secondary:not([hidden]){display:block}.md-sidebar--secondary .md-sidebar__scrollwrap{touch-action:pan-y}}.md-sidebar__scrollwrap{scrollbar-gutter:stable;-webkit-backface-visibility:hidden;backface-visibility:hidden;margin:0 .2rem;overflow-y:auto;scrollbar-color:var(--md-default-fg-color--lighter) #0000;scrollbar-width:thin}.md-sidebar__scrollwrap::-webkit-scrollbar{height:.2rem;width:.2rem}.md-sidebar__scrollwrap:focus-within,.md-sidebar__scrollwrap:hover{scrollbar-color:var(--md-accent-fg-color) #0000}.md-sidebar__scrollwrap:focus-within::-webkit-scrollbar-thumb,.md-sidebar__scrollwrap:hover::-webkit-scrollbar-thumb{background-color:var(--md-default-fg-color--lighter)}.md-sidebar__scrollwrap:focus-within::-webkit-scrollbar-thumb:hover,.md-sidebar__scrollwrap:hover::-webkit-scrollbar-thumb:hover{background-color:var(--md-accent-fg-color)}@supports selector(::-webkit-scrollbar){.md-sidebar__scrollwrap{scrollbar-gutter:auto}[dir=ltr] .md-sidebar__inner{padding-right:calc(100% - 11.5rem)}[dir=rtl] .md-sidebar__inner{padding-left:calc(100% - 11.5rem)}}@media screen and (max-width:76.234375em){.md-overlay{background-color:#0000008a;height:0;opacity:0;position:fixed;top:0;transition:width 0ms .25s,height 0ms .25s,opacity .25s;width:0;z-index:5}[data-md-toggle=drawer]:checked~.md-overlay{height:100%;opacity:1;transition:width 0ms,height 0ms,opacity .25s;width:100%}}@keyframes facts{0%{height:0}to{height:.65rem}}@keyframes fact{0%{opacity:0;transform:translateY(100%)}50%{opacity:0}to{opacity:1;transform:translateY(0)}}:root{--md-source-forks-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 16 16"><path d="M5 5.372v.878c0 .414.336.75.75.75h4.5a.75.75 0 0 0 .75-.75v-.878a2.25 2.25 0 1 1 1.5 0v.878a2.25 2.25 0 0 1-2.25 2.25h-1.5v2.128a2.251 2.251 0 1 1-1.5 0V8.5h-1.5A2.25 2.25 0 0 1 3.5 6.25v-.878a2.25 2.25 0 1 1 1.5 0ZM5 3.25a.75.75 0 1 0-1.5 0 .75.75 0 0 0 1.5 0Zm6.75.75a.75.75 0 1 0 0-1.5.75.75 0 0 0 0 1.5Zm-3 8.75a.75.75 0 1 0-1.5 0 .75.75 0 0 0 1.5 0Z"/></svg>');--md-source-repositories-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 16 16"><path d="M2 2.5A2.5 2.5 0 0 1 4.5 0h8.75a.75.75 0 0 1 .75.75v12.5a.75.75 0 0 1-.75.75h-2.5a.75.75 0 0 1 0-1.5h1.75v-2h-8a1 1 0 0 0-.714 1.7.75.75 0 1 1-1.072 1.05A2.495 2.495 0 0 1 2 11.5Zm10.5-1h-8a1 1 0 0 0-1 1v6.708A2.486 2.486 0 0 1 4.5 9h8ZM5 12.25a.25.25 0 0 1 .25-.25h3.5a.25.25 0 0 1 .25.25v3.25a.25.25 0 0 1-.4.2l-1.45-1.087a.249.249 0 0 0-.3 0L5.4 15.7a.25.25 0 0 1-.4-.2Z"/></svg>');--md-source-stars-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 16 16"><path d="M8 .25a.75.75 0 0 1 .673.418l1.882 3.815 4.21.612a.75.75 0 0 1 .416 1.279l-3.046 2.97.719 4.192a.751.751 0 0 1-1.088.791L8 12.347l-3.766 1.98a.75.75 0 0 1-1.088-.79l.72-4.194L.818 6.374a.75.75 0 0 1 .416-1.28l4.21-.611L7.327.668A.75.75 0 0 1 8 .25Zm0 2.445L6.615 5.5a.75.75 0 0 1-.564.41l-3.097.45 2.24 2.184a.75.75 0 0 1 .216.664l-.528 3.084 2.769-1.456a.75.75 0 0 1 .698 0l2.77 1.456-.53-3.084a.75.75 0 0 1 .216-.664l2.24-2.183-3.096-.45a.75.75 0 0 1-.564-.41L8 2.694Z"/></svg>');--md-source-version-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 16 16"><path d="M1 7.775V2.75C1 1.784 1.784 1 2.75 1h5.025c.464 0 .91.184 1.238.513l6.25 6.25a1.75 1.75 0 0 1 0 2.474l-5.026 5.026a1.75 1.75 0 0 1-2.474 0l-6.25-6.25A1.752 1.752 0 0 1 1 7.775Zm1.5 0c0 .066.026.13.073.177l6.25 6.25a.25.25 0 0 0 .354 0l5.025-5.025a.25.25 0 0 0 0-.354l-6.25-6.25a.25.25 0 0 0-.177-.073H2.75a.25.25 0 0 0-.25.25ZM6 5a1 1 0 1 1 0 2 1 1 0 0 1 0-2Z"/></svg>')}.md-source{-webkit-backface-visibility:hidden;backface-visibility:hidden;display:block;font-size:.65rem;line-height:1.2;outline-color:var(--md-accent-fg-color);transition:opacity .25s;white-space:nowrap}.md-source:hover{opacity:.7}.md-source__icon{display:inline-block;height:2.4rem;vertical-align:middle;width:2rem}[dir=ltr] .md-source__icon svg{margin-left:.6rem}[dir=rtl] .md-source__icon svg{margin-right:.6rem}.md-source__icon svg{margin-top:.6rem}[dir=ltr] .md-source__icon+.md-source__repository{padding-left:2rem}[dir=rtl] .md-source__icon+.md-source__repository{padding-right:2rem}[dir=ltr] .md-source__icon+.md-source__repository{margin-left:-2rem}[dir=rtl] .md-source__icon+.md-source__repository{margin-right:-2rem}[dir=ltr] .md-source__repository{margin-left:.6rem}[dir=rtl] .md-source__repository{margin-right:.6rem}.md-source__repository{display:inline-block;max-width:calc(100% - 1.2rem);overflow:hidden;text-overflow:ellipsis;vertical-align:middle}.md-source__facts{display:flex;font-size:.55rem;gap:.4rem;list-style-type:none;margin:.1rem 0 0;opacity:.75;overflow:hidden;padding:0;width:100%}.md-source__repository--active .md-source__facts{animation:facts .25s ease-in}.md-source__fact{overflow:hidden;text-overflow:ellipsis}.md-source__repository--active .md-source__fact{animation:fact .4s ease-out}[dir=ltr] .md-source__fact:before{margin-right:.1rem}[dir=rtl] .md-source__fact:before{margin-left:.1rem}.md-source__fact:before{background-color:currentcolor;content:"";display:inline-block;height:.6rem;-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;vertical-align:text-top;width:.6rem}.md-source__fact:nth-child(1n+2){flex-shrink:0}.md-source__fact--version:before{-webkit-mask-image:var(--md-source-version-icon);mask-image:var(--md-source-version-icon)}.md-source__fact--stars:before{-webkit-mask-image:var(--md-source-stars-icon);mask-image:var(--md-source-stars-icon)}.md-source__fact--forks:before{-webkit-mask-image:var(--md-source-forks-icon);mask-image:var(--md-source-forks-icon)}.md-source__fact--repositories:before{-webkit-mask-image:var(--md-source-repositories-icon);mask-image:var(--md-source-repositories-icon)}:root{--md-status:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M11 9h2V7h-2m1 13c-4.41 0-8-3.59-8-8s3.59-8 8-8 8 3.59 8 8-3.59 8-8 8m0-18A10 10 0 0 0 2 12a10 10 0 0 0 10 10 10 10 0 0 0 10-10A10 10 0 0 0 12 2m-1 15h2v-6h-2v6Z"/></svg>');--md-status--new:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="m23 12-2.44-2.78.34-3.68-3.61-.82-1.89-3.18L12 3 8.6 1.54 6.71 4.72l-3.61.81.34 3.68L1 12l2.44 2.78-.34 3.69 3.61.82 1.89 3.18L12 21l3.4 1.46 1.89-3.18 3.61-.82-.34-3.68L23 12m-10 5h-2v-2h2v2m0-4h-2V7h2v6Z"/></svg>');--md-status--deprecated:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M9 3v1H4v2h1v13a2 2 0 0 0 2 2h10a2 2 0 0 0 2-2V6h1V4h-5V3H9m0 5h2v9H9V8m4 0h2v9h-2V8Z"/></svg>');--md-status--encrypted:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 1 3 5v6c0 5.55 3.84 10.74 9 12 5.16-1.26 9-6.45 9-12V5l-9-4m0 6c1.4 0 2.8 1.1 2.8 2.5V11c.6 0 1.2.6 1.2 1.3v3.5c0 .6-.6 1.2-1.3 1.2H9.2c-.6 0-1.2-.6-1.2-1.3v-3.5c0-.6.6-1.2 1.2-1.2V9.5C9.2 8.1 10.6 7 12 7m0 1.2c-.8 0-1.5.5-1.5 1.3V11h3V9.5c0-.8-.7-1.3-1.5-1.3Z"/></svg>')}.md-status:after{background-color:var(--md-default-fg-color--light);content:"";display:inline-block;height:1.125em;-webkit-mask-image:var(--md-status);mask-image:var(--md-status);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;vertical-align:text-bottom;width:1.125em}.md-status:hover:after{background-color:currentcolor}.md-status--new:after{-webkit-mask-image:var(--md-status--new);mask-image:var(--md-status--new)}.md-status--deprecated:after{-webkit-mask-image:var(--md-status--deprecated);mask-image:var(--md-status--deprecated)}.md-status--encrypted:after{-webkit-mask-image:var(--md-status--encrypted);mask-image:var(--md-status--encrypted)}.md-tabs{background-color:var(--md-primary-fg-color);color:var(--md-primary-bg-color);display:block;line-height:1.3;overflow:auto;width:100%;z-index:3}@media print{.md-tabs{display:none}}@media screen and (max-width:76.234375em){.md-tabs{display:none}}.md-tabs[hidden]{pointer-events:none}[dir=ltr] .md-tabs__list{margin-left:.2rem}[dir=rtl] .md-tabs__list{margin-right:.2rem}.md-tabs__list{contain:content;display:flex;list-style:none;margin:0;overflow:auto;padding:0;scrollbar-width:none;white-space:nowrap}.md-tabs__list::-webkit-scrollbar{display:none}.md-tabs__item{height:2.4rem;padding-left:.6rem;padding-right:.6rem}.md-tabs__item--active .md-tabs__link{color:inherit;opacity:1}.md-tabs__link{-webkit-backface-visibility:hidden;backface-visibility:hidden;display:flex;font-size:.7rem;margin-top:.8rem;opacity:.7;outline-color:var(--md-accent-fg-color);outline-offset:.2rem;transition:transform .4s cubic-bezier(.1,.7,.1,1),opacity .25s}.md-tabs__link:focus,.md-tabs__link:hover{color:inherit;opacity:1}[dir=ltr] .md-tabs__link svg{margin-right:.4rem}[dir=rtl] .md-tabs__link svg{margin-left:.4rem}.md-tabs__link svg{fill:currentcolor;height:1.3em}.md-tabs__item:nth-child(2) .md-tabs__link{transition-delay:20ms}.md-tabs__item:nth-child(3) .md-tabs__link{transition-delay:40ms}.md-tabs__item:nth-child(4) .md-tabs__link{transition-delay:60ms}.md-tabs__item:nth-child(5) .md-tabs__link{transition-delay:80ms}.md-tabs__item:nth-child(6) .md-tabs__link{transition-delay:.1s}.md-tabs__item:nth-child(7) .md-tabs__link{transition-delay:.12s}.md-tabs__item:nth-child(8) .md-tabs__link{transition-delay:.14s}.md-tabs__item:nth-child(9) .md-tabs__link{transition-delay:.16s}.md-tabs__item:nth-child(10) .md-tabs__link{transition-delay:.18s}.md-tabs__item:nth-child(11) .md-tabs__link{transition-delay:.2s}.md-tabs__item:nth-child(12) .md-tabs__link{transition-delay:.22s}.md-tabs__item:nth-child(13) .md-tabs__link{transition-delay:.24s}.md-tabs__item:nth-child(14) .md-tabs__link{transition-delay:.26s}.md-tabs__item:nth-child(15) .md-tabs__link{transition-delay:.28s}.md-tabs__item:nth-child(16) .md-tabs__link{transition-delay:.3s}.md-tabs[hidden] .md-tabs__link{opacity:0;transform:translateY(50%);transition:transform 0ms .1s,opacity .1s}:root{--md-tag-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="m5.41 21 .71-4h-4l.35-2h4l1.06-6h-4l.35-2h4l.71-4h2l-.71 4h6l.71-4h2l-.71 4h4l-.35 2h-4l-1.06 6h4l-.35 2h-4l-.71 4h-2l.71-4h-6l-.71 4h-2M9.53 9l-1.06 6h6l1.06-6h-6Z"/></svg>')}.md-typeset .md-tags:not([hidden]){display:inline-flex;flex-wrap:wrap;gap:.5em;margin-bottom:.75em;margin-top:-.125em}.md-typeset .md-tag{align-items:center;background:var(--md-default-fg-color--lightest);border-radius:2.4rem;display:inline-flex;font-size:.64rem;font-size:min(.8em,.64rem);font-weight:700;gap:.5em;letter-spacing:normal;line-height:1.6;padding:.3125em .78125em}.md-typeset .md-tag[href]{-webkit-tap-highlight-color:transparent;color:inherit;outline:none;transition:color 125ms,background-color 125ms}.md-typeset .md-tag[href]:focus,.md-typeset .md-tag[href]:hover{background-color:var(--md-accent-fg-color);color:var(--md-accent-bg-color)}[id]>.md-typeset .md-tag{vertical-align:text-top}.md-typeset .md-tag-icon:before{background-color:var(--md-default-fg-color--lighter);content:"";display:inline-block;height:1.2em;-webkit-mask-image:var(--md-tag-icon);mask-image:var(--md-tag-icon);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;transition:background-color 125ms;vertical-align:text-bottom;width:1.2em}.md-typeset .md-tag-icon[href]:focus:before,.md-typeset .md-tag-icon[href]:hover:before{background-color:var(--md-accent-bg-color)}@keyframes pulse{0%{transform:scale(.95)}75%{transform:scale(1)}to{transform:scale(.95)}}:root{--md-annotation-bg-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 2A10 10 0 0 0 2 12a10 10 0 0 0 10 10 10 10 0 0 0 10-10A10 10 0 0 0 12 2Z"/></svg>');--md-annotation-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M17 13h-4v4h-2v-4H7v-2h4V7h2v4h4m-5-9A10 10 0 0 0 2 12a10 10 0 0 0 10 10 10 10 0 0 0 10-10A10 10 0 0 0 12 2Z"/></svg>');--md-tooltip-width:20rem}.md-tooltip{-webkit-backface-visibility:hidden;backface-visibility:hidden;background-color:var(--md-default-bg-color);border-radius:.1rem;box-shadow:var(--md-shadow-z2);color:var(--md-default-fg-color);font-family:var(--md-text-font-family);left:clamp(var(--md-tooltip-0,0rem) + .8rem,var(--md-tooltip-x),100vw + var(--md-tooltip-0,0rem) + .8rem - var(--md-tooltip-width) - 2 * .8rem);max-width:calc(100vw - 1.6rem);opacity:0;position:absolute;top:var(--md-tooltip-y);transform:translateY(-.4rem);transition:transform 0ms .25s,opacity .25s,z-index .25s;width:var(--md-tooltip-width);z-index:0}.md-tooltip--active{opacity:1;transform:translateY(0);transition:transform .25s cubic-bezier(.1,.7,.1,1),opacity .25s,z-index 0ms;z-index:2}.focus-visible>.md-tooltip,.md-tooltip:target{outline:var(--md-accent-fg-color) auto}.md-tooltip__inner{font-size:.64rem;padding:.8rem}.md-tooltip__inner.md-typeset>:first-child{margin-top:0}.md-tooltip__inner.md-typeset>:last-child{margin-bottom:0}.md-annotation{font-weight:400;outline:none;vertical-align:text-bottom;white-space:normal}[dir=rtl] .md-annotation{direction:rtl}code .md-annotation{font-family:var(--md-code-font-family);font-size:inherit}.md-annotation:not([hidden]){display:inline-block;line-height:1.25}.md-annotation__index{border-radius:.01px;cursor:pointer;display:inline-block;margin-left:.4ch;margin-right:.4ch;outline:none;overflow:hidden;position:relative;-webkit-user-select:none;user-select:none;vertical-align:text-top;z-index:0}.md-annotation .md-annotation__index{transition:z-index .25s}@media screen{.md-annotation__index{width:2.2ch}[data-md-visible]>.md-annotation__index{animation:pulse 2s infinite}.md-annotation__index:before{background:var(--md-default-bg-color);-webkit-mask-image:var(--md-annotation-bg-icon);mask-image:var(--md-annotation-bg-icon)}.md-annotation__index:after,.md-annotation__index:before{content:"";height:2.2ch;-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;position:absolute;top:-.1ch;width:2.2ch;z-index:-1}.md-annotation__index:after{background-color:var(--md-default-fg-color--lighter);-webkit-mask-image:var(--md-annotation-icon);mask-image:var(--md-annotation-icon);transform:scale(1.0001);transition:background-color .25s,transform .25s}.md-tooltip--active+.md-annotation__index:after{transform:rotate(45deg)}.md-tooltip--active+.md-annotation__index:after,:hover>.md-annotation__index:after{background-color:var(--md-accent-fg-color)}}.md-tooltip--active+.md-annotation__index{animation-play-state:paused;transition-duration:0ms;z-index:2}.md-annotation__index [data-md-annotation-id]{display:inline-block}@media print{.md-annotation__index [data-md-annotation-id]{background:var(--md-default-fg-color--lighter);border-radius:2ch;color:var(--md-default-bg-color);font-weight:700;padding:0 .6ch;white-space:nowrap}.md-annotation__index [data-md-annotation-id]:after{content:attr(data-md-annotation-id)}}.md-typeset .md-annotation-list{counter-reset:xxx;list-style:none}.md-typeset .md-annotation-list li{position:relative}[dir=ltr] .md-typeset .md-annotation-list li:before{left:-2.125em}[dir=rtl] .md-typeset .md-annotation-list li:before{right:-2.125em}.md-typeset .md-annotation-list li:before{background:var(--md-default-fg-color--lighter);border-radius:2ch;color:var(--md-default-bg-color);content:counter(xxx);counter-increment:xxx;font-size:.8875em;font-weight:700;height:2ch;line-height:1.25;min-width:2ch;padding:0 .6ch;position:absolute;text-align:center;top:.25em}[dir=ltr] .md-top{margin-left:50%}[dir=rtl] .md-top{margin-right:50%}.md-top{background-color:var(--md-default-bg-color);border-radius:1.6rem;box-shadow:var(--md-shadow-z2);color:var(--md-default-fg-color--light);cursor:pointer;display:block;font-size:.7rem;outline:none;padding:.4rem .8rem;position:fixed;top:3.2rem;transform:translate(-50%);transition:color 125ms,background-color 125ms,transform 125ms cubic-bezier(.4,0,.2,1),opacity 125ms;z-index:2}@media print{.md-top{display:none}}[dir=rtl] .md-top{transform:translate(50%)}.md-top[hidden]{opacity:0;pointer-events:none;transform:translate(-50%,.2rem);transition-duration:0ms}[dir=rtl] .md-top[hidden]{transform:translate(50%,.2rem)}.md-top:focus,.md-top:hover{background-color:var(--md-accent-fg-color);color:var(--md-accent-bg-color)}.md-top svg{display:inline-block;vertical-align:-.5em}@keyframes hoverfix{0%{pointer-events:none}}:root{--md-version-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 320 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M137.4 374.6c12.5 12.5 32.8 12.5 45.3 0l128-128c9.2-9.2 11.9-22.9 6.9-34.9S301 191.9 288 191.9L32 192c-12.9 0-24.6 7.8-29.6 19.8s-2.2 25.7 6.9 34.9l128 128z"/></svg>')}.md-version{flex-shrink:0;font-size:.8rem;height:2.4rem}[dir=ltr] .md-version__current{margin-left:1.4rem;margin-right:.4rem}[dir=rtl] .md-version__current{margin-left:.4rem;margin-right:1.4rem}.md-version__current{color:inherit;cursor:pointer;outline:none;position:relative;top:.05rem}[dir=ltr] .md-version__current:after{margin-left:.4rem}[dir=rtl] .md-version__current:after{margin-right:.4rem}.md-version__current:after{background-color:currentcolor;content:"";display:inline-block;height:.6rem;-webkit-mask-image:var(--md-version-icon);mask-image:var(--md-version-icon);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;width:.4rem}.md-version__list{background-color:var(--md-default-bg-color);border-radius:.1rem;box-shadow:var(--md-shadow-z2);color:var(--md-default-fg-color);list-style-type:none;margin:.2rem .8rem;max-height:0;opacity:0;overflow:auto;padding:0;position:absolute;scroll-snap-type:y mandatory;top:.15rem;transition:max-height 0ms .5s,opacity .25s .25s;z-index:3}.md-version:focus-within .md-version__list,.md-version:hover .md-version__list{max-height:10rem;opacity:1;transition:max-height 0ms,opacity .25s}@media (hover:none),(pointer:coarse){.md-version:hover .md-version__list{animation:hoverfix .25s forwards}.md-version:focus-within .md-version__list{animation:none}}.md-version__item{line-height:1.8rem}[dir=ltr] .md-version__link{padding-left:.6rem;padding-right:1.2rem}[dir=rtl] .md-version__link{padding-left:1.2rem;padding-right:.6rem}.md-version__link{cursor:pointer;display:block;outline:none;scroll-snap-align:start;transition:color .25s,background-color .25s;white-space:nowrap;width:100%}.md-version__link:focus,.md-version__link:hover{color:var(--md-accent-fg-color)}.md-version__link:focus{background-color:var(--md-default-fg-color--lightest)}:root{--md-admonition-icon--note:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 2C6.47 2 2 6.47 2 12s4.47 10 10 10 10-4.47 10-10S17.53 2 12 2m3.1 5.07c.14 0 .28.05.4.16l1.27 1.27c.23.22.23.57 0 .78l-1 1-2.05-2.05 1-1c.1-.11.24-.16.38-.16m-1.97 1.74 2.06 2.06-6.06 6.06H7.07v-2.06l6.06-6.06Z"/></svg>');--md-admonition-icon--abstract:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M17 9H7V7h10m0 6H7v-2h10m-3 6H7v-2h7M12 3a1 1 0 0 1 1 1 1 1 0 0 1-1 1 1 1 0 0 1-1-1 1 1 0 0 1 1-1m7 0h-4.18C14.4 1.84 13.3 1 12 1c-1.3 0-2.4.84-2.82 2H5a2 2 0 0 0-2 2v14a2 2 0 0 0 2 2h14a2 2 0 0 0 2-2V5a2 2 0 0 0-2-2Z"/></svg>');--md-admonition-icon--info:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M13 9h-2V7h2m0 10h-2v-6h2m-1-9A10 10 0 0 0 2 12a10 10 0 0 0 10 10 10 10 0 0 0 10-10A10 10 0 0 0 12 2Z"/></svg>');--md-admonition-icon--tip:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M17.66 11.2c-.23-.3-.51-.56-.77-.82-.67-.6-1.43-1.03-2.07-1.66C13.33 7.26 13 4.85 13.95 3c-.95.23-1.78.75-2.49 1.32-2.59 2.08-3.61 5.75-2.39 8.9.04.1.08.2.08.33 0 .22-.15.42-.35.5-.23.1-.47.04-.66-.12a.58.58 0 0 1-.14-.17c-1.13-1.43-1.31-3.48-.55-5.12C5.78 10 4.87 12.3 5 14.47c.06.5.12 1 .29 1.5.14.6.41 1.2.71 1.73 1.08 1.73 2.95 2.97 4.96 3.22 2.14.27 4.43-.12 6.07-1.6 1.83-1.66 2.47-4.32 1.53-6.6l-.13-.26c-.21-.46-.77-1.26-.77-1.26m-3.16 6.3c-.28.24-.74.5-1.1.6-1.12.4-2.24-.16-2.9-.82 1.19-.28 1.9-1.16 2.11-2.05.17-.8-.15-1.46-.28-2.23-.12-.74-.1-1.37.17-2.06.19.38.39.76.63 1.06.77 1 1.98 1.44 2.24 2.8.04.14.06.28.06.43.03.82-.33 1.72-.93 2.27Z"/></svg>');--md-admonition-icon--success:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M21 7 9 19l-5.5-5.5 1.41-1.41L9 16.17 19.59 5.59 21 7Z"/></svg>');--md-admonition-icon--question:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="m15.07 11.25-.9.92C13.45 12.89 13 13.5 13 15h-2v-.5c0-1.11.45-2.11 1.17-2.83l1.24-1.26c.37-.36.59-.86.59-1.41a2 2 0 0 0-2-2 2 2 0 0 0-2 2H8a4 4 0 0 1 4-4 4 4 0 0 1 4 4 3.2 3.2 0 0 1-.93 2.25M13 19h-2v-2h2M12 2A10 10 0 0 0 2 12a10 10 0 0 0 10 10 10 10 0 0 0 10-10c0-5.53-4.5-10-10-10Z"/></svg>');--md-admonition-icon--warning:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M13 14h-2V9h2m0 9h-2v-2h2M1 21h22L12 2 1 21Z"/></svg>');--md-admonition-icon--failure:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M19 6.41 17.59 5 12 10.59 6.41 5 5 6.41 10.59 12 5 17.59 6.41 19 12 13.41 17.59 19 19 17.59 13.41 12 19 6.41Z"/></svg>');--md-admonition-icon--danger:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="m11.5 20 4.86-9.73H13V4l-5 9.73h3.5V20M12 2c2.75 0 5.1 1 7.05 2.95C21 6.9 22 9.25 22 12s-1 5.1-2.95 7.05C17.1 21 14.75 22 12 22s-5.1-1-7.05-2.95C3 17.1 2 14.75 2 12s1-5.1 2.95-7.05C6.9 3 9.25 2 12 2Z"/></svg>');--md-admonition-icon--bug:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M11 13h2v1h-2v-1m10-8v6c0 5.5-3.8 10.7-9 12-5.2-1.3-9-6.5-9-12V5l9-4 9 4m-4 5h-2.2c-.2-.6-.6-1.1-1.1-1.5l1.2-1.2-.7-.7L12.8 8H12c-.2 0-.5 0-.7.1L9.9 6.6l-.8.8 1.2 1.2c-.5.3-.9.8-1.1 1.4H7v1h2v1H7v1h2v1H7v1h2.2c.4 1.2 1.5 2 2.8 2s2.4-.8 2.8-2H17v-1h-2v-1h2v-1h-2v-1h2v-1m-6 2h2v-1h-2v1Z"/></svg>');--md-admonition-icon--example:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M7 2v2h1v14a4 4 0 0 0 4 4 4 4 0 0 0 4-4V4h1V2H7m4 14c-.6 0-1-.4-1-1s.4-1 1-1 1 .4 1 1-.4 1-1 1m2-4c-.6 0-1-.4-1-1s.4-1 1-1 1 .4 1 1-.4 1-1 1m1-5h-4V4h4v3Z"/></svg>');--md-admonition-icon--quote:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M14 17h3l2-4V7h-6v6h3M6 17h3l2-4V7H5v6h3l-2 4Z"/></svg>')}.md-typeset .admonition,.md-typeset details{background-color:var(--md-admonition-bg-color);border:.075rem solid #448aff;border-radius:.2rem;box-shadow:var(--md-shadow-z1);color:var(--md-admonition-fg-color);display:flow-root;font-size:.64rem;margin:1.5625em 0;padding:0 .6rem;page-break-inside:avoid;transition:box-shadow 125ms}@media print{.md-typeset .admonition,.md-typeset details{box-shadow:none}}.md-typeset .admonition:focus-within,.md-typeset details:focus-within{box-shadow:0 0 0 .2rem #448aff1a}.md-typeset .admonition>*,.md-typeset details>*{box-sizing:border-box}.md-typeset .admonition .admonition,.md-typeset .admonition details,.md-typeset details .admonition,.md-typeset details details{margin-bottom:1em;margin-top:1em}.md-typeset .admonition .md-typeset__scrollwrap,.md-typeset details .md-typeset__scrollwrap{margin:1em -.6rem}.md-typeset .admonition .md-typeset__table,.md-typeset details .md-typeset__table{padding:0 .6rem}.md-typeset .admonition>.tabbed-set:only-child,.md-typeset details>.tabbed-set:only-child{margin-top:0}html .md-typeset .admonition>:last-child,html .md-typeset details>:last-child{margin-bottom:.6rem}[dir=ltr] .md-typeset .admonition-title,[dir=ltr] .md-typeset summary{padding-left:2rem;padding-right:.6rem}[dir=rtl] .md-typeset .admonition-title,[dir=rtl] .md-typeset summary{padding-left:.6rem;padding-right:2rem}[dir=ltr] .md-typeset .admonition-title,[dir=ltr] .md-typeset summary{border-left-width:.2rem}[dir=rtl] .md-typeset .admonition-title,[dir=rtl] .md-typeset summary{border-right-width:.2rem}[dir=ltr] .md-typeset .admonition-title,[dir=ltr] .md-typeset summary{border-top-left-radius:.1rem}[dir=ltr] .md-typeset .admonition-title,[dir=ltr] .md-typeset summary,[dir=rtl] .md-typeset .admonition-title,[dir=rtl] .md-typeset summary{border-top-right-radius:.1rem}[dir=rtl] .md-typeset .admonition-title,[dir=rtl] .md-typeset summary{border-top-left-radius:.1rem}.md-typeset .admonition-title,.md-typeset summary{background-color:#448aff1a;border:none;font-weight:700;margin:0 -.6rem;padding-bottom:.4rem;padding-top:.4rem;position:relative}html .md-typeset .admonition-title:last-child,html .md-typeset summary:last-child{margin-bottom:0}[dir=ltr] .md-typeset .admonition-title:before,[dir=ltr] .md-typeset summary:before{left:.6rem}[dir=rtl] .md-typeset .admonition-title:before,[dir=rtl] .md-typeset summary:before{right:.6rem}.md-typeset .admonition-title:before,.md-typeset summary:before{background-color:#448aff;content:"";height:1rem;-webkit-mask-image:var(--md-admonition-icon--note);mask-image:var(--md-admonition-icon--note);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;position:absolute;top:.625em;width:1rem}.md-typeset .admonition-title code,.md-typeset summary code{box-shadow:0 0 0 .05rem var(--md-default-fg-color--lightest)}.md-typeset .admonition.note,.md-typeset details.note{border-color:#448aff}.md-typeset .admonition.note:focus-within,.md-typeset details.note:focus-within{box-shadow:0 0 0 .2rem #448aff1a}.md-typeset .note>.admonition-title,.md-typeset .note>summary{background-color:#448aff1a}.md-typeset .note>.admonition-title:before,.md-typeset .note>summary:before{background-color:#448aff;-webkit-mask-image:var(--md-admonition-icon--note);mask-image:var(--md-admonition-icon--note)}.md-typeset .note>.admonition-title:after,.md-typeset .note>summary:after{color:#448aff}.md-typeset .admonition.abstract,.md-typeset details.abstract{border-color:#00b0ff}.md-typeset .admonition.abstract:focus-within,.md-typeset details.abstract:focus-within{box-shadow:0 0 0 .2rem #00b0ff1a}.md-typeset .abstract>.admonition-title,.md-typeset .abstract>summary{background-color:#00b0ff1a}.md-typeset .abstract>.admonition-title:before,.md-typeset .abstract>summary:before{background-color:#00b0ff;-webkit-mask-image:var(--md-admonition-icon--abstract);mask-image:var(--md-admonition-icon--abstract)}.md-typeset .abstract>.admonition-title:after,.md-typeset .abstract>summary:after{color:#00b0ff}.md-typeset .admonition.info,.md-typeset details.info{border-color:#00b8d4}.md-typeset .admonition.info:focus-within,.md-typeset details.info:focus-within{box-shadow:0 0 0 .2rem #00b8d41a}.md-typeset .info>.admonition-title,.md-typeset .info>summary{background-color:#00b8d41a}.md-typeset .info>.admonition-title:before,.md-typeset .info>summary:before{background-color:#00b8d4;-webkit-mask-image:var(--md-admonition-icon--info);mask-image:var(--md-admonition-icon--info)}.md-typeset .info>.admonition-title:after,.md-typeset .info>summary:after{color:#00b8d4}.md-typeset .admonition.tip,.md-typeset details.tip{border-color:#00bfa5}.md-typeset .admonition.tip:focus-within,.md-typeset details.tip:focus-within{box-shadow:0 0 0 .2rem #00bfa51a}.md-typeset .tip>.admonition-title,.md-typeset .tip>summary{background-color:#00bfa51a}.md-typeset .tip>.admonition-title:before,.md-typeset .tip>summary:before{background-color:#00bfa5;-webkit-mask-image:var(--md-admonition-icon--tip);mask-image:var(--md-admonition-icon--tip)}.md-typeset .tip>.admonition-title:after,.md-typeset .tip>summary:after{color:#00bfa5}.md-typeset .admonition.success,.md-typeset details.success{border-color:#00c853}.md-typeset .admonition.success:focus-within,.md-typeset details.success:focus-within{box-shadow:0 0 0 .2rem #00c8531a}.md-typeset .success>.admonition-title,.md-typeset .success>summary{background-color:#00c8531a}.md-typeset .success>.admonition-title:before,.md-typeset .success>summary:before{background-color:#00c853;-webkit-mask-image:var(--md-admonition-icon--success);mask-image:var(--md-admonition-icon--success)}.md-typeset .success>.admonition-title:after,.md-typeset .success>summary:after{color:#00c853}.md-typeset .admonition.question,.md-typeset details.question{border-color:#64dd17}.md-typeset .admonition.question:focus-within,.md-typeset details.question:focus-within{box-shadow:0 0 0 .2rem #64dd171a}.md-typeset .question>.admonition-title,.md-typeset .question>summary{background-color:#64dd171a}.md-typeset .question>.admonition-title:before,.md-typeset .question>summary:before{background-color:#64dd17;-webkit-mask-image:var(--md-admonition-icon--question);mask-image:var(--md-admonition-icon--question)}.md-typeset .question>.admonition-title:after,.md-typeset .question>summary:after{color:#64dd17}.md-typeset .admonition.warning,.md-typeset details.warning{border-color:#ff9100}.md-typeset .admonition.warning:focus-within,.md-typeset details.warning:focus-within{box-shadow:0 0 0 .2rem #ff91001a}.md-typeset .warning>.admonition-title,.md-typeset .warning>summary{background-color:#ff91001a}.md-typeset .warning>.admonition-title:before,.md-typeset .warning>summary:before{background-color:#ff9100;-webkit-mask-image:var(--md-admonition-icon--warning);mask-image:var(--md-admonition-icon--warning)}.md-typeset .warning>.admonition-title:after,.md-typeset .warning>summary:after{color:#ff9100}.md-typeset .admonition.failure,.md-typeset details.failure{border-color:#ff5252}.md-typeset .admonition.failure:focus-within,.md-typeset details.failure:focus-within{box-shadow:0 0 0 .2rem #ff52521a}.md-typeset .failure>.admonition-title,.md-typeset .failure>summary{background-color:#ff52521a}.md-typeset .failure>.admonition-title:before,.md-typeset .failure>summary:before{background-color:#ff5252;-webkit-mask-image:var(--md-admonition-icon--failure);mask-image:var(--md-admonition-icon--failure)}.md-typeset .failure>.admonition-title:after,.md-typeset .failure>summary:after{color:#ff5252}.md-typeset .admonition.danger,.md-typeset details.danger{border-color:#ff1744}.md-typeset .admonition.danger:focus-within,.md-typeset details.danger:focus-within{box-shadow:0 0 0 .2rem #ff17441a}.md-typeset .danger>.admonition-title,.md-typeset .danger>summary{background-color:#ff17441a}.md-typeset .danger>.admonition-title:before,.md-typeset .danger>summary:before{background-color:#ff1744;-webkit-mask-image:var(--md-admonition-icon--danger);mask-image:var(--md-admonition-icon--danger)}.md-typeset .danger>.admonition-title:after,.md-typeset .danger>summary:after{color:#ff1744}.md-typeset .admonition.bug,.md-typeset details.bug{border-color:#f50057}.md-typeset .admonition.bug:focus-within,.md-typeset details.bug:focus-within{box-shadow:0 0 0 .2rem #f500571a}.md-typeset .bug>.admonition-title,.md-typeset .bug>summary{background-color:#f500571a}.md-typeset .bug>.admonition-title:before,.md-typeset .bug>summary:before{background-color:#f50057;-webkit-mask-image:var(--md-admonition-icon--bug);mask-image:var(--md-admonition-icon--bug)}.md-typeset .bug>.admonition-title:after,.md-typeset .bug>summary:after{color:#f50057}.md-typeset .admonition.example,.md-typeset details.example{border-color:#7c4dff}.md-typeset .admonition.example:focus-within,.md-typeset details.example:focus-within{box-shadow:0 0 0 .2rem #7c4dff1a}.md-typeset .example>.admonition-title,.md-typeset .example>summary{background-color:#7c4dff1a}.md-typeset .example>.admonition-title:before,.md-typeset .example>summary:before{background-color:#7c4dff;-webkit-mask-image:var(--md-admonition-icon--example);mask-image:var(--md-admonition-icon--example)}.md-typeset .example>.admonition-title:after,.md-typeset .example>summary:after{color:#7c4dff}.md-typeset .admonition.quote,.md-typeset details.quote{border-color:#9e9e9e}.md-typeset .admonition.quote:focus-within,.md-typeset details.quote:focus-within{box-shadow:0 0 0 .2rem #9e9e9e1a}.md-typeset .quote>.admonition-title,.md-typeset .quote>summary{background-color:#9e9e9e1a}.md-typeset .quote>.admonition-title:before,.md-typeset .quote>summary:before{background-color:#9e9e9e;-webkit-mask-image:var(--md-admonition-icon--quote);mask-image:var(--md-admonition-icon--quote)}.md-typeset .quote>.admonition-title:after,.md-typeset .quote>summary:after{color:#9e9e9e}:root{--md-footnotes-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M19 7v4H5.83l3.58-3.59L8 6l-6 6 6 6 1.41-1.42L5.83 13H21V7h-2Z"/></svg>')}.md-typeset .footnote{color:var(--md-default-fg-color--light);font-size:.64rem}[dir=ltr] .md-typeset .footnote>ol{margin-left:0}[dir=rtl] .md-typeset .footnote>ol{margin-right:0}.md-typeset .footnote>ol>li{transition:color 125ms}.md-typeset .footnote>ol>li:target{color:var(--md-default-fg-color)}.md-typeset .footnote>ol>li:focus-within .footnote-backref{opacity:1;transform:translateX(0);transition:none}.md-typeset .footnote>ol>li:hover .footnote-backref,.md-typeset .footnote>ol>li:target .footnote-backref{opacity:1;transform:translateX(0)}.md-typeset .footnote>ol>li>:first-child{margin-top:0}.md-typeset .footnote-ref{font-size:.75em;font-weight:700}html .md-typeset .footnote-ref{outline-offset:.1rem}.md-typeset [id^="fnref:"]:target>.footnote-ref{outline:auto}.md-typeset .footnote-backref{color:var(--md-typeset-a-color);display:inline-block;font-size:0;opacity:0;transform:translateX(.25rem);transition:color .25s,transform .25s .25s,opacity 125ms .25s;vertical-align:text-bottom}@media print{.md-typeset .footnote-backref{color:var(--md-typeset-a-color);opacity:1;transform:translateX(0)}}[dir=rtl] .md-typeset .footnote-backref{transform:translateX(-.25rem)}.md-typeset .footnote-backref:hover{color:var(--md-accent-fg-color)}.md-typeset .footnote-backref:before{background-color:currentcolor;content:"";display:inline-block;height:.8rem;-webkit-mask-image:var(--md-footnotes-icon);mask-image:var(--md-footnotes-icon);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;width:.8rem}[dir=rtl] .md-typeset .footnote-backref:before svg{transform:scaleX(-1)}[dir=ltr] .md-typeset .headerlink{margin-left:.5rem}[dir=rtl] .md-typeset .headerlink{margin-right:.5rem}.md-typeset .headerlink{color:var(--md-default-fg-color--lighter);display:inline-block;opacity:0;transition:color .25s,opacity 125ms}@media print{.md-typeset .headerlink{display:none}}.md-typeset .headerlink:focus,.md-typeset :hover>.headerlink,.md-typeset :target>.headerlink{opacity:1;transition:color .25s,opacity 125ms}.md-typeset .headerlink:focus,.md-typeset .headerlink:hover,.md-typeset :target>.headerlink{color:var(--md-accent-fg-color)}.md-typeset :target{--md-scroll-margin:3.6rem;--md-scroll-offset:0rem;scroll-margin-top:calc(var(--md-scroll-margin) - var(--md-scroll-offset))}@media screen and (min-width:76.25em){.md-header--lifted~.md-container .md-typeset :target{--md-scroll-margin:6rem}}.md-typeset h1:target,.md-typeset h2:target,.md-typeset h3:target{--md-scroll-offset:0.2rem}.md-typeset h4:target{--md-scroll-offset:0.15rem}.md-typeset div.arithmatex{overflow:auto}@media screen and (max-width:44.984375em){.md-typeset div.arithmatex{margin:0 -.8rem}}.md-typeset div.arithmatex>*{margin-left:auto!important;margin-right:auto!important;padding:0 .8rem;touch-action:auto;width:-webkit-min-content;width:min-content}.md-typeset div.arithmatex>* mjx-container{margin:0!important}.md-typeset del.critic{background-color:var(--md-typeset-del-color)}.md-typeset del.critic,.md-typeset ins.critic{-webkit-box-decoration-break:clone;box-decoration-break:clone}.md-typeset ins.critic{background-color:var(--md-typeset-ins-color)}.md-typeset .critic.comment{-webkit-box-decoration-break:clone;box-decoration-break:clone;color:var(--md-code-hl-comment-color)}.md-typeset .critic.comment:before{content:"/* "}.md-typeset .critic.comment:after{content:" */"}.md-typeset .critic.block{box-shadow:none;display:block;margin:1em 0;overflow:auto;padding-left:.8rem;padding-right:.8rem}.md-typeset .critic.block>:first-child{margin-top:.5em}.md-typeset .critic.block>:last-child{margin-bottom:.5em}:root{--md-details-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M8.59 16.58 13.17 12 8.59 7.41 10 6l6 6-6 6-1.41-1.42Z"/></svg>')}.md-typeset details{display:flow-root;overflow:visible;padding-top:0}.md-typeset details[open]>summary:after{transform:rotate(90deg)}.md-typeset details:not([open]){box-shadow:none;padding-bottom:0}.md-typeset details:not([open])>summary{border-radius:.1rem}[dir=ltr] .md-typeset summary{padding-right:1.8rem}[dir=rtl] .md-typeset summary{padding-left:1.8rem}[dir=ltr] .md-typeset summary{border-top-left-radius:.1rem}[dir=ltr] .md-typeset summary,[dir=rtl] .md-typeset summary{border-top-right-radius:.1rem}[dir=rtl] .md-typeset summary{border-top-left-radius:.1rem}.md-typeset summary{cursor:pointer;display:block;min-height:1rem;overflow:hidden}.md-typeset summary.focus-visible{outline-color:var(--md-accent-fg-color);outline-offset:.2rem}.md-typeset summary:not(.focus-visible){-webkit-tap-highlight-color:transparent;outline:none}[dir=ltr] .md-typeset summary:after{right:.4rem}[dir=rtl] .md-typeset summary:after{left:.4rem}.md-typeset summary:after{background-color:currentcolor;content:"";height:1rem;-webkit-mask-image:var(--md-details-icon);mask-image:var(--md-details-icon);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;position:absolute;top:.625em;transform:rotate(0deg);transition:transform .25s;width:1rem}[dir=rtl] .md-typeset summary:after{transform:rotate(180deg)}.md-typeset summary::marker{display:none}.md-typeset summary::-webkit-details-marker{display:none}.md-typeset .emojione,.md-typeset .gemoji,.md-typeset .twemoji{display:inline-flex;height:1.125em;vertical-align:text-top}.md-typeset .emojione svg,.md-typeset .gemoji svg,.md-typeset .twemoji svg{fill:currentcolor;max-height:100%;width:1.125em}.highlight .o,.highlight .ow{color:var(--md-code-hl-operator-color)}.highlight .p{color:var(--md-code-hl-punctuation-color)}.highlight .cpf,.highlight .l,.highlight .s,.highlight .s1,.highlight .s2,.highlight .sb,.highlight .sc,.highlight .si,.highlight .ss{color:var(--md-code-hl-string-color)}.highlight .cp,.highlight .se,.highlight .sh,.highlight .sr,.highlight .sx{color:var(--md-code-hl-special-color)}.highlight .il,.highlight .m,.highlight .mb,.highlight .mf,.highlight .mh,.highlight .mi,.highlight .mo{color:var(--md-code-hl-number-color)}.highlight .k,.highlight .kd,.highlight .kn,.highlight .kp,.highlight .kr,.highlight .kt{color:var(--md-code-hl-keyword-color)}.highlight .kc,.highlight .n{color:var(--md-code-hl-name-color)}.highlight .bp,.highlight .nb,.highlight .no{color:var(--md-code-hl-constant-color)}.highlight .nc,.highlight .ne,.highlight .nf,.highlight .nn{color:var(--md-code-hl-function-color)}.highlight .nd,.highlight .ni,.highlight .nl,.highlight .nt{color:var(--md-code-hl-keyword-color)}.highlight .c,.highlight .c1,.highlight .ch,.highlight .cm,.highlight .cs,.highlight .sd{color:var(--md-code-hl-comment-color)}.highlight .na,.highlight .nv,.highlight .vc,.highlight .vg,.highlight .vi{color:var(--md-code-hl-variable-color)}.highlight .ge,.highlight .gh,.highlight .go,.highlight .gp,.highlight .gr,.highlight .gs,.highlight .gt,.highlight .gu{color:var(--md-code-hl-generic-color)}.highlight .gd,.highlight .gi{border-radius:.1rem;margin:0 -.125em;padding:0 .125em}.highlight .gd{background-color:var(--md-typeset-del-color)}.highlight .gi{background-color:var(--md-typeset-ins-color)}.highlight .hll{background-color:var(--md-code-hl-color--light);box-shadow:2px 0 0 0 var(--md-code-hl-color) inset;display:block;margin:0 -1.1764705882em;padding:0 1.1764705882em}.highlight span.filename{background-color:var(--md-code-bg-color);border-bottom:.05rem solid var(--md-default-fg-color--lightest);border-top-left-radius:.1rem;border-top-right-radius:.1rem;display:flow-root;font-size:.85em;font-weight:700;margin-top:1em;padding:.6617647059em 1.1764705882em;position:relative}.highlight span.filename+pre{margin-top:0}.highlight span.filename+pre>code{border-top-left-radius:0;border-top-right-radius:0}.highlight [data-linenos]:before{background-color:var(--md-code-bg-color);box-shadow:-.05rem 0 var(--md-default-fg-color--lightest) inset;color:var(--md-default-fg-color--light);content:attr(data-linenos);float:left;left:-1.1764705882em;margin-left:-1.1764705882em;margin-right:1.1764705882em;padding-left:1.1764705882em;position:sticky;-webkit-user-select:none;user-select:none;z-index:3}.highlight code a[id]{position:absolute;visibility:hidden}.highlight code[data-md-copying] .hll{display:contents}.highlight code[data-md-copying] .md-annotation{display:none}.highlighttable{display:flow-root}.highlighttable tbody,.highlighttable td{display:block;padding:0}.highlighttable tr{display:flex}.highlighttable pre{margin:0}.highlighttable th.filename{flex-grow:1;padding:0;text-align:left}.highlighttable th.filename span.filename{margin-top:0}.highlighttable .linenos{background-color:var(--md-code-bg-color);border-bottom-left-radius:.1rem;border-top-left-radius:.1rem;font-size:.85em;padding:.7720588235em 0 .7720588235em 1.1764705882em;-webkit-user-select:none;user-select:none}.highlighttable .linenodiv{box-shadow:-.05rem 0 var(--md-default-fg-color--lightest) inset;padding-right:.5882352941em}.highlighttable .linenodiv pre{color:var(--md-default-fg-color--light);text-align:right}.highlighttable .code{flex:1;min-width:0}.linenodiv a{color:inherit}.md-typeset .highlighttable{direction:ltr;margin:1em 0}.md-typeset .highlighttable>tbody>tr>.code>div>pre>code{border-bottom-left-radius:0;border-top-left-radius:0}.md-typeset .highlight+.result{border:.05rem solid var(--md-code-bg-color);border-bottom-left-radius:.1rem;border-bottom-right-radius:.1rem;border-top-width:.1rem;margin-top:-1.125em;overflow:visible;padding:0 1em}.md-typeset .highlight+.result:after{clear:both;content:"";display:block}@media screen and (max-width:44.984375em){.md-content__inner>.highlight{margin:1em -.8rem}.md-content__inner>.highlight>.filename,.md-content__inner>.highlight>.highlighttable>tbody>tr>.code>div>pre>code,.md-content__inner>.highlight>.highlighttable>tbody>tr>.filename span.filename,.md-content__inner>.highlight>.highlighttable>tbody>tr>.linenos,.md-content__inner>.highlight>pre>code{border-radius:0}.md-content__inner>.highlight+.result{border-left-width:0;border-radius:0;border-right-width:0;margin-left:-.8rem;margin-right:-.8rem}}.md-typeset .keys kbd:after,.md-typeset .keys kbd:before{-moz-osx-font-smoothing:initial;-webkit-font-smoothing:initial;color:inherit;margin:0;position:relative}.md-typeset .keys span{color:var(--md-default-fg-color--light);padding:0 .2em}.md-typeset .keys .key-alt:before,.md-typeset .keys .key-left-alt:before,.md-typeset .keys .key-right-alt:before{content:"⎇";padding-right:.4em}.md-typeset .keys .key-command:before,.md-typeset .keys .key-left-command:before,.md-typeset .keys .key-right-command:before{content:"⌘";padding-right:.4em}.md-typeset .keys .key-control:before,.md-typeset .keys .key-left-control:before,.md-typeset .keys .key-right-control:before{content:"⌃";padding-right:.4em}.md-typeset .keys .key-left-meta:before,.md-typeset .keys .key-meta:before,.md-typeset .keys .key-right-meta:before{content:"◆";padding-right:.4em}.md-typeset .keys .key-left-option:before,.md-typeset .keys .key-option:before,.md-typeset .keys .key-right-option:before{content:"⌥";padding-right:.4em}.md-typeset .keys .key-left-shift:before,.md-typeset .keys .key-right-shift:before,.md-typeset .keys .key-shift:before{content:"⇧";padding-right:.4em}.md-typeset .keys .key-left-super:before,.md-typeset .keys .key-right-super:before,.md-typeset .keys .key-super:before{content:"❖";padding-right:.4em}.md-typeset .keys .key-left-windows:before,.md-typeset .keys .key-right-windows:before,.md-typeset .keys .key-windows:before{content:"⊞";padding-right:.4em}.md-typeset .keys .key-arrow-down:before{content:"↓";padding-right:.4em}.md-typeset .keys .key-arrow-left:before{content:"←";padding-right:.4em}.md-typeset .keys .key-arrow-right:before{content:"→";padding-right:.4em}.md-typeset .keys .key-arrow-up:before{content:"↑";padding-right:.4em}.md-typeset .keys .key-backspace:before{content:"⌫";padding-right:.4em}.md-typeset .keys .key-backtab:before{content:"⇤";padding-right:.4em}.md-typeset .keys .key-caps-lock:before{content:"⇪";padding-right:.4em}.md-typeset .keys .key-clear:before{content:"⌧";padding-right:.4em}.md-typeset .keys .key-context-menu:before{content:"☰";padding-right:.4em}.md-typeset .keys .key-delete:before{content:"⌦";padding-right:.4em}.md-typeset .keys .key-eject:before{content:"⏏";padding-right:.4em}.md-typeset .keys .key-end:before{content:"⤓";padding-right:.4em}.md-typeset .keys .key-escape:before{content:"⎋";padding-right:.4em}.md-typeset .keys .key-home:before{content:"⤒";padding-right:.4em}.md-typeset .keys .key-insert:before{content:"⎀";padding-right:.4em}.md-typeset .keys .key-page-down:before{content:"⇟";padding-right:.4em}.md-typeset .keys .key-page-up:before{content:"⇞";padding-right:.4em}.md-typeset .keys .key-print-screen:before{content:"⎙";padding-right:.4em}.md-typeset .keys .key-tab:after{content:"⇥";padding-left:.4em}.md-typeset .keys .key-num-enter:after{content:"⌤";padding-left:.4em}.md-typeset .keys .key-enter:after{content:"⏎";padding-left:.4em}:root{--md-tabbed-icon--prev:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M15.41 16.58 10.83 12l4.58-4.59L14 6l-6 6 6 6 1.41-1.42Z"/></svg>');--md-tabbed-icon--next:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M8.59 16.58 13.17 12 8.59 7.41 10 6l6 6-6 6-1.41-1.42Z"/></svg>')}.md-typeset .tabbed-set{border-radius:.1rem;display:flex;flex-flow:column wrap;margin:1em 0;position:relative}.md-typeset .tabbed-set>input{height:0;opacity:0;position:absolute;width:0}.md-typeset .tabbed-set>input:target{--md-scroll-offset:0.625em}.md-typeset .tabbed-set>input.focus-visible~.tabbed-labels:before{background-color:var(--md-accent-fg-color)}.md-typeset .tabbed-labels{-ms-overflow-style:none;box-shadow:0 -.05rem var(--md-default-fg-color--lightest) inset;display:flex;max-width:100%;overflow:auto;scrollbar-width:none}@media print{.md-typeset .tabbed-labels{display:contents}}@media screen{.js .md-typeset .tabbed-labels{position:relative}.js .md-typeset .tabbed-labels:before{background:var(--md-default-fg-color);bottom:0;content:"";display:block;height:2px;left:0;position:absolute;transform:translateX(var(--md-indicator-x));transition:width 225ms,background-color .25s,transform .25s;transition-timing-function:cubic-bezier(.4,0,.2,1);width:var(--md-indicator-width)}}.md-typeset .tabbed-labels::-webkit-scrollbar{display:none}.md-typeset .tabbed-labels>label{border-bottom:.1rem solid #0000;border-radius:.1rem .1rem 0 0;color:var(--md-default-fg-color--light);cursor:pointer;flex-shrink:0;font-size:.64rem;font-weight:700;padding:.78125em 1.25em .625em;scroll-margin-inline-start:1rem;transition:background-color .25s,color .25s;white-space:nowrap;width:auto}@media print{.md-typeset .tabbed-labels>label:first-child{order:1}.md-typeset .tabbed-labels>label:nth-child(2){order:2}.md-typeset .tabbed-labels>label:nth-child(3){order:3}.md-typeset .tabbed-labels>label:nth-child(4){order:4}.md-typeset .tabbed-labels>label:nth-child(5){order:5}.md-typeset .tabbed-labels>label:nth-child(6){order:6}.md-typeset .tabbed-labels>label:nth-child(7){order:7}.md-typeset .tabbed-labels>label:nth-child(8){order:8}.md-typeset .tabbed-labels>label:nth-child(9){order:9}.md-typeset .tabbed-labels>label:nth-child(10){order:10}.md-typeset .tabbed-labels>label:nth-child(11){order:11}.md-typeset .tabbed-labels>label:nth-child(12){order:12}.md-typeset .tabbed-labels>label:nth-child(13){order:13}.md-typeset .tabbed-labels>label:nth-child(14){order:14}.md-typeset .tabbed-labels>label:nth-child(15){order:15}.md-typeset .tabbed-labels>label:nth-child(16){order:16}.md-typeset .tabbed-labels>label:nth-child(17){order:17}.md-typeset .tabbed-labels>label:nth-child(18){order:18}.md-typeset .tabbed-labels>label:nth-child(19){order:19}.md-typeset .tabbed-labels>label:nth-child(20){order:20}}.md-typeset .tabbed-labels>label:hover{color:var(--md-default-fg-color)}.md-typeset .tabbed-content{width:100%}@media print{.md-typeset .tabbed-content{display:contents}}.md-typeset .tabbed-block{display:none}@media print{.md-typeset .tabbed-block{display:block}.md-typeset .tabbed-block:first-child{order:1}.md-typeset .tabbed-block:nth-child(2){order:2}.md-typeset .tabbed-block:nth-child(3){order:3}.md-typeset .tabbed-block:nth-child(4){order:4}.md-typeset .tabbed-block:nth-child(5){order:5}.md-typeset .tabbed-block:nth-child(6){order:6}.md-typeset .tabbed-block:nth-child(7){order:7}.md-typeset .tabbed-block:nth-child(8){order:8}.md-typeset .tabbed-block:nth-child(9){order:9}.md-typeset .tabbed-block:nth-child(10){order:10}.md-typeset .tabbed-block:nth-child(11){order:11}.md-typeset .tabbed-block:nth-child(12){order:12}.md-typeset .tabbed-block:nth-child(13){order:13}.md-typeset .tabbed-block:nth-child(14){order:14}.md-typeset .tabbed-block:nth-child(15){order:15}.md-typeset .tabbed-block:nth-child(16){order:16}.md-typeset .tabbed-block:nth-child(17){order:17}.md-typeset .tabbed-block:nth-child(18){order:18}.md-typeset .tabbed-block:nth-child(19){order:19}.md-typeset .tabbed-block:nth-child(20){order:20}}.md-typeset .tabbed-block>.highlight:first-child>pre,.md-typeset .tabbed-block>pre:first-child{margin:0}.md-typeset .tabbed-block>.highlight:first-child>pre>code,.md-typeset .tabbed-block>pre:first-child>code{border-top-left-radius:0;border-top-right-radius:0}.md-typeset .tabbed-block>.highlight:first-child>.filename{border-top-left-radius:0;border-top-right-radius:0;margin:0}.md-typeset .tabbed-block>.highlight:first-child>.highlighttable{margin:0}.md-typeset .tabbed-block>.highlight:first-child>.highlighttable>tbody>tr>.filename span.filename,.md-typeset .tabbed-block>.highlight:first-child>.highlighttable>tbody>tr>.linenos{border-top-left-radius:0;border-top-right-radius:0;margin:0}.md-typeset .tabbed-block>.highlight:first-child>.highlighttable>tbody>tr>.code>div>pre>code{border-top-left-radius:0;border-top-right-radius:0}.md-typeset .tabbed-block>.highlight:first-child+.result{margin-top:-.125em}.md-typeset .tabbed-block>.tabbed-set{margin:0}.md-typeset .tabbed-button{align-self:center;border-radius:100%;color:var(--md-default-fg-color--light);cursor:pointer;display:block;height:.9rem;margin-top:.1rem;pointer-events:auto;transition:background-color .25s;width:.9rem}.md-typeset .tabbed-button:hover{background-color:var(--md-accent-fg-color--transparent);color:var(--md-accent-fg-color)}.md-typeset .tabbed-button:after{background-color:currentcolor;content:"";display:block;height:100%;-webkit-mask-image:var(--md-tabbed-icon--prev);mask-image:var(--md-tabbed-icon--prev);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;transition:background-color .25s,transform .25s;width:100%}.md-typeset .tabbed-control{background:linear-gradient(to right,var(--md-default-bg-color) 60%,#0000);display:flex;height:1.9rem;justify-content:start;pointer-events:none;position:absolute;transition:opacity 125ms;width:1.2rem}[dir=rtl] .md-typeset .tabbed-control{transform:rotate(180deg)}.md-typeset .tabbed-control[hidden]{opacity:0}.md-typeset .tabbed-control--next{background:linear-gradient(to left,var(--md-default-bg-color) 60%,#0000);justify-content:end;right:0}.md-typeset .tabbed-control--next .tabbed-button:after{-webkit-mask-image:var(--md-tabbed-icon--next);mask-image:var(--md-tabbed-icon--next)}@media screen and (max-width:44.984375em){[dir=ltr] .md-content__inner>.tabbed-set .tabbed-labels{padding-left:.8rem}[dir=rtl] .md-content__inner>.tabbed-set .tabbed-labels{padding-right:.8rem}.md-content__inner>.tabbed-set .tabbed-labels{margin:0 -.8rem;max-width:100vw;scroll-padding-inline-start:.8rem}[dir=ltr] .md-content__inner>.tabbed-set .tabbed-labels:after{padding-right:.8rem}[dir=rtl] .md-content__inner>.tabbed-set .tabbed-labels:after{padding-left:.8rem}.md-content__inner>.tabbed-set .tabbed-labels:after{content:""}[dir=ltr] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--prev{padding-left:.8rem}[dir=rtl] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--prev{padding-right:.8rem}[dir=ltr] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--prev{margin-left:-.8rem}[dir=rtl] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--prev{margin-right:-.8rem}.md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--prev{width:2rem}[dir=ltr] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--next{padding-right:.8rem}[dir=rtl] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--next{padding-left:.8rem}[dir=ltr] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--next{margin-right:-.8rem}[dir=rtl] .md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--next{margin-left:-.8rem}.md-content__inner>.tabbed-set .tabbed-labels~.tabbed-control--next{width:2rem}}@media screen{.md-typeset .tabbed-set>input:first-child:checked~.tabbed-labels>:first-child,.md-typeset .tabbed-set>input:nth-child(10):checked~.tabbed-labels>:nth-child(10),.md-typeset .tabbed-set>input:nth-child(11):checked~.tabbed-labels>:nth-child(11),.md-typeset .tabbed-set>input:nth-child(12):checked~.tabbed-labels>:nth-child(12),.md-typeset .tabbed-set>input:nth-child(13):checked~.tabbed-labels>:nth-child(13),.md-typeset .tabbed-set>input:nth-child(14):checked~.tabbed-labels>:nth-child(14),.md-typeset .tabbed-set>input:nth-child(15):checked~.tabbed-labels>:nth-child(15),.md-typeset .tabbed-set>input:nth-child(16):checked~.tabbed-labels>:nth-child(16),.md-typeset .tabbed-set>input:nth-child(17):checked~.tabbed-labels>:nth-child(17),.md-typeset .tabbed-set>input:nth-child(18):checked~.tabbed-labels>:nth-child(18),.md-typeset .tabbed-set>input:nth-child(19):checked~.tabbed-labels>:nth-child(19),.md-typeset .tabbed-set>input:nth-child(2):checked~.tabbed-labels>:nth-child(2),.md-typeset .tabbed-set>input:nth-child(20):checked~.tabbed-labels>:nth-child(20),.md-typeset .tabbed-set>input:nth-child(3):checked~.tabbed-labels>:nth-child(3),.md-typeset .tabbed-set>input:nth-child(4):checked~.tabbed-labels>:nth-child(4),.md-typeset .tabbed-set>input:nth-child(5):checked~.tabbed-labels>:nth-child(5),.md-typeset .tabbed-set>input:nth-child(6):checked~.tabbed-labels>:nth-child(6),.md-typeset .tabbed-set>input:nth-child(7):checked~.tabbed-labels>:nth-child(7),.md-typeset .tabbed-set>input:nth-child(8):checked~.tabbed-labels>:nth-child(8),.md-typeset .tabbed-set>input:nth-child(9):checked~.tabbed-labels>:nth-child(9){color:var(--md-default-fg-color)}.md-typeset .no-js .tabbed-set>input:first-child:checked~.tabbed-labels>:first-child,.md-typeset .no-js .tabbed-set>input:nth-child(10):checked~.tabbed-labels>:nth-child(10),.md-typeset .no-js .tabbed-set>input:nth-child(11):checked~.tabbed-labels>:nth-child(11),.md-typeset .no-js .tabbed-set>input:nth-child(12):checked~.tabbed-labels>:nth-child(12),.md-typeset .no-js .tabbed-set>input:nth-child(13):checked~.tabbed-labels>:nth-child(13),.md-typeset .no-js .tabbed-set>input:nth-child(14):checked~.tabbed-labels>:nth-child(14),.md-typeset .no-js .tabbed-set>input:nth-child(15):checked~.tabbed-labels>:nth-child(15),.md-typeset .no-js .tabbed-set>input:nth-child(16):checked~.tabbed-labels>:nth-child(16),.md-typeset .no-js .tabbed-set>input:nth-child(17):checked~.tabbed-labels>:nth-child(17),.md-typeset .no-js .tabbed-set>input:nth-child(18):checked~.tabbed-labels>:nth-child(18),.md-typeset .no-js .tabbed-set>input:nth-child(19):checked~.tabbed-labels>:nth-child(19),.md-typeset .no-js .tabbed-set>input:nth-child(2):checked~.tabbed-labels>:nth-child(2),.md-typeset .no-js .tabbed-set>input:nth-child(20):checked~.tabbed-labels>:nth-child(20),.md-typeset .no-js .tabbed-set>input:nth-child(3):checked~.tabbed-labels>:nth-child(3),.md-typeset .no-js .tabbed-set>input:nth-child(4):checked~.tabbed-labels>:nth-child(4),.md-typeset .no-js .tabbed-set>input:nth-child(5):checked~.tabbed-labels>:nth-child(5),.md-typeset .no-js .tabbed-set>input:nth-child(6):checked~.tabbed-labels>:nth-child(6),.md-typeset .no-js .tabbed-set>input:nth-child(7):checked~.tabbed-labels>:nth-child(7),.md-typeset .no-js .tabbed-set>input:nth-child(8):checked~.tabbed-labels>:nth-child(8),.md-typeset .no-js .tabbed-set>input:nth-child(9):checked~.tabbed-labels>:nth-child(9),.no-js .md-typeset .tabbed-set>input:first-child:checked~.tabbed-labels>:first-child,.no-js .md-typeset .tabbed-set>input:nth-child(10):checked~.tabbed-labels>:nth-child(10),.no-js .md-typeset .tabbed-set>input:nth-child(11):checked~.tabbed-labels>:nth-child(11),.no-js .md-typeset .tabbed-set>input:nth-child(12):checked~.tabbed-labels>:nth-child(12),.no-js .md-typeset .tabbed-set>input:nth-child(13):checked~.tabbed-labels>:nth-child(13),.no-js .md-typeset .tabbed-set>input:nth-child(14):checked~.tabbed-labels>:nth-child(14),.no-js .md-typeset .tabbed-set>input:nth-child(15):checked~.tabbed-labels>:nth-child(15),.no-js .md-typeset .tabbed-set>input:nth-child(16):checked~.tabbed-labels>:nth-child(16),.no-js .md-typeset .tabbed-set>input:nth-child(17):checked~.tabbed-labels>:nth-child(17),.no-js .md-typeset .tabbed-set>input:nth-child(18):checked~.tabbed-labels>:nth-child(18),.no-js .md-typeset .tabbed-set>input:nth-child(19):checked~.tabbed-labels>:nth-child(19),.no-js .md-typeset .tabbed-set>input:nth-child(2):checked~.tabbed-labels>:nth-child(2),.no-js .md-typeset .tabbed-set>input:nth-child(20):checked~.tabbed-labels>:nth-child(20),.no-js .md-typeset .tabbed-set>input:nth-child(3):checked~.tabbed-labels>:nth-child(3),.no-js .md-typeset .tabbed-set>input:nth-child(4):checked~.tabbed-labels>:nth-child(4),.no-js .md-typeset .tabbed-set>input:nth-child(5):checked~.tabbed-labels>:nth-child(5),.no-js .md-typeset .tabbed-set>input:nth-child(6):checked~.tabbed-labels>:nth-child(6),.no-js .md-typeset .tabbed-set>input:nth-child(7):checked~.tabbed-labels>:nth-child(7),.no-js .md-typeset .tabbed-set>input:nth-child(8):checked~.tabbed-labels>:nth-child(8),.no-js .md-typeset .tabbed-set>input:nth-child(9):checked~.tabbed-labels>:nth-child(9){border-color:var(--md-default-fg-color)}}.md-typeset .tabbed-set>input:first-child.focus-visible~.tabbed-labels>:first-child,.md-typeset .tabbed-set>input:nth-child(10).focus-visible~.tabbed-labels>:nth-child(10),.md-typeset .tabbed-set>input:nth-child(11).focus-visible~.tabbed-labels>:nth-child(11),.md-typeset .tabbed-set>input:nth-child(12).focus-visible~.tabbed-labels>:nth-child(12),.md-typeset .tabbed-set>input:nth-child(13).focus-visible~.tabbed-labels>:nth-child(13),.md-typeset .tabbed-set>input:nth-child(14).focus-visible~.tabbed-labels>:nth-child(14),.md-typeset .tabbed-set>input:nth-child(15).focus-visible~.tabbed-labels>:nth-child(15),.md-typeset .tabbed-set>input:nth-child(16).focus-visible~.tabbed-labels>:nth-child(16),.md-typeset .tabbed-set>input:nth-child(17).focus-visible~.tabbed-labels>:nth-child(17),.md-typeset .tabbed-set>input:nth-child(18).focus-visible~.tabbed-labels>:nth-child(18),.md-typeset .tabbed-set>input:nth-child(19).focus-visible~.tabbed-labels>:nth-child(19),.md-typeset .tabbed-set>input:nth-child(2).focus-visible~.tabbed-labels>:nth-child(2),.md-typeset .tabbed-set>input:nth-child(20).focus-visible~.tabbed-labels>:nth-child(20),.md-typeset .tabbed-set>input:nth-child(3).focus-visible~.tabbed-labels>:nth-child(3),.md-typeset .tabbed-set>input:nth-child(4).focus-visible~.tabbed-labels>:nth-child(4),.md-typeset .tabbed-set>input:nth-child(5).focus-visible~.tabbed-labels>:nth-child(5),.md-typeset .tabbed-set>input:nth-child(6).focus-visible~.tabbed-labels>:nth-child(6),.md-typeset .tabbed-set>input:nth-child(7).focus-visible~.tabbed-labels>:nth-child(7),.md-typeset .tabbed-set>input:nth-child(8).focus-visible~.tabbed-labels>:nth-child(8),.md-typeset .tabbed-set>input:nth-child(9).focus-visible~.tabbed-labels>:nth-child(9){color:var(--md-accent-fg-color)}.md-typeset .tabbed-set>input:first-child:checked~.tabbed-content>:first-child,.md-typeset .tabbed-set>input:nth-child(10):checked~.tabbed-content>:nth-child(10),.md-typeset .tabbed-set>input:nth-child(11):checked~.tabbed-content>:nth-child(11),.md-typeset .tabbed-set>input:nth-child(12):checked~.tabbed-content>:nth-child(12),.md-typeset .tabbed-set>input:nth-child(13):checked~.tabbed-content>:nth-child(13),.md-typeset .tabbed-set>input:nth-child(14):checked~.tabbed-content>:nth-child(14),.md-typeset .tabbed-set>input:nth-child(15):checked~.tabbed-content>:nth-child(15),.md-typeset .tabbed-set>input:nth-child(16):checked~.tabbed-content>:nth-child(16),.md-typeset .tabbed-set>input:nth-child(17):checked~.tabbed-content>:nth-child(17),.md-typeset .tabbed-set>input:nth-child(18):checked~.tabbed-content>:nth-child(18),.md-typeset .tabbed-set>input:nth-child(19):checked~.tabbed-content>:nth-child(19),.md-typeset .tabbed-set>input:nth-child(2):checked~.tabbed-content>:nth-child(2),.md-typeset .tabbed-set>input:nth-child(20):checked~.tabbed-content>:nth-child(20),.md-typeset .tabbed-set>input:nth-child(3):checked~.tabbed-content>:nth-child(3),.md-typeset .tabbed-set>input:nth-child(4):checked~.tabbed-content>:nth-child(4),.md-typeset .tabbed-set>input:nth-child(5):checked~.tabbed-content>:nth-child(5),.md-typeset .tabbed-set>input:nth-child(6):checked~.tabbed-content>:nth-child(6),.md-typeset .tabbed-set>input:nth-child(7):checked~.tabbed-content>:nth-child(7),.md-typeset .tabbed-set>input:nth-child(8):checked~.tabbed-content>:nth-child(8),.md-typeset .tabbed-set>input:nth-child(9):checked~.tabbed-content>:nth-child(9){display:block}:root{--md-tasklist-icon:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M1 12C1 5.925 5.925 1 12 1s11 4.925 11 11-4.925 11-11 11S1 18.075 1 12Zm16.28-2.72a.751.751 0 0 0-.018-1.042.751.751 0 0 0-1.042-.018l-5.97 5.97-2.47-2.47a.751.751 0 0 0-1.042.018.751.751 0 0 0-.018 1.042l3 3a.75.75 0 0 0 1.06 0Z"/></svg>');--md-tasklist-icon--checked:url('data:image/svg+xml;charset=utf-8,<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M1 12C1 5.925 5.925 1 12 1s11 4.925 11 11-4.925 11-11 11S1 18.075 1 12Zm16.28-2.72a.751.751 0 0 0-.018-1.042.751.751 0 0 0-1.042-.018l-5.97 5.97-2.47-2.47a.751.751 0 0 0-1.042.018.751.751 0 0 0-.018 1.042l3 3a.75.75 0 0 0 1.06 0Z"/></svg>')}.md-typeset .task-list-item{list-style-type:none;position:relative}[dir=ltr] .md-typeset .task-list-item [type=checkbox]{left:-2em}[dir=rtl] .md-typeset .task-list-item [type=checkbox]{right:-2em}.md-typeset .task-list-item [type=checkbox]{position:absolute;top:.45em}.md-typeset .task-list-control [type=checkbox]{opacity:0;z-index:-1}[dir=ltr] .md-typeset .task-list-indicator:before{left:-1.5em}[dir=rtl] .md-typeset .task-list-indicator:before{right:-1.5em}.md-typeset .task-list-indicator:before{background-color:var(--md-default-fg-color--lightest);content:"";height:1.25em;-webkit-mask-image:var(--md-tasklist-icon);mask-image:var(--md-tasklist-icon);-webkit-mask-position:center;mask-position:center;-webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;-webkit-mask-size:contain;mask-size:contain;position:absolute;top:.15em;width:1.25em}.md-typeset [type=checkbox]:checked+.task-list-indicator:before{background-color:#00e676;-webkit-mask-image:var(--md-tasklist-icon--checked);mask-image:var(--md-tasklist-icon--checked)}:root>*{--md-mermaid-font-family:var(--md-text-font-family),sans-serif;--md-mermaid-edge-color:var(--md-code-fg-color);--md-mermaid-node-bg-color:var(--md-accent-fg-color--transparent);--md-mermaid-node-fg-color:var(--md-accent-fg-color);--md-mermaid-label-bg-color:var(--md-default-bg-color);--md-mermaid-label-fg-color:var(--md-code-fg-color);--md-mermaid-sequence-actor-bg-color:var(--md-mermaid-label-bg-color);--md-mermaid-sequence-actor-fg-color:var(--md-mermaid-label-fg-color);--md-mermaid-sequence-actor-border-color:var(--md-mermaid-node-fg-color);--md-mermaid-sequence-actor-line-color:var(--md-default-fg-color--lighter);--md-mermaid-sequence-actorman-bg-color:var(--md-mermaid-label-bg-color);--md-mermaid-sequence-actorman-line-color:var(--md-mermaid-node-fg-color);--md-mermaid-sequence-box-bg-color:var(--md-mermaid-node-bg-color);--md-mermaid-sequence-box-fg-color:var(--md-mermaid-edge-color);--md-mermaid-sequence-label-bg-color:var(--md-mermaid-node-bg-color);--md-mermaid-sequence-label-fg-color:var(--md-mermaid-node-fg-color);--md-mermaid-sequence-loop-bg-color:var(--md-mermaid-node-bg-color);--md-mermaid-sequence-loop-fg-color:var(--md-mermaid-edge-color);--md-mermaid-sequence-loop-border-color:var(--md-mermaid-node-fg-color);--md-mermaid-sequence-message-fg-color:var(--md-mermaid-edge-color);--md-mermaid-sequence-message-line-color:var(--md-mermaid-edge-color);--md-mermaid-sequence-note-bg-color:var(--md-mermaid-label-bg-color);--md-mermaid-sequence-note-fg-color:var(--md-mermaid-edge-color);--md-mermaid-sequence-note-border-color:var(--md-mermaid-label-fg-color);--md-mermaid-sequence-number-bg-color:var(--md-mermaid-node-fg-color);--md-mermaid-sequence-number-fg-color:var(--md-accent-bg-color)}.mermaid{line-height:normal;margin:1em 0}@media screen and (min-width:45em){[dir=ltr] .md-typeset .inline{float:left}[dir=rtl] .md-typeset .inline{float:right}[dir=ltr] .md-typeset .inline{margin-right:.8rem}[dir=rtl] .md-typeset .inline{margin-left:.8rem}.md-typeset .inline{margin-bottom:.8rem;margin-top:0;width:11.7rem}[dir=ltr] .md-typeset .inline.end{float:right}[dir=rtl] .md-typeset .inline.end{float:left}[dir=ltr] .md-typeset .inline.end{margin-left:.8rem;margin-right:0}[dir=rtl] .md-typeset .inline.end{margin-left:0;margin-right:.8rem}}
\ No newline at end of file
diff --git a/assets/stylesheets/main.fad675c6.min.css.map b/assets/stylesheets/main.fad675c6.min.css.map
deleted file mode 100644
index d7829fda71..0000000000
--- a/assets/stylesheets/main.fad675c6.min.css.map
+++ /dev/null
@@ -1 +0,0 @@
-{"version":3,"sources":["src/templates/assets/stylesheets/main/components/_meta.scss","../../../../src/templates/assets/stylesheets/main.scss","src/templates/assets/stylesheets/main/_resets.scss","src/templates/assets/stylesheets/main/_colors.scss","src/templates/assets/stylesheets/main/_icons.scss","src/templates/assets/stylesheets/main/_typeset.scss","src/templates/assets/stylesheets/utilities/_break.scss","src/templates/assets/stylesheets/main/components/_author.scss","src/templates/assets/stylesheets/main/components/_banner.scss","src/templates/assets/stylesheets/main/components/_base.scss","src/templates/assets/stylesheets/main/components/_clipboard.scss","src/templates/assets/stylesheets/main/components/_consent.scss","src/templates/assets/stylesheets/main/components/_content.scss","src/templates/assets/stylesheets/main/components/_dialog.scss","src/templates/assets/stylesheets/main/components/_feedback.scss","src/templates/assets/stylesheets/main/components/_footer.scss","src/templates/assets/stylesheets/main/components/_form.scss","src/templates/assets/stylesheets/main/components/_header.scss","node_modules/material-design-color/material-color.scss","src/templates/assets/stylesheets/main/components/_nav.scss","src/templates/assets/stylesheets/main/components/_pagination.scss","src/templates/assets/stylesheets/main/components/_post.scss","src/templates/assets/stylesheets/main/components/_progress.scss","src/templates/assets/stylesheets/main/components/_search.scss","src/templates/assets/stylesheets/main/components/_select.scss","src/templates/assets/stylesheets/main/components/_sidebar.scss","src/templates/assets/stylesheets/main/components/_source.scss","src/templates/assets/stylesheets/main/components/_status.scss","src/templates/assets/stylesheets/main/components/_tabs.scss","src/templates/assets/stylesheets/main/components/_tag.scss","src/templates/assets/stylesheets/main/components/_tooltip.scss","src/templates/assets/stylesheets/main/components/_top.scss","src/templates/assets/stylesheets/main/components/_version.scss","src/templates/assets/stylesheets/main/extensions/markdown/_admonition.scss","src/templates/assets/stylesheets/main/extensions/markdown/_footnotes.scss","src/templates/assets/stylesheets/main/extensions/markdown/_toc.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_arithmatex.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_critic.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_details.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_emoji.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_highlight.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_keys.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_tabbed.scss","src/templates/assets/stylesheets/main/extensions/pymdownx/_tasklist.scss","src/templates/assets/stylesheets/main/integrations/_mermaid.scss","src/templates/assets/stylesheets/main/_modifiers.scss"],"names":[],"mappings":"AA0CE,gBC6xCF,CC3yCA,KAEE,6BAAA,CAAA,0BAAA,CAAA,qBAAA,CADA,qBDzBF,CC8BA,iBAGE,kBD3BF,CC8BE,gCANF,iBAOI,yBDzBF,CACF,CC6BA,KACE,QD1BF,CC8BA,qBAIE,uCD3BF,CC+BA,EACE,aAAA,CACA,oBD5BF,CCgCA,GAME,QAAA,CALA,kBAAA,CACA,aAAA,CACA,aAAA,CAEA,gBAAA,CADA,SD3BF,CCiCA,MACE,aD9BF,CCkCA,QAEE,eD/BF,CCmCA,IACE,iBDhCF,CCoCA,MAEE,uBAAA,CADA,gBDhCF,CCqCA,MAEE,eAAA,CACA,kBDlCF,CCsCA,OAKE,gBAAA,CACA,QAAA,CAHA,mBAAA,CACA,iBAAA,CAFA,QAAA,CADA,SD9BF,CCuCA,MACE,QAAA,CACA,YDpCF,CErDA,MAIE,6BAAA,CACA,oCAAA,CACA,mCAAA,CACA,0BAAA,CACA,sCAAA,CAGA,4BAAA,CACA,2CAAA,CACA,yBAAA,CACA,qCFmDF,CE7CA,+BAIE,kBF6CF,CE1CE,oHAEE,YF4CJ,CEnCA,qCAIE,eAAA,CAGA,+BAAA,CACA,sCAAA,CACA,wCAAA,CACA,yCAAA,CACA,0BAAA,CACA,sCAAA,CACA,wCAAA,CACA,yCAAA,CAGA,0BAAA,CACA,0BAAA,CAGA,0BAAA,CACA,mCAAA,CACA,iCAAA,CACA,kCAAA,CACA,mCAAA,CACA,mCAAA,CACA,kCAAA,CACA,iCAAA,CACA,+CAAA,CACA,6DAAA,CACA,gEAAA,CACA,4DAAA,CACA,4DAAA,CACA,6DAAA,CAGA,6CAAA,CAGA,+CAAA,CAGA,gCAAA,CACA,gCAAA,CAGA,8BAAA,CACA,kCAAA,CACA,qCAAA,CAGA,iCAAA,CAGA,kCAAA,CACA,gDAAA,CAGA,mDAAA,CACA,mDAAA,CAGA,+BAAA,CACA,0BAAA,CAGA,yBAAA,CACA,qCAAA,CACA,uCAAA,CACA,8BAAA,CACA,oCAAA,CAGA,8DAAA,CAKA,8DAAA,CAKA,0DFOF,CG9HE,aAIE,iBAAA,CAHA,aAAA,CAEA,aAAA,CADA,YHmIJ,CIxIA,KACE,kCAAA,CACA,iCAAA,CAGA,uGAAA,CAKA,mFJyIF,CInIA,iBAIE,mCAAA,CACA,6BAAA,CAFA,sCJwIF,CIlIA,aAIE,4BAAA,CADA,sCJsIF,CI7HA,MACE,0NAAA,CACA,mNAAA,CACA,oNJgIF,CIzHA,YAGE,gCAAA,CAAA,kBAAA,CAFA,eAAA,CACA,eJ6HF,CIxHE,aAPF,YAQI,gBJ2HF,CACF,CIxHE,uGAME,iBAAA,CAAA,cJ0HJ,CItHE,eAKE,uCAAA,CAHA,aAAA,CAEA,eAAA,CAHA,iBJ6HJ,CIpHE,8BAPE,eAAA,CAGA,qBJ+HJ,CI3HE,eAEE,kBAAA,CAEA,eAAA,CAHA,oBJ0HJ,CIlHE,eAEE,gBAAA,CACA,eAAA,CAEA,qBAAA,CADA,eAAA,CAHA,mBJwHJ,CIhHE,kBACE,eJkHJ,CI9GE,eAEE,eAAA,CACA,qBAAA,CAFA,YJkHJ,CI5GE,8BAKE,uCAAA,CAFA,cAAA,CACA,eAAA,CAEA,qBAAA,CAJA,eJkHJ,CI1GE,eACE,wBJ4GJ,CIxGE,eAGE,+DAAA,CAFA,iBAAA,CACA,cJ2GJ,CItGE,cACE,+BAAA,CACA,qBJwGJ,CIrGI,mCAEE,sBJsGN,CIlGI,wCACE,+BJoGN,CIjGM,kDACE,uDJmGR,CI9FI,mBACE,kBAAA,CACA,iCJgGN,CI5FI,4BACE,uCAAA,CACA,oBJ8FN,CIzFE,iDAIE,6BAAA,CACA,aAAA,CAFA,2BJ6FJ,CIxFI,aARF,iDASI,oBJ6FJ,CACF,CIzFE,iBAIE,wCAAA,CACA,mBAAA,CACA,kCAAA,CAAA,0BAAA,CAJA,eAAA,CADA,uBAAA,CAEA,qBJ8FJ,CIxFI,qCAEE,uCAAA,CADA,YJ2FN,CIrFE,gBAEE,iBAAA,CACA,eAAA,CAFA,iBJyFJ,CIpFI,qBASE,kCAAA,CAAA,0BAAA,CADA,eAAA,CAPA,aAAA,CAEA,QAAA,CAIA,uCAAA,CAHA,aAAA,CAFA,oCAAA,CASA,yDAAA,CADA,oBAAA,CAJA,iBAAA,CADA,iBJ4FN,CInFM,2BACE,+CJqFR,CIjFM,wCAEE,YAAA,CADA,WJoFR,CI/EM,8CACE,oDJiFR,CI9EQ,oDACE,0CJgFV,CIzEE,gBAOE,4CAAA,CACA,mBAAA,CACA,mKACE,CANF,gCAAA,CAHA,oBAAA,CAEA,eAAA,CADA,uBAAA,CAIA,uBAAA,CADA,qBJ+EJ,CIpEE,iBAGE,6CAAA,CACA,kCAAA,CAAA,0BAAA,CAHA,aAAA,CACA,qBJwEJ,CIlEE,iBAGE,6DAAA,CADA,WAAA,CADA,oBJsEJ,CIjEI,oBAGE,wEAQE,2CAAA,CACA,mBAAA,CACA,8BAAA,CAJA,gCAAA,CACA,mBAAA,CAFA,eAAA,CAHA,UAAA,CAEA,cAAA,CADA,mBAAA,CAFA,iBAAA,CACA,WJyEN,CACF,CI5DE,kBACE,WJ8DJ,CI1DE,oDAEE,qBJ4DJ,CI9DE,oDAEE,sBJ4DJ,CIxDE,iCACE,kBJ6DJ,CI9DE,iCACE,mBJ6DJ,CI9DE,iCAIE,2DJ0DJ,CI9DE,iCAIE,4DJ0DJ,CI9DE,uBAGE,uCAAA,CADA,aAAA,CAAA,cJ4DJ,CItDE,eACE,oBJwDJ,CIpDE,kDAGE,kBJsDJ,CIzDE,kDAGE,mBJsDJ,CIzDE,8BAEE,SJuDJ,CInDI,0DACE,iBJsDN,CIlDI,oCACE,2BJqDN,CIlDM,0CACE,2BJqDR,CIhDI,wDACE,kBJoDN,CIrDI,wDACE,mBJoDN,CIrDI,oCAEE,kBJmDN,CIhDM,kGAEE,aJoDR,CIhDM,0DACE,eJmDR,CI/CM,4HAEE,kBJkDR,CIpDM,4HAEE,mBJkDR,CIpDM,oFACE,kBAAA,CAAA,eJmDR,CI5CE,yBAEE,mBJ8CJ,CIhDE,yBAEE,oBJ8CJ,CIhDE,eACE,mBAAA,CAAA,cJ+CJ,CI1CE,kDAIE,WAAA,CADA,cJ6CJ,CIrCI,4BAEE,oBJuCN,CInCI,6BAEE,oBJqCN,CIjCI,kCACE,YJmCN,CI9BE,mBACE,iBAAA,CAGA,eAAA,CADA,cAAA,CAEA,iBAAA,CAHA,yBAAA,CAAA,sBAAA,CAAA,iBJmCJ,CI7BI,uBACE,aJ+BN,CI1BE,uBAGE,iBAAA,CADA,eAAA,CADA,eJ8BJ,CIxBE,mBACE,cJ0BJ,CItBE,+BAME,2CAAA,CACA,iDAAA,CACA,mBAAA,CAPA,oBAAA,CAGA,gBAAA,CAFA,cAAA,CACA,aAAA,CAEA,iBJ2BJ,CIrBI,aAXF,+BAYI,aJwBJ,CACF,CInBI,iCACE,gBJqBN,CIdM,8FACE,YJgBR,CIZM,4FACE,eJcR,CITI,8FACE,eJWN,CIRM,kHACE,gBJUR,CILI,kCAGE,eAAA,CAFA,cAAA,CACA,sBAAA,CAEA,kBJON,CIHI,kCAGE,qDAAA,CAFA,sBAAA,CACA,kBJMN,CIDI,wCACE,iCJGN,CIAM,8CACE,qDAAA,CACA,sDJER,CIGI,iCACE,iBJDN,CIME,wCACE,cJJJ,CIOI,wDAIE,gBJCN,CILI,wDAIE,iBJCN,CILI,8CAME,UAAA,CALA,oBAAA,CAEA,YAAA,CAKA,oDAAA,CAAA,4CAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAHA,iCAAA,CAFA,0BAAA,CAHA,WJGN,CISI,oDACE,oDJPN,CIWI,mEACE,kDAAA,CACA,yDAAA,CAAA,iDJTN,CIaI,oEACE,kDAAA,CACA,0DAAA,CAAA,kDJXN,CIgBE,wBACE,iBAAA,CACA,eAAA,CACA,iBJdJ,CIkBE,mBACE,oBAAA,CAEA,kBAAA,CADA,eJfJ,CImBI,aANF,mBAOI,aJhBJ,CACF,CImBI,8BACE,aAAA,CAEA,QAAA,CACA,eAAA,CAFA,UJfN,CKhWI,0CD8XF,uBACE,iBJ1BF,CI6BE,4BACE,eJ3BJ,CACF,CM/hBE,uBAEE,aAAA,CACA,aAAA,CAEA,aAAA,CACA,eAAA,CALA,iBAAA,CAMA,sCACE,CAJF,YNoiBJ,CM5hBI,2BAEE,kBAAA,CADA,aN+hBN,CM1hBI,6BAME,+CAAA,CAFA,yCAAA,CAHA,eAAA,CACA,eAAA,CACA,kBAAA,CAEA,iBN6hBN,CMxhBI,6BAEE,aAAA,CADA,YN2hBN,CMrhBE,wBACE,kBNuhBJ,CMphBI,4BACE,mCAAA,CACA,uBNshBN,CMlhBI,4DAEE,oBAAA,CADA,SNqhBN,CMjhBM,oEACE,mBNmhBR,COzkBA,WAGE,0CAAA,CADA,+BAAA,CADA,aP8kBF,COzkBE,aANF,WAOI,YP4kBF,CACF,COzkBE,oBAEE,2CAAA,CADA,gCP4kBJ,COvkBE,kBAGE,eAAA,CADA,iBAAA,CADA,eP2kBJ,COrkBE,6BACE,WP0kBJ,CO3kBE,6BACE,UP0kBJ,CO3kBE,mBAEE,aAAA,CACA,cAAA,CACA,uBPukBJ,COpkBI,0BACE,YPskBN,COlkBI,yBACE,UPokBN,CQzmBA,KASE,cAAA,CARA,WAAA,CACA,iBR6mBF,CKzcI,oCGtKJ,KAaI,gBRsmBF,CACF,CK9cI,oCGtKJ,KAkBI,cRsmBF,CACF,CQjmBA,KASE,2CAAA,CAPA,YAAA,CACA,qBAAA,CAKA,eAAA,CAHA,eAAA,CAJA,iBAAA,CAGA,URumBF,CQ/lBE,aAZF,KAaI,aRkmBF,CACF,CK/cI,0CGhJF,yBAII,cR+lBJ,CACF,CQtlBA,SAEE,gBAAA,CAAA,iBAAA,CADA,eR0lBF,CQrlBA,cACE,YAAA,CACA,qBAAA,CACA,WRwlBF,CQrlBE,aANF,cAOI,aRwlBF,CACF,CQplBA,SACE,WRulBF,CQplBE,gBACE,YAAA,CACA,WAAA,CACA,iBRslBJ,CQjlBA,aACE,eAAA,CACA,sBRolBF,CQ3kBA,WACE,YR8kBF,CQzkBA,WAGE,QAAA,CACA,SAAA,CAHA,iBAAA,CACA,OR8kBF,CQzkBE,uCACE,aR2kBJ,CQvkBE,+BAEE,uCAAA,CADA,kBR0kBJ,CQpkBA,SASE,2CAAA,CACA,mBAAA,CAFA,gCAAA,CADA,gBAAA,CADA,YAAA,CAMA,SAAA,CADA,uCAAA,CANA,mBAAA,CAJA,cAAA,CAYA,2BAAA,CATA,UR8kBF,CQlkBE,eAEE,SAAA,CAIA,uBAAA,CAHA,oEACE,CAHF,URukBJ,CQzjBA,MACE,WR4jBF,CSrtBA,MACE,+PTutBF,CSjtBA,cASE,mBAAA,CAFA,0CAAA,CACA,cAAA,CAFA,YAAA,CAIA,uCAAA,CACA,oBAAA,CAVA,iBAAA,CAEA,UAAA,CADA,QAAA,CAUA,qBAAA,CAPA,WAAA,CADA,ST4tBF,CSjtBE,aAfF,cAgBI,YTotBF,CACF,CSjtBE,kCAEE,uCAAA,CADA,YTotBJ,CS/sBE,qBACE,uCTitBJ,CS7sBE,wCACE,+BT+sBJ,CS1sBE,oBAME,6BAAA,CADA,UAAA,CAJA,aAAA,CAEA,cAAA,CACA,aAAA,CAGA,2CAAA,CAAA,mCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CARA,aTotBJ,CSxsBE,sBACE,cT0sBJ,CSvsBI,2BACE,2CTysBN,CSnsBI,kEAEE,uDAAA,CADA,+BTssBN,CU5wBA,mBACE,GACE,SAAA,CACA,0BV+wBF,CU5wBA,GACE,SAAA,CACA,uBV8wBF,CACF,CU1wBA,mBACE,GACE,SV4wBF,CUzwBA,GACE,SV2wBF,CACF,CUhwBE,qBASE,2BAAA,CADA,mCAAA,CAAA,2BAAA,CAFA,0BAAA,CADA,WAAA,CAEA,SAAA,CANA,cAAA,CACA,KAAA,CAEA,UAAA,CADA,SVwwBJ,CU9vBE,mBAcE,mDAAA,CANA,2CAAA,CACA,QAAA,CACA,mBAAA,CARA,QAAA,CASA,kDACE,CAPF,eAAA,CAEA,aAAA,CADA,SAAA,CALA,cAAA,CAGA,UAAA,CADA,SVywBJ,CU1vBE,kBACE,aV4vBJ,CUxvBE,sBACE,YAAA,CACA,YV0vBJ,CUvvBI,oCACE,aVyvBN,CUpvBE,sBACE,mBVsvBJ,CUnvBI,6CACE,cVqvBN,CK/oBI,0CKvGA,6CAKI,aAAA,CAEA,gBAAA,CACA,iBAAA,CAFA,UVuvBN,CACF,CUhvBE,kBACE,cVkvBJ,CWn1BA,YACE,WAAA,CAIA,WXm1BF,CWh1BE,mBAEE,qBAAA,CADA,iBXm1BJ,CKtrBI,sCMtJE,4EACE,kBX+0BN,CW30BI,0JACE,mBX60BN,CW90BI,8EACE,kBX60BN,CACF,CWx0BI,0BAGE,UAAA,CAFA,aAAA,CACA,YX20BN,CWt0BI,+BACE,eXw0BN,CWl0BE,8BACE,WXu0BJ,CWx0BE,8BACE,UXu0BJ,CWx0BE,8BAIE,iBXo0BJ,CWx0BE,8BAIE,kBXo0BJ,CWx0BE,oBAGE,cAAA,CADA,SXs0BJ,CWj0BI,aAPF,oBAQI,YXo0BJ,CACF,CWj0BI,gCACE,yCXm0BN,CW/zBI,wBACE,cAAA,CACA,kBXi0BN,CW9zBM,kCACE,oBXg0BR,CYj4BA,qBAeE,WZk4BF,CYj5BA,qBAeE,UZk4BF,CYj5BA,WAOE,2CAAA,CACA,mBAAA,CANA,YAAA,CAOA,8BAAA,CALA,iBAAA,CAMA,SAAA,CALA,mBAAA,CACA,mBAAA,CALA,cAAA,CAaA,0BAAA,CAHA,wCACE,CATF,SZ84BF,CY/3BE,aAlBF,WAmBI,YZk4BF,CACF,CY/3BE,mBAEE,SAAA,CADA,mBAAA,CAKA,uBAAA,CAHA,kEZk4BJ,CY33BE,kBAEE,gCAAA,CADA,eZ83BJ,Cah6BA,aACE,gBAAA,CACA,iBbm6BF,Cah6BE,sBAGE,WAAA,CADA,QAAA,CADA,Sbo6BJ,Ca95BE,oBAEE,eAAA,CADA,ebi6BJ,Ca55BE,oBACE,iBb85BJ,Ca15BE,mBAEE,YAAA,CACA,cAAA,CACA,6BAAA,CAHA,iBb+5BJ,Caz5BI,iDACE,yCb25BN,Cav5BI,6BACE,iBby5BN,Cap5BE,mBAGE,uCAAA,CACA,cAAA,CAHA,aAAA,CACA,cAAA,CAGA,sBbs5BJ,Can5BI,gDACE,+Bbq5BN,Caj5BI,4BACE,0CAAA,CACA,mBbm5BN,Ca94BE,mBAEE,SAAA,CADA,iBAAA,CAKA,2BAAA,CAHA,8Dbi5BJ,Ca34BI,qBAEE,aAAA,CADA,eb84BN,Caz4BI,6BACE,SAAA,CACA,uBb24BN,Ccz9BA,WAEE,0CAAA,CADA,+Bd69BF,Ccz9BE,aALF,WAMI,Yd49BF,CACF,Ccz9BE,kBACE,6BAAA,CAEA,aAAA,CADA,ad49BJ,Ccx9BI,gCACE,Yd09BN,Ccr9BE,iBAOE,eAAA,CANA,YAAA,CAKA,cAAA,CAGA,mBAAA,CAAA,eAAA,CADA,cAAA,CAGA,uCAAA,CADA,eAAA,CAEA,uBdm9BJ,Cch9BI,8CACE,Udk9BN,Cc98BI,+BACE,oBdg9BN,CKl0BI,0CSvIE,uBACE,ad48BN,Ccz8BM,yCACE,Yd28BR,CACF,Cct8BI,iCACE,gBdy8BN,Cc18BI,iCACE,iBdy8BN,Cc18BI,uBAEE,gBdw8BN,Ccr8BM,iCACE,edu8BR,Ccj8BE,kBACE,WAAA,CAIA,eAAA,CADA,mBAAA,CAFA,6BAAA,CACA,cAAA,CAGA,kBdm8BJ,Cc/7BE,mBAEE,YAAA,CADA,adk8BJ,Cc77BE,sBACE,gBAAA,CACA,Ud+7BJ,Cc17BA,gBACE,gDd67BF,Cc17BE,uBACE,YAAA,CACA,cAAA,CACA,6BAAA,CACA,ad47BJ,Ccx7BE,kCACE,sCd07BJ,Ccv7BI,gFACE,+Bdy7BN,Ccj7BA,cAKE,wCAAA,CADA,gBAAA,CADA,iBAAA,CADA,eAAA,CADA,Udw7BF,CK54BI,mCS7CJ,cASI,Udo7BF,CACF,Cch7BE,yBACE,sCdk7BJ,Cc36BA,WACE,mBAAA,CACA,SAAA,CAEA,cAAA,CADA,qBd+6BF,CK35BI,mCSvBJ,WAQI,ed86BF,CACF,Cc36BE,iBACE,oBAAA,CAEA,aAAA,CACA,iBAAA,CAFA,Yd+6BJ,Cc16BI,wBACE,ed46BN,Ccx6BI,qBAGE,iBAAA,CAFA,gBAAA,CACA,mBd26BN,CejlCE,uBAME,kBAAA,CACA,mBAAA,CAHA,gCAAA,CACA,cAAA,CAJA,oBAAA,CAEA,eAAA,CADA,kBAAA,CAMA,gEfolCJ,Ce9kCI,gCAEE,2CAAA,CACA,uCAAA,CAFA,gCfklCN,Ce5kCI,0DAEE,0CAAA,CACA,sCAAA,CAFA,+BfglCN,CezkCE,gCAKE,4Bf8kCJ,CenlCE,gEAME,6Bf6kCJ,CenlCE,gCAME,4Bf6kCJ,CenlCE,sBAIE,6DAAA,CAGA,8BAAA,CAJA,eAAA,CAFA,aAAA,CACA,eAAA,CAMA,sCf2kCJ,CetkCI,wDACE,6CAAA,CACA,8BfwkCN,CepkCI,+BACE,UfskCN,CgBznCA,WAOE,2CAAA,CAGA,8CACE,CALF,gCAAA,CADA,aAAA,CAHA,MAAA,CADA,eAAA,CACA,OAAA,CACA,KAAA,CACA,ShBgoCF,CgBrnCE,aAfF,WAgBI,YhBwnCF,CACF,CgBrnCE,mBAIE,2BAAA,CAHA,iEhBwnCJ,CgBjnCE,mBACE,kDACE,CAEF,kEhBinCJ,CgB3mCE,kBAEE,kBAAA,CADA,YAAA,CAEA,ehB6mCJ,CgBzmCE,mBAKE,kBAAA,CAEA,cAAA,CAHA,YAAA,CAIA,uCAAA,CALA,aAAA,CAFA,iBAAA,CAQA,uBAAA,CAHA,qBAAA,CAJA,ShBknCJ,CgBxmCI,yBACE,UhB0mCN,CgBtmCI,iCACE,oBhBwmCN,CgBpmCI,uCAEE,uCAAA,CADA,YhBumCN,CgBlmCI,2BAEE,YAAA,CADA,ahBqmCN,CKv/BI,0CW/GA,2BAMI,YhBomCN,CACF,CgBjmCM,8DAIE,iBAAA,CAHA,aAAA,CAEA,aAAA,CADA,UhBqmCR,CKrhCI,mCWzEA,iCAII,YhB8lCN,CACF,CgB3lCM,wCACE,YhB6lCR,CgBzlCM,+CACE,oBhB2lCR,CKhiCI,sCWtDA,iCAII,YhBslCN,CACF,CgBjlCE,kBAEE,YAAA,CACA,cAAA,CAFA,iBAAA,CAIA,8DACE,CAFF,kBhBolCJ,CgB9kCI,oCAGE,SAAA,CADA,mBAAA,CAKA,6BAAA,CAHA,8DACE,CAJF,UhBolCN,CgB3kCM,8CACE,8BhB6kCR,CgBxkCI,8BACE,ehB0kCN,CgBrkCE,4BAGE,gBAAA,CAAA,kBhBykCJ,CgB5kCE,4BAGE,iBAAA,CAAA,iBhBykCJ,CgB5kCE,kBACE,WAAA,CAGA,eAAA,CAFA,aAAA,CAGA,kBhBukCJ,CgBpkCI,4CAGE,SAAA,CADA,mBAAA,CAKA,8BAAA,CAHA,8DACE,CAJF,UhB0kCN,CgBjkCM,sDACE,6BhBmkCR,CgB/jCM,8DAGE,SAAA,CADA,mBAAA,CAKA,uBAAA,CAHA,8DACE,CAJF,ShBqkCR,CgB1jCI,uCAGE,WAAA,CAFA,iBAAA,CACA,UhB6jCN,CgBvjCE,mBACE,YAAA,CACA,aAAA,CACA,cAAA,CAEA,+CACE,CAFF,kBhB0jCJ,CgBpjCI,8DACE,WAAA,CACA,SAAA,CACA,oChBsjCN,CgB7iCI,yBACE,QhB+iCN,CgB1iCE,mBACE,YhB4iCJ,CKxmCI,mCW2DF,6BAQI,gBhB4iCJ,CgBpjCA,6BAQI,iBhB4iCJ,CgBpjCA,mBAKI,aAAA,CAEA,iBAAA,CADA,ahB8iCJ,CACF,CKhnCI,sCW2DF,6BAaI,kBhB4iCJ,CgBzjCA,6BAaI,mBhB4iCJ,CACF,CD3xCA,SAGE,uCAAA,CAFA,eAAA,CACA,eC+xCF,CD3xCE,eACE,mBAAA,CACA,cAAA,CAGA,eAAA,CADA,QAAA,CADA,SC+xCJ,CDzxCE,sCAEE,WAAA,CADA,iBAAA,CAAA,kBC4xCJ,CDvxCE,eACE,+BCyxCJ,CDtxCI,0CACE,+BCwxCN,CDlxCA,UAKE,wBkBaa,ClBZb,oBAAA,CAFA,UAAA,CAHA,oBAAA,CAEA,eAAA,CADA,0BAAA,CAAA,2BCyxCF,CkB3zCA,MACE,0MAAA,CACA,gMAAA,CACA,yNlB8zCF,CkBxzCA,QACE,eAAA,CACA,elB2zCF,CkBxzCE,eAKE,uCAAA,CAJA,aAAA,CAGA,eAAA,CADA,eAAA,CADA,eAAA,CAIA,sBlB0zCJ,CkBvzCI,+BACE,YlByzCN,CkBtzCM,mCAEE,WAAA,CADA,UlByzCR,CkBjzCQ,sFAME,iBAAA,CALA,aAAA,CAGA,aAAA,CADA,cAAA,CAEA,kBAAA,CAHA,UlBuzCV,CkB5yCE,cAGE,eAAA,CADA,QAAA,CADA,SlBgzCJ,CkB1yCE,cAGE,sBAAA,CAFA,YAAA,CACA,SAAA,CAEA,iBAAA,CAEA,uBAAA,CADA,sBlB6yCJ,CkBzyCI,sBACE,uClB2yCN,CkBpyCM,6EAEE,+BlBsyCR,CkBjyCI,2BAIE,iBlBgyCN,CkB5xCI,4CACE,gBlB8xCN,CkB/xCI,4CACE,iBlB8xCN,CkB1xCI,kBAGE,iBAAA,CAFA,aAAA,CACA,YlB6xCN,CkBxxCI,sGACE,+BAAA,CACA,clB0xCN,CkBtxCI,4BACE,uCAAA,CACA,oBlBwxCN,CkBpxCI,0CACE,YlBsxCN,CkBnxCM,yDAKE,6BAAA,CAJA,aAAA,CAEA,WAAA,CACA,qCAAA,CAAA,6BAAA,CAFA,UlBwxCR,CkBjxCM,kDACE,YlBmxCR,CkB7wCE,iCACE,YlB+wCJ,CkB5wCI,6CACE,WAAA,CAGA,WlB4wCN,CkBvwCE,cACE,alBywCJ,CkBrwCE,gBACE,YlBuwCJ,CKruCI,0Ca3BA,0CASE,2CAAA,CAHA,YAAA,CACA,qBAAA,CACA,WAAA,CALA,MAAA,CADA,iBAAA,CACA,OAAA,CACA,KAAA,CACA,SlBswCJ,CkB3vCI,+DACE,eAAA,CACA,elB6vCN,CkBzvCI,gCAQE,qDAAA,CAHA,uCAAA,CAEA,cAAA,CALA,aAAA,CAEA,kBAAA,CADA,wBAAA,CAFA,iBAAA,CAKA,kBlB6vCN,CkBxvCM,wDAGE,UlB8vCR,CkBjwCM,wDAGE,WlB8vCR,CkBjwCM,8CAIE,aAAA,CAEA,aAAA,CACA,YAAA,CANA,iBAAA,CACA,SAAA,CAGA,YlB4vCR,CkBvvCQ,oDAKE,6BAAA,CADA,UAAA,CAHA,aAAA,CAEA,WAAA,CAGA,2CAAA,CAAA,mCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAPA,UlBgwCV,CkBpvCM,8CAGE,2CAAA,CACA,gEACE,CAJF,eAAA,CAKA,4BAAA,CAJA,kBlByvCR,CkBlvCQ,2DACE,YlBovCV,CkB/uCM,8CAGE,2CAAA,CADA,gCAAA,CADA,elBmvCR,CkB7uCM,yCAIE,aAAA,CAFA,UAAA,CAIA,YAAA,CADA,aAAA,CAJA,iBAAA,CACA,WAAA,CACA,SlBkvCR,CkB1uCI,+BACE,MlB4uCN,CkBxuCI,+BACE,4DlB0uCN,CkBvuCM,qDACE,+BlByuCR,CkBtuCQ,sHACE,+BlBwuCV,CkBluCI,+BAEE,YAAA,CADA,mBlBquCN,CkBjuCM,mCACE,elBmuCR,CkB/tCM,6CACE,SlBiuCR,CkB7tCM,uDAGE,mBlBguCR,CkBnuCM,uDAGE,kBlBguCR,CkBnuCM,6CAIE,gBAAA,CAFA,aAAA,CADA,YlBkuCR,CkB5tCQ,mDAKE,6BAAA,CADA,UAAA,CAHA,aAAA,CAEA,WAAA,CAGA,2CAAA,CAAA,mCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAPA,UlBquCV,CkBrtCM,+CACE,mBlButCR,CkB/sCM,4CAEE,wBAAA,CADA,elBktCR,CkB9sCQ,oEACE,mBlBgtCV,CkBjtCQ,oEACE,oBlBgtCV,CkB5sCQ,4EACE,iBlB8sCV,CkB/sCQ,4EACE,kBlB8sCV,CkB1sCQ,oFACE,mBlB4sCV,CkB7sCQ,oFACE,oBlB4sCV,CkBxsCQ,4FACE,mBlB0sCV,CkB3sCQ,4FACE,oBlB0sCV,CkBnsCE,mBACE,wBlBqsCJ,CkBjsCE,wBACE,YAAA,CACA,SAAA,CAIA,0BAAA,CAHA,oElBosCJ,CkB9rCI,kCACE,2BlBgsCN,CkB3rCE,gCACE,SAAA,CAIA,uBAAA,CAHA,qElB8rCJ,CkBxrCI,8CAEE,kCAAA,CAAA,0BlByrCN,CACF,CKx3CI,0CauMA,0CACE,YlBorCJ,CkBjrCI,yDACE,UlBmrCN,CkB/qCI,wDACE,YlBirCN,CkB7qCI,kDACE,YlB+qCN,CkB1qCE,gBAIE,iDAAA,CADA,gCAAA,CAFA,aAAA,CACA,elB8qCJ,CACF,CKr7CM,+DagRF,6CACE,YlBwqCJ,CkBrqCI,4DACE,UlBuqCN,CkBnqCI,2DACE,YlBqqCN,CkBjqCI,qDACE,YlBmqCN,CACF,CK76CI,mCa7JJ,QA6aI,oBlBiqCF,CkB3pCI,kCAME,qCAAA,CACA,qDAAA,CANA,eAAA,CACA,KAAA,CAGA,SlB6pCN,CkBxpCM,6CACE,uBlB0pCR,CkBtpCM,gDACE,YlBwpCR,CkBnpCI,2CACE,kBlBspCN,CkBvpCI,2CACE,mBlBspCN,CkBvpCI,iCAEE,oBlBqpCN,CkB9oCI,yDACE,kBlBgpCN,CkBjpCI,yDACE,iBlBgpCN,CACF,CKt8CI,sCa7JJ,QAydI,oBAAA,CACA,oDlB8oCF,CkBxoCI,gCAME,qCAAA,CACA,qDAAA,CANA,eAAA,CACA,KAAA,CAGA,SlB0oCN,CkBroCM,8CACE,uBlBuoCR,CkBnoCM,8CACE,YlBqoCR,CkBhoCI,yCACE,kBlBmoCN,CkBpoCI,yCACE,mBlBmoCN,CkBpoCI,+BAEE,oBlBkoCN,CkB3nCI,uDACE,kBlB6nCN,CkB9nCI,uDACE,iBlB6nCN,CkBxnCE,wBACE,YAAA,CACA,sBAAA,CAEA,SAAA,CACA,6FACE,CAHF,mBlB4nCJ,CkBpnCI,sCACE,elBsnCN,CkBjnCE,iFACE,sBAAA,CAEA,SAAA,CACA,4FACE,CAHF,kBlBqnCJ,CkB5mCE,iDACE,elB8mCJ,CkB1mCE,6CACE,YlB4mCJ,CkBxmCE,uBACE,aAAA,CACA,elB0mCJ,CkBvmCI,kCACE,elBymCN,CkBrmCI,qCACE,elBumCN,CkBpmCM,0CACE,uClBsmCR,CkBlmCM,6DACE,mBlBomCR,CkBhmCM,yFAEE,YlBkmCR,CkB7lCI,yCAEE,kBlBimCN,CkBnmCI,yCAEE,mBlBimCN,CkBnmCI,+BACE,aAAA,CAGA,SAAA,CADA,kBlBgmCN,CkB5lCM,2DACE,SlB8lCR,CkBxlCE,cAGE,kBAAA,CADA,YAAA,CAEA,gCAAA,CAHA,WlB6lCJ,CkBvlCI,oBACE,uDlBylCN,CkBrlCI,oBAME,6BAAA,CACA,kBAAA,CAFA,UAAA,CAJA,oBAAA,CAEA,WAAA,CAMA,2CAAA,CAAA,mCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAJA,yBAAA,CAJA,qBAAA,CAFA,UlBimCN,CkBplCM,8BACE,wBlBslCR,CkBllCM,kKAEE,uBlBmlCR,CkBrkCI,2EACE,YlB0kCN,CkBvkCM,oDACE,alBykCR,CkBtkCQ,kEAKE,qCAAA,CACA,qDAAA,CAFA,YAAA,CAHA,eAAA,CACA,KAAA,CACA,SlB2kCV,CkBrkCU,0FACE,mBlBukCZ,CkBlkCQ,0EACE,QlBokCV,CkB/jCM,8DACE,kBlBikCR,CkBlkCM,8DACE,mBlBikCR,CkB7jCM,kDACE,uClB+jCR,CkBzjCI,2CACE,sBAAA,CAEA,SAAA,CADA,kBlB4jCN,CkBnjCI,mFACE,elBqjCN,CkBljCM,iGACE,SlBojCR,CkB/iCI,qFAIE,mDlBkjCN,CkBtjCI,qFAIE,oDlBkjCN,CkBtjCI,2EACE,aAAA,CACA,oBAAA,CAGA,SAAA,CAFA,kBlBmjCN,CkB9iCM,yFAEE,gBAAA,CADA,gBlBijCR,CkB5iCM,0FACE,YlB8iCR,CACF,CmB5wDA,eAKE,eAAA,CACA,eAAA,CAJA,SnBmxDF,CmB5wDE,gCANA,kBAAA,CAFA,YAAA,CAGA,sBnB0xDF,CmBrxDE,iBAOE,mBAAA,CAFA,aAAA,CADA,gBAAA,CAEA,iBnB+wDJ,CmB1wDE,wBAEE,qDAAA,CADA,uCnB6wDJ,CmBxwDE,qBACE,6CnB0wDJ,CmBrwDI,sDAEE,uDAAA,CADA,+BnBwwDN,CmBpwDM,8DACE,+BnBswDR,CmBjwDI,mCACE,uCAAA,CACA,oBnBmwDN,CmB/vDI,yBAKE,iBAAA,CADA,yCAAA,CAHA,aAAA,CAEA,eAAA,CADA,YnBowDN,CoBpzDE,eAGE,+DAAA,CADA,oBAAA,CADA,qBpByzDJ,CKpoDI,0CetLF,eAOI,YpBuzDJ,CACF,CoBjzDM,6BACE,oBpBmzDR,CoB7yDE,kBACE,YAAA,CACA,qBAAA,CACA,SAAA,CACA,qBpB+yDJ,CoBxyDI,0BACE,sBpB0yDN,CoBvyDM,gEACE,+BpByyDR,CoBnyDE,gBAEE,uCAAA,CADA,epBsyDJ,CoBjyDE,kBACE,oBpBmyDJ,CoBhyDI,mCAGE,kBAAA,CAFA,YAAA,CACA,SAAA,CAEA,iBpBkyDN,CoB9xDI,oCAIE,kBAAA,CAHA,mBAAA,CACA,kBAAA,CACA,SAAA,CAGA,QAAA,CADA,iBpBiyDN,CoB5xDI,0DACE,kBpB8xDN,CoB/xDI,0DACE,iBpB8xDN,CoB1xDI,iDACE,uBAAA,CAEA,YpB2xDN,CoBtxDE,4BACE,YpBwxDJ,CoBjxDA,YAGE,kBAAA,CAFA,YAAA,CAIA,eAAA,CAHA,SAAA,CAIA,eAAA,CAFA,UpBsxDF,CoBjxDE,yBACE,WpBmxDJ,CoB5wDA,kBACE,YpB+wDF,CKvsDI,0CezEJ,kBAKI,wBpB+wDF,CACF,CoB5wDE,qCACE,WpB8wDJ,CKluDI,sCe7CF,+CAKI,kBpB8wDJ,CoBnxDA,+CAKI,mBpB8wDJ,CACF,CKptDI,0CerDJ,6BAMI,SAAA,CAFA,eAAA,CACA,UpB2wDF,CoBxwDE,qDACE,gBpB0wDJ,CoBvwDE,gDACE,SpBywDJ,CoBtwDE,4CACE,iBAAA,CAAA,kBpBwwDJ,CoBrwDE,2CAEE,WAAA,CADA,cpBwwDJ,CoBpwDE,2CACE,mBAAA,CACA,cAAA,CACA,SAAA,CACA,oBAAA,CAAA,iBpBswDJ,CoBnwDE,2CACE,SpBqwDJ,CoBlwDE,qCAEE,WAAA,CACA,eAAA,CAFA,epBswDJ,CACF,CqBh7DA,MACE,qBAAA,CACA,yBrBm7DF,CqB76DA,aAME,qCAAA,CADA,cAAA,CAEA,0FACE,CAPF,cAAA,CACA,KAAA,CAaA,mDAAA,CACA,qBAAA,CAJA,wFACE,CATF,UAAA,CADA,SrBu7DF,CsBl8DA,MACE,igBtBq8DF,CsB/7DA,WACE,iBtBk8DF,CKpyDI,mCiB/JJ,WAKI,etBk8DF,CACF,CsB/7DE,kBACE,YtBi8DJ,CsB77DE,oBAEE,SAAA,CADA,StBg8DJ,CK7xDI,0CiBpKF,8BAkBI,YtB67DJ,CsB/8DA,8BAkBI,atB67DJ,CsB/8DA,oBAYI,2CAAA,CACA,kBAAA,CAJA,WAAA,CACA,eAAA,CACA,mBAAA,CALA,iBAAA,CACA,SAAA,CAUA,uBAAA,CAHA,4CACE,CAPF,UtBu8DJ,CsB17DI,+DACE,SAAA,CACA,oCtB47DN,CACF,CKn0DI,mCiBjJF,8BAyCI,MtBs7DJ,CsB/9DA,8BAyCI,OtBs7DJ,CsB/9DA,oBAoCI,0BAAA,CADA,cAAA,CADA,QAAA,CAHA,cAAA,CACA,KAAA,CAKA,sDACE,CALF,OtB87DJ,CsBn7DI,+DAME,YAAA,CACA,SAAA,CACA,4CACE,CARF,UtBw7DN,CACF,CKl0DI,0CiBxGA,+DAII,mBtB06DN,CACF,CKh3DM,+DiB/DF,+DASI,mBtB06DN,CACF,CKr3DM,+DiB/DF,+DAcI,mBtB06DN,CACF,CsBr6DE,kBAEE,kCAAA,CAAA,0BtBs6DJ,CKp1DI,0CiBpFF,4BAmBI,MtBk6DJ,CsBr7DA,4BAmBI,OtBk6DJ,CsBr7DA,kBAUI,QAAA,CAEA,SAAA,CADA,eAAA,CALA,cAAA,CACA,KAAA,CAWA,wBAAA,CALA,qGACE,CALF,OAAA,CADA,StB66DJ,CsB/5DI,4BACE,yBtBi6DN,CsB75DI,6DAEE,WAAA,CACA,SAAA,CAMA,uBAAA,CALA,sGACE,CAJF,UtBm6DN,CACF,CK/3DI,mCiBjEF,4BA2CI,WtB65DJ,CsBx8DA,4BA2CI,UtB65DJ,CsBx8DA,kBA6CI,eAAA,CAHA,iBAAA,CAIA,8CAAA,CAFA,atB45DJ,CACF,CK95DM,+DiBOF,6DAII,atBu5DN,CACF,CK74DI,sCiBfA,6DASI,atBu5DN,CACF,CsBl5DE,iBAIE,2CAAA,CACA,0BAAA,CAFA,aAAA,CAFA,iBAAA,CAKA,2CACE,CALF,StBw5DJ,CK15DI,mCiBAF,iBAaI,0BAAA,CACA,mBAAA,CAFA,atBo5DJ,CsB/4DI,uBACE,0BtBi5DN,CACF,CsB74DI,4DAEE,2CAAA,CACA,6BAAA,CACA,8BAAA,CAHA,gCtBk5DN,CsB14DE,4BAKE,mBAAA,CAAA,oBtB+4DJ,CsBp5DE,4BAKE,mBAAA,CAAA,oBtB+4DJ,CsBp5DE,kBAQE,gBAAA,CAFA,eAAA,CAFA,WAAA,CAHA,iBAAA,CAMA,sBAAA,CAJA,UAAA,CADA,StBk5DJ,CsBz4DI,+BACE,qBtB24DN,CsBv4DI,kEAEE,uCtBw4DN,CsBp4DI,6BACE,YtBs4DN,CK16DI,0CiBaF,kBA8BI,eAAA,CADA,aAAA,CADA,UtBu4DJ,CACF,CKp8DI,mCiBgCF,4BAmCI,mBtBu4DJ,CsB16DA,4BAmCI,oBtBu4DJ,CsB16DA,kBAqCI,aAAA,CADA,etBs4DJ,CsBl4DI,+BACE,uCtBo4DN,CsBh4DI,mCACE,gCtBk4DN,CsB93DI,6DACE,kBtBg4DN,CsB73DM,8EACE,uCtB+3DR,CsB33DM,0EACE,WtB63DR,CACF,CsBv3DE,iBAIE,cAAA,CAHA,oBAAA,CAEA,aAAA,CAEA,kCACE,CAJF,YtB43DJ,CsBp3DI,uBACE,UtBs3DN,CsBl3DI,yCAGE,UtBq3DN,CsBx3DI,yCAGE,WtBq3DN,CsBx3DI,+BACE,iBAAA,CACA,SAAA,CAEA,StBo3DN,CsBj3DM,6CACE,oBtBm3DR,CK19DI,0CiB+FA,yCAcI,UtBk3DN,CsBh4DE,yCAcI,WtBk3DN,CsBh4DE,+BAaI,StBm3DN,CsB/2DM,+CACE,YtBi3DR,CACF,CKt/DI,mCiBkHA,+BAwBI,mBtBg3DN,CsB72DM,8CACE,YtB+2DR,CACF,CsBz2DE,8BAGE,WtB62DJ,CsBh3DE,8BAGE,UtB62DJ,CsBh3DE,oBAKE,mBAAA,CAJA,iBAAA,CACA,SAAA,CAEA,StB42DJ,CKl/DI,0CiBkIF,8BAUI,WtB22DJ,CsBr3DA,8BAUI,UtB22DJ,CsBr3DA,oBASI,StB42DJ,CACF,CsBx2DI,uCACE,iBtB82DN,CsB/2DI,uCACE,kBtB82DN,CsB/2DI,6BAEE,uCAAA,CACA,SAAA,CAIA,oBAAA,CAHA,+DtB22DN,CsBr2DM,iDAEE,uCAAA,CADA,YtBw2DR,CsBn2DM,gGAGE,SAAA,CADA,mBAAA,CAEA,kBtBo2DR,CsBj2DQ,sGACE,UtBm2DV,CsB51DE,8BAOE,mBAAA,CAAA,oBtBm2DJ,CsB12DE,8BAOE,mBAAA,CAAA,oBtBm2DJ,CsB12DE,oBAIE,kBAAA,CAKA,yCAAA,CANA,YAAA,CAKA,eAAA,CAFA,WAAA,CAKA,SAAA,CAVA,iBAAA,CACA,KAAA,CAUA,uBAAA,CAFA,kBAAA,CALA,UtBq2DJ,CK5iEI,mCiBkMF,8BAgBI,mBtB+1DJ,CsB/2DA,8BAgBI,oBtB+1DJ,CsB/2DA,oBAiBI,etB81DJ,CACF,CsB31DI,+DACE,SAAA,CACA,0BtB61DN,CsBx1DE,6BAKE,+BtB21DJ,CsBh2DE,0DAME,gCtB01DJ,CsBh2DE,6BAME,+BtB01DJ,CsBh2DE,mBAIE,eAAA,CAHA,iBAAA,CAEA,UAAA,CADA,StB81DJ,CK3iEI,0CiB2MF,mBAWI,QAAA,CADA,UtB21DJ,CACF,CKpkEI,mCiB8NF,mBAiBI,SAAA,CADA,UAAA,CAEA,sBtB01DJ,CsBv1DI,8DACE,8BAAA,CACA,StBy1DN,CACF,CsBp1DE,uBASE,kCAAA,CAAA,0BAAA,CAFA,2CAAA,CANA,WAAA,CACA,eAAA,CAIA,kBtBq1DJ,CsB/0DI,iEAZF,uBAaI,uBtBk1DJ,CACF,CKjnEM,+DiBiRJ,uBAkBI,atBk1DJ,CACF,CKhmEI,sCiB2PF,uBAuBI,atBk1DJ,CACF,CKrmEI,mCiB2PF,uBA4BI,YAAA,CAEA,yDAAA,CADA,oBtBm1DJ,CsB/0DI,kEACE,etBi1DN,CsB70DI,6BACE,+CtB+0DN,CsB30DI,0CAEE,YAAA,CADA,WtB80DN,CsBz0DI,gDACE,oDtB20DN,CsBx0DM,sDACE,0CtB00DR,CACF,CsBn0DA,kBACE,gCAAA,CACA,qBtBs0DF,CsBn0DE,wBAKE,qDAAA,CADA,uCAAA,CAFA,gBAAA,CACA,kBAAA,CAFA,eAAA,CAKA,uBtBq0DJ,CKzoEI,mCiB8TF,kCAUI,mBtBq0DJ,CsB/0DA,kCAUI,oBtBq0DJ,CACF,CsBj0DE,wBAGE,eAAA,CADA,QAAA,CADA,SAAA,CAIA,wBAAA,CAAA,gBtBk0DJ,CsB9zDE,wBACE,yDtBg0DJ,CsB7zDI,oCACE,etB+zDN,CsB1zDE,wBACE,aAAA,CACA,YAAA,CAEA,uBAAA,CADA,gCtB6zDJ,CsBzzDI,4DACE,uDtB2zDN,CsBvzDI,gDACE,mBtByzDN,CsBpzDE,gCAKE,cAAA,CADA,aAAA,CAEA,YAAA,CALA,eAAA,CAMA,uBAAA,CALA,KAAA,CACA,StB0zDJ,CsBnzDI,wCACE,YtBqzDN,CsBhzDI,wDACE,YtBkzDN,CsB9yDI,oCAGE,+BAAA,CADA,gBAAA,CADA,mBAAA,CAGA,2CtBgzDN,CK3rEI,mCiBuYA,8CAUI,mBtB8yDN,CsBxzDE,8CAUI,oBtB8yDN,CACF,CsB1yDI,oFAEE,uDAAA,CADA,+BtB6yDN,CsBvyDE,sCACE,2CtByyDJ,CsBpyDE,2BAGE,eAAA,CADA,eAAA,CADA,iBtBwyDJ,CK5sEI,mCiBmaF,qCAOI,mBtBsyDJ,CsB7yDA,qCAOI,oBtBsyDJ,CACF,CsBlyDE,kCAEE,MtBwyDJ,CsB1yDE,kCAEE,OtBwyDJ,CsB1yDE,wBAME,uCAAA,CAFA,aAAA,CACA,YAAA,CAJA,iBAAA,CAEA,YtBuyDJ,CKtsEI,0CiB4ZF,wBAUI,YtBoyDJ,CACF,CsBjyDI,8BAKE,6BAAA,CADA,UAAA,CAHA,oBAAA,CAEA,WAAA,CAGA,+CAAA,CAAA,uCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAPA,UtB0yDN,CsBhyDM,wCACE,oBtBkyDR,CsB5xDE,8BAGE,uCAAA,CAFA,gBAAA,CACA,etB+xDJ,CsB3xDI,iCAKE,gCAAA,CAHA,eAAA,CACA,eAAA,CACA,eAAA,CAHA,etBiyDN,CsB1xDM,sCACE,oBtB4xDR,CsBvxDI,iCAKE,gCAAA,CAHA,gBAAA,CACA,eAAA,CACA,eAAA,CAHA,atB6xDN,CsBtxDM,sCACE,oBtBwxDR,CsBlxDE,yBAKE,gCAAA,CAJA,aAAA,CAEA,gBAAA,CACA,iBAAA,CAFA,atBuxDJ,CsBhxDE,uBAGE,wBAAA,CAFA,+BAAA,CACA,yBtBmxDJ,CuBv7EA,WACE,iBAAA,CACA,SvB07EF,CuBv7EE,kBAOE,2CAAA,CACA,mBAAA,CACA,8BAAA,CAHA,gCAAA,CAHA,QAAA,CAEA,gBAAA,CADA,YAAA,CAMA,SAAA,CATA,iBAAA,CACA,sBAAA,CAaA,mCAAA,CAJA,oEvB07EJ,CuBn7EI,6EACE,gBAAA,CACA,SAAA,CAKA,+BAAA,CAJA,8EvBs7EN,CuB96EI,wBAWE,+BAAA,CAAA,8CAAA,CAFA,6BAAA,CAAA,8BAAA,CACA,YAAA,CAFA,UAAA,CAHA,QAAA,CAFA,QAAA,CAIA,kBAAA,CADA,iBAAA,CALA,iBAAA,CACA,KAAA,CAEA,OvBu7EN,CuB36EE,iBAOE,mBAAA,CAFA,eAAA,CACA,oBAAA,CAHA,QAAA,CAFA,kBAAA,CAGA,aAAA,CAFA,SvBk7EJ,CuBz6EE,iBACE,kBvB26EJ,CuBv6EE,2BAGE,kBAAA,CAAA,oBvB66EJ,CuBh7EE,2BAGE,mBAAA,CAAA,mBvB66EJ,CuBh7EE,iBAIE,cAAA,CAHA,aAAA,CAIA,YAAA,CAIA,uBAAA,CAHA,2CACE,CALF,UvB86EJ,CuBp6EI,8CACE,+BvBs6EN,CuBl6EI,uBACE,qDvBo6EN,CwBx/EA,YAIE,qBAAA,CADA,aAAA,CAGA,gBAAA,CALA,eAAA,CACA,UAAA,CAGA,axB4/EF,CwBx/EE,aATF,YAUI,YxB2/EF,CACF,CK70EI,0CmB3KF,+BAeI,axBs/EJ,CwBrgFA,+BAeI,cxBs/EJ,CwBrgFA,qBAUI,2CAAA,CAHA,aAAA,CAEA,WAAA,CALA,cAAA,CACA,KAAA,CASA,uBAAA,CAHA,iEACE,CAJF,aAAA,CAFA,SxB+/EJ,CwBn/EI,mEACE,8BAAA,CACA,6BxBq/EN,CwBl/EM,6EACE,8BxBo/ER,CwB/+EI,6CAEE,QAAA,CAAA,MAAA,CACA,QAAA,CAEA,eAAA,CAJA,iBAAA,CACA,OAAA,CAEA,qBAAA,CAFA,KxBo/EN,CACF,CK53EI,sCmBtKJ,YAuDI,QxB++EF,CwB5+EE,mBACE,WxB8+EJ,CwB1+EE,6CACE,UxB4+EJ,CACF,CwBx+EE,uBACE,YAAA,CACA,OxB0+EJ,CK34EI,mCmBjGF,uBAMI,QxB0+EJ,CwBv+EI,8BACE,WxBy+EN,CwBr+EI,qCACE,axBu+EN,CwBn+EI,+CACE,kBxBq+EN,CACF,CwBh+EE,wBAUE,uBAAA,CANA,kCAAA,CAAA,0BAAA,CAHA,cAAA,CACA,eAAA,CASA,yDAAA,CAFA,oBxB+9EJ,CwB19EI,2CAEE,YAAA,CADA,WxB69EN,CwBx9EI,mEACE,+CxB09EN,CwBv9EM,qHACE,oDxBy9ER,CwBt9EQ,iIACE,0CxBw9EV,CwBz8EE,wCAGE,wBACE,qBxBy8EJ,CwBr8EE,6BACE,kCxBu8EJ,CwBx8EE,6BACE,iCxBu8EJ,CACF,CKn6EI,0CmB5BF,YAME,0BAAA,CADA,QAAA,CAEA,SAAA,CANA,cAAA,CACA,KAAA,CAMA,sDACE,CALF,OAAA,CADA,SxBw8EF,CwB77EE,4CAEE,WAAA,CACA,SAAA,CACA,4CACE,CAJF,UxBk8EJ,CACF,CyB/mFA,iBACE,GACE,QzBinFF,CyB9mFA,GACE,azBgnFF,CACF,CyB5mFA,gBACE,GACE,SAAA,CACA,0BzB8mFF,CyB3mFA,IACE,SzB6mFF,CyB1mFA,GACE,SAAA,CACA,uBzB4mFF,CACF,CyBpmFA,MACE,+eAAA,CACA,ygBAAA,CACA,mmBAAA,CACA,sfzBsmFF,CyBhmFA,WAOE,kCAAA,CAAA,0BAAA,CANA,aAAA,CACA,gBAAA,CACA,eAAA,CAEA,uCAAA,CAGA,uBAAA,CAJA,kBzBsmFF,CyB/lFE,iBACE,UzBimFJ,CyB7lFE,iBACE,oBAAA,CAEA,aAAA,CACA,qBAAA,CAFA,UzBimFJ,CyB5lFI,+BACE,iBzB+lFN,CyBhmFI,+BACE,kBzB+lFN,CyBhmFI,qBAEE,gBzB8lFN,CyB1lFI,kDACE,iBzB6lFN,CyB9lFI,kDACE,kBzB6lFN,CyB9lFI,kDAEE,iBzB4lFN,CyB9lFI,kDAEE,kBzB4lFN,CyBvlFE,iCAGE,iBzB4lFJ,CyB/lFE,iCAGE,kBzB4lFJ,CyB/lFE,uBACE,oBAAA,CACA,6BAAA,CAEA,eAAA,CACA,sBAAA,CACA,qBzBylFJ,CyBrlFE,kBACE,YAAA,CAMA,gBAAA,CALA,SAAA,CAMA,oBAAA,CAHA,gBAAA,CAIA,WAAA,CAHA,eAAA,CAFA,SAAA,CADA,UzB6lFJ,CyBplFI,iDACE,4BzBslFN,CyBjlFE,iBACE,eAAA,CACA,sBzBmlFJ,CyBhlFI,gDACE,2BzBklFN,CyB9kFI,kCAIE,kBzBslFN,CyB1lFI,kCAIE,iBzBslFN,CyB1lFI,wBAOE,6BAAA,CADA,UAAA,CALA,oBAAA,CAEA,YAAA,CAKA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CALA,uBAAA,CAHA,WzBwlFN,CyB5kFI,iCACE,azB8kFN,CyB1kFI,iCACE,gDAAA,CAAA,wCzB4kFN,CyBxkFI,+BACE,8CAAA,CAAA,sCzB0kFN,CyBtkFI,+BACE,8CAAA,CAAA,sCzBwkFN,CyBpkFI,sCACE,qDAAA,CAAA,6CzBskFN,C0B7tFA,MACE,mSAAA,CACA,oVAAA,CACA,mOAAA,CACA,qZ1BguFF,C0BvtFE,iBAME,kDAAA,CADA,UAAA,CAJA,oBAAA,CAEA,cAAA,CAIA,mCAAA,CAAA,2BAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CANA,0BAAA,CAFA,a1BkuFJ,C0BttFE,uBACE,6B1BwtFJ,C0BptFE,sBACE,wCAAA,CAAA,gC1BstFJ,C0BltFE,6BACE,+CAAA,CAAA,uC1BotFJ,C0BhtFE,4BACE,8CAAA,CAAA,sC1BktFJ,C2B7vFA,SASE,2CAAA,CADA,gCAAA,CAJA,aAAA,CAGA,eAAA,CADA,aAAA,CADA,UAAA,CAFA,S3BowFF,C2B3vFE,aAZF,SAaI,Y3B8vFF,CACF,CKnlFI,0CsBzLJ,SAkBI,Y3B8vFF,CACF,C2B3vFE,iBACE,mB3B6vFJ,C2BzvFE,yBAIE,iB3BgwFJ,C2BpwFE,yBAIE,kB3BgwFJ,C2BpwFE,eAQE,eAAA,CAPA,YAAA,CAMA,eAAA,CAJA,QAAA,CAEA,aAAA,CAHA,SAAA,CAWA,oBAAA,CAPA,kB3B8vFJ,C2BpvFI,kCACE,Y3BsvFN,C2BjvFE,eACE,aAAA,CACA,kBAAA,CAAA,mB3BmvFJ,C2BhvFI,sCACE,aAAA,CACA,S3BkvFN,C2B5uFE,eAOE,kCAAA,CAAA,0BAAA,CANA,YAAA,CAEA,eAAA,CADA,gBAAA,CAMA,UAAA,CAJA,uCAAA,CACA,oBAAA,CAIA,8D3B6uFJ,C2BxuFI,0CACE,aAAA,CACA,S3B0uFN,C2BtuFI,6BAEE,kB3ByuFN,C2B3uFI,6BAEE,iB3ByuFN,C2B3uFI,mBAGE,iBAAA,CAFA,Y3B0uFN,C2BnuFM,2CACE,qB3BquFR,C2BtuFM,2CACE,qB3BwuFR,C2BzuFM,2CACE,qB3B2uFR,C2B5uFM,2CACE,qB3B8uFR,C2B/uFM,2CACE,oB3BivFR,C2BlvFM,2CACE,qB3BovFR,C2BrvFM,2CACE,qB3BuvFR,C2BxvFM,2CACE,qB3B0vFR,C2B3vFM,4CACE,qB3B6vFR,C2B9vFM,4CACE,oB3BgwFR,C2BjwFM,4CACE,qB3BmwFR,C2BpwFM,4CACE,qB3BswFR,C2BvwFM,4CACE,qB3BywFR,C2B1wFM,4CACE,qB3B4wFR,C2B7wFM,4CACE,oB3B+wFR,C2BzwFI,gCACE,SAAA,CAIA,yBAAA,CAHA,wC3B4wFN,C4B/2FA,MACE,wS5Bk3FF,C4Bz2FE,mCACE,mBAAA,CACA,cAAA,CACA,QAAA,CAEA,mBAAA,CADA,kB5B62FJ,C4Bx2FE,oBAGE,kBAAA,CAOA,+CAAA,CACA,oBAAA,CAVA,mBAAA,CAIA,gBAAA,CACA,0BAAA,CACA,eAAA,CALA,QAAA,CAOA,qBAAA,CADA,eAAA,CAJA,wB5Bi3FJ,C4Bv2FI,0BAGE,uCAAA,CAFA,aAAA,CACA,YAAA,CAEA,6C5By2FN,C4Bp2FM,gEAEE,0CAAA,CADA,+B5Bu2FR,C4Bj2FI,yBACE,uB5Bm2FN,C4B31FI,gCAME,oDAAA,CADA,UAAA,CAJA,oBAAA,CAEA,YAAA,CAKA,qCAAA,CAAA,6BAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAJA,iCAAA,CAHA,0BAAA,CAFA,W5Bs2FN,C4Bz1FI,wFACE,0C5B21FN,C6Br6FA,iBACE,GACE,oB7Bw6FF,C6Br6FA,IACE,kB7Bu6FF,C6Bp6FA,GACE,oB7Bs6FF,CACF,C6B95FA,MACE,0NAAA,CACA,uPAAA,CACA,wB7Bg6FF,C6B15FA,YA6BE,kCAAA,CAAA,0BAAA,CAVA,2CAAA,CACA,mBAAA,CACA,8BAAA,CAHA,gCAAA,CADA,sCAAA,CAdA,+IACE,CAYF,8BAAA,CAMA,SAAA,CArBA,iBAAA,CACA,uBAAA,CAyBA,4BAAA,CAJA,uDACE,CATF,6BAAA,CADA,S7B85FF,C6B54FE,oBAEE,SAAA,CAKA,uBAAA,CAJA,2EACE,CAHF,S7Bi5FJ,C6Bv4FE,8CACE,sC7By4FJ,C6Br4FE,mBAEE,gBAAA,CADA,a7Bw4FJ,C6Bp4FI,2CACE,Y7Bs4FN,C6Bl4FI,0CACE,e7Bo4FN,C6B53FA,eACE,eAAA,CAGA,YAAA,CADA,0BAAA,CADA,kB7Bi4FF,C6B53FE,yBACE,a7B83FJ,C6B13FE,oBACE,sCAAA,CACA,iB7B43FJ,C6Bx3FE,6BACE,oBAAA,CAGA,gB7Bw3FJ,C6Bp3FE,sBAmBE,mBAAA,CAbA,cAAA,CAHA,oBAAA,CACA,gBAAA,CAAA,iBAAA,CAIA,YAAA,CAUA,eAAA,CAjBA,iBAAA,CAMA,wBAAA,CAAA,gBAAA,CAFA,uBAAA,CAHA,S7B83FJ,C6Bp3FI,qCACE,uB7Bs3FN,C6B72FI,cAtBF,sBAuBI,W7Bg3FJ,C6B72FI,wCACE,2B7B+2FN,C6B32FI,6BAOE,qCAAA,CACA,+CAAA,CAAA,uC7Bg3FN,C6Bt2FI,yDAZE,UAAA,CADA,YAAA,CAIA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAVA,iBAAA,CACA,SAAA,CAEA,WAAA,CADA,U7Bo4FN,C6Br3FI,4BAOE,oDAAA,CAMA,4CAAA,CAAA,oCAAA,CADA,uBAAA,CAJA,+C7B62FN,C6Bl2FM,gDACE,uB7Bo2FR,C6Bh2FM,mFACE,0C7Bk2FR,CACF,C6B71FI,0CAGE,2BAAA,CADA,uBAAA,CADA,S7Bi2FN,C6B31FI,8CACE,oB7B61FN,C6B11FM,aAJF,8CASI,8CAAA,CACA,iBAAA,CAHA,gCAAA,CADA,eAAA,CADA,cAAA,CAGA,kB7B+1FN,C6B11FM,oDACE,mC7B41FR,CACF,C6Bh1FE,gCAEE,iBAAA,CADA,e7Bo1FJ,C6Bh1FI,mCACE,iB7Bk1FN,C6B/0FM,oDAGE,a7B61FR,C6Bh2FM,oDAGE,c7B61FR,C6Bh2FM,0CAcE,8CAAA,CACA,iBAAA,CALA,gCAAA,CAEA,oBAAA,CACA,qBAAA,CANA,iBAAA,CACA,eAAA,CAHA,UAAA,CAIA,gBAAA,CALA,aAAA,CAEA,cAAA,CALA,iBAAA,CAUA,iBAAA,CATA,S7B81FR,C8BplGA,kBAME,e9BgmGF,C8BtmGA,kBAME,gB9BgmGF,C8BtmGA,QAUE,2CAAA,CACA,oBAAA,CAEA,8BAAA,CALA,uCAAA,CACA,cAAA,CALA,aAAA,CAGA,eAAA,CAKA,YAAA,CAPA,mBAAA,CAJA,cAAA,CACA,UAAA,CAiBA,yBAAA,CALA,mGACE,CAZF,S9BmmGF,C8BhlGE,aAtBF,QAuBI,Y9BmlGF,CACF,C8BhlGE,kBACE,wB9BklGJ,C8B9kGE,gBAEE,SAAA,CADA,mBAAA,CAGA,+BAAA,CADA,uB9BilGJ,C8B7kGI,0BACE,8B9B+kGN,C8B1kGE,4BAEE,0CAAA,CADA,+B9B6kGJ,C8BxkGE,YACE,oBAAA,CACA,oB9B0kGJ,C+B/nGA,oBACE,GACE,mB/BkoGF,CACF,C+B1nGA,MACE,wf/B4nGF,C+BtnGA,YACE,aAAA,CAEA,eAAA,CADA,a/B0nGF,C+BtnGE,+BAOE,kBAAA,CAAA,kB/BunGJ,C+B9nGE,+BAOE,iBAAA,CAAA,mB/BunGJ,C+B9nGE,qBAQE,aAAA,CACA,cAAA,CACA,YAAA,CATA,iBAAA,CAKA,U/BwnGJ,C+BjnGI,qCAIE,iB/BynGN,C+B7nGI,qCAIE,kB/BynGN,C+B7nGI,2BAME,6BAAA,CADA,UAAA,CAJA,oBAAA,CAEA,YAAA,CAIA,yCAAA,CAAA,iCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CARA,W/B2nGN,C+B9mGE,kBAUE,2CAAA,CACA,mBAAA,CACA,8BAAA,CAJA,gCAAA,CACA,oBAAA,CAHA,kBAAA,CAFA,YAAA,CASA,SAAA,CANA,aAAA,CAFA,SAAA,CAJA,iBAAA,CAgBA,4BAAA,CAfA,UAAA,CAYA,+CACE,CAZF,S/B4nGJ,C+B3mGI,+EACE,gBAAA,CACA,SAAA,CACA,sC/B6mGN,C+BvmGI,qCAEE,oCACE,gC/BwmGN,C+BpmGI,2CACE,c/BsmGN,CACF,C+BjmGE,kBACE,kB/BmmGJ,C+B/lGE,4BAGE,kBAAA,CAAA,oB/BsmGJ,C+BzmGE,4BAGE,mBAAA,CAAA,mB/BsmGJ,C+BzmGE,kBAKE,cAAA,CAJA,aAAA,CAKA,YAAA,CAIA,uBAAA,CAHA,2CACE,CAJF,kBAAA,CAFA,U/BumGJ,C+B5lGI,gDACE,+B/B8lGN,C+B1lGI,wBACE,qD/B4lGN,CgC5rGA,MAEI,uWAAA,CAAA,8WAAA,CAAA,sPAAA,CAAA,8xBAAA,CAAA,0MAAA,CAAA,gbAAA,CAAA,gMAAA,CAAA,iQAAA,CAAA,0VAAA,CAAA,6aAAA,CAAA,8SAAA,CAAA,gMhCqtGJ,CgCzsGE,4CAME,8CAAA,CACA,4BAAA,CACA,mBAAA,CACA,8BAAA,CAJA,mCAAA,CAJA,iBAAA,CAGA,gBAAA,CADA,iBAAA,CADA,eAAA,CASA,uBAAA,CADA,2BhC6sGJ,CgCzsGI,aAdF,4CAeI,ehC4sGJ,CACF,CgCzsGI,sEACE,gChC2sGN,CgCtsGI,gDACE,qBhCwsGN,CgCpsGI,gIAEE,iBAAA,CADA,chCusGN,CgClsGI,4FACE,iBhCosGN,CgChsGI,kFACE,ehCksGN,CgC9rGI,0FACE,YhCgsGN,CgC5rGI,8EACE,mBhC8rGN,CgCzrGE,sEAGE,iBAAA,CAAA,mBhCmsGJ,CgCtsGE,sEAGE,kBAAA,CAAA,kBhCmsGJ,CgCtsGE,sEASE,uBhC6rGJ,CgCtsGE,sEASE,wBhC6rGJ,CgCtsGE,sEAUE,4BhC4rGJ,CgCtsGE,4IAWE,6BhC2rGJ,CgCtsGE,sEAWE,4BhC2rGJ,CgCtsGE,kDAOE,0BAAA,CACA,WAAA,CAFA,eAAA,CADA,eAAA,CAHA,oBAAA,CAAA,iBAAA,CADA,iBhCqsGJ,CgCxrGI,kFACE,ehC0rGN,CgCtrGI,oFAOE,UhC4rGN,CgCnsGI,oFAOE,WhC4rGN,CgCnsGI,gEAME,wBfkIU,CenIV,UAAA,CADA,WAAA,CAIA,kDAAA,CAAA,0CAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAVA,iBAAA,CACA,UAAA,CACA,UhCgsGN,CgCprGI,4DACE,4DhCsrGN,CgCxqGE,sDACE,oBhC2qGJ,CgCxqGI,gFACE,gChC0qGN,CgCrqGE,8DACE,0BhCwqGJ,CgCrqGI,4EACE,wBAlBG,CAmBH,kDAAA,CAAA,0ChCuqGN,CgCnqGI,0EACE,ahCqqGN,CgC1rGE,8DACE,oBhC6rGJ,CgC1rGI,wFACE,gChC4rGN,CgCvrGE,sEACE,0BhC0rGJ,CgCvrGI,oFACE,wBAlBG,CAmBH,sDAAA,CAAA,8ChCyrGN,CgCrrGI,kFACE,ahCurGN,CgC5sGE,sDACE,oBhC+sGJ,CgC5sGI,gFACE,gChC8sGN,CgCzsGE,8DACE,0BhC4sGJ,CgCzsGI,4EACE,wBAlBG,CAmBH,kDAAA,CAAA,0ChC2sGN,CgCvsGI,0EACE,ahCysGN,CgC9tGE,oDACE,oBhCiuGJ,CgC9tGI,8EACE,gChCguGN,CgC3tGE,4DACE,0BhC8tGJ,CgC3tGI,0EACE,wBAlBG,CAmBH,iDAAA,CAAA,yChC6tGN,CgCztGI,wEACE,ahC2tGN,CgChvGE,4DACE,oBhCmvGJ,CgChvGI,sFACE,gChCkvGN,CgC7uGE,oEACE,0BhCgvGJ,CgC7uGI,kFACE,wBAlBG,CAmBH,qDAAA,CAAA,6ChC+uGN,CgC3uGI,gFACE,ahC6uGN,CgClwGE,8DACE,oBhCqwGJ,CgClwGI,wFACE,gChCowGN,CgC/vGE,sEACE,0BhCkwGJ,CgC/vGI,oFACE,wBAlBG,CAmBH,sDAAA,CAAA,8ChCiwGN,CgC7vGI,kFACE,ahC+vGN,CgCpxGE,4DACE,oBhCuxGJ,CgCpxGI,sFACE,gChCsxGN,CgCjxGE,oEACE,0BhCoxGJ,CgCjxGI,kFACE,wBAlBG,CAmBH,qDAAA,CAAA,6ChCmxGN,CgC/wGI,gFACE,ahCixGN,CgCtyGE,4DACE,oBhCyyGJ,CgCtyGI,sFACE,gChCwyGN,CgCnyGE,oEACE,0BhCsyGJ,CgCnyGI,kFACE,wBAlBG,CAmBH,qDAAA,CAAA,6ChCqyGN,CgCjyGI,gFACE,ahCmyGN,CgCxzGE,0DACE,oBhC2zGJ,CgCxzGI,oFACE,gChC0zGN,CgCrzGE,kEACE,0BhCwzGJ,CgCrzGI,gFACE,wBAlBG,CAmBH,oDAAA,CAAA,4ChCuzGN,CgCnzGI,8EACE,ahCqzGN,CgC10GE,oDACE,oBhC60GJ,CgC10GI,8EACE,gChC40GN,CgCv0GE,4DACE,0BhC00GJ,CgCv0GI,0EACE,wBAlBG,CAmBH,iDAAA,CAAA,yChCy0GN,CgCr0GI,wEACE,ahCu0GN,CgC51GE,4DACE,oBhC+1GJ,CgC51GI,sFACE,gChC81GN,CgCz1GE,oEACE,0BhC41GJ,CgCz1GI,kFACE,wBAlBG,CAmBH,qDAAA,CAAA,6ChC21GN,CgCv1GI,gFACE,ahCy1GN,CgC92GE,wDACE,oBhCi3GJ,CgC92GI,kFACE,gChCg3GN,CgC32GE,gEACE,0BhC82GJ,CgC32GI,8EACE,wBAlBG,CAmBH,mDAAA,CAAA,2ChC62GN,CgCz2GI,4EACE,ahC22GN,CiC/gHA,MACE,wMjCkhHF,CiCzgHE,sBAEE,uCAAA,CADA,gBjC6gHJ,CiCzgHI,mCACE,ajC2gHN,CiC5gHI,mCACE,cjC2gHN,CiCvgHM,4BACE,sBjCygHR,CiCtgHQ,mCACE,gCjCwgHV,CiCpgHQ,2DACE,SAAA,CAEA,uBAAA,CADA,ejCugHV,CiClgHQ,yGACE,SAAA,CACA,uBjCogHV,CiChgHQ,yCACE,YjCkgHV,CiC3/GE,0BACE,eAAA,CACA,ejC6/GJ,CiC1/GI,+BACE,oBjC4/GN,CiCv/GE,gDACE,YjCy/GJ,CiCr/GE,8BAIE,+BAAA,CAHA,oBAAA,CAEA,WAAA,CAGA,SAAA,CAKA,4BAAA,CAJA,4DACE,CAHF,0BjCy/GJ,CiCh/GI,aAdF,8BAeI,+BAAA,CACA,SAAA,CACA,uBjCm/GJ,CACF,CiCh/GI,wCACE,6BjCk/GN,CiC9+GI,oCACE,+BjCg/GN,CiC5+GI,qCAKE,6BAAA,CADA,UAAA,CAHA,oBAAA,CAEA,YAAA,CAGA,2CAAA,CAAA,mCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAPA,WjCq/GN,CiCx+GQ,mDACE,oBjC0+GV,CkCxlHE,kCAEE,iBlC8lHJ,CkChmHE,kCAEE,kBlC8lHJ,CkChmHE,wBAGE,yCAAA,CAFA,oBAAA,CAGA,SAAA,CACA,mClC2lHJ,CkCtlHI,aAVF,wBAWI,YlCylHJ,CACF,CkCrlHE,6FAEE,SAAA,CACA,mClCulHJ,CkCjlHE,4FAEE,+BlCmlHJ,CkC/kHE,oBACE,yBAAA,CACA,uBAAA,CAGA,yElC+kHJ,CKh9GI,sC6BrHE,qDACE,uBlCwkHN,CACF,CkCnkHE,kEACE,yBlCqkHJ,CkCjkHE,sBACE,0BlCmkHJ,CmC9nHE,2BACE,anCioHJ,CK58GI,0C8BtLF,2BAKI,enCioHJ,CACF,CmC9nHI,6BAGE,0BAAA,CAAA,2BAAA,CADA,eAAA,CAEA,iBAAA,CAHA,yBAAA,CAAA,iBnCmoHN,CmC7nHM,2CACE,kBnC+nHR,CoChpHE,uBACE,4CpCopHJ,CoC/oHE,8CAJE,kCAAA,CAAA,0BpCupHJ,CoCnpHE,uBACE,4CpCkpHJ,CoC7oHE,4BAEE,kCAAA,CAAA,0BAAA,CADA,qCpCgpHJ,CoC5oHI,mCACE,apC8oHN,CoC1oHI,kCACE,apC4oHN,CoCvoHE,0BAKE,eAAA,CAJA,aAAA,CAEA,YAAA,CACA,aAAA,CAFA,kBAAA,CAAA,mBpC4oHJ,CoCtoHI,uCACE,epCwoHN,CoCpoHI,sCACE,kBpCsoHN,CqCnrHA,MACE,8LrCsrHF,CqC7qHE,oBAGE,iBAAA,CAEA,gBAAA,CADA,arC+qHJ,CqC3qHI,wCACE,uBrC6qHN,CqCzqHI,gCAEE,eAAA,CADA,gBrC4qHN,CqCrqHM,wCACE,mBrCuqHR,CqCjqHE,8BAKE,oBrCqqHJ,CqC1qHE,8BAKE,mBrCqqHJ,CqC1qHE,8BAUE,4BrCgqHJ,CqC1qHE,4DAWE,6BrC+pHJ,CqC1qHE,8BAWE,4BrC+pHJ,CqC1qHE,oBASE,cAAA,CANA,aAAA,CACA,eAAA,CAIA,erCkqHJ,CqC5pHI,kCACE,uCAAA,CACA,oBrC8pHN,CqC1pHI,wCAEE,uCAAA,CADA,YrC6pHN,CqCxpHI,oCASE,WrC8pHN,CqCvqHI,oCASE,UrC8pHN,CqCvqHI,0BAME,6BAAA,CADA,UAAA,CADA,WAAA,CAMA,yCAAA,CAAA,iCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAZA,iBAAA,CACA,UAAA,CAMA,sBAAA,CADA,yBAAA,CAJA,UrCoqHN,CqCvpHM,oCACE,wBrCypHR,CqCppHI,4BACE,YrCspHN,CqCjpHI,4CACE,YrCmpHN,CsC7uHE,+DACE,mBAAA,CACA,cAAA,CACA,uBtCgvHJ,CsC7uHI,2EAGE,iBAAA,CADA,eAAA,CADA,atCivHN,CuCvvHE,6BACE,sCvC0vHJ,CuCvvHE,cACE,yCvCyvHJ,CuC7uHE,sIACE,oCvC+uHJ,CuCvuHE,2EACE,qCvCyuHJ,CuC/tHE,wGACE,oCvCiuHJ,CuCxtHE,yFACE,qCvC0tHJ,CuCrtHE,6BACE,kCvCutHJ,CuCjtHE,6CACE,sCvCmtHJ,CuC5sHE,4DACE,sCvC8sHJ,CuCvsHE,4DACE,qCvCysHJ,CuChsHE,yFACE,qCvCksHJ,CuC1rHE,2EACE,sCvC4rHJ,CuCjrHE,wHACE,qCvCmrHJ,CuC9qHE,8BAGE,mBAAA,CADA,gBAAA,CADA,gBvCkrHJ,CuC7qHE,eACE,4CvC+qHJ,CuC5qHE,eACE,4CvC8qHJ,CuC1qHE,gBAIE,+CAAA,CACA,kDAAA,CAJA,aAAA,CAEA,wBAAA,CADA,wBvC+qHJ,CuCxqHE,yBAOE,wCAAA,CACA,+DAAA,CACA,4BAAA,CACA,6BAAA,CARA,iBAAA,CAGA,eAAA,CACA,eAAA,CAFA,cAAA,CADA,oCAAA,CAFA,iBvCmrHJ,CuCvqHI,6BACE,YvCyqHN,CuCtqHM,kCACE,wBAAA,CACA,yBvCwqHR,CuClqHE,iCAaE,wCAAA,CACA,+DAAA,CAJA,uCAAA,CACA,0BAAA,CALA,UAAA,CAJA,oBAAA,CAOA,2BAAA,CADA,2BAAA,CADA,2BAAA,CANA,eAAA,CAWA,wBAAA,CAAA,gBAAA,CAPA,SvC2qHJ,CuCzpHE,sBACE,iBAAA,CACA,iBvC2pHJ,CuCnpHI,sCACE,gBvCqpHN,CuCjpHI,gDACE,YvCmpHN,CuCzoHA,gBACE,iBvC4oHF,CuCxoHE,yCACE,aAAA,CACA,SvC0oHJ,CuCroHE,mBACE,YvCuoHJ,CuCloHE,oBACE,QvCooHJ,CuChoHE,4BACE,WAAA,CACA,SAAA,CACA,evCkoHJ,CuC/nHI,0CACE,YvCioHN,CuC3nHE,yBAKE,wCAAA,CAEA,+BAAA,CADA,4BAAA,CAHA,eAAA,CADA,oDAAA,CAEA,wBAAA,CAAA,gBvCgoHJ,CuCznHE,2BAEE,+DAAA,CADA,2BvC4nHJ,CuCxnHI,+BACE,uCAAA,CACA,gBvC0nHN,CuCrnHE,sBACE,MAAA,CACA,WvCunHJ,CuClnHA,aACE,avCqnHF,CuC3mHE,4BAEE,aAAA,CADA,YvC+mHJ,CuC3mHI,wDAEE,2BAAA,CADA,wBvC8mHN,CuCxmHE,+BAKE,2CAAA,CAEA,+BAAA,CADA,gCAAA,CADA,sBAAA,CAHA,mBAAA,CACA,gBAAA,CAFA,avCgnHJ,CuCvmHI,qCAEE,UAAA,CACA,UAAA,CAFA,avC2mHN,CK7uHI,0CkCiJF,8BACE,iBvCgmHF,CuCtlHE,wSAGE,evC4lHJ,CuCxlHE,sCAEE,mBAAA,CACA,eAAA,CADA,oBAAA,CADA,kBAAA,CAAA,mBvC4lHJ,CACF,CwCp7HI,yDAIE,+BAAA,CACA,8BAAA,CAFA,aAAA,CADA,QAAA,CADA,iBxC07HN,CwCl7HI,uBAEE,uCAAA,CADA,cxCq7HN,CwCh4HM,iHAEE,WAlDkB,CAiDlB,kBxC24HR,CwC54HM,6HAEE,WAlDkB,CAiDlB,kBxCu5HR,CwCx5HM,6HAEE,WAlDkB,CAiDlB,kBxCm6HR,CwCp6HM,oHAEE,WAlDkB,CAiDlB,kBxC+6HR,CwCh7HM,0HAEE,WAlDkB,CAiDlB,kBxC27HR,CwC57HM,uHAEE,WAlDkB,CAiDlB,kBxCu8HR,CwCx8HM,uHAEE,WAlDkB,CAiDlB,kBxCm9HR,CwCp9HM,6HAEE,WAlDkB,CAiDlB,kBxC+9HR,CwCh+HM,yCAEE,WAlDkB,CAiDlB,kBxCm+HR,CwCp+HM,yCAEE,WAlDkB,CAiDlB,kBxCu+HR,CwCx+HM,0CAEE,WAlDkB,CAiDlB,kBxC2+HR,CwC5+HM,uCAEE,WAlDkB,CAiDlB,kBxC++HR,CwCh/HM,wCAEE,WAlDkB,CAiDlB,kBxCm/HR,CwCp/HM,sCAEE,WAlDkB,CAiDlB,kBxCu/HR,CwCx/HM,wCAEE,WAlDkB,CAiDlB,kBxC2/HR,CwC5/HM,oCAEE,WAlDkB,CAiDlB,kBxC+/HR,CwChgIM,2CAEE,WAlDkB,CAiDlB,kBxCmgIR,CwCpgIM,qCAEE,WAlDkB,CAiDlB,kBxCugIR,CwCxgIM,oCAEE,WAlDkB,CAiDlB,kBxC2gIR,CwC5gIM,kCAEE,WAlDkB,CAiDlB,kBxC+gIR,CwChhIM,qCAEE,WAlDkB,CAiDlB,kBxCmhIR,CwCphIM,mCAEE,WAlDkB,CAiDlB,kBxCuhIR,CwCxhIM,qCAEE,WAlDkB,CAiDlB,kBxC2hIR,CwC5hIM,wCAEE,WAlDkB,CAiDlB,kBxC+hIR,CwChiIM,sCAEE,WAlDkB,CAiDlB,kBxCmiIR,CwCpiIM,2CAEE,WAlDkB,CAiDlB,kBxCuiIR,CwC5hIM,iCAEE,WAPkB,CAMlB,iBxC+hIR,CwChiIM,uCAEE,WAPkB,CAMlB,iBxCmiIR,CwCpiIM,mCAEE,WAPkB,CAMlB,iBxCuiIR,CyCznIA,MACE,qMAAA,CACA,mMzC4nIF,CyCnnIE,wBAKE,mBAAA,CAHA,YAAA,CACA,qBAAA,CACA,YAAA,CAHA,iBzC0nIJ,CyChnII,8BAGE,QAAA,CACA,SAAA,CAHA,iBAAA,CACA,OzConIN,CyC/mIM,qCACE,0BzCinIR,CyCplIM,kEACE,0CzCslIR,CyChlIE,2BAKE,uBAAA,CADA,+DAAA,CAHA,YAAA,CACA,cAAA,CACA,aAAA,CAGA,oBzCklIJ,CyC/kII,aATF,2BAUI,gBzCklIJ,CACF,CyC/kII,cAGE,+BACE,iBzC+kIN,CyC5kIM,sCAQE,qCAAA,CANA,QAAA,CAKA,UAAA,CAHA,aAAA,CAEA,UAAA,CAHA,MAAA,CAFA,iBAAA,CAaA,2CAAA,CALA,2DACE,CAGF,kDAAA,CARA,+BzColIR,CACF,CyCtkII,8CACE,YzCwkIN,CyCpkII,iCASE,+BAAA,CACA,6BAAA,CAJA,uCAAA,CAEA,cAAA,CAPA,aAAA,CAGA,gBAAA,CACA,eAAA,CAFA,8BAAA,CAWA,+BAAA,CAHA,2CACE,CALF,kBAAA,CALA,UzCglIN,CyCjkIM,aAII,6CACE,OzCgkIV,CyCjkIQ,8CACE,OzCmkIV,CyCpkIQ,8CACE,OzCskIV,CyCvkIQ,8CACE,OzCykIV,CyC1kIQ,8CACE,OzC4kIV,CyC7kIQ,8CACE,OzC+kIV,CyChlIQ,8CACE,OzCklIV,CyCnlIQ,8CACE,OzCqlIV,CyCtlIQ,8CACE,OzCwlIV,CyCzlIQ,+CACE,QzC2lIV,CyC5lIQ,+CACE,QzC8lIV,CyC/lIQ,+CACE,QzCimIV,CyClmIQ,+CACE,QzComIV,CyCrmIQ,+CACE,QzCumIV,CyCxmIQ,+CACE,QzC0mIV,CyC3mIQ,+CACE,QzC6mIV,CyC9mIQ,+CACE,QzCgnIV,CyCjnIQ,+CACE,QzCmnIV,CyCpnIQ,+CACE,QzCsnIV,CyCvnIQ,+CACE,QzCynIV,CACF,CyCpnIM,uCACE,gCzCsnIR,CyChnIE,4BACE,UzCknIJ,CyC/mII,aAJF,4BAKI,gBzCknIJ,CACF,CyC9mIE,0BACE,YzCgnIJ,CyC7mII,aAJF,0BAKI,azCgnIJ,CyC5mIM,sCACE,OzC8mIR,CyC/mIM,uCACE,OzCinIR,CyClnIM,uCACE,OzConIR,CyCrnIM,uCACE,OzCunIR,CyCxnIM,uCACE,OzC0nIR,CyC3nIM,uCACE,OzC6nIR,CyC9nIM,uCACE,OzCgoIR,CyCjoIM,uCACE,OzCmoIR,CyCpoIM,uCACE,OzCsoIR,CyCvoIM,wCACE,QzCyoIR,CyC1oIM,wCACE,QzC4oIR,CyC7oIM,wCACE,QzC+oIR,CyChpIM,wCACE,QzCkpIR,CyCnpIM,wCACE,QzCqpIR,CyCtpIM,wCACE,QzCwpIR,CyCzpIM,wCACE,QzC2pIR,CyC5pIM,wCACE,QzC8pIR,CyC/pIM,wCACE,QzCiqIR,CyClqIM,wCACE,QzCoqIR,CyCrqIM,wCACE,QzCuqIR,CACF,CyCjqII,+FAEE,QzCmqIN,CyChqIM,yGACE,wBAAA,CACA,yBzCmqIR,CyC1pIM,2DAEE,wBAAA,CACA,yBAAA,CAFA,QzC8pIR,CyCvpIM,iEACE,QzCypIR,CyCtpIQ,qLAGE,wBAAA,CACA,yBAAA,CAFA,QzC0pIV,CyCppIQ,6FACE,wBAAA,CACA,yBzCspIV,CyCjpIM,yDACE,kBzCmpIR,CyC9oII,sCACE,QzCgpIN,CyC3oIE,2BAEE,iBAAA,CAOA,kBAAA,CAHA,uCAAA,CAEA,cAAA,CAPA,aAAA,CAGA,YAAA,CACA,gBAAA,CAEA,mBAAA,CAGA,gCAAA,CAPA,WzCopIJ,CyC1oII,iCAEE,uDAAA,CADA,+BzC6oIN,CyCxoII,iCAKE,6BAAA,CADA,UAAA,CAHA,aAAA,CAEA,WAAA,CAMA,8CAAA,CAAA,sCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CANA,+CACE,CALF,UzCkpIN,CyCnoIE,4BAOE,yEACE,CANF,YAAA,CAGA,aAAA,CAFA,qBAAA,CAGA,mBAAA,CALA,iBAAA,CAYA,wBAAA,CATA,YzCyoIJ,CyC7nII,sCACE,wBzC+nIN,CyC3nII,oCACE,SzC6nIN,CyCznII,kCAGE,wEACE,CAFF,mBAAA,CADA,OzC6nIN,CyCnnIM,uDACE,8CAAA,CAAA,sCzCqnIR,CK3uII,0CoCoIF,wDAEE,kBzC6mIF,CyC/mIA,wDAEE,mBzC6mIF,CyC/mIA,8CAGE,eAAA,CAFA,eAAA,CAGA,iCzC2mIF,CyCvmIE,8DACE,mBzC0mIJ,CyC3mIE,8DACE,kBzC0mIJ,CyC3mIE,oDAEE,UzCymIJ,CyCrmIE,8EAEE,kBzCwmIJ,CyC1mIE,8EAEE,mBzCwmIJ,CyC1mIE,8EAGE,kBzCumIJ,CyC1mIE,8EAGE,mBzCumIJ,CyC1mIE,oEACE,UzCymIJ,CyCnmIE,8EAEE,mBzCsmIJ,CyCxmIE,8EAEE,kBzCsmIJ,CyCxmIE,8EAGE,mBzCqmIJ,CyCxmIE,8EAGE,kBzCqmIJ,CyCxmIE,oEACE,UzCumIJ,CACF,CyCzlIE,cAHF,olDAII,gCzC4lIF,CyCzlIE,g8GACE,uCzC2lIJ,CACF,CyCtlIA,4sDACE,+BzCylIF,CyCrlIA,wmDACE,azCwlIF,C0C38IA,MACE,8WAAA,CACA,uX1C88IF,C0Cr8IE,4BAEE,oBAAA,CADA,iB1Cy8IJ,C0Cp8II,sDAGE,S1Cs8IN,C0Cz8II,sDAGE,U1Cs8IN,C0Cz8II,4CACE,iBAAA,CACA,S1Cu8IN,C0Cj8IE,+CAEE,SAAA,CADA,U1Co8IJ,C0C/7IE,kDAOE,W1Cq8IJ,C0C58IE,kDAOE,Y1Cq8IJ,C0C58IE,wCAME,qDAAA,CADA,UAAA,CADA,aAAA,CAIA,0CAAA,CAAA,kCAAA,CACA,4BAAA,CAAA,oBAAA,CACA,6BAAA,CAAA,qBAAA,CACA,yBAAA,CAAA,iBAAA,CAVA,iBAAA,CACA,SAAA,CACA,Y1Cy8IJ,C0C77IE,gEACE,wBzB2Wa,CyB1Wb,mDAAA,CAAA,2C1C+7IJ,C2C/+IA,QACE,8DAAA,CAGA,+CAAA,CACA,iEAAA,CACA,oDAAA,CACA,sDAAA,CACA,mDAAA,CAGA,qEAAA,CACA,qEAAA,CACA,wEAAA,CACA,0EAAA,CACA,wEAAA,CACA,yEAAA,CACA,kEAAA,CACA,+DAAA,CACA,oEAAA,CACA,oEAAA,CACA,mEAAA,CACA,gEAAA,CACA,uEAAA,CACA,mEAAA,CACA,qEAAA,CACA,oEAAA,CACA,gEAAA,CACA,wEAAA,CACA,qEAAA,CACA,+D3C8+IF,C2Cx+IA,SAEE,kBAAA,CADA,Y3C4+IF,CK32II,mCuChKA,8BACE,U5CmhJJ,C4CphJE,8BACE,W5CmhJJ,C4CphJE,8BAGE,kB5CihJJ,C4CphJE,8BAGE,iB5CihJJ,C4CphJE,oBAKE,mBAAA,CADA,YAAA,CAFA,a5CkhJJ,C4C5gJI,kCACE,W5C+gJN,C4ChhJI,kCACE,U5C+gJN,C4ChhJI,kCAEE,iBAAA,CAAA,c5C8gJN,C4ChhJI,kCAEE,aAAA,CAAA,kB5C8gJN,CACF","file":"main.css"}
\ No newline at end of file
diff --git a/assets/stylesheets/palette.06af60db.min.css b/assets/stylesheets/palette.06af60db.min.css
new file mode 100644
index 0000000000..a640d38abb
--- /dev/null
+++ b/assets/stylesheets/palette.06af60db.min.css
@@ -0,0 +1 @@
+@media screen{[data-md-color-scheme=slate]{--md-default-fg-color:hsla(var(--md-hue),15%,90%,0.82);--md-default-fg-color--light:hsla(var(--md-hue),15%,90%,0.56);--md-default-fg-color--lighter:hsla(var(--md-hue),15%,90%,0.32);--md-default-fg-color--lightest:hsla(var(--md-hue),15%,90%,0.12);--md-default-bg-color:hsla(var(--md-hue),15%,14%,1);--md-default-bg-color--light:hsla(var(--md-hue),15%,14%,0.54);--md-default-bg-color--lighter:hsla(var(--md-hue),15%,14%,0.26);--md-default-bg-color--lightest:hsla(var(--md-hue),15%,14%,0.07);--md-code-fg-color:hsla(var(--md-hue),18%,86%,0.82);--md-code-bg-color:hsla(var(--md-hue),15%,18%,1);--md-code-hl-color:#2977ff;--md-code-hl-color--light:#2977ff1a;--md-code-hl-number-color:#e6695b;--md-code-hl-special-color:#f06090;--md-code-hl-function-color:#c973d9;--md-code-hl-constant-color:#9383e2;--md-code-hl-keyword-color:#6791e0;--md-code-hl-string-color:#2fb170;--md-code-hl-name-color:var(--md-code-fg-color);--md-code-hl-operator-color:var(--md-default-fg-color--light);--md-code-hl-punctuation-color:var(--md-default-fg-color--light);--md-code-hl-comment-color:var(--md-default-fg-color--light);--md-code-hl-generic-color:var(--md-default-fg-color--light);--md-code-hl-variable-color:var(--md-default-fg-color--light);--md-typeset-color:var(--md-default-fg-color);--md-typeset-a-color:var(--md-primary-fg-color);--md-typeset-kbd-color:hsla(var(--md-hue),15%,90%,0.12);--md-typeset-kbd-accent-color:hsla(var(--md-hue),15%,90%,0.2);--md-typeset-kbd-border-color:hsla(var(--md-hue),15%,14%,1);--md-typeset-mark-color:#4287ff4d;--md-typeset-table-color:hsla(var(--md-hue),15%,95%,0.12);--md-typeset-table-color--light:hsla(var(--md-hue),15%,95%,0.035);--md-admonition-fg-color:var(--md-default-fg-color);--md-admonition-bg-color:var(--md-default-bg-color);--md-footer-bg-color:hsla(var(--md-hue),15%,10%,0.87);--md-footer-bg-color--dark:hsla(var(--md-hue),15%,8%,1);--md-shadow-z1:0 0.2rem 0.5rem #0000000d,0 0 0.05rem #0000001a;--md-shadow-z2:0 0.2rem 0.5rem #00000040,0 0 0.05rem #00000040;--md-shadow-z3:0 0.2rem 0.5rem #0006,0 0 0.05rem #00000059;color-scheme:dark}[data-md-color-scheme=slate] img[src$="#gh-light-mode-only"],[data-md-color-scheme=slate] img[src$="#only-light"]{display:none}[data-md-color-scheme=slate][data-md-color-primary=pink]{--md-typeset-a-color:#ed5487}[data-md-color-scheme=slate][data-md-color-primary=purple]{--md-typeset-a-color:#c46fd3}[data-md-color-scheme=slate][data-md-color-primary=deep-purple]{--md-typeset-a-color:#a47bea}[data-md-color-scheme=slate][data-md-color-primary=indigo]{--md-typeset-a-color:#5488e8}[data-md-color-scheme=slate][data-md-color-primary=teal]{--md-typeset-a-color:#00ccb8}[data-md-color-scheme=slate][data-md-color-primary=green]{--md-typeset-a-color:#71c174}[data-md-color-scheme=slate][data-md-color-primary=deep-orange]{--md-typeset-a-color:#ff764d}[data-md-color-scheme=slate][data-md-color-primary=brown]{--md-typeset-a-color:#c1775c}[data-md-color-scheme=slate][data-md-color-primary=black],[data-md-color-scheme=slate][data-md-color-primary=blue-grey],[data-md-color-scheme=slate][data-md-color-primary=grey],[data-md-color-scheme=slate][data-md-color-primary=white]{--md-typeset-a-color:#5e8bde}[data-md-color-switching] *,[data-md-color-switching] :after,[data-md-color-switching] :before{transition-duration:0ms!important}}[data-md-color-accent=red]{--md-accent-fg-color:#ff1947;--md-accent-fg-color--transparent:#ff19471a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=pink]{--md-accent-fg-color:#f50056;--md-accent-fg-color--transparent:#f500561a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=purple]{--md-accent-fg-color:#df41fb;--md-accent-fg-color--transparent:#df41fb1a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=deep-purple]{--md-accent-fg-color:#7c4dff;--md-accent-fg-color--transparent:#7c4dff1a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=indigo]{--md-accent-fg-color:#526cfe;--md-accent-fg-color--transparent:#526cfe1a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=blue]{--md-accent-fg-color:#4287ff;--md-accent-fg-color--transparent:#4287ff1a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=light-blue]{--md-accent-fg-color:#0091eb;--md-accent-fg-color--transparent:#0091eb1a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=cyan]{--md-accent-fg-color:#00bad6;--md-accent-fg-color--transparent:#00bad61a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=teal]{--md-accent-fg-color:#00bda4;--md-accent-fg-color--transparent:#00bda41a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=green]{--md-accent-fg-color:#00c753;--md-accent-fg-color--transparent:#00c7531a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=light-green]{--md-accent-fg-color:#63de17;--md-accent-fg-color--transparent:#63de171a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=lime]{--md-accent-fg-color:#b0eb00;--md-accent-fg-color--transparent:#b0eb001a;--md-accent-bg-color:#000000de;--md-accent-bg-color--light:#0000008a}[data-md-color-accent=yellow]{--md-accent-fg-color:#ffd500;--md-accent-fg-color--transparent:#ffd5001a;--md-accent-bg-color:#000000de;--md-accent-bg-color--light:#0000008a}[data-md-color-accent=amber]{--md-accent-fg-color:#fa0;--md-accent-fg-color--transparent:#ffaa001a;--md-accent-bg-color:#000000de;--md-accent-bg-color--light:#0000008a}[data-md-color-accent=orange]{--md-accent-fg-color:#ff9100;--md-accent-fg-color--transparent:#ff91001a;--md-accent-bg-color:#000000de;--md-accent-bg-color--light:#0000008a}[data-md-color-accent=deep-orange]{--md-accent-fg-color:#ff6e42;--md-accent-fg-color--transparent:#ff6e421a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-primary=red]{--md-primary-fg-color:#ef5552;--md-primary-fg-color--light:#e57171;--md-primary-fg-color--dark:#e53734;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=pink]{--md-primary-fg-color:#e92063;--md-primary-fg-color--light:#ec417a;--md-primary-fg-color--dark:#c3185d;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=purple]{--md-primary-fg-color:#ab47bd;--md-primary-fg-color--light:#bb69c9;--md-primary-fg-color--dark:#8c24a8;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=deep-purple]{--md-primary-fg-color:#7e56c2;--md-primary-fg-color--light:#9574cd;--md-primary-fg-color--dark:#673ab6;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=indigo]{--md-primary-fg-color:#4051b5;--md-primary-fg-color--light:#5d6cc0;--md-primary-fg-color--dark:#303fa1;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=blue]{--md-primary-fg-color:#2094f3;--md-primary-fg-color--light:#42a5f5;--md-primary-fg-color--dark:#1975d2;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=light-blue]{--md-primary-fg-color:#02a6f2;--md-primary-fg-color--light:#28b5f6;--md-primary-fg-color--dark:#0287cf;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=cyan]{--md-primary-fg-color:#00bdd6;--md-primary-fg-color--light:#25c5da;--md-primary-fg-color--dark:#0097a8;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=teal]{--md-primary-fg-color:#009485;--md-primary-fg-color--light:#26a699;--md-primary-fg-color--dark:#007a6c;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=green]{--md-primary-fg-color:#4cae4f;--md-primary-fg-color--light:#68bb6c;--md-primary-fg-color--dark:#398e3d;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=light-green]{--md-primary-fg-color:#8bc34b;--md-primary-fg-color--light:#9ccc66;--md-primary-fg-color--dark:#689f38;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=lime]{--md-primary-fg-color:#cbdc38;--md-primary-fg-color--light:#d3e156;--md-primary-fg-color--dark:#b0b52c;--md-primary-bg-color:#000000de;--md-primary-bg-color--light:#0000008a}[data-md-color-primary=yellow]{--md-primary-fg-color:#ffec3d;--md-primary-fg-color--light:#ffee57;--md-primary-fg-color--dark:#fbc02d;--md-primary-bg-color:#000000de;--md-primary-bg-color--light:#0000008a}[data-md-color-primary=amber]{--md-primary-fg-color:#ffc105;--md-primary-fg-color--light:#ffc929;--md-primary-fg-color--dark:#ffa200;--md-primary-bg-color:#000000de;--md-primary-bg-color--light:#0000008a}[data-md-color-primary=orange]{--md-primary-fg-color:#ffa724;--md-primary-fg-color--light:#ffa724;--md-primary-fg-color--dark:#fa8900;--md-primary-bg-color:#000000de;--md-primary-bg-color--light:#0000008a}[data-md-color-primary=deep-orange]{--md-primary-fg-color:#ff6e42;--md-primary-fg-color--light:#ff8a66;--md-primary-fg-color--dark:#f4511f;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=brown]{--md-primary-fg-color:#795649;--md-primary-fg-color--light:#8d6e62;--md-primary-fg-color--dark:#5d4037;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=grey]{--md-primary-fg-color:#757575;--md-primary-fg-color--light:#9e9e9e;--md-primary-fg-color--dark:#616161;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3;--md-typeset-a-color:#4051b5}[data-md-color-primary=blue-grey]{--md-primary-fg-color:#546d78;--md-primary-fg-color--light:#607c8a;--md-primary-fg-color--dark:#455a63;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3;--md-typeset-a-color:#4051b5}[data-md-color-primary=light-green]:not([data-md-color-scheme=slate]){--md-typeset-a-color:#72ad2e}[data-md-color-primary=lime]:not([data-md-color-scheme=slate]){--md-typeset-a-color:#8b990a}[data-md-color-primary=yellow]:not([data-md-color-scheme=slate]){--md-typeset-a-color:#b8a500}[data-md-color-primary=amber]:not([data-md-color-scheme=slate]){--md-typeset-a-color:#d19d00}[data-md-color-primary=orange]:not([data-md-color-scheme=slate]){--md-typeset-a-color:#e68a00}[data-md-color-primary=white]{--md-primary-fg-color:hsla(var(--md-hue),0%,100%,1);--md-primary-fg-color--light:hsla(var(--md-hue),0%,100%,0.7);--md-primary-fg-color--dark:hsla(var(--md-hue),0%,0%,0.07);--md-primary-bg-color:hsla(var(--md-hue),0%,0%,0.87);--md-primary-bg-color--light:hsla(var(--md-hue),0%,0%,0.54);--md-typeset-a-color:#4051b5}[data-md-color-primary=white] .md-button{color:var(--md-typeset-a-color)}[data-md-color-primary=white] .md-button--primary{background-color:var(--md-typeset-a-color);border-color:var(--md-typeset-a-color);color:hsla(var(--md-hue),0%,100%,1)}@media screen and (min-width:60em){[data-md-color-primary=white] .md-search__form{background-color:hsla(var(--md-hue),0%,0%,.07)}[data-md-color-primary=white] .md-search__form:hover{background-color:hsla(var(--md-hue),0%,0%,.32)}[data-md-color-primary=white] .md-search__input+.md-search__icon{color:hsla(var(--md-hue),0%,0%,.87)}}@media screen and (min-width:76.25em){[data-md-color-primary=white] .md-tabs{border-bottom:.05rem solid #00000012}}[data-md-color-primary=black]{--md-primary-fg-color:hsla(var(--md-hue),15%,9%,1);--md-primary-fg-color--light:hsla(var(--md-hue),15%,9%,0.54);--md-primary-fg-color--dark:hsla(var(--md-hue),15%,9%,1);--md-primary-bg-color:hsla(var(--md-hue),15%,100%,1);--md-primary-bg-color--light:hsla(var(--md-hue),15%,100%,0.7);--md-typeset-a-color:#4051b5}[data-md-color-primary=black] .md-button{color:var(--md-typeset-a-color)}[data-md-color-primary=black] .md-button--primary{background-color:var(--md-typeset-a-color);border-color:var(--md-typeset-a-color);color:hsla(var(--md-hue),0%,100%,1)}[data-md-color-primary=black] .md-header{background-color:hsla(var(--md-hue),15%,9%,1)}@media screen and (max-width:59.984375em){[data-md-color-primary=black] .md-nav__source{background-color:hsla(var(--md-hue),15%,11%,.87)}}@media screen and (max-width:76.234375em){html [data-md-color-primary=black] .md-nav--primary .md-nav__title[for=__drawer]{background-color:hsla(var(--md-hue),15%,9%,1)}}@media screen and (min-width:76.25em){[data-md-color-primary=black] .md-tabs{background-color:hsla(var(--md-hue),15%,9%,1)}}
\ No newline at end of file
diff --git a/assets/stylesheets/palette.06af60db.min.css.map b/assets/stylesheets/palette.06af60db.min.css.map
new file mode 100644
index 0000000000..efb568c52a
--- /dev/null
+++ b/assets/stylesheets/palette.06af60db.min.css.map
@@ -0,0 +1 @@
+{"version":3,"sources":["src/templates/assets/stylesheets/palette/_scheme.scss","../../../../src/templates/assets/stylesheets/palette.scss","src/templates/assets/stylesheets/palette/_accent.scss","src/templates/assets/stylesheets/palette/_primary.scss","src/templates/assets/stylesheets/utilities/_break.scss"],"names":[],"mappings":"AA2BA,cAGE,6BAME,sDAAA,CACA,6DAAA,CACA,+DAAA,CACA,gEAAA,CACA,mDAAA,CACA,6DAAA,CACA,+DAAA,CACA,gEAAA,CAGA,mDAAA,CACA,gDAAA,CAGA,0BAAA,CACA,mCAAA,CAGA,iCAAA,CACA,kCAAA,CACA,mCAAA,CACA,mCAAA,CACA,kCAAA,CACA,iCAAA,CACA,+CAAA,CACA,6DAAA,CACA,gEAAA,CACA,4DAAA,CACA,4DAAA,CACA,6DAAA,CAGA,6CAAA,CAGA,+CAAA,CAGA,uDAAA,CACA,6DAAA,CACA,2DAAA,CAGA,iCAAA,CAGA,yDAAA,CACA,iEAAA,CAGA,mDAAA,CACA,mDAAA,CAGA,qDAAA,CACA,uDAAA,CAGA,8DAAA,CAKA,8DAAA,CAKA,0DAAA,CAvEA,iBCeF,CD6DE,kHAEE,YC3DJ,CDkFE,yDACE,4BChFJ,CD+EE,2DACE,4BC7EJ,CD4EE,gEACE,4BC1EJ,CDyEE,2DACE,4BCvEJ,CDsEE,yDACE,4BCpEJ,CDmEE,0DACE,4BCjEJ,CDgEE,gEACE,4BC9DJ,CD6DE,0DACE,4BC3DJ,CD0DE,2OACE,4BC/CJ,CDsDA,+FAGE,iCCpDF,CACF,CC/CE,2BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCD2CN,CCrDE,4BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDkDN,CC5DE,8BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDyDN,CCnEE,mCACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDgEN,CC1EE,8BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDuEN,CCjFE,4BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCD8EN,CCxFE,kCACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDqFN,CC/FE,4BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCD4FN,CCtGE,4BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDmGN,CC7GE,6BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCD0GN,CCpHE,mCACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDiHN,CC3HE,4BACE,4BAAA,CACA,2CAAA,CAIE,8BAAA,CACA,qCD2HN,CClIE,8BACE,4BAAA,CACA,2CAAA,CAIE,8BAAA,CACA,qCDkIN,CCzIE,6BACE,yBAAA,CACA,2CAAA,CAIE,8BAAA,CACA,qCDyIN,CChJE,8BACE,4BAAA,CACA,2CAAA,CAIE,8BAAA,CACA,qCDgJN,CCvJE,mCACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDoJN,CEzJE,4BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCFsJN,CEjKE,6BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCF8JN,CEzKE,+BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCFsKN,CEjLE,oCACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCF8KN,CEzLE,+BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCFsLN,CEjME,6BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCF8LN,CEzME,mCACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCFsMN,CEjNE,6BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCF8MN,CEzNE,6BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCFsNN,CEjOE,8BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCF8NN,CEzOE,oCACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCFsON,CEjPE,6BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAIE,+BAAA,CACA,sCFiPN,CEzPE,+BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAIE,+BAAA,CACA,sCFyPN,CEjQE,8BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAIE,+BAAA,CACA,sCFiQN,CEzQE,+BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAIE,+BAAA,CACA,sCFyQN,CEjRE,oCACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCF8QN,CEzRE,8BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCFsRN,CEjSE,6BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCAAA,CAKA,4BF0RN,CE1SE,kCACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCAAA,CAKA,4BFmSN,CEpRE,sEACE,4BFuRJ,CExRE,+DACE,4BF2RJ,CE5RE,iEACE,4BF+RJ,CEhSE,gEACE,4BFmSJ,CEpSE,iEACE,4BFuSJ,CE9RA,8BACE,mDAAA,CACA,4DAAA,CACA,0DAAA,CACA,oDAAA,CACA,2DAAA,CAGA,4BF+RF,CE5RE,yCACE,+BF8RJ,CE3RI,kDAEE,0CAAA,CACA,sCAAA,CAFA,mCF+RN,CG3MI,mCD1EA,+CACE,8CFwRJ,CErRI,qDACE,8CFuRN,CElRE,iEACE,mCFoRJ,CACF,CGtNI,sCDvDA,uCACE,oCFgRJ,CACF,CEvQA,8BACE,kDAAA,CACA,4DAAA,CACA,wDAAA,CACA,oDAAA,CACA,6DAAA,CAGA,4BFwQF,CErQE,yCACE,+BFuQJ,CEpQI,kDAEE,0CAAA,CACA,sCAAA,CAFA,mCFwQN,CEjQE,yCACE,6CFmQJ,CG5NI,0CDhCA,8CACE,gDF+PJ,CACF,CGjOI,0CDvBA,iFACE,6CF2PJ,CACF,CGzPI,sCDKA,uCACE,6CFuPJ,CACF","file":"palette.css"}
\ No newline at end of file
diff --git a/assets/stylesheets/palette.356b1318.min.css b/assets/stylesheets/palette.356b1318.min.css
deleted file mode 100644
index 141eb5f9dc..0000000000
--- a/assets/stylesheets/palette.356b1318.min.css
+++ /dev/null
@@ -1 +0,0 @@
-@media screen{[data-md-color-scheme=slate]{--md-default-fg-color:hsla(var(--md-hue),15%,90%,0.82);--md-default-fg-color--light:hsla(var(--md-hue),15%,90%,0.56);--md-default-fg-color--lighter:hsla(var(--md-hue),15%,90%,0.32);--md-default-fg-color--lightest:hsla(var(--md-hue),15%,90%,0.12);--md-default-bg-color:hsla(var(--md-hue),15%,14%,1);--md-default-bg-color--light:hsla(var(--md-hue),15%,14%,0.54);--md-default-bg-color--lighter:hsla(var(--md-hue),15%,14%,0.26);--md-default-bg-color--lightest:hsla(var(--md-hue),15%,14%,0.07);--md-code-fg-color:hsla(var(--md-hue),18%,86%,0.82);--md-code-bg-color:hsla(var(--md-hue),15%,18%,1);--md-code-hl-color--light:#4287ff26;--md-code-hl-number-color:#e6695b;--md-code-hl-special-color:#f06090;--md-code-hl-function-color:#c973d9;--md-code-hl-constant-color:#9383e2;--md-code-hl-keyword-color:#6791e0;--md-code-hl-string-color:#2fb170;--md-code-hl-name-color:var(--md-code-fg-color);--md-code-hl-operator-color:var(--md-default-fg-color--light);--md-code-hl-punctuation-color:var(--md-default-fg-color--light);--md-code-hl-comment-color:var(--md-default-fg-color--light);--md-code-hl-generic-color:var(--md-default-fg-color--light);--md-code-hl-variable-color:var(--md-default-fg-color--light);--md-typeset-color:var(--md-default-fg-color);--md-typeset-a-color:var(--md-primary-fg-color);--md-typeset-kbd-color:hsla(var(--md-hue),15%,90%,0.12);--md-typeset-kbd-accent-color:hsla(var(--md-hue),15%,90%,0.2);--md-typeset-kbd-border-color:hsla(var(--md-hue),15%,14%,1);--md-typeset-mark-color:#4287ff4d;--md-typeset-table-color:hsla(var(--md-hue),15%,95%,0.12);--md-typeset-table-color--light:hsla(var(--md-hue),15%,95%,0.035);--md-admonition-fg-color:var(--md-default-fg-color);--md-admonition-bg-color:var(--md-default-bg-color);--md-footer-bg-color:hsla(var(--md-hue),15%,10%,0.87);--md-footer-bg-color--dark:hsla(var(--md-hue),15%,8%,1);--md-shadow-z1:0 0.2rem 0.5rem #0000000d,0 0 0.05rem #0000001a;--md-shadow-z2:0 0.2rem 0.5rem #00000040,0 0 0.05rem #00000040;--md-shadow-z3:0 0.2rem 0.5rem #0006,0 0 0.05rem #00000059;color-scheme:dark}[data-md-color-scheme=slate] img[src$="#gh-light-mode-only"],[data-md-color-scheme=slate] img[src$="#only-light"]{display:none}[data-md-color-scheme=slate][data-md-color-primary=pink]{--md-typeset-a-color:#ed5487}[data-md-color-scheme=slate][data-md-color-primary=purple]{--md-typeset-a-color:#c46fd3}[data-md-color-scheme=slate][data-md-color-primary=deep-purple]{--md-typeset-a-color:#a47bea}[data-md-color-scheme=slate][data-md-color-primary=indigo]{--md-typeset-a-color:#5488e8}[data-md-color-scheme=slate][data-md-color-primary=teal]{--md-typeset-a-color:#00ccb8}[data-md-color-scheme=slate][data-md-color-primary=green]{--md-typeset-a-color:#71c174}[data-md-color-scheme=slate][data-md-color-primary=deep-orange]{--md-typeset-a-color:#ff764d}[data-md-color-scheme=slate][data-md-color-primary=brown]{--md-typeset-a-color:#c1775c}[data-md-color-scheme=slate][data-md-color-primary=black],[data-md-color-scheme=slate][data-md-color-primary=blue-grey],[data-md-color-scheme=slate][data-md-color-primary=grey],[data-md-color-scheme=slate][data-md-color-primary=white]{--md-typeset-a-color:#5e8bde}[data-md-color-switching] *,[data-md-color-switching] :after,[data-md-color-switching] :before{transition-duration:0ms!important}}[data-md-color-accent=red]{--md-accent-fg-color:#ff1947;--md-accent-fg-color--transparent:#ff19471a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=pink]{--md-accent-fg-color:#f50056;--md-accent-fg-color--transparent:#f500561a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=purple]{--md-accent-fg-color:#df41fb;--md-accent-fg-color--transparent:#df41fb1a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=deep-purple]{--md-accent-fg-color:#7c4dff;--md-accent-fg-color--transparent:#7c4dff1a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=indigo]{--md-accent-fg-color:#526cfe;--md-accent-fg-color--transparent:#526cfe1a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=blue]{--md-accent-fg-color:#4287ff;--md-accent-fg-color--transparent:#4287ff1a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=light-blue]{--md-accent-fg-color:#0091eb;--md-accent-fg-color--transparent:#0091eb1a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=cyan]{--md-accent-fg-color:#00bad6;--md-accent-fg-color--transparent:#00bad61a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=teal]{--md-accent-fg-color:#00bda4;--md-accent-fg-color--transparent:#00bda41a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=green]{--md-accent-fg-color:#00c753;--md-accent-fg-color--transparent:#00c7531a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=light-green]{--md-accent-fg-color:#63de17;--md-accent-fg-color--transparent:#63de171a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-accent=lime]{--md-accent-fg-color:#b0eb00;--md-accent-fg-color--transparent:#b0eb001a;--md-accent-bg-color:#000000de;--md-accent-bg-color--light:#0000008a}[data-md-color-accent=yellow]{--md-accent-fg-color:#ffd500;--md-accent-fg-color--transparent:#ffd5001a;--md-accent-bg-color:#000000de;--md-accent-bg-color--light:#0000008a}[data-md-color-accent=amber]{--md-accent-fg-color:#fa0;--md-accent-fg-color--transparent:#ffaa001a;--md-accent-bg-color:#000000de;--md-accent-bg-color--light:#0000008a}[data-md-color-accent=orange]{--md-accent-fg-color:#ff9100;--md-accent-fg-color--transparent:#ff91001a;--md-accent-bg-color:#000000de;--md-accent-bg-color--light:#0000008a}[data-md-color-accent=deep-orange]{--md-accent-fg-color:#ff6e42;--md-accent-fg-color--transparent:#ff6e421a;--md-accent-bg-color:#fff;--md-accent-bg-color--light:#ffffffb3}[data-md-color-primary=red]{--md-primary-fg-color:#ef5552;--md-primary-fg-color--light:#e57171;--md-primary-fg-color--dark:#e53734;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=pink]{--md-primary-fg-color:#e92063;--md-primary-fg-color--light:#ec417a;--md-primary-fg-color--dark:#c3185d;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=purple]{--md-primary-fg-color:#ab47bd;--md-primary-fg-color--light:#bb69c9;--md-primary-fg-color--dark:#8c24a8;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=deep-purple]{--md-primary-fg-color:#7e56c2;--md-primary-fg-color--light:#9574cd;--md-primary-fg-color--dark:#673ab6;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=indigo]{--md-primary-fg-color:#4051b5;--md-primary-fg-color--light:#5d6cc0;--md-primary-fg-color--dark:#303fa1;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=blue]{--md-primary-fg-color:#2094f3;--md-primary-fg-color--light:#42a5f5;--md-primary-fg-color--dark:#1975d2;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=light-blue]{--md-primary-fg-color:#02a6f2;--md-primary-fg-color--light:#28b5f6;--md-primary-fg-color--dark:#0287cf;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=cyan]{--md-primary-fg-color:#00bdd6;--md-primary-fg-color--light:#25c5da;--md-primary-fg-color--dark:#0097a8;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=teal]{--md-primary-fg-color:#009485;--md-primary-fg-color--light:#26a699;--md-primary-fg-color--dark:#007a6c;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=green]{--md-primary-fg-color:#4cae4f;--md-primary-fg-color--light:#68bb6c;--md-primary-fg-color--dark:#398e3d;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=light-green]{--md-primary-fg-color:#8bc34b;--md-primary-fg-color--light:#9ccc66;--md-primary-fg-color--dark:#689f38;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=lime]{--md-primary-fg-color:#cbdc38;--md-primary-fg-color--light:#d3e156;--md-primary-fg-color--dark:#b0b52c;--md-primary-bg-color:#000000de;--md-primary-bg-color--light:#0000008a}[data-md-color-primary=yellow]{--md-primary-fg-color:#ffec3d;--md-primary-fg-color--light:#ffee57;--md-primary-fg-color--dark:#fbc02d;--md-primary-bg-color:#000000de;--md-primary-bg-color--light:#0000008a}[data-md-color-primary=amber]{--md-primary-fg-color:#ffc105;--md-primary-fg-color--light:#ffc929;--md-primary-fg-color--dark:#ffa200;--md-primary-bg-color:#000000de;--md-primary-bg-color--light:#0000008a}[data-md-color-primary=orange]{--md-primary-fg-color:#ffa724;--md-primary-fg-color--light:#ffa724;--md-primary-fg-color--dark:#fa8900;--md-primary-bg-color:#000000de;--md-primary-bg-color--light:#0000008a}[data-md-color-primary=deep-orange]{--md-primary-fg-color:#ff6e42;--md-primary-fg-color--light:#ff8a66;--md-primary-fg-color--dark:#f4511f;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=brown]{--md-primary-fg-color:#795649;--md-primary-fg-color--light:#8d6e62;--md-primary-fg-color--dark:#5d4037;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3}[data-md-color-primary=grey]{--md-primary-fg-color:#757575;--md-primary-fg-color--light:#9e9e9e;--md-primary-fg-color--dark:#616161;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3;--md-typeset-a-color:#4051b5}[data-md-color-primary=blue-grey]{--md-primary-fg-color:#546d78;--md-primary-fg-color--light:#607c8a;--md-primary-fg-color--dark:#455a63;--md-primary-bg-color:#fff;--md-primary-bg-color--light:#ffffffb3;--md-typeset-a-color:#4051b5}[data-md-color-primary=light-green]:not([data-md-color-scheme=slate]){--md-typeset-a-color:#72ad2e}[data-md-color-primary=lime]:not([data-md-color-scheme=slate]){--md-typeset-a-color:#8b990a}[data-md-color-primary=yellow]:not([data-md-color-scheme=slate]){--md-typeset-a-color:#b8a500}[data-md-color-primary=amber]:not([data-md-color-scheme=slate]){--md-typeset-a-color:#d19d00}[data-md-color-primary=orange]:not([data-md-color-scheme=slate]){--md-typeset-a-color:#e68a00}[data-md-color-primary=white]{--md-primary-fg-color:hsla(var(--md-hue),0%,100%,1);--md-primary-fg-color--light:hsla(var(--md-hue),0%,100%,0.7);--md-primary-fg-color--dark:hsla(var(--md-hue),0%,0%,0.07);--md-primary-bg-color:hsla(var(--md-hue),0%,0%,0.87);--md-primary-bg-color--light:hsla(var(--md-hue),0%,0%,0.54);--md-typeset-a-color:#4051b5}[data-md-color-primary=white] .md-button{color:var(--md-typeset-a-color)}[data-md-color-primary=white] .md-button--primary{background-color:var(--md-typeset-a-color);border-color:var(--md-typeset-a-color);color:hsla(var(--md-hue),0%,100%,1)}@media screen and (min-width:60em){[data-md-color-primary=white] .md-search__form{background-color:hsla(var(--md-hue),0%,0%,.07)}[data-md-color-primary=white] .md-search__form:hover{background-color:hsla(var(--md-hue),0%,0%,.32)}[data-md-color-primary=white] .md-search__input+.md-search__icon{color:hsla(var(--md-hue),0%,0%,.87)}}@media screen and (min-width:76.25em){[data-md-color-primary=white] .md-tabs{border-bottom:.05rem solid #00000012}}[data-md-color-primary=black]{--md-primary-fg-color:hsla(var(--md-hue),15%,9%,1);--md-primary-fg-color--light:hsla(var(--md-hue),15%,9%,0.54);--md-primary-fg-color--dark:hsla(var(--md-hue),15%,9%,1);--md-primary-bg-color:hsla(var(--md-hue),15%,100%,1);--md-primary-bg-color--light:hsla(var(--md-hue),15%,100%,0.7);--md-typeset-a-color:#4051b5}[data-md-color-primary=black] .md-button{color:var(--md-typeset-a-color)}[data-md-color-primary=black] .md-button--primary{background-color:var(--md-typeset-a-color);border-color:var(--md-typeset-a-color);color:hsla(var(--md-hue),0%,100%,1)}[data-md-color-primary=black] .md-header{background-color:hsla(var(--md-hue),15%,9%,1)}@media screen and (max-width:59.984375em){[data-md-color-primary=black] .md-nav__source{background-color:hsla(var(--md-hue),15%,11%,.87)}}@media screen and (max-width:76.234375em){html [data-md-color-primary=black] .md-nav--primary .md-nav__title[for=__drawer]{background-color:hsla(var(--md-hue),15%,9%,1)}}@media screen and (min-width:76.25em){[data-md-color-primary=black] .md-tabs{background-color:hsla(var(--md-hue),15%,9%,1)}}
\ No newline at end of file
diff --git a/assets/stylesheets/palette.356b1318.min.css.map b/assets/stylesheets/palette.356b1318.min.css.map
deleted file mode 100644
index 4627b68764..0000000000
--- a/assets/stylesheets/palette.356b1318.min.css.map
+++ /dev/null
@@ -1 +0,0 @@
-{"version":3,"sources":["src/templates/assets/stylesheets/palette/_scheme.scss","../../../../src/templates/assets/stylesheets/palette.scss","src/templates/assets/stylesheets/palette/_accent.scss","src/templates/assets/stylesheets/palette/_primary.scss","src/templates/assets/stylesheets/utilities/_break.scss"],"names":[],"mappings":"AA2BA,cAGE,6BAME,sDAAA,CACA,6DAAA,CACA,+DAAA,CACA,gEAAA,CACA,mDAAA,CACA,6DAAA,CACA,+DAAA,CACA,gEAAA,CAGA,mDAAA,CACA,gDAAA,CAGA,mCAAA,CACA,iCAAA,CACA,kCAAA,CACA,mCAAA,CACA,mCAAA,CACA,kCAAA,CACA,iCAAA,CACA,+CAAA,CACA,6DAAA,CACA,gEAAA,CACA,4DAAA,CACA,4DAAA,CACA,6DAAA,CAGA,6CAAA,CAGA,+CAAA,CAGA,uDAAA,CACA,6DAAA,CACA,2DAAA,CAGA,iCAAA,CAGA,yDAAA,CACA,iEAAA,CAGA,mDAAA,CACA,mDAAA,CAGA,qDAAA,CACA,uDAAA,CAGA,8DAAA,CAKA,8DAAA,CAKA,0DAAA,CApEA,iBCcF,CD2DE,kHAEE,YCzDJ,CDgFE,yDACE,4BC9EJ,CD6EE,2DACE,4BC3EJ,CD0EE,gEACE,4BCxEJ,CDuEE,2DACE,4BCrEJ,CDoEE,yDACE,4BClEJ,CDiEE,0DACE,4BC/DJ,CD8DE,gEACE,4BC5DJ,CD2DE,0DACE,4BCzDJ,CDwDE,2OACE,4BC7CJ,CDoDA,+FAGE,iCClDF,CACF,CC9CE,2BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCD0CN,CCpDE,4BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDiDN,CC3DE,8BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDwDN,CClEE,mCACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCD+DN,CCzEE,8BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDsEN,CChFE,4BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCD6EN,CCvFE,kCACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDoFN,CC9FE,4BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCD2FN,CCrGE,4BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDkGN,CC5GE,6BACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDyGN,CCnHE,mCACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDgHN,CC1HE,4BACE,4BAAA,CACA,2CAAA,CAIE,8BAAA,CACA,qCD0HN,CCjIE,8BACE,4BAAA,CACA,2CAAA,CAIE,8BAAA,CACA,qCDiIN,CCxIE,6BACE,yBAAA,CACA,2CAAA,CAIE,8BAAA,CACA,qCDwIN,CC/IE,8BACE,4BAAA,CACA,2CAAA,CAIE,8BAAA,CACA,qCD+IN,CCtJE,mCACE,4BAAA,CACA,2CAAA,CAOE,yBAAA,CACA,qCDmJN,CExJE,4BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCFqJN,CEhKE,6BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCF6JN,CExKE,+BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCFqKN,CEhLE,oCACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCF6KN,CExLE,+BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCFqLN,CEhME,6BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCF6LN,CExME,mCACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCFqMN,CEhNE,6BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCF6MN,CExNE,6BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCFqNN,CEhOE,8BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCF6NN,CExOE,oCACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCFqON,CEhPE,6BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAIE,+BAAA,CACA,sCFgPN,CExPE,+BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAIE,+BAAA,CACA,sCFwPN,CEhQE,8BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAIE,+BAAA,CACA,sCFgQN,CExQE,+BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAIE,+BAAA,CACA,sCFwQN,CEhRE,oCACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCF6QN,CExRE,8BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCFqRN,CEhSE,6BACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCAAA,CAKA,4BFyRN,CEzSE,kCACE,6BAAA,CACA,oCAAA,CACA,mCAAA,CAOE,0BAAA,CACA,sCAAA,CAKA,4BFkSN,CEnRE,sEACE,4BFsRJ,CEvRE,+DACE,4BF0RJ,CE3RE,iEACE,4BF8RJ,CE/RE,gEACE,4BFkSJ,CEnSE,iEACE,4BFsSJ,CE7RA,8BACE,mDAAA,CACA,4DAAA,CACA,0DAAA,CACA,oDAAA,CACA,2DAAA,CAGA,4BF8RF,CE3RE,yCACE,+BF6RJ,CE1RI,kDAEE,0CAAA,CACA,sCAAA,CAFA,mCF8RN,CG1MI,mCD1EA,+CACE,8CFuRJ,CEpRI,qDACE,8CFsRN,CEjRE,iEACE,mCFmRJ,CACF,CGrNI,sCDvDA,uCACE,oCF+QJ,CACF,CEtQA,8BACE,kDAAA,CACA,4DAAA,CACA,wDAAA,CACA,oDAAA,CACA,6DAAA,CAGA,4BFuQF,CEpQE,yCACE,+BFsQJ,CEnQI,kDAEE,0CAAA,CACA,sCAAA,CAFA,mCFuQN,CEhQE,yCACE,6CFkQJ,CG3NI,0CDhCA,8CACE,gDF8PJ,CACF,CGhOI,0CDvBA,iFACE,6CF0PJ,CACF,CGxPI,sCDKA,uCACE,6CFsPJ,CACF","file":"palette.css"}
\ No newline at end of file
diff --git a/com.unity.ml-agents.extensions/index.html b/com.unity.ml-agents.extensions/index.html
index 7c8672da4c..9b132480e2 100644
--- a/com.unity.ml-agents.extensions/index.html
+++ b/com.unity.ml-agents.extensions/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1103,11 +1120,19 @@ <h2 id="need-help">Need Help?</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1140,10 +1165,10 @@ <h2 id="need-help">Need Help?</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/com.unity.ml-agents/index.html b/com.unity.ml-agents/index.html
index 26b9aa2993..d63d5ba394 100644
--- a/com.unity.ml-agents/index.html
+++ b/com.unity.ml-agents/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1229,11 +1246,19 @@ <h2 id="helpful-links">Helpful links</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1266,10 +1291,10 @@ <h2 id="helpful-links">Helpful links</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/doxygen/Readme/index.html b/doxygen/Readme/index.html
index 9694bbbbec..73304f8e42 100644
--- a/doxygen/Readme/index.html
+++ b/doxygen/Readme/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../.." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../../Versioning/" class="md-nav__link">
         
@@ -954,11 +971,19 @@ <h1 id="doxygen-files">Doxygen files</h1>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -991,10 +1016,10 @@ <h1 id="doxygen-files">Doxygen files</h1>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "../..", "features": [], "search": "../../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "../..", "features": [], "search": "../../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/index.html b/index.html
index bb40830617..0eb10aca10 100644
--- a/index.html
+++ b/index.html
@@ -20,7 +20,7 @@
       
       
       <link rel="icon" href="images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -28,10 +28,10 @@
       
     
     
-      <link rel="stylesheet" href="assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -72,9 +72,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -129,21 +126,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -151,9 +148,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -162,6 +159,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -205,7 +205,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -250,7 +250,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -267,6 +267,7 @@
     
   
   
+  
     <li class="md-nav__item md-nav__item--active">
       
       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
@@ -384,6 +385,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="ML-Agents-Overview/" class="md-nav__link">
         
@@ -403,6 +405,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Installation/" class="md-nav__link">
         
@@ -422,6 +425,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -441,7 +445,7 @@
   
   
   
-    
+  
     
     
     
@@ -450,7 +454,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -477,6 +480,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Background-Machine-Learning/" class="md-nav__link">
         
@@ -497,6 +501,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Background-PyTorch/" class="md-nav__link">
         
@@ -517,6 +522,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Background-Unity/" class="md-nav__link">
         
@@ -537,6 +543,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="ELO-Rating-System/" class="md-nav__link">
         
@@ -564,7 +571,7 @@
   
   
   
-    
+  
     
     
     
@@ -573,7 +580,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -600,6 +606,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Python-Gym-API/" class="md-nav__link">
         
@@ -620,6 +627,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Python-PettingZoo-API/" class="md-nav__link">
         
@@ -640,6 +648,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Python-LLAPI/" class="md-nav__link">
         
@@ -667,7 +676,7 @@
   
   
   
-    
+  
     
     
     
@@ -676,7 +685,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -703,6 +711,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -723,6 +732,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -743,6 +753,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -763,6 +774,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -790,7 +802,7 @@
   
   
   
-    
+  
     
     
     
@@ -799,7 +811,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -826,6 +837,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Training-Plugins/" class="md-nav__link">
         
@@ -846,6 +858,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -866,6 +879,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Hugging-Face-Integration/" class="md-nav__link">
         
@@ -893,7 +907,7 @@
   
   
   
-    
+  
     
     
     
@@ -902,7 +916,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -929,6 +942,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="FAQ/" class="md-nav__link">
         
@@ -949,6 +963,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Limitations/" class="md-nav__link">
         
@@ -969,6 +984,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Migrating/" class="md-nav__link">
         
@@ -989,6 +1005,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="Versioning/" class="md-nav__link">
         
@@ -1162,18 +1179,20 @@ <h2 id="releases-documentation">Releases &amp; Documentation</h2>
 to get started with the latest release of ML-Agents.</strong></p>
 <p><strong>You can also check out our new <a href="https://unity-technologies.github.io/ml-agents/">web docs</a>!</strong></p>
 <p>The table below lists all our releases, including our <code>main</code> branch which is
-under active development and may be unstable. A few helpful guidelines:
-- The <a href="Versioning/">Versioning page</a> overviews how we manage our GitHub
-  releases and the versioning process for each of the ML-Agents components.
-- The <a href="https://github.com/Unity-Technologies/ml-agents/releases">Releases page</a>
-  contains details of the changes between releases.
-- The <a href="Migrating/">Migration page</a> contains details on how to upgrade
-  from earlier releases of the ML-Agents Toolkit.
-- The <strong>Documentation</strong> links in the table below include installation and usage
+under active development and may be unstable. A few helpful guidelines:</p>
+<ul>
+<li>The <a href="Versioning/">Versioning page</a> overviews how we manage our GitHub
+  releases and the versioning process for each of the ML-Agents components.</li>
+<li>The <a href="https://github.com/Unity-Technologies/ml-agents/releases">Releases page</a>
+  contains details of the changes between releases.</li>
+<li>The <a href="Migrating/">Migration page</a> contains details on how to upgrade
+  from earlier releases of the ML-Agents Toolkit.</li>
+<li>The <strong>Documentation</strong> links in the table below include installation and usage
   instructions specific to each release. Remember to always use the
-  documentation that corresponds to the release version you're using.
-- The <code>com.unity.ml-agents</code> package is <a href="https://docs.unity3d.com/2020.1/Documentation/Manual/pack-safe.html">verified</a>
-  for Unity 2020.1 and later. Verified packages releases are numbered 1.0.x.</p>
+  documentation that corresponds to the release version you're using.</li>
+<li>The <code>com.unity.ml-agents</code> package is <a href="https://docs.unity3d.com/2020.1/Documentation/Manual/pack-safe.html">verified</a>
+  for Unity 2020.1 and later. Verified packages releases are numbered 1.0.x.</li>
+</ul>
 <table>
 <thead>
 <tr>
@@ -1290,7 +1309,7 @@ <h2 id="additional-resources">Additional Resources</h2>
 <h3 id="more-from-unity">More from Unity</h3>
 <ul>
 <li><a href="https://unity.com/products/sentis">Unity Sentis</a></li>
-<li><a href="https://blog.unity.com/engine-platform/introducing-unity-muse-and-unity-sentis-ai">Introductin Unity Muse and Sentis</a></li>
+<li><a href="https://blog.unity.com/engine-platform/introducing-unity-muse-and-unity-sentis-ai">Introducing Unity Muse and Sentis</a></li>
 </ul>
 <h2 id="community-and-feedback">Community and Feedback</h2>
 <p>The ML-Agents Toolkit is an open-source project and we encourage and welcome
@@ -1322,11 +1341,19 @@ <h2 id="privacy">Privacy</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1359,10 +1386,10 @@ <h2 id="privacy">Privacy</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": ".", "features": [], "search": "assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": ".", "features": [], "search": "assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/ml-agents-envs/index.html b/ml-agents-envs/index.html
index c27d47a1c2..d29240f5e6 100644
--- a/ml-agents-envs/index.html
+++ b/ml-agents-envs/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1019,11 +1036,19 @@ <h2 id="limitations">Limitations</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1056,10 +1081,10 @@ <h2 id="limitations">Limitations</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/ml-agents/index.html b/ml-agents/index.html
index fef630093f..7ec138b186 100644
--- a/ml-agents/index.html
+++ b/ml-agents/index.html
@@ -18,7 +18,7 @@
       
       
       <link rel="icon" href="../images/unity-logo-black.png">
-      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.4.12">
+      <meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.17">
     
     
       
@@ -26,10 +26,10 @@
       
     
     
-      <link rel="stylesheet" href="../assets/stylesheets/main.fad675c6.min.css">
+      <link rel="stylesheet" href="../assets/stylesheets/main.bcfcd587.min.css">
       
         
-        <link rel="stylesheet" href="../assets/stylesheets/palette.356b1318.min.css">
+        <link rel="stylesheet" href="../assets/stylesheets/palette.06af60db.min.css">
       
       
 
@@ -70,9 +70,6 @@
     <body dir="ltr" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo">
   
     
-    
-      <script>var palette=__md_get("__palette");if(palette&&"object"==typeof palette.color)for(var key of Object.keys(palette.color))document.body.setAttribute("data-md-color-"+key,palette.color[key])</script>
-    
     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
     <label class="md-overlay" for="__drawer"></label>
@@ -127,21 +124,21 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="blue" data-md-color-accent="indigo"  aria-hidden="true"  type="radio" name="__palette" id="__palette_0">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_2">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="light-blue"  aria-hidden="true"  type="radio" name="__palette" id="__palette_1">
     
   
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_3">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="default" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to dark mode"  type="radio" name="__palette" id="__palette_2">
     
-      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_2" hidden>
+      <label class="md-header__button md-icon" title="Switch to dark mode" for="__palette_3" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a4 4 0 0 0-4 4 4 4 0 0 0 4 4 4 4 0 0 0 4-4 4 4 0 0 0-4-4m0 10a6 6 0 0 1-6-6 6 6 0 0 1 6-6 6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -149,9 +146,9 @@
     
     
     
-    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_4">
+    <input class="md-option" data-md-color-media="" data-md-color-scheme="slate" data-md-color-primary="indigo" data-md-color-accent="indigo"  aria-label="Switch to light mode"  type="radio" name="__palette" id="__palette_3">
     
-      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_3" hidden>
+      <label class="md-header__button md-icon" title="Switch to light mode" for="__palette_0" hidden>
         <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 18c-.89 0-1.74-.2-2.5-.55C11.56 16.5 13 14.42 13 12c0-2.42-1.44-4.5-3.5-5.45C10.26 6.2 11.11 6 12 6a6 6 0 0 1 6 6 6 6 0 0 1-6 6m8-9.31V4h-4.69L12 .69 8.69 4H4v4.69L.69 12 4 15.31V20h4.69L12 23.31 15.31 20H20v-4.69L23.31 12 20 8.69Z"/></svg>
       </label>
     
@@ -160,6 +157,9 @@
       
     
     
+      <script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
+    
+    
     
       <label class="md-header__button md-icon" for="__search">
         
@@ -203,7 +203,7 @@
         <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -248,7 +248,7 @@
       <a href="https://github.com/Unity-Technologies/ml-agents" title="Go to repository" class="md-source" data-md-component="source">
   <div class="md-source__icon md-icon">
     
-    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.4.2 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
+    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512"><!--! Font Awesome Free 6.5.1 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) Copyright 2023 Fonticons, Inc.--><path d="M439.55 236.05 244 40.45a28.87 28.87 0 0 0-40.81 0l-40.66 40.63 51.52 51.52c27.06-9.14 52.68 16.77 43.39 43.68l49.66 49.66c34.23-11.8 61.18 31 35.47 56.69-26.49 26.49-70.21-2.87-56-37.34L240.22 199v121.85c25.3 12.54 22.26 41.85 9.08 55a34.34 34.34 0 0 1-48.55 0c-17.57-17.6-11.07-46.91 11.25-56v-123c-20.8-8.51-24.6-30.74-18.64-45L142.57 101 8.45 235.14a28.86 28.86 0 0 0 0 40.81l195.61 195.6a28.86 28.86 0 0 0 40.8 0l194.69-194.69a28.86 28.86 0 0 0 0-40.81z"/></svg>
   </div>
   <div class="md-source__repository">
     GitHub
@@ -263,6 +263,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href=".." class="md-nav__link">
         
@@ -282,6 +283,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Overview/" class="md-nav__link">
         
@@ -301,6 +303,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Installation/" class="md-nav__link">
         
@@ -320,6 +323,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ML-Agents-Toolkit-Documentation/" class="md-nav__link">
         
@@ -339,7 +343,7 @@
   
   
   
-    
+  
     
     
     
@@ -348,7 +352,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
         
           
@@ -375,6 +378,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Machine-Learning/" class="md-nav__link">
         
@@ -395,6 +399,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-PyTorch/" class="md-nav__link">
         
@@ -415,6 +420,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Background-Unity/" class="md-nav__link">
         
@@ -435,6 +441,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../ELO-Rating-System/" class="md-nav__link">
         
@@ -462,7 +469,7 @@
   
   
   
-    
+  
     
     
     
@@ -471,7 +478,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6" >
         
           
@@ -498,6 +504,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API/" class="md-nav__link">
         
@@ -518,6 +525,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API/" class="md-nav__link">
         
@@ -538,6 +546,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI/" class="md-nav__link">
         
@@ -565,7 +574,7 @@
   
   
   
-    
+  
     
     
     
@@ -574,7 +583,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7" >
         
           
@@ -601,6 +609,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-Gym-API-Documentation/" class="md-nav__link">
         
@@ -621,6 +630,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-PettingZoo-API-Documentation/" class="md-nav__link">
         
@@ -641,6 +651,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-LLAPI-Documentation/" class="md-nav__link">
         
@@ -661,6 +672,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Python-On-Off-Policy-Trainer-Documentation/" class="md-nav__link">
         
@@ -688,7 +700,7 @@
   
   
   
-    
+  
     
     
     
@@ -697,7 +709,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8" >
         
           
@@ -724,6 +735,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Training-Plugins/" class="md-nav__link">
         
@@ -744,6 +756,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Tutorial-Custom-Trainer-Plugin/" class="md-nav__link">
         
@@ -764,6 +777,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Hugging-Face-Integration/" class="md-nav__link">
         
@@ -791,7 +805,7 @@
   
   
   
-    
+  
     
     
     
@@ -800,7 +814,6 @@
       
         
         
-        
         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_9" >
         
           
@@ -827,6 +840,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../FAQ/" class="md-nav__link">
         
@@ -847,6 +861,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Limitations/" class="md-nav__link">
         
@@ -867,6 +882,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Migrating/" class="md-nav__link">
         
@@ -887,6 +903,7 @@
   
   
   
+  
     <li class="md-nav__item">
       <a href="../Versioning/" class="md-nav__link">
         
@@ -1008,11 +1025,19 @@ <h2 id="limitations">Limitations</h2>
 
 
 
+
+
+
+
+
+
+
                 
               </article>
             </div>
           
           
+<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
         </div>
         
       </main>
@@ -1045,10 +1070,10 @@ <h2 id="limitations">Limitations</h2>
     </div>
     
     
-    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.f886a092.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
+    <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
     
     
-      <script src="../assets/javascripts/bundle.6c14ae12.min.js"></script>
+      <script src="../assets/javascripts/bundle.1e8ae164.min.js"></script>
       
     
   </body>
diff --git a/search/search_index.json b/search/search_index.json
index ca6587e27a..b0531ccdc0 100644
--- a/search/search_index.json
+++ b/search/search_index.json
@@ -1 +1 @@
-{"config":{"lang":["en"],"separator":"[\\s\\-]+","pipeline":["stopWordFilter"]},"docs":[{"location":"","title":"Home","text":""},{"location":"#unity-ml-agents-toolkit","title":"Unity ML-Agents Toolkit","text":"<p>(latest release) (all releases)</p> <p>The Unity Machine Learning Agents Toolkit (ML-Agents) is an open-source project that enables games and simulations to serve as environments for training intelligent agents. We provide implementations (based on PyTorch) of state-of-the-art algorithms to enable game developers and hobbyists to easily train intelligent agents for 2D, 3D and VR/AR games. Researchers can also use the provided simple-to-use Python API to train Agents using reinforcement learning, imitation learning, neuroevolution, or any other methods. These trained agents can be used for multiple purposes, including controlling NPC behavior (in a variety of settings such as multi-agent and adversarial), automated testing of game builds and evaluating different game design decisions pre-release. The ML-Agents Toolkit is mutually beneficial for both game developers and AI researchers as it provides a central platform where advances in AI can be evaluated on Unity\u2019s rich environments and then made accessible to the wider research and game developer communities.</p>"},{"location":"#features","title":"Features","text":"<ul> <li>17+ example Unity environments</li> <li>Support for multiple environment configurations and training scenarios</li> <li>Flexible Unity SDK that can be integrated into your game or custom Unity scene</li> <li>Support for training single-agent, multi-agent cooperative, and multi-agent   competitive scenarios via several Deep Reinforcement Learning algorithms (PPO, SAC, MA-POCA, self-play).</li> <li>Support for learning from demonstrations through two Imitation Learning algorithms (BC and GAIL).</li> <li>Quickly and easily add your own custom training algorithm and/or components.</li> <li>Easily definable Curriculum Learning scenarios for complex tasks</li> <li>Train robust agents using environment randomization</li> <li>Flexible agent control with On Demand Decision Making</li> <li>Train using multiple concurrent Unity environment instances</li> <li>Utilizes the Sentis to   provide native cross-platform support</li> <li>Unity environment control from Python</li> <li>Wrap Unity learning environments as a gym environment</li> <li>Wrap Unity learning environments as a PettingZoo environment</li> </ul> <p>See our ML-Agents Overview page for detailed descriptions of all these features. Or go straight to our web docs.</p>"},{"location":"#releases-documentation","title":"Releases &amp; Documentation","text":"<p>Our latest, stable release is <code>Release 21</code>. Click here to get started with the latest release of ML-Agents.</p> <p>You can also check out our new web docs!</p> <p>The table below lists all our releases, including our <code>main</code> branch which is under active development and may be unstable. A few helpful guidelines: - The Versioning page overviews how we manage our GitHub   releases and the versioning process for each of the ML-Agents components. - The Releases page   contains details of the changes between releases. - The Migration page contains details on how to upgrade   from earlier releases of the ML-Agents Toolkit. - The Documentation links in the table below include installation and usage   instructions specific to each release. Remember to always use the   documentation that corresponds to the release version you're using. - The <code>com.unity.ml-agents</code> package is verified   for Unity 2020.1 and later. Verified packages releases are numbered 1.0.x.</p> Version Release Date Source Documentation Download Python Package Unity Package develop (unstable) -- source docs download -- -- Release 21 October 9, 2023 source docs download 1.0.0 3.0.0 <p>If you are a researcher interested in a discussion of Unity as an AI platform, see a pre-print of our reference paper on Unity and the ML-Agents Toolkit.</p> <p>If you use Unity or the ML-Agents Toolkit to conduct research, we ask that you cite the following paper as a reference:</p> <pre><code>@article{juliani2020,\n  title={Unity: A general platform for intelligent agents},\n  author={Juliani, Arthur and Berges, Vincent-Pierre and Teng, Ervin and Cohen, Andrew and Harper, Jonathan and Elion, Chris and Goy, Chris and Gao, Yuan and Henry, Hunter and Mattar, Marwan and Lange, Danny},\n  journal={arXiv preprint arXiv:1809.02627},\n  url={https://arxiv.org/pdf/1809.02627.pdf},\n  year={2020}\n}\n</code></pre> <p>Additionally, if you use the MA-POCA trainer in your research, we ask that you cite the following paper as a reference:</p> <pre><code>@article{cohen2022,\n  title={On the Use and Misuse of Absorbing States in Multi-agent Reinforcement Learning},\n  author={Cohen, Andrew and Teng, Ervin and Berges, Vincent-Pierre and Dong, Ruo-Ping and Henry, Hunter and Mattar, Marwan and Zook, Alexander and Ganguly, Sujoy},\n  journal={RL in Games Workshop AAAI 2022},\n  url={http://aaai-rlg.mlanctot.info/papers/AAAI22-RLG_paper_32.pdf},\n  year={2022}\n}\n</code></pre>"},{"location":"#additional-resources","title":"Additional Resources","text":"<p>We have a Unity Learn course, ML-Agents: Hummingbirds, that provides a gentle introduction to Unity and the ML-Agents Toolkit.</p> <p>We've also partnered with CodeMonkeyUnity to create a series of tutorial videos on how to implement and use the ML-Agents Toolkit.</p> <p>We have also published a series of blog posts that are relevant for ML-Agents:</p> <ul> <li>(July 12, 2021)   ML-Agents plays Dodgeball</li> <li>(May 5, 2021)   ML-Agents v2.0 release: Now supports training complex cooperative behaviors</li> <li>(December 28, 2020)   Happy holidays from the Unity ML-Agents team!</li> <li>(November 20, 2020)   How Eidos-Montr\u00e9al created Grid Sensors to improve observations for training agents</li> <li>(November 11, 2020)   2020 AI@Unity interns shoutout</li> <li>(May 12, 2020)   Announcing ML-Agents Unity Package v1.0!</li> <li>(February 28, 2020)   Training intelligent adversaries using self-play with ML-Agents</li> <li>(November 11, 2019)   Training your agents 7 times faster with ML-Agents</li> <li>(October 21, 2019)   The AI@Unity interns help shape the world</li> <li>(April 15, 2019)   Unity ML-Agents Toolkit v0.8: Faster training on real games</li> <li>(March 1, 2019)   Unity ML-Agents Toolkit v0.7: A leap towards cross-platform inference</li> <li>(December 17, 2018)   ML-Agents Toolkit v0.6: Improved usability of Brains and Imitation Learning</li> <li>(October 2, 2018)   Puppo, The Corgi: Cuteness Overload with the Unity ML-Agents Toolkit</li> <li>(September 11, 2018)   ML-Agents Toolkit v0.5, new resources for AI researchers available now</li> <li>(June 26, 2018)   Solving sparse-reward tasks with Curiosity</li> <li>(June 19, 2018)   Unity ML-Agents Toolkit v0.4 and Udacity Deep Reinforcement Learning Nanodegree</li> <li>(May 24, 2018)   Imitation Learning in Unity: The Workflow</li> <li>(March 15, 2018)   ML-Agents Toolkit v0.3 Beta released: Imitation Learning, feedback-driven features, and more</li> <li>(December 11, 2017)   Using Machine Learning Agents in a real game: a beginner\u2019s guide</li> <li>(December 8, 2017)   Introducing ML-Agents Toolkit v0.2: Curriculum Learning, new environments, and more</li> <li>(September 19, 2017)   Introducing: Unity Machine Learning Agents Toolkit</li> <li>Overviewing reinforcement learning concepts   (multi-armed bandit   and   Q-learning)</li> </ul>"},{"location":"#more-from-unity","title":"More from Unity","text":"<ul> <li>Unity Sentis</li> <li>Introductin Unity Muse and Sentis</li> </ul>"},{"location":"#community-and-feedback","title":"Community and Feedback","text":"<p>The ML-Agents Toolkit is an open-source project and we encourage and welcome contributions. If you wish to contribute, be sure to review our contribution guidelines and code of conduct.</p> <p>For problems with the installation and setup of the ML-Agents Toolkit, or discussions about how to best setup or train your agents, please create a new thread on the Unity ML-Agents forum and make sure to include as much detail as possible. If you run into any other problems using the ML-Agents Toolkit or have a specific feature request, please submit a GitHub issue.</p> <p>Please tell us which samples you would like to see shipped with the ML-Agents Unity package by replying to this forum thread.</p> <p>Your opinion matters a great deal to us. Only by hearing your thoughts on the Unity ML-Agents Toolkit can we continue to improve and grow. Please take a few minutes to let us know about it.</p> <p>For any other questions or feedback, connect directly with the ML-Agents team at ml-agents@unity3d.com.</p>"},{"location":"#privacy","title":"Privacy","text":"<p>In order to improve the developer experience for Unity ML-Agents Toolkit, we have added in-editor analytics. Please refer to \"Information that is passively collected by Unity\" in the Unity Privacy Policy.</p>"},{"location":"API-Reference/","title":"API Reference","text":"<p>Our developer-facing C# classes have been documented to be compatible with Doxygen for auto-generating HTML documentation.</p> <p>To generate the API reference, download Doxygen and run the following command within the <code>docs/</code> directory:</p> <pre><code>doxygen dox-ml-agents.conf\n</code></pre> <p><code>dox-ml-agents.conf</code> is a Doxygen configuration file for the ML-Agents Toolkit that includes the classes that have been properly formatted. The generated HTML files will be placed in the <code>html/</code> subdirectory. Open <code>index.html</code> within that subdirectory to navigate to the API reference home. Note that <code>html/</code> is already included in the repository's <code>.gitignore</code> file.</p> <p>In the near future, we aim to expand our documentation to include the Python classes.</p>"},{"location":"Background-Machine-Learning/","title":"Background: Machine Learning","text":"<p>Given that a number of users of the ML-Agents Toolkit might not have a formal machine learning background, this page provides an overview to facilitate the understanding of the ML-Agents Toolkit. However, we will not attempt to provide a thorough treatment of machine learning as there are fantastic resources online.</p> <p>Machine learning, a branch of artificial intelligence, focuses on learning patterns from data. The three main classes of machine learning algorithms include: unsupervised learning, supervised learning and reinforcement learning. Each class of algorithm learns from a different type of data. The following paragraphs provide an overview for each of these classes of machine learning, as well as introductory examples.</p>"},{"location":"Background-Machine-Learning/#unsupervised-learning","title":"Unsupervised Learning","text":"<p>The goal of unsupervised learning is to group or cluster similar items in a data set. For example, consider the players of a game. We may want to group the players depending on how engaged they are with the game. This would enable us to target different groups (e.g. for highly-engaged players we might invite them to be beta testers for new features, while for unengaged players we might email them helpful tutorials). Say that we wish to split our players into two groups. We would first define basic attributes of the players, such as the number of hours played, total money spent on in-app purchases and number of levels completed. We can then feed this data set (three attributes for every player) to an unsupervised learning algorithm where we specify the number of groups to be two. The algorithm would then split the data set of players into two groups where the players within each group would be similar to each other. Given the attributes we used to describe each player, in this case, the output would be a split of all the players into two groups, where one group would semantically represent the engaged players and the second group would semantically represent the unengaged players.</p> <p>With unsupervised learning, we did not provide specific examples of which players are considered engaged and which are considered unengaged. We just defined the appropriate attributes and relied on the algorithm to uncover the two groups on its own. This type of data set is typically called an unlabeled data set as it is lacking these direct labels. Consequently, unsupervised learning can be helpful in situations where these labels can be expensive or hard to produce. In the next paragraph, we overview supervised learning algorithms which accept input labels in addition to attributes.</p>"},{"location":"Background-Machine-Learning/#supervised-learning","title":"Supervised Learning","text":"<p>In supervised learning, we do not want to just group similar items but directly learn a mapping from each item to the group (or class) that it belongs to. Returning to our earlier example of clustering players, let's say we now wish to predict which of our players are about to churn (that is stop playing the game for the next 30 days). We can look into our historical records and create a data set that contains attributes of our players in addition to a label indicating whether they have churned or not. Note that the player attributes we use for this churn prediction task may be different from the ones we used for our earlier clustering task. We can then feed this data set (attributes and label for each player) into a supervised learning algorithm which would learn a mapping from the player attributes to a label indicating whether that player will churn or not. The intuition is that the supervised learning algorithm will learn which values of these attributes typically correspond to players who have churned and not churned (for example, it may learn that players who spend very little and play for very short periods will most likely churn). Now given this learned model, we can provide it the attributes of a new player (one that recently started playing the game) and it would output a predicted label for that player. This prediction is the algorithms expectation of whether the player will churn or not. We can now use these predictions to target the players who are expected to churn and entice them to continue playing the game.</p> <p>As you may have noticed, for both supervised and unsupervised learning, there are two tasks that need to be performed: attribute selection and model selection. Attribute selection (also called feature selection) pertains to selecting how we wish to represent the entity of interest, in this case, the player. Model selection, on the other hand, pertains to selecting the algorithm (and its parameters) that perform the task well. Both of these tasks are active areas of machine learning research and, in practice, require several iterations to achieve good performance.</p> <p>We now switch to reinforcement learning, the third class of machine learning algorithms, and arguably the one most relevant for the ML-Agents Toolkit.</p>"},{"location":"Background-Machine-Learning/#reinforcement-learning","title":"Reinforcement Learning","text":"<p>Reinforcement learning can be viewed as a form of learning for sequential decision making that is commonly associated with controlling robots (but is, in fact, much more general). Consider an autonomous firefighting robot that is tasked with navigating into an area, finding the fire and neutralizing it. At any given moment, the robot perceives the environment through its sensors (e.g. camera, heat, touch), processes this information and produces an action (e.g. move to the left, rotate the water hose, turn on the water). In other words, it is continuously making decisions about how to interact in this environment given its view of the world (i.e. sensors input) and objective (i.e. neutralizing the fire). Teaching a robot to be a successful firefighting machine is precisely what reinforcement learning is designed to do.</p> <p>More specifically, the goal of reinforcement learning is to learn a policy, which is essentially a mapping from observations to actions. An observation is what the robot can measure from its environment (in this case, all its sensory inputs) and an action, in its most raw form, is a change to the configuration of the robot (e.g. position of its base, position of its water hose and whether the hose is on or off).</p> <p>The last remaining piece of the reinforcement learning task is the reward signal. The robot is trained to learn a policy that maximizes its overall rewards. When training a robot to be a mean firefighting machine, we provide it with rewards (positive and negative) indicating how well it is doing on completing the task. Note that the robot does not know how to put out fires before it is trained. It learns the objective because it receives a large positive reward when it puts out the fire and a small negative reward for every passing second. The fact that rewards are sparse (i.e. may not be provided at every step, but only when a robot arrives at a success or failure situation), is a defining characteristic of reinforcement learning and precisely why learning good policies can be difficult (and/or time-consuming) for complex environments.</p> <p>Learning a policy usually requires many trials and iterative policy updates. More specifically, the robot is placed in several fire situations and over time learns an optimal policy which allows it to put out fires more effectively. Obviously, we cannot expect to train a robot repeatedly in the real world, particularly when fires are involved. This is precisely why the use of Unity as a simulator serves as the perfect training grounds for learning such behaviors. While our discussion of reinforcement learning has centered around robots, there are strong parallels between robots and characters in a game. In fact, in many ways, one can view a non-playable character (NPC) as a virtual robot, with its own observations about the environment, its own set of actions and a specific objective. Thus it is natural to explore how we can train behaviors within Unity using reinforcement learning. This is precisely what the ML-Agents Toolkit offers. The video linked below includes a reinforcement learning demo showcasing training character behaviors using the ML-Agents Toolkit.</p> <p> </p> <p>Similar to both unsupervised and supervised learning, reinforcement learning also involves two tasks: attribute selection and model selection. Attribute selection is defining the set of observations for the robot that best help it complete its objective, while model selection is defining the form of the policy (mapping from observations to actions) and its parameters. In practice, training behaviors is an iterative process that may require changing the attribute and model choices.</p>"},{"location":"Background-Machine-Learning/#training-and-inference","title":"Training and Inference","text":"<p>One common aspect of all three branches of machine learning is that they all involve a training phase and an inference phase. While the details of the training and inference phases are different for each of the three, at a high-level, the training phase involves building a model using the provided data, while the inference phase involves applying this model to new, previously unseen, data. More specifically:</p> <ul> <li>For our unsupervised learning example, the training phase learns the optimal   two clusters based on the data describing existing players, while the   inference phase assigns a new player to one of these two clusters.</li> <li>For our supervised learning example, the training phase learns the mapping   from player attributes to player label (whether they churned or not), and the   inference phase predicts whether a new player will churn or not based on that   learned mapping.</li> <li>For our reinforcement learning example, the training phase learns the optimal   policy through guided trials, and in the inference phase, the agent observes   and takes actions in the wild using its learned policy.</li> </ul> <p>To briefly summarize: all three classes of algorithms involve training and inference phases in addition to attribute and model selections. What ultimately separates them is the type of data available to learn from. In unsupervised learning our data set was a collection of attributes, in supervised learning our data set was a collection of attribute-label pairs, and, lastly, in reinforcement learning our data set was a collection of observation-action-reward tuples.</p>"},{"location":"Background-Machine-Learning/#deep-learning","title":"Deep Learning","text":"<p>Deep learning is a family of algorithms that can be used to address any of the problems introduced above. More specifically, they can be used to solve both attribute and model selection tasks. Deep learning has gained popularity in recent years due to its outstanding performance on several challenging machine learning tasks. One example is AlphaGo, a computer Go program, that leverages deep learning, that was able to beat Lee Sedol (a Go world champion).</p> <p>A key characteristic of deep learning algorithms is their ability to learn very complex functions from large amounts of training data. This makes them a natural choice for reinforcement learning tasks when a large amount of data can be generated, say through the use of a simulator or engine such as Unity. By generating hundreds of thousands of simulations of the environment within Unity, we can learn policies for very complex environments (a complex environment is one where the number of observations an agent perceives and the number of actions they can take are large). Many of the algorithms we provide in ML-Agents use some form of deep learning, built on top of the open-source library, PyTorch.</p>"},{"location":"Background-PyTorch/","title":"Background: PyTorch","text":"<p>As discussed in our machine learning background page, many of the algorithms we provide in the ML-Agents Toolkit leverage some form of deep learning. More specifically, our implementations are built on top of the open-source library PyTorch. In this page we provide a brief overview of PyTorch and TensorBoard that we leverage within the ML-Agents Toolkit.</p>"},{"location":"Background-PyTorch/#pytorch","title":"PyTorch","text":"<p>PyTorch is an open source library for performing computations using data flow graphs, the underlying representation of deep learning models. It facilitates training and inference on CPUs and GPUs in a desktop, server, or mobile device. Within the ML-Agents Toolkit, when you train the behavior of an agent, the output is a model (.onnx) file that you can then associate with an Agent. Unless you implement a new algorithm, the use of PyTorch is mostly abstracted away and behind the scenes.</p>"},{"location":"Background-PyTorch/#tensorboard","title":"TensorBoard","text":"<p>One component of training models with PyTorch is setting the values of certain model attributes (called hyperparameters). Finding the right values of these hyperparameters can require a few iterations. Consequently, we leverage a visualization tool called TensorBoard. It allows the visualization of certain agent attributes (e.g. reward) throughout training which can be helpful in both building intuitions for the different hyperparameters and setting the optimal values for your Unity environment. We provide more details on setting the hyperparameters in the Training ML-Agents page. If you are unfamiliar with TensorBoard we recommend our guide on using TensorBoard with ML-Agents or this tutorial.</p>"},{"location":"Background-Unity/","title":"Background: Unity","text":"<p>If you are not familiar with the Unity Engine, we highly recommend the Unity Manual and Tutorials page. The Roll-a-ball tutorial is a fantastic resource to learn all the basic concepts of Unity to get started with the ML-Agents Toolkit:</p> <ul> <li>Editor</li> <li>Scene</li> <li>GameObject</li> <li>Rigidbody</li> <li>Camera</li> <li>Scripting</li> <li>Physics</li> <li>Ordering of event functions   (e.g. FixedUpdate, Update)</li> <li>Prefabs</li> </ul>"},{"location":"CODE_OF_CONDUCT/","title":"Contributor Covenant Code of Conduct","text":""},{"location":"CODE_OF_CONDUCT/#our-pledge","title":"Our Pledge","text":"<p>In the interest of fostering an open and welcoming environment, we as contributors and maintainers pledge to making participation in our project and our community a harassment-free experience for everyone, regardless of age, body size, disability, ethnicity, gender identity and expression, level of experience, nationality, personal appearance, race, religion, or sexual identity and orientation.</p>"},{"location":"CODE_OF_CONDUCT/#our-standards","title":"Our Standards","text":"<p>Examples of behavior that contributes to creating a positive environment include:</p> <ul> <li>Using welcoming and inclusive language</li> <li>Being respectful of differing viewpoints and experiences</li> <li>Gracefully accepting constructive criticism</li> <li>Focusing on what is best for the community</li> <li>Showing empathy towards other community members</li> </ul> <p>Examples of unacceptable behavior by participants include:</p> <ul> <li>The use of sexualized language or imagery and unwelcome sexual attention or   advances</li> <li>Trolling, insulting/derogatory comments, and personal or political attacks</li> <li>Public or private harassment</li> <li>Publishing others' private information, such as a physical or electronic   address, without explicit permission</li> <li>Other conduct which could reasonably be considered inappropriate in a   professional setting</li> </ul>"},{"location":"CODE_OF_CONDUCT/#our-responsibilities","title":"Our Responsibilities","text":"<p>Project maintainers are responsible for clarifying the standards of acceptable behavior and are expected to take appropriate and fair corrective action in response to any instances of unacceptable behavior.</p> <p>Project maintainers have the right and responsibility to remove, edit, or reject comments, commits, code, wiki edits, issues, and other contributions that are not aligned to this Code of Conduct, or to ban temporarily or permanently any contributor for other behaviors that they deem inappropriate, threatening, offensive, or harmful.</p>"},{"location":"CODE_OF_CONDUCT/#scope","title":"Scope","text":"<p>This Code of Conduct applies both within project spaces and in public spaces when an individual is representing the project or its community. Examples of representing a project or community include using an official project e-mail address, posting via an official social media account, or acting as an appointed representative at an online or offline event. Representation of a project may be further defined and clarified by project maintainers.</p>"},{"location":"CODE_OF_CONDUCT/#enforcement","title":"Enforcement","text":"<p>Instances of abusive, harassing, or otherwise unacceptable behavior may be reported by contacting the project team at ml-agents@unity3d.com. All complaints will be reviewed and investigated and will result in a response that is deemed necessary and appropriate to the circumstances. The project team is obligated to maintain confidentiality with regard to the reporter of an incident. Further details of specific enforcement policies may be posted separately.</p> <p>Project maintainers who do not follow or enforce the Code of Conduct in good faith may face temporary or permanent repercussions as determined by other members of the project's leadership.</p>"},{"location":"CODE_OF_CONDUCT/#attribution","title":"Attribution","text":"<p>This Code of Conduct is adapted from the Contributor Covenant, version 1.4, available at https://www.contributor-covenant.org/version/1/4/code-of-conduct/</p>"},{"location":"CONTRIBUTING/","title":"How to Contribute to ML-Agents","text":""},{"location":"CONTRIBUTING/#1fork-the-repository","title":"1.Fork the repository","text":"<p>Fork the ML-Agents repository by clicking on the \"Fork\" button in the top right corner of the GitHub page. This creates a copy of the repository under your GitHub account.</p>"},{"location":"CONTRIBUTING/#2-set-up-your-development-environment","title":"2. Set up your development environment","text":"<p>Clone the forked repository to your local machine using Git. Install the necessary dependencies and follow the instructions provided in the project's documentation to set up your development environment properly.</p>"},{"location":"CONTRIBUTING/#3-choose-an-issue-or-feature","title":"3. Choose an issue or feature","text":"<p>Browse the project's issue tracker or discussions to find an open issue or feature that you would like to contribute to. Read the guidelines and comments associated with the issue to understand the requirements and constraints.</p>"},{"location":"CONTRIBUTING/#4-make-your-changes","title":"4. Make your changes","text":"<p>Create a new branch for your changes based on the main branch of the ML-Agents repository. Implement your code changes or add new features as necessary. Ensure that your code follows the project's coding style and conventions.</p> <ul> <li>Example: Let's say you want to add support for a new type of reward function in the ML-Agents framework. You can create a new branch named feature/reward-function to implement this feature.</li> </ul>"},{"location":"CONTRIBUTING/#5-test-your-changes","title":"5. Test your changes","text":"<p>Run the appropriate tests to ensure your changes work as intended. If necessary, add new tests to cover your code and verify that it doesn't introduce regressions.</p> <ul> <li>Example: For the reward function feature, you would write tests to check different scenarios and expected outcomes of the new reward function.</li> </ul>"},{"location":"CONTRIBUTING/#6-submit-a-pull-request","title":"6. Submit a pull request","text":"<p>Push your branch to your forked repository and submit a pull request (PR) to the ML-Agents main repository. Provide a clear and concise description of your changes, explaining the problem you solved or the feature you added.</p> <ul> <li>Example: In the pull request description, you would explain how the new reward function works, its benefits, and any relevant implementation details.</li> </ul>"},{"location":"CONTRIBUTING/#7-respond-to-feedback","title":"7. Respond to feedback","text":"<p>Be responsive to any feedback or comments provided by the project maintainers. Address the feedback by making necessary revisions to your code and continue the discussion if required.</p>"},{"location":"CONTRIBUTING/#8-continuous-integration-and-code-review","title":"8. Continuous integration and code review","text":"<p>The ML-Agents project utilizes automated continuous integration (CI) systems to run tests on pull requests. Address any issues flagged by the CI system and actively participate in the code review process by addressing comments from reviewers.</p>"},{"location":"CONTRIBUTING/#9-merge-your-changes","title":"9. Merge your changes","text":"<p>Once your pull request has been approved and meets all the project's requirements, a project maintainer will merge your changes into the main repository. Congratulations, your contribution has been successfully integrated!</p> <p>Remember to always adhere to the project's code of conduct, be respectful, and follow any specific contribution guidelines provided by the ML-Agents project. Happy contributing!</p>"},{"location":"Custom-SideChannels/","title":"Custom Side Channels","text":"<p>You can create your own side channel in C# and Python and use it to communicate custom data structures between the two. This can be useful for situations in which the data to be sent is too complex or structured for the built-in <code>EnvironmentParameters</code>, or is not related to any specific agent, and therefore inappropriate as an agent observation.</p>"},{"location":"Custom-SideChannels/#overview","title":"Overview","text":"<p>In order to use a side channel, it must be implemented as both Unity and Python classes.</p>"},{"location":"Custom-SideChannels/#unity-side","title":"Unity side","text":"<p>The side channel will have to implement the <code>SideChannel</code> abstract class and the following method.</p> <ul> <li><code>OnMessageReceived(IncomingMessage msg)</code> : You must implement this method and   read the data from IncomingMessage. The data must be read in the order that it   was written.</li> </ul> <p>The side channel must also assign a <code>ChannelId</code> property in the constructor. The <code>ChannelId</code> is a Guid (or UUID in Python) used to uniquely identify a side channel. This Guid must be the same on C# and Python. There can only be one side channel of a certain id during communication.</p> <p>To send data from C# to Python, create an <code>OutgoingMessage</code> instance, add data to it, call the <code>base.QueueMessageToSend(msg)</code> method inside the side channel, and call the <code>OutgoingMessage.Dispose()</code> method.</p> <p>To register a side channel on the Unity side, call <code>SideChannelManager.RegisterSideChannel</code> with the side channel as only argument.</p>"},{"location":"Custom-SideChannels/#python-side","title":"Python side","text":"<p>The side channel will have to implement the <code>SideChannel</code> abstract class. You must implement :</p> <ul> <li><code>on_message_received(self, msg: \"IncomingMessage\") -&gt; None</code> : You must   implement this method and read the data from IncomingMessage. The data must be   read in the order that it was written.</li> </ul> <p>The side channel must also assign a <code>channel_id</code> property in the constructor. The <code>channel_id</code> is a UUID (referred in C# as Guid) used to uniquely identify a side channel. This number must be the same on C# and Python. There can only be one side channel of a certain id during communication.</p> <p>To assign the <code>channel_id</code> call the abstract class constructor with the appropriate <code>channel_id</code> as follows:</p> <pre><code>super().__init__(my_channel_id)\n</code></pre> <p>To send a byte array from Python to C#, create an <code>OutgoingMessage</code> instance, add data to it, and call the <code>super().queue_message_to_send(msg)</code> method inside the side channel.</p> <p>To register a side channel on the Python side, pass the side channel as argument when creating the <code>UnityEnvironment</code> object. One of the arguments of the constructor (<code>side_channels</code>) is a list of side channels.</p>"},{"location":"Custom-SideChannels/#example-implementation","title":"Example implementation","text":"<p>Below is a simple implementation of a side channel that will exchange ASCII encoded strings between a Unity environment and Python.</p>"},{"location":"Custom-SideChannels/#example-unity-c-code","title":"Example Unity C# code","text":"<p>The first step is to create the <code>StringLogSideChannel</code> class within the Unity project. Here is an implementation of a <code>StringLogSideChannel</code> that will listen for messages from python and print them to the Unity debug log, as well as send error messages from Unity to python.</p> <pre><code>using UnityEngine;\nusing Unity.MLAgents;\nusing Unity.MLAgents.SideChannels;\nusing System.Text;\nusing System;\n\npublic class StringLogSideChannel : SideChannel\n{\n    public StringLogSideChannel()\n    {\n        ChannelId = new Guid(\"621f0a70-4f87-11ea-a6bf-784f4387d1f7\");\n    }\n\n    protected override void OnMessageReceived(IncomingMessage msg)\n    {\n        var receivedString = msg.ReadString();\n        Debug.Log(\"From Python : \" + receivedString);\n    }\n\n    public void SendDebugStatementToPython(string logString, string stackTrace, LogType type)\n    {\n        if (type == LogType.Error)\n        {\n            var stringToSend = type.ToString() + \": \" + logString + \"\\n\" + stackTrace;\n            using (var msgOut = new OutgoingMessage())\n            {\n                msgOut.WriteString(stringToSend);\n                QueueMessageToSend(msgOut);\n            }\n        }\n    }\n}\n</code></pre> <p>Once we have defined our custom side channel class, we need to ensure that it is instantiated and registered. This can typically be done wherever the logic of the side channel makes sense to be associated, for example on a MonoBehaviour object that might need to access data from the side channel. Here we show a simple MonoBehaviour object which instantiates and registers the new side channel. If you have not done it already, make sure that the MonoBehaviour which registers the side channel is attached to a GameObject which will be live in your Unity scene.</p> <pre><code>using UnityEngine;\nusing Unity.MLAgents;\n\n\npublic class RegisterStringLogSideChannel : MonoBehaviour\n{\n\n    StringLogSideChannel stringChannel;\n    public void Awake()\n    {\n        // We create the Side Channel\n        stringChannel = new StringLogSideChannel();\n\n        // When a Debug.Log message is created, we send it to the stringChannel\n        Application.logMessageReceived += stringChannel.SendDebugStatementToPython;\n\n        // The channel must be registered with the SideChannelManager class\n        SideChannelManager.RegisterSideChannel(stringChannel);\n    }\n\n    public void OnDestroy()\n    {\n        // De-register the Debug.Log callback\n        Application.logMessageReceived -= stringChannel.SendDebugStatementToPython;\n        if (Academy.IsInitialized){\n            SideChannelManager.UnregisterSideChannel(stringChannel);\n        }\n    }\n\n    public void Update()\n    {\n        // Optional : If the space bar is pressed, raise an error !\n        if (Input.GetKeyDown(KeyCode.Space))\n        {\n            Debug.LogError(\"This is a fake error. Space bar was pressed in Unity.\");\n        }\n    }\n}\n</code></pre>"},{"location":"Custom-SideChannels/#example-python-code","title":"Example Python code","text":"<p>Now that we have created the necessary Unity C# classes, we can create their Python counterparts.</p> <pre><code>from mlagents_envs.environment import UnityEnvironment\nfrom mlagents_envs.side_channel.side_channel import (\n    SideChannel,\n    IncomingMessage,\n    OutgoingMessage,\n)\nimport numpy as np\nimport uuid\n\n\n# Create the StringLogChannel class\nclass StringLogChannel(SideChannel):\n\n    def __init__(self) -&gt; None:\n        super().__init__(uuid.UUID(\"621f0a70-4f87-11ea-a6bf-784f4387d1f7\"))\n\n    def on_message_received(self, msg: IncomingMessage) -&gt; None:\n        \"\"\"\n        Note: We must implement this method of the SideChannel interface to\n        receive messages from Unity\n        \"\"\"\n        # We simply read a string from the message and print it.\n        print(msg.read_string())\n\n    def send_string(self, data: str) -&gt; None:\n        # Add the string to an OutgoingMessage\n        msg = OutgoingMessage()\n        msg.write_string(data)\n        # We call this method to queue the data we want to send\n        super().queue_message_to_send(msg)\n</code></pre> <p>We can then instantiate the new side channel, launch a <code>UnityEnvironment</code> with that side channel active, and send a series of messages to the Unity environment from Python using it.</p> <pre><code># Create the channel\nstring_log = StringLogChannel()\n\n# We start the communication with the Unity Editor and pass the string_log side channel as input\nenv = UnityEnvironment(side_channels=[string_log])\nenv.reset()\nstring_log.send_string(\"The environment was reset\")\n\ngroup_name = list(env.behavior_specs.keys())[0]  # Get the first group_name\ngroup_spec = env.behavior_specs[group_name]\nfor i in range(1000):\n    decision_steps, terminal_steps = env.get_steps(group_name)\n    # We send data to Unity : A string with the number of Agent at each\n    string_log.send_string(\n        f\"Step {i} occurred with {len(decision_steps)} deciding agents and \"\n        f\"{len(terminal_steps)} terminal agents\"\n    )\n    env.step()  # Move the simulation forward\n\nenv.close()\n</code></pre> <p>Now, if you run this script and press <code>Play</code> the Unity Editor when prompted, the console in the Unity Editor will display a message at every Python step. Additionally, if you press the Space Bar in the Unity Engine, a message will appear in the terminal.</p>"},{"location":"ELO-Rating-System/","title":"ELO Rating System","text":"<p>In adversarial games, the cumulative environment reward may not be a meaningful metric by which to track learning progress.</p> <p>This is because the cumulative reward is entirely dependent on the skill of the opponent.</p> <p>An agent at a particular skill level will get more or less reward against a worse or better agent, respectively.</p> <p>Instead, it's better to use ELO rating system, a method to calculate the relative skill level between two players in a zero-sum game.</p> <p>If the training performs correctly, this value should steadily increase.</p>"},{"location":"ELO-Rating-System/#what-is-a-zero-sum-game","title":"What is a zero-sum game?","text":"<p>A zero-sum game is a game where each player's gain or loss of utility is exactly balanced by the gain or loss of the utility of the opponent.</p> <p>Simply explained, we face a zero-sum game when one agent gets +1.0, its opponent gets -1.0 reward.</p> <p>For instance, Tennis is a zero-sum game: if you win the point you get +1.0 and your opponent gets -1.0 reward.</p>"},{"location":"ELO-Rating-System/#how-works-the-elo-rating-system","title":"How works the ELO Rating System","text":"<ul> <li> <p>Each player has an initial ELO score. It's defined in the <code>initial_elo</code> trainer config hyperparameter.</p> </li> <li> <p>The difference in rating between the two players serves as the predictor of the outcomes of a match.</p> </li> </ul> <p> For instance, if player A has an Elo score of 2100 and player B has an ELO score of 1800 the chance that player A wins is 85% against 15% for player b.</p> <ul> <li>We calculate the expected score of each player using this formula:</li> </ul> <p></p> <ul> <li>At the end of the game, based on the outcome we update the player\u2019s actual Elo score, we use a linear adjustment proportional to the amount by which the player over-performed or under-performed. The winning player takes points from the losing one:</li> <li>If the higher-rated player wins \u2192 a few points will be taken from the lower-rated player.</li> <li>If the lower-rated player wins \u2192 a lot of points will be taken from the high-rated player.</li> <li> <p>If it\u2019s a draw \u2192 the lower-rated player gains a few points from higher.</p> </li> <li> <p>We update players rating using this formula:</p> </li> </ul> <p></p>"},{"location":"ELO-Rating-System/#the-tennis-example","title":"The Tennis example","text":"<ul> <li>We start to train our agents.</li> <li>Both of them have the same skills. So ELO score for each of them that we defined using parameter <code>initial_elo = 1200.0</code>.</li> </ul> <p>We calculate the expected score E: Ea = 0.5 Eb = 0.5</p> <p>So it means that each player has 50% chances of winning the point.</p> <p>If A wins, the new rating R would be:</p> <p>Ra = 1200 + 16 * (1 - 0.5) \u2192 1208</p> <p>Rb = 1200 + 16 * (0 - 0.5) \u2192 1192</p> <p>Player A has now an ELO score of 1208 and Player B an ELO score of 1192. Therefore, Player A is now a little bit better than Player B.</p>"},{"location":"FAQ/","title":"Frequently Asked Questions","text":""},{"location":"FAQ/#installation-problems","title":"Installation problems","text":""},{"location":"FAQ/#environment-permission-error","title":"Environment Permission Error","text":"<p>If you directly import your Unity environment without building it in the editor, you might need to give it additional permissions to execute it.</p> <p>If you receive such a permission error on macOS, run:</p> <pre><code>chmod -R 755 *.app\n</code></pre> <p>or on Linux:</p> <pre><code>chmod -R 755 *.x86_64\n</code></pre> <p>On Windows, you can find instructions.</p>"},{"location":"FAQ/#environment-connection-timeout","title":"Environment Connection Timeout","text":"<p>If you are able to launch the environment from <code>UnityEnvironment</code> but then receive a timeout error like this:</p> <pre><code>UnityAgentsException: The Communicator was unable to connect. Please make sure the External process is ready to accept communication with Unity.\n</code></pre> <p>There may be a number of possible causes:</p> <ul> <li>Cause: There may be no agent in the scene</li> <li>Cause: On OSX, the firewall may be preventing communication with the   environment. Solution: Add the built environment binary to the list of   exceptions on the firewall by following   instructions.</li> <li>Cause: An error happened in the Unity Environment preventing communication.   Solution: Look into the   log files generated by the   Unity Environment to figure what error happened.</li> <li>Cause: You have assigned <code>HTTP_PROXY</code> and <code>HTTPS_PROXY</code> values in your   environment variables. Solution: Remove these values and try again.</li> <li>Cause: You are running in a headless environment (e.g. remotely connected   to a server). Solution: Pass <code>--no-graphics</code> to <code>mlagents-learn</code>, or   <code>no_graphics=True</code> to <code>RemoteRegistryEntry.make()</code> or the <code>UnityEnvironment</code>   initializer. If you need graphics for visual observations, you will need to   set up <code>xvfb</code> (or equivalent).</li> </ul>"},{"location":"FAQ/#communication-port-still-in-use","title":"Communication port {} still in use","text":"<p>If you receive an exception <code>\"Couldn't launch new environment because communication port {} is still in use. \"</code>, you can change the worker number in the Python script when calling</p> <pre><code>UnityEnvironment(file_name=filename, worker_id=X)\n</code></pre>"},{"location":"FAQ/#mean-reward-nan","title":"Mean reward : nan","text":"<p>If you receive a message <code>Mean reward : nan</code> when attempting to train a model using PPO, this is due to the episodes of the Learning Environment not terminating. In order to address this, set <code>Max Steps</code> for the Agents within the Scene Inspector to a value greater than 0. Alternatively, it is possible to manually set <code>done</code> conditions for episodes from within scripts for custom episode-terminating events.</p>"},{"location":"FAQ/#file-name-cannot-be-opened-because-the-developer-cannot-be-verified","title":"\"File name\" cannot be opened because the developer cannot be verified.","text":"<p>If you have downloaded the repository using the github website on macOS 10.15 (Catalina) or later, you may see this error when attempting to play scenes in the Unity project. Workarounds include installing the package using the Unity Package Manager (this is the officially supported approach - see here), or following the instructions here to verify the relevant files on your machine on a file-by-file basis.</p>"},{"location":"Getting-Started/","title":"Getting Started Guide","text":"<p>This guide walks through the end-to-end process of opening one of our example environments in Unity, training an Agent in it, and embedding the trained model into the Unity environment. After reading this tutorial, you should be able to train any of the example environments. If you are not familiar with the Unity Engine, view our Background: Unity page for helpful pointers. Additionally, if you're not familiar with machine learning, view our Background: Machine Learning page for a brief overview and helpful pointers.</p> <p></p> <p>For this guide, we'll use the 3D Balance Ball environment which contains a number of agent cubes and balls (which are all copies of each other). Each agent cube tries to keep its ball from falling by rotating either horizontally or vertically. In this environment, an agent cube is an Agent that receives a reward for every step that it balances the ball. An agent is also penalized with a negative reward for dropping the ball. The goal of the training process is to have the agents learn to balance the ball on their head.</p> <p>Let's get started!</p>"},{"location":"Getting-Started/#installation","title":"Installation","text":"<p>If you haven't already, follow the installation instructions. Afterwards, open the Unity Project that contains all the example environments:</p> <ol> <li>Open the Package Manager Window by navigating to <code>Window -&gt; Package Manager</code>    in the menu.</li> <li>Navigate to the ML-Agents Package and click on it.</li> <li>Find the <code>3D Ball</code> sample and click <code>Import</code>.</li> <li>In the Project window, go to the    <code>Assets/ML-Agents/Examples/3DBall/Scenes</code> folder and open the <code>3DBall</code> scene    file.</li> </ol>"},{"location":"Getting-Started/#understanding-a-unity-environment","title":"Understanding a Unity Environment","text":"<p>An agent is an autonomous actor that observes and interacts with an environment. In the context of Unity, an environment is a scene containing one or more Agent objects, and, of course, the other entities that an agent interacts with.</p> <p></p> <p>Note: In Unity, the base object of everything in a scene is the GameObject. The GameObject is essentially a container for everything else, including behaviors, graphics, physics, etc. To see the components that make up a GameObject, select the GameObject in the Scene window, and open the Inspector window. The Inspector shows every component on a GameObject.</p> <p>The first thing you may notice after opening the 3D Balance Ball scene is that it contains not one, but several agent cubes. Each agent cube in the scene is an independent agent, but they all share the same Behavior. 3D Balance Ball does this to speed up training since all twelve agents contribute to training in parallel.</p>"},{"location":"Getting-Started/#agent","title":"Agent","text":"<p>The Agent is the actor that observes and takes actions in the environment. In the 3D Balance Ball environment, the Agent components are placed on the twelve \"Agent\" GameObjects. The base Agent object has a few properties that affect its behavior:</p> <ul> <li>Behavior Parameters \u2014 Every Agent must have a Behavior. The Behavior   determines how an Agent makes decisions.</li> <li>Max Step \u2014 Defines how many simulation steps can occur before the Agent's   episode ends. In 3D Balance Ball, an Agent restarts after 5000 steps.</li> </ul>"},{"location":"Getting-Started/#behavior-parameters-vector-observation-space","title":"Behavior Parameters : Vector Observation Space","text":"<p>Before making a decision, an agent collects its observation about its state in the world. The vector observation is a vector of floating point numbers which contain relevant information for the agent to make decisions.</p> <p>The Behavior Parameters of the 3D Balance Ball example uses a <code>Space Size</code> of 8. This means that the feature vector containing the Agent's observations contains eight elements: the <code>x</code> and <code>z</code> components of the agent cube's rotation and the <code>x</code>, <code>y</code>, and <code>z</code> components of the ball's relative position and velocity.</p>"},{"location":"Getting-Started/#behavior-parameters-actions","title":"Behavior Parameters : Actions","text":"<p>An Agent is given instructions in the form of actions. ML-Agents Toolkit classifies actions into two types: continuous and discrete. The 3D Balance Ball example is programmed to use continuous actions, which are a vector of floating-point numbers that can vary continuously. More specifically, it uses a <code>Space Size</code> of 2 to control the amount of <code>x</code> and <code>z</code> rotations to apply to itself to keep the ball balanced on its head.</p>"},{"location":"Getting-Started/#running-a-pre-trained-model","title":"Running a pre-trained model","text":"<p>We include pre-trained models for our agents (<code>.onnx</code> files) and we use the Sentis to run these models inside Unity. In this section, we will use the pre-trained model for the 3D Ball example.</p> <ol> <li>In the Project window, go to the    <code>Assets/ML-Agents/Examples/3DBall/Prefabs</code> folder. Expand <code>3DBall</code> and click    on the <code>Agent</code> prefab. You should see the <code>Agent</code> prefab in the Inspector    window.</li> </ol> <p>Note: The platforms in the <code>3DBall</code> scene were created using the <code>3DBall</code>    prefab. Instead of updating all 12 platforms individually, you can update the    <code>3DBall</code> prefab instead.</p> <p></p> <ol> <li>In the Project window, drag the 3DBall Model located in    <code>Assets/ML-Agents/Examples/3DBall/TFModels</code> into the <code>Model</code> property under    <code>Behavior Parameters (Script)</code> component in the Agent GameObject    Inspector window.</li> </ol> <p></p> <ol> <li>You should notice that each <code>Agent</code> under each <code>3DBall</code> in the Hierarchy    windows now contains 3DBall as <code>Model</code> on the <code>Behavior Parameters</code>.    Note : You can modify multiple game objects in a scene by selecting them    all at once using the search bar in the Scene Hierarchy.</li> <li>Set the Inference Device to use for this model as <code>CPU</code>.</li> <li>Click the Play button in the Unity Editor and you will see the platforms    balance the balls using the pre-trained model.</li> </ol>"},{"location":"Getting-Started/#training-a-new-model-with-reinforcement-learning","title":"Training a new model with Reinforcement Learning","text":"<p>While we provide pre-trained models for the agents in this environment, any environment you make yourself will require training agents from scratch to generate a new model file. In this section we will demonstrate how to use the reinforcement learning algorithms that are part of the ML-Agents Python package to accomplish this. We have provided a convenient command <code>mlagents-learn</code> which accepts arguments used to configure both training and inference phases.</p>"},{"location":"Getting-Started/#training-the-environment","title":"Training the environment","text":"<ol> <li>Open a command or terminal window.</li> <li>Navigate to the folder where you cloned the <code>ml-agents</code> repository. Note:    If you followed the default installation, then you should    be able to run <code>mlagents-learn</code> from any directory.</li> <li>Run <code>mlagents-learn config/ppo/3DBall.yaml --run-id=first3DBallRun</code>.</li> <li><code>config/ppo/3DBall.yaml</code> is the path to a default training      configuration file that we provide. The <code>config/ppo</code> folder includes training configuration      files for all our example environments, including 3DBall.</li> <li><code>run-id</code> is a unique name for this training session.</li> <li>When the message \"Start training by pressing the Play button in the Unity    Editor\" is displayed on the screen, you can press the Play button in    Unity to start training in the Editor.</li> </ol> <p>If <code>mlagents-learn</code> runs correctly and starts training, you should see something like this:</p> <pre><code>INFO:mlagents_envs:\n'Ball3DAcademy' started successfully!\nUnity Academy name: Ball3DAcademy\n\nINFO:mlagents_envs:Connected new brain:\nUnity brain name: 3DBallLearning\n        Number of Visual Observations (per agent): 0\n        Vector Observation space size (per agent): 8\n        Number of stacked Vector Observation: 1\nINFO:mlagents_envs:Hyperparameters for the PPO Trainer of brain 3DBallLearning:\n        batch_size:          64\n        beta:                0.001\n        buffer_size:         12000\n        epsilon:             0.2\n        gamma:               0.995\n        hidden_units:        128\n        lambd:               0.99\n        learning_rate:       0.0003\n        max_steps:           5.0e4\n        normalize:           True\n        num_epoch:           3\n        num_layers:          2\n        time_horizon:        1000\n        sequence_length:     64\n        summary_freq:        1000\n        use_recurrent:       False\n        memory_size:         256\n        use_curiosity:       False\n        curiosity_strength:  0.01\n        curiosity_enc_size:  128\n        output_path: ./results/first3DBallRun/3DBallLearning\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 1000. Mean Reward: 1.242. Std of Reward: 0.746. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 2000. Mean Reward: 1.319. Std of Reward: 0.693. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 3000. Mean Reward: 1.804. Std of Reward: 1.056. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 4000. Mean Reward: 2.151. Std of Reward: 1.432. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 5000. Mean Reward: 3.175. Std of Reward: 2.250. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 6000. Mean Reward: 4.898. Std of Reward: 4.019. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 7000. Mean Reward: 6.716. Std of Reward: 5.125. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 8000. Mean Reward: 12.124. Std of Reward: 11.929. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 9000. Mean Reward: 18.151. Std of Reward: 16.871. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 10000. Mean Reward: 27.284. Std of Reward: 28.667. Training.\n</code></pre> <p>Note how the <code>Mean Reward</code> value printed to the screen increases as training progresses. This is a positive sign that training is succeeding.</p> <p>Note: You can train using an executable rather than the Editor. To do so, follow the instructions in Using an Executable.</p>"},{"location":"Getting-Started/#observing-training-progress","title":"Observing Training Progress","text":"<p>Once you start training using <code>mlagents-learn</code> in the way described in the previous section, the <code>ml-agents</code> directory will contain a <code>results</code> directory. In order to observe the training process in more detail, you can use TensorBoard. From the command line run:</p> <pre><code>tensorboard --logdir results\n</code></pre> <p>Then navigate to <code>localhost:6006</code> in your browser to view the TensorBoard summary statistics as shown below. For the purposes of this section, the most important statistic is <code>Environment/Cumulative Reward</code> which should increase throughout training, eventually converging close to <code>100</code> which is the maximum reward the agent can accumulate.</p> <p></p>"},{"location":"Getting-Started/#embedding-the-model-into-the-unity-environment","title":"Embedding the model into the Unity Environment","text":"<p>Once the training process completes, and the training process saves the model (denoted by the <code>Saved Model</code> message) you can add it to the Unity project and use it with compatible Agents (the Agents that generated the model). Note: Do not just close the Unity Window once the <code>Saved Model</code> message appears. Either wait for the training process to close the window or press <code>Ctrl+C</code> at the command-line prompt. If you close the window manually, the <code>.onnx</code> file containing the trained model is not exported into the ml-agents folder.</p> <p>If you've quit the training early using <code>Ctrl+C</code> and want to resume training, run the same command again, appending the <code>--resume</code> flag:</p> <pre><code>mlagents-learn config/ppo/3DBall.yaml --run-id=first3DBallRun --resume\n</code></pre> <p>Your trained model will be at <code>results/&lt;run-identifier&gt;/&lt;behavior_name&gt;.onnx</code> where <code>&lt;behavior_name&gt;</code> is the name of the <code>Behavior Name</code> of the agents corresponding to the model. This file corresponds to your model's latest checkpoint. You can now embed this trained model into your Agents by following the steps below, which is similar to the steps described above.</p> <ol> <li>Move your model file into    <code>Project/Assets/ML-Agents/Examples/3DBall/TFModels/</code>.</li> <li>Open the Unity Editor, and select the 3DBall scene as described above.</li> <li>Select the 3DBall prefab Agent object.</li> <li>Drag the <code>&lt;behavior_name&gt;.onnx</code> file from the Project window of the Editor to    the Model placeholder in the Ball3DAgent inspector window.</li> <li>Press the Play button at the top of the Editor.</li> </ol>"},{"location":"Getting-Started/#next-steps","title":"Next Steps","text":"<ul> <li>For more information on the ML-Agents Toolkit, in addition to helpful   background, check out the ML-Agents Toolkit Overview   page.</li> <li>For a \"Hello World\" introduction to creating your own Learning Environment,   check out the   Making a New Learning Environment page.</li> <li>For an overview on the more complex example environments that are provided in   this toolkit, check out the   Example Environments page.</li> <li>For more information on the various training options available, check out the   Training ML-Agents page.</li> </ul>"},{"location":"Glossary/","title":"ML-Agents Toolkit Glossary","text":"<ul> <li>Academy - Singleton object which controls timing, reset, and   training/inference settings of the environment.</li> <li>Action - The carrying-out of a decision on the part of an agent within the   environment.</li> <li>Agent - Unity Component which produces observations and takes actions in   the environment. Agents actions are determined by decisions produced by a   Policy.</li> <li>Decision - The specification produced by a Policy for an action to be   carried out given an observation.</li> <li>Editor - The Unity Editor, which may include any pane (e.g. Hierarchy,   Scene, Inspector).</li> <li>Environment - The Unity scene which contains Agents.</li> <li>Experience - Corresponds to a tuple of [Agent observations, actions,   rewards] of a single Agent obtained after a Step.</li> <li>External Coordinator - ML-Agents class responsible for communication with   outside processes (in this case, the Python API).</li> <li>FixedUpdate - Unity method called each time the game engine is stepped.   ML-Agents logic should be placed here.</li> <li>Frame - An instance of rendering the main camera for the display.   Corresponds to each <code>Update</code> call of the game engine.</li> <li>Observation - Partial information describing the state of the environment   available to a given agent. (e.g. Vector, Visual)</li> <li>Policy - The decision making mechanism for producing decisions from   observations, typically a neural network model.</li> <li>Reward - Signal provided at every step used to indicate desirability of an   agent\u2019s action within the current state of the environment.</li> <li>State - The underlying properties of the environment (including all agents   within it) at a given time.</li> <li>Step - Corresponds to an atomic change of the engine that happens between   Agent decisions.</li> <li>Trainer - Python class which is responsible for training a given group of   Agents.</li> <li>Update - Unity function called each time a frame is rendered. ML-Agents   logic should not be placed here.</li> </ul>"},{"location":"Hugging-Face-Integration/","title":"The Hugging Face Integration","text":"<p>The Hugging Face Hub \ud83e\udd17 is a central place where anyone can share and download models.</p> <p>It allows you to: - Host your trained models. - Download trained models from the community. - Visualize your agents playing directly on your browser.</p> <p>You can see the list of ml-agents models here.</p> <p>We wrote a complete tutorial to learn to train your first agent using ML-Agents and publish it to the Hub:</p> <ul> <li>A short tutorial where you teach Huggy the Dog to fetch the stick and then play with him directly in your browser</li> <li>A more in-depth tutorial</li> </ul>"},{"location":"Hugging-Face-Integration/#download-a-model-from-the-hub","title":"Download a model from the Hub","text":"<p>You can simply download a model from the Hub using <code>mlagents-load-from-hf</code>.</p> <p>You need to define two parameters:</p> <ul> <li><code>--repo-id</code>: the name of the Hugging Face repo you want to download.</li> <li><code>--local-dir</code>: the path to download the model.</li> </ul> <p>For instance, I want to load the model with model-id \"ThomasSimonini/MLAgents-Pyramids\" and put it in the downloads directory:</p> <pre><code>mlagents-load-from-hf --repo-id=\"ThomasSimonini/MLAgents-Pyramids\" --local-dir=\"./downloads\"\n</code></pre>"},{"location":"Hugging-Face-Integration/#upload-a-model-to-the-hub","title":"Upload a model to the Hub","text":"<p>You can simply upload a model to the Hub using <code>mlagents-push-to-hf</code></p> <p>You need to define four parameters:</p> <ul> <li><code>--run-id</code>: the name of the training run id.</li> <li><code>--local-dir</code>: where the model was saved</li> <li><code>--repo-id</code>: the name of the Hugging Face repo you want to create or update. It\u2019s always / If the repo does not exist it will be created automatically <li><code>--commit-message</code>: since HF repos are git repositories you need to give a commit message.</li> <p>For instance, I want to upload my model trained with run-id \"SnowballTarget1\" to the repo-id: ThomasSimonini/ppo-SnowballTarget:</p> <pre><code>  mlagents-push-to-hf --run-id=\"SnowballTarget1\" --local-dir=\"./results/SnowballTarget1\" --repo-id=\"ThomasSimonini/ppo-SnowballTarget\" --commit-message=\"First Push\"\n</code></pre>"},{"location":"Hugging-Face-Integration/#visualize-an-agent-playing","title":"Visualize an agent playing","text":"<p>You can watch your agent playing directly in your browser (if the environment is from the ML-Agents official environments)</p> <ul> <li>Step 1: Go to https://huggingface.co/unity and select the environment demo.</li> <li>Step 2: Find your model_id in the list.</li> <li>Step 3: Select your .nn /.onnx file.</li> <li>Step 4: Click on Watch the agent play</li> </ul>"},{"location":"Installation-Anaconda-Windows/","title":"Installing ML-Agents Toolkit for Windows (Deprecated)","text":"<p>:warning: Note: We no longer use this guide ourselves and so it may not work correctly. We've decided to keep it up just in case it is helpful to you.</p> <p>The ML-Agents Toolkit supports Windows 10. While it might be possible to run the ML-Agents Toolkit using other versions of Windows, it has not been tested on other versions. Furthermore, the ML-Agents Toolkit has not been tested on a Windows VM such as Bootcamp or Parallels.</p> <p>To use the ML-Agents Toolkit, you install Python and the required Python packages as outlined below. This guide also covers how set up GPU-based training (for advanced users). GPU-based training is not currently required for the ML-Agents Toolkit. However, training on a GPU might be required by future versions and features.</p>"},{"location":"Installation-Anaconda-Windows/#step-1-install-python-via-anaconda","title":"Step 1: Install Python via Anaconda","text":"<p>Download and install Anaconda for Windows. By using Anaconda, you can manage separate environments for different distributions of Python. Python 3.7.2 or higher is required as we no longer support Python 2. In this guide, we are using Python version 3.7 and Anaconda version 5.1 (64-bit or 32-bit direct links).</p> <p> </p> <p>We recommend the default advanced installation options. However, select the options appropriate for your specific situation.</p> <p> </p> <p>After installation, you must open Anaconda Navigator to finish the setup. From the Windows search bar, type anaconda navigator. You can close Anaconda Navigator after it opens.</p> <p>If environment variables were not created, you will see error \"conda is not recognized as internal or external command\" when you type <code>conda</code> into the command line. To solve this you will need to set the environment variable correctly.</p> <p>Type <code>environment variables</code> in the search bar (this can be reached by hitting the Windows key or the bottom left Windows button). You should see an option called Edit the system environment variables.</p> <p> </p> <p>From here, click the Environment Variables button. Double click \"Path\" under System variable to edit the \"Path\" variable, click New to add the following new paths.</p> <pre><code>%UserProfile%\\Anaconda3\\Scripts\n%UserProfile%\\Anaconda3\\Scripts\\conda.exe\n%UserProfile%\\Anaconda3\n%UserProfile%\\Anaconda3\\python.exe\n</code></pre>"},{"location":"Installation-Anaconda-Windows/#step-2-setup-and-activate-a-new-conda-environment","title":"Step 2: Setup and Activate a New Conda Environment","text":"<p>You will create a new Conda environment to be used with the ML-Agents Toolkit. This means that all the packages that you install are localized to just this environment. It will not affect any other installation of Python or other environments. Whenever you want to run ML-Agents, you will need activate this Conda environment.</p> <p>To create a new Conda environment, open a new Anaconda Prompt (Anaconda Prompt in the search bar) and type in the following command:</p> <pre><code>conda create -n ml-agents python=3.7\n</code></pre> <p>You may be asked to install new packages. Type <code>y</code> and press enter (make sure you are connected to the Internet). You must install these required packages. The new Conda environment is called ml-agents and uses Python version 3.7.</p> <p> </p> <p>To use this environment, you must activate it. (To use this environment In the future, you can run the same command). In the same Anaconda Prompt, type in the following command:</p> <pre><code>activate ml-agents\n</code></pre> <p>You should see <code>(ml-agents)</code> prepended on the last line.</p> <p>Next, install <code>tensorflow</code>. Install this package using <code>pip</code> - which is a package management system used to install Python packages. Latest versions of TensorFlow won't work, so you will need to make sure that you install version 1.7.1. In the same Anaconda Prompt, type in the following command (make sure you are connected to the Internet):</p> <pre><code>pip install tensorflow==1.7.1\n</code></pre>"},{"location":"Installation-Anaconda-Windows/#step-3-install-required-python-packages","title":"Step 3: Install Required Python Packages","text":"<p>The ML-Agents Toolkit depends on a number of Python packages. Use <code>pip</code> to install these Python dependencies.</p> <p>If you haven't already, clone the ML-Agents Toolkit Github repository to your local computer. You can do this using Git (download here) and running the following commands in an Anaconda Prompt (if you open a new prompt, be sure to activate the ml-agents Conda environment by typing <code>activate ml-agents</code>):</p> <pre><code>git clone --branch release_21 https://github.com/Unity-Technologies/ml-agents.git\n</code></pre> <p>The <code>--branch release_21</code> option will switch to the tag of the latest stable release. Omitting that will get the <code>main</code> branch which is potentially unstable.</p> <p>If you don't want to use Git, you can find download links on the releases page.</p> <p>The <code>com.unity.ml-agents</code> subdirectory contains the core code to add to your projects. The <code>Project</code> subdirectory contains many example environments to help you get started.</p> <p>The <code>ml-agents</code> subdirectory contains a Python package which provides deep reinforcement learning trainers to use with Unity environments.</p> <p>The <code>ml-agents-envs</code> subdirectory contains a Python API to interface with Unity, which the <code>ml-agents</code> package depends on.</p> <p>The <code>gym-unity</code> subdirectory contains a package to interface with OpenAI Gym.</p> <p>Keep in mind where the files were downloaded, as you will need the trainer config files in this directory when running <code>mlagents-learn</code>. Make sure you are connected to the Internet and then type in the Anaconda Prompt:</p> <pre><code>python -m pip install mlagents==1.0.0\n</code></pre> <p>This will complete the installation of all the required Python packages to run the ML-Agents Toolkit.</p> <p>Sometimes on Windows, when you use pip to install certain Python packages, the pip will get stuck when trying to read the cache of the package. If you see this, you can try:</p> <pre><code>python -m pip install mlagents==1.0.0 --no-cache-dir\n</code></pre> <p>This <code>--no-cache-dir</code> tells the pip to disable the cache.</p>"},{"location":"Installation-Anaconda-Windows/#installing-for-development","title":"Installing for Development","text":"<p>If you intend to make modifications to <code>ml-agents</code> or <code>ml-agents-envs</code>, you should install the packages from the cloned repo rather than from PyPi. To do this, you will need to install <code>ml-agents</code> and <code>ml-agents-envs</code> separately.</p> <p>In our example, the files are located in <code>C:\\Downloads</code>. After you have either cloned or downloaded the files, from the Anaconda Prompt, change to the ml-agents subdirectory inside the ml-agents directory:</p> <pre><code>cd C:\\Downloads\\ml-agents\n</code></pre> <p>From the repo's main directory, now run:</p> <pre><code>cd ml-agents-envs\npip install -e .\ncd ..\ncd ml-agents\npip install -e .\n</code></pre> <p>Running pip with the <code>-e</code> flag will let you make changes to the Python files directly and have those reflected when you run <code>mlagents-learn</code>. It is important to install these packages in this order as the <code>mlagents</code> package depends on <code>mlagents_envs</code>, and installing it in the other order will download <code>mlagents_envs</code> from PyPi.</p>"},{"location":"Installation-Anaconda-Windows/#optional-step-4-gpu-training-using-the-ml-agents-toolkit","title":"(Optional) Step 4: GPU Training using The ML-Agents Toolkit","text":"<p>GPU is not required for the ML-Agents Toolkit and won't speed up the PPO algorithm a lot during training(but something in the future will benefit from GPU). This is a guide for advanced users who want to train using GPUs. Additionally, you will need to check if your GPU is CUDA compatible. Please check Nvidia's page here.</p> <p>Currently for the ML-Agents Toolkit, only CUDA v9.0 and cuDNN v7.0.5 is supported.</p>"},{"location":"Installation-Anaconda-Windows/#install-nvidia-cuda-toolkit","title":"Install Nvidia CUDA toolkit","text":"<p>Download and install the CUDA toolkit 9.0 from Nvidia's archive. The toolkit includes GPU-accelerated libraries, debugging and optimization tools, a C/C++ (Step Visual Studio 2017) compiler and a runtime library and is needed to run the ML-Agents Toolkit. In this guide, we are using version 9.0.176).</p> <p>Before installing, please make sure you close any running instances of Unity or Visual Studio.</p> <p>Run the installer and select the Express option. Note the directory where you installed the CUDA toolkit. In this guide, we installed in the directory <code>C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v9.0</code></p>"},{"location":"Installation-Anaconda-Windows/#install-nvidia-cudnn-library","title":"Install Nvidia cuDNN library","text":"<p>Download and install the cuDNN library from Nvidia. cuDNN is a GPU-accelerated library of primitives for deep neural networks. Before you can download, you will need to sign up for free to the Nvidia Developer Program.</p> <p> </p> <p>Once you've signed up, go back to the cuDNN downloads page. You may or may not be asked to fill out a short survey. When you get to the list cuDNN releases, make sure you are downloading the right version for the CUDA toolkit you installed in Step 1. In this guide, we are using version 7.0.5 for CUDA toolkit version 9.0 (direct link).</p> <p>After you have downloaded the cuDNN files, you will need to extract the files into the CUDA toolkit directory. In the cuDNN zip file, there are three folders called <code>bin</code>, <code>include</code>, and <code>lib</code>.</p> <p> </p> <p>Copy these three folders into the CUDA toolkit directory. The CUDA toolkit directory is located at <code>C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v9.0</code></p> <p> </p>"},{"location":"Installation-Anaconda-Windows/#set-environment-variables","title":"Set Environment Variables","text":"<p>You will need to add one environment variable and two path variables.</p> <p>To set the environment variable, type <code>environment variables</code> in the search bar (this can be reached by hitting the Windows key or the bottom left Windows button). You should see an option called Edit the system environment variables.</p> <p> </p> <p>From here, click the Environment Variables button. Click New to add a new system variable (make sure you do this under System variables and not User variables.</p> <p> </p> <p>For Variable Name, enter <code>CUDA_HOME</code>. For the variable value, put the directory location for the CUDA toolkit. In this guide, the directory location is <code>C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v9.0</code>. Press OK once.</p> <p> </p> <p>To set the two path variables, inside the same Environment Variables window and under the second box called System Variables, find a variable called <code>Path</code> and click Edit. You will add two directories to the list. For this guide, the two entries would look like:</p> <pre><code>C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v9.0\\lib\\x64\nC:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v9.0\\extras\\CUPTI\\libx64\n</code></pre> <p>Make sure to replace the relevant directory location with the one you have installed. Please note that case sensitivity matters.</p> <p> </p>"},{"location":"Installation-Anaconda-Windows/#install-tensorflow-gpu","title":"Install TensorFlow GPU","text":"<p>Next, install <code>tensorflow-gpu</code> using <code>pip</code>. You'll need version 1.7.1. In an Anaconda Prompt with the Conda environment ml-agents activated, type in the following command to uninstall TensorFlow for cpu and install TensorFlow for gpu (make sure you are connected to the Internet):</p> <pre><code>pip uninstall tensorflow\npip install tensorflow-gpu==1.7.1\n</code></pre> <p>Lastly, you should test to see if everything installed properly and that TensorFlow can identify your GPU. In the same Anaconda Prompt, open Python in the Prompt by calling:</p> <pre><code>python\n</code></pre> <p>And then type the following commands:</p> <pre><code>import tensorflow as tf\n\nsess = tf.Session(config=tf.ConfigProto(log_device_placement=True))\n</code></pre> <p>You should see something similar to:</p> <pre><code>Found device 0 with properties ...\n</code></pre>"},{"location":"Installation-Anaconda-Windows/#acknowledgments","title":"Acknowledgments","text":"<p>We would like to thank Jason Weimann and Nitish S. Mutha for writing the original articles which were used to create this guide.</p>"},{"location":"Installation/","title":"Installation","text":"<p>The ML-Agents Toolkit contains several components:</p> <ul> <li>Unity package (<code>com.unity.ml-agents</code>) contains the   Unity C# SDK that will be integrated into your Unity project.  This package contains   a sample to help you get started with ML-Agents.</li> <li>Unity package   (<code>com.unity.ml-agents.extensions</code>)   contains experimental C#/Unity components that are not yet ready to be part   of the base <code>com.unity.ml-agents</code> package. <code>com.unity.ml-agents.extensions</code>   has a direct dependency on <code>com.unity.ml-agents</code>.</li> <li>Two Python packages:</li> <li><code>mlagents</code> contains the machine learning algorithms that     enables you to train behaviors in your Unity scene. Most users of ML-Agents     will only need to directly install <code>mlagents</code>.</li> <li><code>mlagents_envs</code> contains a set of Python APIs to interact with     a Unity scene. It is a foundational layer that facilitates data messaging     between Unity scene and the Python machine learning algorithms.     Consequently, <code>mlagents</code> depends on <code>mlagents_envs</code>.</li> <li>Unity Project that contains several   example environments that highlight the   various features of the toolkit to help you get started.</li> </ul> <p>Consequently, to install and use the ML-Agents Toolkit you will need to:</p> <ul> <li>Install Unity (2022.3 or later)</li> <li>Install Python (3.10.12 or higher)</li> <li>Clone this repository (Recommended for the latest version and bug fixes)</li> <li>Note: If you do not clone the repository, then you will not be   able to access the example environments and training configurations or the   <code>com.unity.ml-agents.extensions</code> package. Additionally, the   Getting Started Guide assumes that you have cloned the   repository.</li> <li>Install the <code>com.unity.ml-agents</code> Unity package</li> <li>Install the <code>com.unity.ml-agents.extensions</code> Unity package (Optional)</li> <li>Install the <code>mlagents-envs</code></li> <li>Install the <code>mlagents</code> Python package</li> </ul>"},{"location":"Installation/#install-unity-20223-or-later","title":"Install Unity 2022.3 or Later","text":"<p>Download and install Unity. We strongly recommend that you install Unity through the Unity Hub as it will enable you to manage multiple Unity versions.</p>"},{"location":"Installation/#install-python-31012","title":"Install Python 3.10.12","text":"<p>We recommend installing Python 3.10.12. If you are using Windows, please install the x86-64 version and not x86. If your Python environment doesn't include <code>pip3</code>, see these instructions on installing it. We also recommend using conda or mamba to manage your python virtual environments.</p>"},{"location":"Installation/#conda-python-setup","title":"Conda python setup","text":"<p>Once conda has been installed in your system, open a terminal and execute the following commands to setup a python 3.10.12 virtual environment and activate it.</p> <pre><code>conda create -n mlagents python=3.10.12 &amp;&amp; conda activate mlagents\n</code></pre>"},{"location":"Installation/#clone-the-ml-agents-toolkit-repository-recommended","title":"Clone the ML-Agents Toolkit Repository (Recommended)","text":"<p>Now that you have installed Unity and Python, you can now install the Unity and Python packages. You do not need to clone the repository to install those packages, but you may choose to clone the repository if you'd like download our example environments and training configurations to experiment with them (some of our tutorials / guides assume you have access to our example environments).</p> <p>NOTE: There are samples shipped with the Unity Package.  You only need to clone the repository if you would like to explore more examples.</p> <pre><code>git clone --branch release_21 https://github.com/Unity-Technologies/ml-agents.git\n</code></pre> <p>The <code>--branch release_21</code> option will switch to the tag of the latest stable release. Omitting that will get the <code>main</code> branch which is potentially unstable.</p>"},{"location":"Installation/#advanced-local-installation-for-development","title":"Advanced: Local Installation for Development","text":"<p>You will need to clone the repository if you plan to modify or extend the ML-Agents Toolkit for your purposes. If you plan to contribute those changes back, make sure to clone the <code>develop</code> branch (by omitting <code>--branch release_21</code> from the command above). See our Contributions Guidelines for more information on contributing to the ML-Agents Toolkit.</p>"},{"location":"Installation/#install-the-comunityml-agents-unity-package","title":"Install the <code>com.unity.ml-agents</code> Unity package","text":"<p>The Unity ML-Agents C# SDK is a Unity Package. You can install the <code>com.unity.ml-agents</code> package directly from the Package Manager registry. Please make sure you enable 'Preview Packages' in the 'Advanced' dropdown in order to find the latest Preview release of the package.</p> <p>NOTE: If you do not see the ML-Agents package listed in the Package Manager please follow the advanced installation instructions below.</p>"},{"location":"Installation/#advanced-local-installation-for-development_1","title":"Advanced: Local Installation for Development","text":"<p>You can add the local <code>com.unity.ml-agents</code> package (from the repository that you just cloned) to your project by:</p> <ol> <li>navigating to the menu <code>Window</code> -&gt; <code>Package Manager</code>.</li> <li>In the package manager window click on the <code>+</code> button on the top left of the packages list).</li> <li>Select <code>Add package from disk...</code></li> <li>Navigate into the <code>com.unity.ml-agents</code> folder.</li> <li>Select the <code>package.json</code> file.</li> </ol> <p> </p> <p>If you are going to follow the examples from our documentation, you can open the <code>Project</code> folder in Unity and start tinkering immediately.</p>"},{"location":"Installation/#install-the-comunityml-agentsextensions-unity-package-optional","title":"Install the <code>com.unity.ml-agents.extensions</code> Unity package (Optional)","text":"<p>To install the <code>com.unity.ml-agents.extensions</code> package, you need to first clone the repo and then complete a local installation similar to what was outlined in the previous Advanced: Local Installation for Development section. Complete installation steps can be found in the package documentation.</p>"},{"location":"Installation/#install-the-mlagents-python-package","title":"Install the <code>mlagents</code> Python package","text":"<p>Installing the <code>mlagents</code> Python package involves installing other Python packages that <code>mlagents</code> depends on. So you may run into installation issues if your machine has older versions of any of those dependencies already installed. Consequently, our supported path for installing <code>mlagents</code> is to leverage Python Virtual Environments. Virtual Environments provide a mechanism for isolating the dependencies for each project and are supported on Mac / Windows / Linux. We offer a dedicated guide on Virtual Environments.</p>"},{"location":"Installation/#windows-installing-pytorch","title":"(Windows) Installing PyTorch","text":"<p>On Windows, you'll have to install the PyTorch package separately prior to installing ML-Agents. Activate your virtual environment and run from the command line:</p> <pre><code>pip3 install torch~=1.13.1 -f https://download.pytorch.org/whl/torch_stable.html\n</code></pre> <p>Note that on Windows, you may also need Microsoft's Visual C++ Redistributable if you don't have it already. See the PyTorch installation guide for more installation options and versions.</p>"},{"location":"Installation/#installing-mlagents","title":"Installing <code>mlagents</code>","text":"<p>To install the <code>mlagents</code> Python package, activate your virtual environment and run from the command line:</p> <pre><code>cd /path/to/ml-agents\npython -m pip install ./ml-agents-envs\npython -m pip install ./ml-agents\n</code></pre> <p>Note that this will install <code>mlagents</code> from the cloned repository, not from the PyPi repository. If you installed this correctly, you should be able to run <code>mlagents-learn --help</code>, after which you will see the command line parameters you can use with <code>mlagents-learn</code>.</p> <p>NOTE: Since ML-Agents development has slowed, PyPi releases will be less frequent. However, you can install from PyPi by executing the following command:</p> <pre><code>python -m pip install mlagents==1.0.0\n</code></pre> <p>which will install the latest version of ML-Agents and associated dependencies available on PyPi. Note, you need to have the matching version of the Unity packages with the particular release of the python packages. You can find the release history here</p> <p>By installing the <code>mlagents</code> package, the dependencies listed in the setup.py file are also installed. These include PyTorch.</p>"},{"location":"Installation/#advanced-local-installation-for-development_2","title":"Advanced: Local Installation for Development","text":"<p>If you intend to make modifications to <code>mlagents</code> or <code>mlagents_envs</code>, you should install the packages from the cloned repository rather than from PyPi. To do this, you will need to install <code>mlagents</code> and <code>mlagents_envs</code> separately. From the repository's root directory, run:</p> <pre><code>pip3 install torch -f https://download.pytorch.org/whl/torch_stable.html\npip3 install -e ./ml-agents-envs\npip3 install -e ./ml-agents\n</code></pre> <p>Running pip with the <code>-e</code> flag will let you make changes to the Python files directly and have those reflected when you run <code>mlagents-learn</code>. It is important to install these packages in this order as the <code>mlagents</code> package depends on <code>mlagents_envs</code>, and installing it in the other order will download <code>mlagents_envs</code> from PyPi.</p>"},{"location":"Installation/#next-steps","title":"Next Steps","text":"<p>The Getting Started guide contains several short tutorials on setting up the ML-Agents Toolkit within Unity, running a pre-trained model, in addition to building and training environments.</p>"},{"location":"Installation/#help","title":"Help","text":"<p>If you run into any problems regarding ML-Agents, refer to our FAQ and our Limitations pages. If you can't find anything please submit an issue and make sure to cite relevant information on OS, Python version, and exact error message (whenever possible).</p>"},{"location":"Integrations-Match3/","title":"Match-3 with ML-Agents","text":""},{"location":"Integrations-Match3/#getting-started","title":"Getting started","text":"<p>The C# code for Match-3 exists inside of the Unity package (<code>com.unity.ml-agents</code>). The good first step would be to take a look at how we have implemented the C# code in the example Match-3 scene (located under /Project/Assets/ML-Agents/Examples/match3). Once you have some familiarity, then the next step would be to implement the C# code for Match-3 from the extensions package.</p> <p>Additionally, see below for additional technical specifications on the C# code for Match-3. Please note the Match-3 game isn't human playable as implemented and can be only played via training.</p>"},{"location":"Integrations-Match3/#technical-specifications-for-match-3-with-ml-agents","title":"Technical specifications for Match-3 with ML-Agents","text":""},{"location":"Integrations-Match3/#abstractboard-class","title":"AbstractBoard class","text":"<p>The <code>AbstractBoard</code> is the bridge between ML-Agents and your game. It allows ML-Agents to * ask your game what the current and maximum sizes (rows, columns, and potential piece types) of the board are * ask your game what the \"color\" of a cell is * ask whether the cell is a \"special\" piece type or not * ask your game whether a move is allowed * request that your game make a move</p> <p>These are handled by implementing the abstract methods of <code>AbstractBoard</code>.</p>"},{"location":"Integrations-Match3/#public-abstract-boardsize-getmaxboardsize","title":"<code>public abstract BoardSize GetMaxBoardSize()</code>","text":"<p>Returns the largest <code>BoardSize</code> that the game can use. This is used to determine the sizes of observations and sensors, so don't make it larger than necessary.</p>"},{"location":"Integrations-Match3/#public-virtual-boardsize-getcurrentboardsize","title":"<code>public virtual BoardSize GetCurrentBoardSize()</code>","text":"<p>Returns the current size of the board. Each field on this BoardSize must be less than or equal to the corresponding field returned by <code>GetMaxBoardSize()</code>. This method is optional; if your always use the same size board, you don't need to override it.</p> <p>If the current board size is smaller than the maximum board size, <code>GetCellType()</code> and <code>GetSpecialType()</code> will not be called for cells outside the current board size, and <code>IsValidMove</code> won't be called for moves that would go outside of the current board size.</p>"},{"location":"Integrations-Match3/#public-abstract-int-getcelltypeint-row-int-col","title":"<code>public abstract int GetCellType(int row, int col)</code>","text":"<p>Returns the \"color\" of piece at the given row and column. This should be between 0 and BoardSize.NumCellTypes-1 (inclusive). The actual order of the values doesn't matter.</p>"},{"location":"Integrations-Match3/#public-abstract-int-getspecialtypeint-row-int-col","title":"<code>public abstract int GetSpecialType(int row, int col)</code>","text":"<p>Returns the special type of the piece at the given row and column. This should be between 0 and BoardSize.NumSpecialTypes (inclusive). The actual order of the values doesn't matter.</p>"},{"location":"Integrations-Match3/#public-abstract-bool-ismovevalidmove-m","title":"<code>public abstract bool IsMoveValid(Move m)</code>","text":"<p>Check whether the particular <code>Move</code> is valid for the game. The actual results will depend on the rules of the game, but we provide the <code>SimpleIsMoveValid()</code> method that handles basic match3 rules with no special or immovable pieces.</p>"},{"location":"Integrations-Match3/#public-abstract-bool-makemovemove-m","title":"<code>public abstract bool MakeMove(Move m)</code>","text":"<p>Instruct the game to make the given move. Returns true if the move was made. Note that during training, a move that was marked as invalid may occasionally still be requested. If this happens, it is safe to do nothing and request another move.</p>"},{"location":"Integrations-Match3/#move-struct","title":"<code>Move</code> struct","text":"<p>The Move struct encapsulates a swap of two adjacent cells. You can get the number of potential moves for a board of a given size with. <code>Move.NumPotentialMoves(maxBoardSize)</code>. There are two helper functions to create a new <code>Move</code>: * <code>public static Move FromMoveIndex(int moveIndex, BoardSize maxBoardSize)</code> can be used to iterate over all potential moves for the board by looping from 0 to <code>Move.NumPotentialMoves()</code> * <code>public static Move FromPositionAndDirection(int row, int col, Direction dir, BoardSize maxBoardSize)</code> creates a <code>Move</code> from a row, column, and direction (and board size).</p>"},{"location":"Integrations-Match3/#boardsize-struct","title":"<code>BoardSize</code> struct","text":"<p>Describes the \"size\" of the board, including the number of potential piece types that the board can have. This is returned by the AbstractBoard.GetMaxBoardSize() and GetCurrentBoardSize() methods.</p>"},{"location":"Integrations-Match3/#match3sensor-and-match3sensorcomponent-classes","title":"<code>Match3Sensor</code> and <code>Match3SensorComponent</code> classes","text":"<p>The <code>Match3Sensor</code> generates observations about the state using the <code>AbstractBoard</code> interface. You can choose whether to use vector or \"visual\" observations; in theory, visual observations should perform better because they are 2-dimensional like the board, but we need to experiment more on this.</p> <p>A <code>Match3SensorComponent</code> generates <code>Match3Sensor</code>s (the exact number of sensors depends on your configuration) at runtime, and should be added to the same GameObject as your <code>Agent</code> implementation. You do not need to write any additional code to use them.</p>"},{"location":"Integrations-Match3/#match3actuator-and-match3actuatorcomponent-classes","title":"<code>Match3Actuator</code> and <code>Match3ActuatorComponent</code> classes","text":"<p>The <code>Match3Actuator</code> converts actions from training or inference into a <code>Move</code> that is sent to<code>AbstractBoard.MakeMove()</code> It also checks <code>AbstractBoard.IsMoveValid</code> for each potential move and uses this to set the action mask for Agent.</p> <p>A <code>Match3ActuatorComponent</code> generates a <code>Match3Actuator</code> at runtime, and should be added to the same GameObject as your <code>Agent</code> implementation.  You do not need to write any additional code to use them.</p>"},{"location":"Integrations-Match3/#setting-up-match-3-simulation","title":"Setting up Match-3 simulation","text":"<ul> <li>Implement the <code>AbstractBoard</code> methods to integrate with your game.</li> <li>Give the <code>Agent</code> rewards when it does what you want it to (match multiple pieces in a row, clears pieces of a certain type, etc).</li> <li>Add the <code>Agent</code>, <code>AbstractBoard</code> implementation, <code>Match3SensorComponent</code>, and <code>Match3ActuatorComponent</code> to the same <code>GameObject</code>.</li> <li>Call <code>Agent.RequestDecision()</code> when you're ready for the <code>Agent</code> to make a move on the next <code>Academy</code> step. During the next <code>Academy</code> step, the <code>MakeMove()</code> method on the board will be called.</li> </ul>"},{"location":"Integrations-Match3/#implementation-details","title":"Implementation Details","text":""},{"location":"Integrations-Match3/#action-space","title":"Action Space","text":"<p>The indexing for actions is the same as described in Human Like Playtesting with Deep Learning (for example, Figure 2b). The horizontal moves are enumerated first, then the vertical ones. </p>"},{"location":"Integrations/","title":"Game Integrations","text":"<p>ML-Agents provides some utilities to make it easier to integrate with some common genres of games.</p>"},{"location":"Integrations/#match-3","title":"Match-3","text":"<p>The Match-3 integration provides an abstraction of a match-3 game board and moves, along with a sensor to observe the game state, and an actuator to translate the ML-Agent actions into game moves.</p>"},{"location":"Integrations/#interested-in-more-game-templates","title":"Interested in more game templates?","text":"<p>Do you have a type of game you are interested for ML-Agents? If so, please post a forum issue with <code>[GAME TEMPLATE]</code> in the title.</p>"},{"location":"LICENSE/","title":"LICENSE","text":"<p>Copyright 2017-2021 Unity Technologies</p> <p>Licensed under the Apache License, Version 2.0 (the \"License\"); you may not use this file except in compliance with the License. You may obtain a copy of the License at</p> <pre><code>http://www.apache.org/licenses/LICENSE-2.0\n</code></pre> <p>Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.</p> <pre><code>                             Apache License\n                       Version 2.0, January 2004\n                    http://www.apache.org/licenses/\n</code></pre> <p>TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION</p> <ol> <li> <p>Definitions.</p> <p>\"License\" shall mean the terms and conditions for use, reproduction,   and distribution as defined by Sections 1 through 9 of this document.</p> <p>\"Licensor\" shall mean the copyright owner or entity authorized by   the copyright owner that is granting the License.</p> <p>\"Legal Entity\" shall mean the union of the acting entity and all   other entities that control, are controlled by, or are under common   control with that entity. For the purposes of this definition,   \"control\" means (i) the power, direct or indirect, to cause the   direction or management of such entity, whether by contract or   otherwise, or (ii) ownership of fifty percent (50%) or more of the   outstanding shares, or (iii) beneficial ownership of such entity.</p> <p>\"You\" (or \"Your\") shall mean an individual or Legal Entity   exercising permissions granted by this License.</p> <p>\"Source\" form shall mean the preferred form for making modifications,   including but not limited to software source code, documentation   source, and configuration files.</p> <p>\"Object\" form shall mean any form resulting from mechanical   transformation or translation of a Source form, including but   not limited to compiled object code, generated documentation,   and conversions to other media types.</p> <p>\"Work\" shall mean the work of authorship, whether in Source or   Object form, made available under the License, as indicated by a   copyright notice that is included in or attached to the work   (an example is provided in the Appendix below).</p> <p>\"Derivative Works\" shall mean any work, whether in Source or Object   form, that is based on (or derived from) the Work and for which the   editorial revisions, annotations, elaborations, or other modifications   represent, as a whole, an original work of authorship. For the purposes   of this License, Derivative Works shall not include works that remain   separable from, or merely link (or bind by name) to the interfaces of,   the Work and Derivative Works thereof.</p> <p>\"Contribution\" shall mean any work of authorship, including   the original version of the Work and any modifications or additions   to that Work or Derivative Works thereof, that is intentionally   submitted to Licensor for inclusion in the Work by the copyright owner   or by an individual or Legal Entity authorized to submit on behalf of   the copyright owner. For the purposes of this definition, \"submitted\"   means any form of electronic, verbal, or written communication sent   to the Licensor or its representatives, including but not limited to   communication on electronic mailing lists, source code control systems,   and issue tracking systems that are managed by, or on behalf of, the   Licensor for the purpose of discussing and improving the Work, but   excluding communication that is conspicuously marked or otherwise   designated in writing by the copyright owner as \"Not a Contribution.\"</p> <p>\"Contributor\" shall mean Licensor and any individual or Legal Entity   on behalf of whom a Contribution has been received by Licensor and   subsequently incorporated within the Work.</p> </li> <li> <p>Grant of Copyright License. Subject to the terms and conditions of       this License, each Contributor hereby grants to You a perpetual,       worldwide, non-exclusive, no-charge, royalty-free, irrevocable       copyright license to reproduce, prepare Derivative Works of,       publicly display, publicly perform, sublicense, and distribute the       Work and such Derivative Works in Source or Object form.</p> </li> <li> <p>Grant of Patent License. Subject to the terms and conditions of       this License, each Contributor hereby grants to You a perpetual,       worldwide, non-exclusive, no-charge, royalty-free, irrevocable       (except as stated in this section) patent license to make, have made,       use, offer to sell, sell, import, and otherwise transfer the Work,       where such license applies only to those patent claims licensable       by such Contributor that are necessarily infringed by their       Contribution(s) alone or by combination of their Contribution(s)       with the Work to which such Contribution(s) was submitted. If You       institute patent litigation against any entity (including a       cross-claim or counterclaim in a lawsuit) alleging that the Work       or a Contribution incorporated within the Work constitutes direct       or contributory patent infringement, then any patent licenses       granted to You under this License for that Work shall terminate       as of the date such litigation is filed.</p> </li> <li> <p>Redistribution. You may reproduce and distribute copies of the       Work or Derivative Works thereof in any medium, with or without       modifications, and in Source or Object form, provided that You       meet the following conditions:</p> <p>(a) You must give any other recipients of the Work or       Derivative Works a copy of this License; and</p> <p>(b) You must cause any modified files to carry prominent notices       stating that You changed the files; and</p> <p>(c) You must retain, in the Source form of any Derivative Works       that You distribute, all copyright, patent, trademark, and       attribution notices from the Source form of the Work,       excluding those notices that do not pertain to any part of       the Derivative Works; and</p> <p>(d) If the Work includes a \"NOTICE\" text file as part of its       distribution, then any Derivative Works that You distribute must       include a readable copy of the attribution notices contained       within such NOTICE file, excluding those notices that do not       pertain to any part of the Derivative Works, in at least one       of the following places: within a NOTICE text file distributed       as part of the Derivative Works; within the Source form or       documentation, if provided along with the Derivative Works; or,       within a display generated by the Derivative Works, if and       wherever such third-party notices normally appear. The contents       of the NOTICE file are for informational purposes only and       do not modify the License. You may add Your own attribution       notices within Derivative Works that You distribute, alongside       or as an addendum to the NOTICE text from the Work, provided       that such additional attribution notices cannot be construed       as modifying the License.</p> <p>You may add Your own copyright statement to Your modifications and   may provide additional or different license terms and conditions   for use, reproduction, or distribution of Your modifications, or   for any such Derivative Works as a whole, provided Your use,   reproduction, and distribution of the Work otherwise complies with   the conditions stated in this License.</p> </li> <li> <p>Submission of Contributions. Unless You explicitly state otherwise,       any Contribution intentionally submitted for inclusion in the Work       by You to the Licensor shall be under the terms and conditions of       this License, without any additional terms or conditions.       Notwithstanding the above, nothing herein shall supersede or modify       the terms of any separate license agreement you may have executed       with Licensor regarding such Contributions.</p> </li> <li> <p>Trademarks. This License does not grant permission to use the trade       names, trademarks, service marks, or product names of the Licensor,       except as required for reasonable and customary use in describing the       origin of the Work and reproducing the content of the NOTICE file.</p> </li> <li> <p>Disclaimer of Warranty. Unless required by applicable law or       agreed to in writing, Licensor provides the Work (and each       Contributor provides its Contributions) on an \"AS IS\" BASIS,       WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or       implied, including, without limitation, any warranties or conditions       of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A       PARTICULAR PURPOSE. You are solely responsible for determining the       appropriateness of using or redistributing the Work and assume any       risks associated with Your exercise of permissions under this License.</p> </li> <li> <p>Limitation of Liability. In no event and under no legal theory,       whether in tort (including negligence), contract, or otherwise,       unless required by applicable law (such as deliberate and grossly       negligent acts) or agreed to in writing, shall any Contributor be       liable to You for damages, including any direct, indirect, special,       incidental, or consequential damages of any character arising as a       result of this License or out of the use or inability to use the       Work (including but not limited to damages for loss of goodwill,       work stoppage, computer failure or malfunction, or any and all       other commercial damages or losses), even if such Contributor       has been advised of the possibility of such damages.</p> </li> <li> <p>Accepting Warranty or Additional Liability. While redistributing       the Work or Derivative Works thereof, You may choose to offer,       and charge a fee for, acceptance of support, warranty, indemnity,       or other liability obligations and/or rights consistent with this       License. However, in accepting such obligations, You may act only       on Your own behalf and on Your sole responsibility, not on behalf       of any other Contributor, and only if You agree to indemnify,       defend, and hold each Contributor harmless for any liability       incurred by, or claims asserted against, such Contributor by reason       of your accepting any such warranty or additional liability.</p> </li> </ol> <p>END OF TERMS AND CONDITIONS</p> <p>APPENDIX: How to apply the Apache License to your work.</p> <pre><code>  To apply the Apache License to your work, attach the following\n  boilerplate notice, with the fields enclosed by brackets \"{}\"\n  replaced with your own identifying information. (Don't include\n  the brackets!)  The text should be enclosed in the appropriate\n  comment syntax for the file format. We also recommend that a\n  file or class name and description of purpose be included on the\n  same \"printed page\" as the copyright notice for easier\n  identification within third-party archives.\n</code></pre>"},{"location":"Learning-Environment-Create-New/","title":"Making a New Learning Environment","text":"<p>This tutorial walks through the process of creating a Unity Environment from scratch. We recommend first reading the Getting Started guide to understand the concepts presented here first in an already-built environment.</p> <p></p> <p>In this example, we will create an agent capable of controlling a ball on a platform. We will then train the agent to roll the ball toward the cube while avoiding falling off the platform.</p>"},{"location":"Learning-Environment-Create-New/#overview","title":"Overview","text":"<p>Using the ML-Agents Toolkit in a Unity project involves the following basic steps:</p> <ol> <li>Create an environment for your agents to live in. An environment can range    from a simple physical simulation containing a few objects to an entire game    or ecosystem.</li> <li>Implement your Agent subclasses. An Agent subclass defines the code an Agent    uses to observe its environment, to carry out assigned actions, and to    calculate the rewards used for reinforcement training. You can also implement    optional methods to reset the Agent when it has finished or failed its task.</li> <li>Add your Agent subclasses to appropriate GameObjects, typically, the object    in the scene that represents the Agent in the simulation.</li> </ol> <p>Note: If you are unfamiliar with Unity, refer to the Unity manual if an Editor task isn't explained sufficiently in this tutorial.</p> <p>If you haven't already, follow the installation instructions.</p>"},{"location":"Learning-Environment-Create-New/#set-up-the-unity-project","title":"Set Up the Unity Project","text":"<p>The first task to accomplish is simply creating a new Unity project and importing the ML-Agents assets into it:</p> <ol> <li>Launch Unity Hub and create a new 3D project named \"RollerBall\".</li> <li>Add the ML-Agents Unity package    to your project.</li> </ol> <p>Your Unity Project window should contain the following assets:</p> <p></p>"},{"location":"Learning-Environment-Create-New/#create-the-environment","title":"Create the Environment","text":"<p>Next, we will create a very simple scene to act as our learning environment. The \"physical\" components of the environment include a Plane to act as the floor for the Agent to move around on, a Cube to act as the goal or target for the agent to seek, and a Sphere to represent the Agent itself.</p>"},{"location":"Learning-Environment-Create-New/#create-the-floor-plane","title":"Create the Floor Plane","text":"<ol> <li>Right click in Hierarchy window, select 3D Object &gt; Plane.</li> <li>Name the GameObject \"Floor\".</li> <li>Select the Floor Plane to view its properties in the Inspector window.</li> <li>Set Transform to Position = <code>(0, 0, 0)</code>, Rotation = <code>(0, 0, 0)</code>, Scale =    <code>(1, 1, 1)</code>.</li> </ol>"},{"location":"Learning-Environment-Create-New/#add-the-target-cube","title":"Add the Target Cube","text":"<ol> <li>Right click in Hierarchy window, select 3D Object &gt; Cube.</li> <li>Name the GameObject \"Target\".</li> <li>Select the Target Cube to view its properties in the Inspector window.</li> <li>Set Transform to Position = <code>(3, 0.5, 3)</code>, Rotation = <code>(0, 0, 0)</code>, Scale =    <code>(1, 1, 1)</code>.</li> </ol>"},{"location":"Learning-Environment-Create-New/#add-the-agent-sphere","title":"Add the Agent Sphere","text":"<ol> <li>Right click in Hierarchy window, select 3D Object &gt; Sphere.</li> <li>Name the GameObject \"RollerAgent\".</li> <li>Select the RollerAgent Sphere to view its properties in the Inspector window.</li> <li>Set Transform to Position = <code>(0, 0.5, 0)</code>, Rotation = <code>(0, 0, 0)</code>, Scale =    <code>(1, 1, 1)</code>.</li> <li>Click Add Component.</li> <li>Add the <code>Rigidbody</code> component to the Sphere.</li> </ol>"},{"location":"Learning-Environment-Create-New/#group-into-training-area","title":"Group into Training Area","text":"<p>Group the floor, target and agent under a single, empty, GameObject. This will simplify some of our subsequent steps.</p> <p>To do so:</p> <ol> <li>Right-click on your Project Hierarchy and create a new empty GameObject. Name    it TrainingArea.</li> <li>Reset the TrainingArea\u2019s Transform so that it is at <code>(0,0,0)</code> with Rotation    <code>(0,0,0)</code> and Scale <code>(1,1,1)</code>.</li> <li>Drag the Floor, Target, and RollerAgent GameObjects in the Hierarchy into the    TrainingArea GameObject.</li> </ol> <p></p>"},{"location":"Learning-Environment-Create-New/#implement-an-agent","title":"Implement an Agent","text":"<p>To create the Agent Script:</p> <ol> <li>Select the RollerAgent GameObject to view it in the Inspector window.</li> <li>Click Add Component.</li> <li>Click New Script in the list of components (at the bottom).</li> <li>Name the script \"RollerAgent\".</li> <li>Click Create and Add.</li> </ol> <p>Then, edit the new <code>RollerAgent</code> script:</p> <ol> <li>In the Unity Project window, double-click the <code>RollerAgent</code> script to open it    in your code editor.</li> <li>Import ML-Agent package by adding</li> </ol> <p><code>csharp    using Unity.MLAgents;    using Unity.MLAgents.Sensors;    using Unity.MLAgents.Actuators;</code>    then change the base class from <code>MonoBehaviour</code> to <code>Agent</code>. 1. Delete <code>Update()</code> since we are not using it, but keep <code>Start()</code>.</p> <p>So far, these are the basic steps that you would use to add ML-Agents to any Unity project. Next, we will add the logic that will let our Agent learn to roll to the cube using reinforcement learning. More specifically, we will need to extend three methods from the <code>Agent</code> base class:</p> <ul> <li><code>OnEpisodeBegin()</code></li> <li><code>CollectObservations(VectorSensor sensor)</code></li> <li><code>OnActionReceived(ActionBuffers actionBuffers)</code></li> </ul> <p>We overview each of these in more detail in the dedicated subsections below.</p>"},{"location":"Learning-Environment-Create-New/#initialization-and-resetting-the-agent","title":"Initialization and Resetting the Agent","text":"<p>The process of training in the ML-Agents Toolkit involves running episodes where the Agent (Sphere) attempts to solve the task. Each episode lasts until the Agents solves the task (i.e. reaches the cube), fails (rolls off the platform) or times out (takes too long to solve or fail at the task). At the start of each episode, <code>OnEpisodeBegin()</code> is called to set-up the environment for a new episode. Typically the scene is initialized in a random manner to enable the agent to learn to solve the task under a variety of conditions.</p> <p>In this example, each time the Agent (Sphere) reaches its target (Cube), the episode ends and the target (Cube) is moved to a new random location; and if the Agent rolls off the platform, it will be put back onto the floor. These are all handled in <code>OnEpisodeBegin()</code>.</p> <p>To move the target (Cube), we need a reference to its Transform (which stores a GameObject's position, orientation and scale in the 3D world). To get this reference, add a public field of type <code>Transform</code> to the RollerAgent class. Public fields of a component in Unity get displayed in the Inspector window, allowing you to choose which GameObject to use as the target in the Unity Editor.</p> <p>To reset the Agent's velocity (and later to apply force to move the agent) we need a reference to the Rigidbody component. A Rigidbody is Unity's primary element for physics simulation. (See Physics for full documentation of Unity physics.) Since the Rigidbody component is on the same GameObject as our Agent script, the best way to get this reference is using <code>GameObject.GetComponent&lt;T&gt;()</code>, which we can call in our script's <code>Start()</code> method.</p> <p>So far, our RollerAgent script looks like:</p> <pre><code>using System.Collections.Generic;\nusing UnityEngine;\nusing Unity.MLAgents;\nusing Unity.MLAgents.Sensors;\n\npublic class RollerAgent : Agent\n{\n    Rigidbody rBody;\n    void Start () {\n        rBody = GetComponent&lt;Rigidbody&gt;();\n    }\n\n    public Transform Target;\n    public override void OnEpisodeBegin()\n    {\n       // If the Agent fell, zero its momentum\n        if (this.transform.localPosition.y &lt; 0)\n        {\n            this.rBody.angularVelocity = Vector3.zero;\n            this.rBody.velocity = Vector3.zero;\n            this.transform.localPosition = new Vector3( 0, 0.5f, 0);\n        }\n\n        // Move the target to a new spot\n        Target.localPosition = new Vector3(Random.value * 8 - 4,\n                                           0.5f,\n                                           Random.value * 8 - 4);\n    }\n}\n</code></pre> <p>Next, let's implement the <code>Agent.CollectObservations(VectorSensor sensor)</code> method.</p>"},{"location":"Learning-Environment-Create-New/#observing-the-environment","title":"Observing the Environment","text":"<p>The Agent sends the information we collect to the Brain, which uses it to make a decision. When you train the Agent (or use a trained model), the data is fed into a neural network as a feature vector. For an Agent to successfully learn a task, we need to provide the correct information. A good rule of thumb for deciding what information to collect is to consider what you would need to calculate an analytical solution to the problem.</p> <p>In our case, the information our Agent collects includes the position of the target, the position of the agent itself, and the velocity of the agent. This helps the Agent learn to control its speed so it doesn't overshoot the target and roll off the platform. In total, the agent observation contains 8 values as implemented below:</p> <pre><code>public override void CollectObservations(VectorSensor sensor)\n{\n    // Target and Agent positions\n    sensor.AddObservation(Target.localPosition);\n    sensor.AddObservation(this.transform.localPosition);\n\n    // Agent velocity\n    sensor.AddObservation(rBody.velocity.x);\n    sensor.AddObservation(rBody.velocity.z);\n}\n</code></pre>"},{"location":"Learning-Environment-Create-New/#taking-actions-and-assigning-rewards","title":"Taking Actions and Assigning Rewards","text":"<p>The final part of the Agent code is the <code>Agent.OnActionReceived()</code> method, which receives actions and assigns the reward.</p>"},{"location":"Learning-Environment-Create-New/#actions","title":"Actions","text":"<p>To solve the task of moving towards the target, the Agent (Sphere) needs to be able to move in the <code>x</code> and <code>z</code> directions. As such, the agent needs 2 actions: the first determines the force applied along the x-axis; and the second determines the force applied along the z-axis. (If we allowed the Agent to move in three dimensions, then we would need a third action.)</p> <p>The RollerAgent applies the values from the <code>action[]</code> array to its Rigidbody component <code>rBody</code>, using <code>Rigidbody.AddForce()</code>:</p> <pre><code>Vector3 controlSignal = Vector3.zero;\ncontrolSignal.x = action[0];\ncontrolSignal.z = action[1];\nrBody.AddForce(controlSignal * forceMultiplier);\n</code></pre>"},{"location":"Learning-Environment-Create-New/#rewards","title":"Rewards","text":"<p>Reinforcement learning requires rewards to signal which decisions are good and which are bad. The learning algorithm uses the rewards to determine whether it is giving the Agent the optimal actions. You want to reward an Agent for completing the assigned task. In this case, the Agent is given a reward of 1.0 for reaching the Target cube.</p> <p>Rewards are assigned in <code>OnActionReceived()</code>. The RollerAgent calculates the distance to detect when it reaches the target. When it does, the code calls <code>Agent.SetReward()</code> to assign a reward of 1.0 and marks the agent as finished by calling <code>EndEpisode()</code> on the Agent.</p> <pre><code>float distanceToTarget = Vector3.Distance(this.transform.localPosition, Target.localPosition);\n// Reached target\nif (distanceToTarget &lt; 1.42f)\n{\n    SetReward(1.0f);\n    EndEpisode();\n}\n</code></pre> <p>Finally, if the Agent falls off the platform, end the episode so that it can reset itself:</p> <pre><code>// Fell off platform\nif (this.transform.localPosition.y &lt; 0)\n{\n    EndEpisode();\n}\n</code></pre>"},{"location":"Learning-Environment-Create-New/#onactionreceived","title":"OnActionReceived()","text":"<p>With the action and reward logic outlined above, the final version of <code>OnActionReceived()</code> looks like:</p> <pre><code>public float forceMultiplier = 10;\npublic override void OnActionReceived(ActionBuffers actionBuffers)\n{\n    // Actions, size = 2\n    Vector3 controlSignal = Vector3.zero;\n    controlSignal.x = actionBuffers.ContinuousActions[0];\n    controlSignal.z = actionBuffers.ContinuousActions[1];\n    rBody.AddForce(controlSignal * forceMultiplier);\n\n    // Rewards\n    float distanceToTarget = Vector3.Distance(this.transform.localPosition, Target.localPosition);\n\n    // Reached target\n    if (distanceToTarget &lt; 1.42f)\n    {\n        SetReward(1.0f);\n        EndEpisode();\n    }\n\n    // Fell off platform\n    else if (this.transform.localPosition.y &lt; 0)\n    {\n        EndEpisode();\n    }\n}\n</code></pre> <p>Note the <code>forceMultiplier</code> class variable is defined before the method definition. Since <code>forceMultiplier</code> is public, you can set the value from the Inspector window.</p>"},{"location":"Learning-Environment-Create-New/#final-agent-setup-in-editor","title":"Final Agent Setup in Editor","text":"<p>Now that all the GameObjects and ML-Agent components are in place, it is time to connect everything together in the Unity Editor. This involves adding and setting some of the Agent Component's properties so that they are compatible with our Agent script.</p> <ol> <li>Select the RollerAgent GameObject to show its properties in the Inspector    window.</li> <li>Drag the Target GameObject in the Hierarchy into the <code>Target</code> field in RollerAgent Script.</li> <li>Add a <code>Decision Requester</code> script with the Add Component button.    Set the Decision Period to <code>10</code>. For more information on decisions,    see the Agent documentation</li> <li>Add a <code>Behavior Parameters</code> script with the Add Component button.    Set the Behavior Parameters of the Agent to the following:</li> <li><code>Behavior Name</code>: RollerBall</li> <li><code>Vector Observation</code> &gt; <code>Space Size</code> = 8</li> <li><code>Actions</code> &gt; <code>Continuous Actions</code> = 2</li> </ol> <p>In the inspector, the <code>RollerAgent</code> should look like this now:</p> <p></p> <p>Now you are ready to test the environment before training.</p>"},{"location":"Learning-Environment-Create-New/#testing-the-environment","title":"Testing the Environment","text":"<p>It is always a good idea to first test your environment by controlling the Agent using the keyboard. To do so, you will need to extend the <code>Heuristic()</code> method in the <code>RollerAgent</code> class. For our example, the heuristic will generate an action corresponding to the values of the \"Horizontal\" and \"Vertical\" input axis (which correspond to the keyboard arrow keys):</p> <pre><code>public override void Heuristic(in ActionBuffers actionsOut)\n{\n    var continuousActionsOut = actionsOut.ContinuousActions;\n    continuousActionsOut[0] = Input.GetAxis(\"Horizontal\");\n    continuousActionsOut[1] = Input.GetAxis(\"Vertical\");\n}\n</code></pre> <p>In order for the Agent to use the Heuristic, You will need to set the <code>Behavior Type</code> to <code>Heuristic Only</code> in the <code>Behavior Parameters</code> of the RollerAgent.</p> <p>Press Play to run the scene and use the arrows keys to move the Agent around the platform. Make sure that there are no errors displayed in the Unity Editor Console window and that the Agent resets when it reaches its target or falls from the platform.</p>"},{"location":"Learning-Environment-Create-New/#training-the-environment","title":"Training the Environment","text":"<p>The process is the same as described in the Getting Started Guide.</p> <p>The hyperparameters for training are specified in a configuration file that you pass to the <code>mlagents-learn</code> program. Create a new <code>rollerball_config.yaml</code> file under <code>config/</code> and include the following hyperparameter values:</p> <pre><code>behaviors:\n  RollerBall:\n    trainer_type: ppo\n    hyperparameters:\n      batch_size: 10\n      buffer_size: 100\n      learning_rate: 3.0e-4\n      beta: 5.0e-4\n      epsilon: 0.2\n      lambd: 0.99\n      num_epoch: 3\n      learning_rate_schedule: linear\n      beta_schedule: constant\n      epsilon_schedule: linear\n    network_settings:\n      normalize: false\n      hidden_units: 128\n      num_layers: 2\n    reward_signals:\n      extrinsic:\n        gamma: 0.99\n        strength: 1.0\n    max_steps: 500000\n    time_horizon: 64\n    summary_freq: 10000\n</code></pre> <p>Hyperparameters are explained in the training configuration file documentation</p> <p>Since this example creates a very simple training environment with only a few inputs and outputs, using small batch and buffer sizes speeds up the training considerably. However, if you add more complexity to the environment or change the reward or observation functions, you might also find that training performs better with different hyperparameter values. In addition to setting these hyperparameter values, the Agent DecisionFrequency parameter has a large effect on training time and success. A larger value reduces the number of decisions the training algorithm has to consider and, in this simple environment, speeds up training.</p> <p>To train your agent, run the following command before pressing Play in the Editor:</p> <pre><code>mlagents-learn config/rollerball_config.yaml --run-id=RollerBall\n</code></pre> <p>To monitor the statistics of Agent performance during training, use TensorBoard.</p> <p></p> <p>In particular, the cumulative_reward and value_estimate statistics show how well the Agent is achieving the task. In this example, the maximum reward an Agent can earn is 1.0, so these statistics approach that value when the Agent has successfully solved the problem.</p>"},{"location":"Learning-Environment-Create-New/#optional-multiple-training-areas-within-the-same-scene","title":"Optional: Multiple Training Areas within the Same Scene","text":"<p>In many of the example environments, many copies of the training area are instantiated in the scene. This generally speeds up training, allowing the environment to gather many experiences in parallel. This can be achieved simply by instantiating many Agents with the same <code>Behavior Name</code>. Note that we've already simplified our transition to using multiple areas by creating the <code>TrainingArea</code> GameObject and relying on local positions in <code>RollerAgent.cs</code>. Use the following steps to parallelize your RollerBall environment:</p> <ol> <li>Drag the TrainingArea GameObject, along with its attached GameObjects, into    your Assets browser, turning it into a prefab.</li> <li>You can now instantiate copies of the TrainingArea prefab. Drag them into    your scene, positioning them so that they do not overlap.</li> </ol> <p>Alternatively, you can use the <code>TrainingAreaReplicator</code> to replicate training areas. Use the following steps:</p> <ol> <li>Create a new empty Game Object in the scene.</li> <li>Click on the new object and add a TrainingAreaReplicator component to the empty Game Object through the inspector.</li> <li>Drag the training area to <code>Base Area</code> in the Training Area Replicator.</li> <li>Specify the number of areas to replicate and the separation between areas.</li> <li>Hit play and the areas will be replicated automatically!</li> </ol>"},{"location":"Learning-Environment-Create-New/#optional-training-using-concurrent-unity-instances","title":"Optional: Training Using Concurrent Unity Instances","text":"<p>Another level of parallelization comes by training using concurrent Unity instances. For example,</p> <pre><code>mlagents-learn config/rollerball_config.yaml --run-id=RollerBall --num-envs=2\n</code></pre> <p>will start ML Agents training with two environment instances. Combining multiple training areas within the same scene, with concurrent Unity instances, effectively gives you two levels of parallelism to speed up training. The command line option <code>--num-envs=&lt;n&gt;</code> controls the number of concurrent Unity instances that are executed in parallel during training.</p>"},{"location":"Learning-Environment-Design-Agents/","title":"Agents","text":"<p>Table of Contents:</p> <ul> <li>Decisions</li> <li>Observations and Sensors</li> <li>Generating Observations<ul> <li>Agent.CollectObservations()</li> <li>Observable Fields and Properties</li> <li>ISensor interface and SensorComponents</li> </ul> </li> <li>Vector Observations<ul> <li>One-hot encoding categorical information</li> <li>Normalization</li> <li>Stacking</li> <li>Vector Observation Summary &amp; Best Practices</li> </ul> </li> <li>Visual Observations<ul> <li>Visual Observation Summary &amp; Best Practices</li> </ul> </li> <li>Raycast Observations<ul> <li>RayCast Observation Summary &amp; Best Practices</li> </ul> </li> <li>Variable Length Observations<ul> <li>Variable Length Observation Summary &amp; Best Practices</li> </ul> </li> <li>Goal Signal<ul> <li>Goal Signal Summary &amp; Best Practices</li> </ul> </li> <li>Actions and Actuators</li> <li>Continuous Actions</li> <li>Discrete Actions<ul> <li>Masking Discrete Actions</li> </ul> </li> <li>Actions Summary &amp; Best Practices</li> <li>Rewards</li> <li>Examples</li> <li>Rewards Summary &amp; Best Practices</li> <li>Agent Properties</li> <li>Destroying an Agent</li> <li>Defining Multi-agent Scenarios</li> <li>Teams for Adversarial Scenarios</li> <li>Groups for Cooperative Scenarios</li> <li>Recording Demonstrations</li> </ul> <p>An agent is an entity that can observe its environment, decide on the best course of action using those observations, and execute those actions within its environment. Agents can be created in Unity by extending the <code>Agent</code> class. The most important aspects of creating agents that can successfully learn are the observations the agent collects, and the reward you assign to estimate the value of the agent's current state toward accomplishing its tasks.</p> <p>An Agent passes its observations to its Policy. The Policy then makes a decision and passes the chosen action back to the agent. Your agent code must execute the action, for example, move the agent in one direction or another. In order to train an agent using reinforcement learning, your agent must calculate a reward value at each action. The reward is used to discover the optimal decision-making policy.</p> <p>The <code>Policy</code> class abstracts out the decision making logic from the Agent itself so that you can use the same Policy in multiple Agents. How a Policy makes its decisions depends on the <code>Behavior Parameters</code> associated with the agent. If you set <code>Behavior Type</code> to <code>Heuristic Only</code>, the Agent will use its <code>Heuristic()</code> method to make decisions which can allow you to control the Agent manually or write your own Policy. If the Agent has a <code>Model</code> file, its Policy will use the neural network <code>Model</code> to take decisions.</p> <p>When you create an Agent, you should usually extend the base Agent class. This includes implementing the following methods:</p> <ul> <li><code>Agent.OnEpisodeBegin()</code> \u2014 Called at the beginning of an Agent's episode,   including at the beginning of the simulation.</li> <li><code>Agent.CollectObservations(VectorSensor sensor)</code> \u2014 Called every step that the Agent   requests a decision. This is one possible way for collecting the Agent's   observations of the environment; see Generating Observations   below for more options.</li> <li><code>Agent.OnActionReceived()</code> \u2014 Called every time the Agent receives an action to   take. Receives the action chosen by the Agent. It is also common to assign a   reward in this method.</li> <li><code>Agent.Heuristic()</code> - When the <code>Behavior Type</code> is set to <code>Heuristic Only</code> in   the Behavior Parameters of the Agent, the Agent will use the <code>Heuristic()</code>   method to generate the actions of the Agent. As such, the <code>Heuristic()</code> method   writes to the array of floats provided to the Heuristic method as argument.   Note: Do not create a new float array of action in the <code>Heuristic()</code> method,   as this will prevent writing floats to the original action array.</li> </ul> <p>As a concrete example, here is how the Ball3DAgent class implements these methods:</p> <ul> <li><code>Agent.OnEpisodeBegin()</code> \u2014 Resets the agent cube and ball to their starting   positions. The function randomizes the reset values so that the training   generalizes to more than a specific starting position and agent cube   orientation.</li> <li><code>Agent.CollectObservations(VectorSensor sensor)</code> \u2014 Adds information about the   orientation of the agent cube, the ball velocity, and the relative position   between the ball and the cube. Since the  <code>CollectObservations()</code>   method calls <code>VectorSensor.AddObservation()</code> such that vector size adds up to 8,   the Behavior Parameters of the Agent are set with vector observation space   with a state size of 8.</li> <li><code>Agent.OnActionReceived()</code> \u2014 The action results   in a small change in the agent cube's rotation at each step. In this example,   an Agent receives a small positive reward for each step it keeps the ball on the   agent cube's head and a larger, negative reward for dropping the ball. An   Agent's episode is also ended when it drops the ball so that it will reset   with a new ball for the next simulation step.</li> <li><code>Agent.Heuristic()</code> - Converts the keyboard inputs into actions.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#decisions","title":"Decisions","text":"<p>The observation-decision-action-reward cycle repeats each time the Agent request a decision. Agents will request a decision when <code>Agent.RequestDecision()</code> is called. If you need the Agent to request decisions on its own at regular intervals, add a <code>Decision Requester</code> component to the Agent's GameObject. Making decisions at regular step intervals is generally most appropriate for physics-based simulations. For example, an agent in a robotic simulator that must provide fine-control of joint torques should make its decisions every step of the simulation. In games such as real-time strategy, where many agents make their decisions at regular intervals, the decision timing for each agent can be staggered by setting the <code>DecisionStep</code> parameter in the <code>Decision Requester</code> component for each agent. On the other hand, an agent that only needs to make decisions when certain game or simulation events occur, such as in a turn-based game, should call <code>Agent.RequestDecision()</code> manually.</p>"},{"location":"Learning-Environment-Design-Agents/#observations-and-sensors","title":"Observations and Sensors","text":"<p>In order for an agent to learn, the observations should include all the information an agent needs to accomplish its task. Without sufficient and relevant information, an agent may learn poorly or may not learn at all. A reasonable approach for determining what information should be included is to consider what you would need to calculate an analytical solution to the problem, or what you would expect a human to be able to use to solve the problem.</p>"},{"location":"Learning-Environment-Design-Agents/#generating-observations","title":"Generating Observations","text":"<p>ML-Agents provides multiple ways for an Agent to make observations:   1. Overriding the <code>Agent.CollectObservations()</code> method and passing the     observations to the provided <code>VectorSensor</code>.   1. Adding the <code>[Observable]</code> attribute to fields and properties on the Agent.   1. Implementing the <code>ISensor</code> interface, using a <code>SensorComponent</code> attached to   the Agent to create the <code>ISensor</code>.</p>"},{"location":"Learning-Environment-Design-Agents/#agentcollectobservations","title":"Agent.CollectObservations()","text":"<p>Agent.CollectObservations() is best used for aspects of the environment which are numerical and non-visual. The Policy class calls the <code>CollectObservations(VectorSensor sensor)</code> method of each Agent. Your implementation of this function must call <code>VectorSensor.AddObservation</code> to add vector observations.</p> <p>The <code>VectorSensor.AddObservation</code> method provides a number of overloads for adding common types of data to your observation vector. You can add Integers and booleans directly to the observation vector, as well as some common Unity data types such as <code>Vector2</code>, <code>Vector3</code>, and <code>Quaternion</code>.</p> <p>For examples of various state observation functions, you can look at the example environments included in the ML-Agents SDK. For instance, the 3DBall example uses the rotation of the platform, the relative position of the ball, and the velocity of the ball as its state observation.</p> <pre><code>public GameObject ball;\n\npublic override void CollectObservations(VectorSensor sensor)\n{\n    // Orientation of the cube (2 floats)\n    sensor.AddObservation(gameObject.transform.rotation.z);\n    sensor.AddObservation(gameObject.transform.rotation.x);\n    // Relative position of the ball to the cube (3 floats)\n    sensor.AddObservation(ball.transform.position - gameObject.transform.position);\n    // Velocity of the ball (3 floats)\n    sensor.AddObservation(m_BallRb.velocity);\n    // 8 floats total\n}\n</code></pre> <p>As an experiment, you can remove the velocity components from the observation and retrain the 3DBall agent. While it will learn to balance the ball reasonably well, the performance of the agent without using velocity is noticeably worse.</p> <p>The observations passed to <code>VectorSensor.AddObservation()</code> must always contain the same number of elements must always be in the same order. If the number of observed entities in an environment can vary, you can pad the calls with zeros for any missing entities in a specific observation, or you can limit an agent's observations to a fixed subset. For example, instead of observing every enemy in an environment, you could only observe the closest five.</p> <p>Additionally, when you set up an Agent's <code>Behavior Parameters</code> in the Unity Editor, you must set the Vector Observations &gt; Space Size to equal the number of floats that are written by <code>CollectObservations()</code>.</p>"},{"location":"Learning-Environment-Design-Agents/#observable-fields-and-properties","title":"Observable Fields and Properties","text":"<p>Another approach is to define the relevant observations as fields or properties on your Agent class, and annotate them with an <code>ObservableAttribute</code>. For example, in the Ball3DHardAgent, the difference between positions could be observed by adding a property to the Agent:</p> <pre><code>using Unity.MLAgents.Sensors.Reflection;\n\npublic class Ball3DHardAgent : Agent {\n\n    [Observable(numStackedObservations: 9)]\n    Vector3 PositionDelta\n    {\n        get\n        {\n            return ball.transform.position - gameObject.transform.position;\n        }\n    }\n}\n</code></pre> <p><code>ObservableAttribute</code> currently supports most basic types (e.g. floats, ints, bools), as well as <code>Vector2</code>, <code>Vector3</code>, <code>Vector4</code>, <code>Quaternion</code>, and enums.</p> <p>The behavior of <code>ObservableAttribute</code>s are controlled by the \"Observable Attribute Handling\" in the Agent's <code>Behavior Parameters</code>. The possible values for this are:  * Ignore (default) - All ObservableAttributes on the Agent will be ignored.   If there are no ObservableAttributes on the Agent, this will result in the   fastest  initialization time.  * Exclude Inherited - Only members on the declared class will be examined;   members that are inherited are ignored. This is a reasonable tradeoff between   performance and flexibility.  * Examine All All members on the class will be examined. This can lead to   slower startup times.</p> <p>\"Exclude Inherited\" is generally sufficient, but if your Agent inherits from another Agent implementation that has Observable members, you will need to use \"Examine All\".</p> <p>Internally, ObservableAttribute uses reflection to determine which members of the Agent have ObservableAttributes, and also uses reflection to access the fields or invoke the properties at runtime. This may be slower than using CollectObservations or an ISensor, although this might not be enough to noticeably affect performance.</p> <p>NOTE: you do not need to adjust the Space Size in the Agent's <code>Behavior Parameters</code> when you add <code>[Observable]</code> fields or properties to an Agent, since their size can be computed before they are used.</p>"},{"location":"Learning-Environment-Design-Agents/#isensor-interface-and-sensorcomponents","title":"ISensor interface and SensorComponents","text":"<p>The <code>ISensor</code> interface is generally intended for advanced users. The <code>Write()</code> method is used to actually generate the observation, but some other methods such as returning the shape of the observations must also be implemented.</p> <p>The <code>SensorComponent</code> abstract class is used to create the actual <code>ISensor</code> at runtime. It must be attached to the same <code>GameObject</code> as the <code>Agent</code>, or to a child <code>GameObject</code>.</p> <p>There are several SensorComponents provided in the API, including: - <code>CameraSensorComponent</code> - Uses images from a <code>Camera</code> as observations. - <code>RenderTextureSensorComponent</code> - Uses the content of a <code>RenderTexture</code> as observations. - <code>RayPerceptionSensorComponent</code> - Uses the information from set of ray casts as observations. - <code>Match3SensorComponent</code> - Uses the board of a Match-3 game as observations. - <code>GridSensorComponent</code> - Uses a set of box queries in a grid shape as observations.</p> <p>NOTE: you do not need to adjust the Space Size in the Agent's <code>Behavior Parameters</code> when using <code>SensorComponents</code>s.</p> <p>Internally, both <code>Agent.CollectObservations</code> and <code>[Observable]</code> attribute use an ISensors to write observations, although this is mostly abstracted from the user.</p>"},{"location":"Learning-Environment-Design-Agents/#vector-observations","title":"Vector Observations","text":"<p>Both <code>Agent.CollectObservations()</code> and <code>ObservableAttribute</code>s produce vector observations, which are represented at lists of <code>float</code>s. <code>ISensor</code>s can produce both vector observations and visual observations, which are multi-dimensional arrays of floats.</p> <p>Below are some additional considerations when dealing with vector observations:</p>"},{"location":"Learning-Environment-Design-Agents/#one-hot-encoding-categorical-information","title":"One-hot encoding categorical information","text":"<p>Type enumerations should be encoded in the one-hot style. That is, add an element to the feature vector for each element of enumeration, setting the element representing the observed member to one and set the rest to zero. For example, if your enumeration contains [Sword, Shield, Bow] and the agent observes that the current item is a Bow, you would add the elements: 0, 0, 1 to the feature vector. The following code example illustrates how to add.</p> <pre><code>enum ItemType { Sword, Shield, Bow, LastItem }\npublic override void CollectObservations(VectorSensor sensor)\n{\n    for (int ci = 0; ci &lt; (int)ItemType.LastItem; ci++)\n    {\n        sensor.AddObservation((int)currentItem == ci ? 1.0f : 0.0f);\n    }\n}\n</code></pre> <p><code>VectorSensor</code> also provides a two-argument function <code>AddOneHotObservation()</code> as a shortcut for one-hot style observations. The following example is identical to the previous one.</p> <pre><code>enum ItemType { Sword, Shield, Bow, LastItem }\nconst int NUM_ITEM_TYPES = (int)ItemType.LastItem + 1;\n\npublic override void CollectObservations(VectorSensor sensor)\n{\n    // The first argument is the selection index; the second is the\n    // number of possibilities\n    sensor.AddOneHotObservation((int)currentItem, NUM_ITEM_TYPES);\n}\n</code></pre> <p><code>ObservableAttribute</code> has built-in support for enums. Note that you don't need the <code>LastItem</code> placeholder in this case:</p> <pre><code>enum ItemType { Sword, Shield, Bow }\n\npublic class HeroAgent : Agent\n{\n    [Observable]\n    ItemType m_CurrentItem;\n}\n</code></pre>"},{"location":"Learning-Environment-Design-Agents/#normalization","title":"Normalization","text":"<p>For the best results when training, you should normalize the components of your feature vector to the range [-1, +1] or [0, 1]. When you normalize the values, the PPO neural network can often converge to a solution faster. Note that it isn't always necessary to normalize to these recommended ranges, but it is considered a best practice when using neural networks. The greater the variation in ranges between the components of your observation, the more likely that training will be affected.</p> <p>To normalize a value to [0, 1], you can use the following formula:</p> <pre><code>normalizedValue = (currentValue - minValue)/(maxValue - minValue)\n</code></pre> <p>:warning: For vectors, you should apply the above formula to each component (x, y, and z). Note that this is not the same as using the <code>Vector3.normalized</code> property or <code>Vector3.Normalize()</code> method in Unity (and similar for <code>Vector2</code>).</p> <p>Rotations and angles should also be normalized. For angles between 0 and 360 degrees, you can use the following formulas:</p> <pre><code>Quaternion rotation = transform.rotation;\nVector3 normalized = rotation.eulerAngles / 180.0f - Vector3.one;  // [-1,1]\nVector3 normalized = rotation.eulerAngles / 360.0f;  // [0,1]\n</code></pre> <p>For angles that can be outside the range [0,360], you can either reduce the angle, or, if the number of turns is significant, increase the maximum value used in your normalization formula.</p>"},{"location":"Learning-Environment-Design-Agents/#stacking","title":"Stacking","text":"<p>Stacking refers to repeating observations from previous steps as part of a larger observation. For example, consider an Agent that generates these observations in four steps</p> <pre><code>step 1: [0.1]\nstep 2: [0.2]\nstep 3: [0.3]\nstep 4: [0.4]\n</code></pre> <p>If we use a stack size of 3, the observations would instead be:</p> <pre><code>step 1: [0.1, 0.0, 0.0]\nstep 2: [0.2, 0.1, 0.0]\nstep 3: [0.3, 0.2, 0.1]\nstep 4: [0.4, 0.3, 0.2]\n</code></pre> <p>(The observations are padded with zeroes for the first <code>stackSize-1</code> steps). This is a simple way to give an Agent limited \"memory\" without the complexity of adding a recurrent neural network (RNN).</p> <p>The steps for enabling stacking depends on how you generate observations: * For Agent.CollectObservations(), set \"Stacked Vectors\" on the Agent's   <code>Behavior Parameters</code> to a value greater than 1. * For ObservableAttribute, set the <code>numStackedObservations</code> parameter in the   constructor, e.g. <code>[Observable(numStackedObservations: 2)]</code>. * For <code>ISensor</code>s, wrap them in a <code>StackingSensor</code> (which is also an <code>ISensor</code>).   Generally, this should happen in the <code>CreateSensor()</code> method of your   <code>SensorComponent</code>.</p>"},{"location":"Learning-Environment-Design-Agents/#vector-observation-summary-best-practices","title":"Vector Observation Summary &amp; Best Practices","text":"<ul> <li>Vector Observations should include all variables relevant for allowing the   agent to take the optimally informed decision, and ideally no extraneous   information.</li> <li>In cases where Vector Observations need to be remembered or compared over   time, either an RNN should be used in the model, or the <code>Stacked Vectors</code>   value in the agent GameObject's <code>Behavior Parameters</code> should be changed.</li> <li>Categorical variables such as type of object (Sword, Shield, Bow) should be   encoded in one-hot fashion (i.e. <code>3</code> -&gt; <code>0, 0, 1</code>). This can be done   automatically using the <code>AddOneHotObservation()</code> method of the <code>VectorSensor</code>,   or using <code>[Observable]</code> on an enum field or property of the Agent.</li> <li>In general, all inputs should be normalized to be in the range 0 to +1 (or -1   to 1). For example, the <code>x</code> position information of an agent where the maximum   possible value is <code>maxValue</code> should be recorded as   <code>VectorSensor.AddObservation(transform.position.x / maxValue);</code> rather than   <code>VectorSensor.AddObservation(transform.position.x);</code>.</li> <li>Positional information of relevant GameObjects should be encoded in relative   coordinates wherever possible. This is often relative to the agent position.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#visual-observations","title":"Visual Observations","text":"<p>Visual observations are generally provided to agent via either a <code>CameraSensor</code> or <code>RenderTextureSensor</code>. These collect image information and transforms it into a 3D Tensor which can be fed into the convolutional neural network (CNN) of the agent policy. For more information on CNNs, see this guide. This allows agents to learn from spatial regularities in the observation images. It is possible to use visual and vector observations with the same agent.</p> <p>Agents using visual observations can capture state of arbitrary complexity and are useful when the state is difficult to describe numerically. However, they are also typically less efficient and slower to train, and sometimes don't succeed at all as compared to vector observations. As such, they should only be used when it is not possible to properly define the problem using vector or ray-cast observations.</p> <p>Visual observations can be derived from Cameras or RenderTextures within your scene. To add a visual observation to an Agent, add either a Camera Sensor Component or RenderTextures Sensor Component to the Agent. Then drag the camera or render texture you want to add to the <code>Camera</code> or <code>RenderTexture</code> field. You can have more than one camera or render texture and even use a combination of both attached to an Agent. For each visual observation, set the width and height of the image (in pixels) and whether or not the observation is color or grayscale.</p> <p></p> <p>or</p> <p></p> <p>Each Agent that uses the same Policy must have the same number of visual observations, and they must all have the same resolutions (including whether or not they are grayscale). Additionally, each Sensor Component on an Agent must have a unique name so that they can be sorted deterministically (the name must be unique for that Agent, but multiple Agents can have a Sensor Component with the same name).</p> <p>Visual observations also support stacking, by specifying <code>Observation Stacks</code> to a value greater than 1. The visual observations from the last <code>stackSize</code> steps will be stacked on the last dimension (channel dimension).</p> <p>When using <code>RenderTexture</code> visual observations, a handy feature for debugging is adding a <code>Canvas</code>, then adding a <code>Raw Image</code> with it's texture set to the Agent's <code>RenderTexture</code>. This will render the agent observation on the game screen.</p> <p></p> <p>The GridWorld environment is an example on how to use a RenderTexture for both debugging and observation. Note that in this example, a Camera is rendered to a RenderTexture, which is then used for observations and debugging. To update the RenderTexture, the Camera must be asked to render every time a decision is requested within the game code. When using Cameras as observations directly, this is done automatically by the Agent.</p> <p></p>"},{"location":"Learning-Environment-Design-Agents/#visual-observation-summary-best-practices","title":"Visual Observation Summary &amp; Best Practices","text":"<ul> <li>To collect visual observations, attach <code>CameraSensor</code> or <code>RenderTextureSensor</code>   components to the agent GameObject.</li> <li>Visual observations should generally only be used when vector observations are   not sufficient.</li> <li>Image size should be kept as small as possible, without the loss of needed   details for decision making.</li> <li>Images should be made grayscale in situations where color information is not   needed for making informed decisions.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#raycast-observations","title":"Raycast Observations","text":"<p>Raycasts are another possible method for providing observations to an agent. This can be easily implemented by adding a <code>RayPerceptionSensorComponent3D</code> (or <code>RayPerceptionSensorComponent2D</code>) to the Agent GameObject.</p> <p>During observations, several rays (or spheres, depending on settings) are cast into the physics world, and the objects that are hit determine the observation vector that is produced.</p> <p></p> <p>Both sensor components have several settings:</p> <ul> <li>Detectable Tags A list of strings corresponding to the types of objects that   the Agent should be able to distinguish between. For example, in the WallJump   example, we use \"wall\", \"goal\", and \"block\" as the list of objects to detect.</li> <li>Rays Per Direction Determines the number of rays that are cast. One ray is   always cast forward, and this many rays are cast to the left and right.</li> <li>Max Ray Degrees The angle (in degrees) for the outermost rays. 90 degrees   corresponds to the left and right of the agent.</li> <li>Sphere Cast Radius The size of the sphere used for sphere casting. If set to   0, rays will be used instead of spheres. Rays may be more efficient,   especially in complex scenes.</li> <li>Ray Length The length of the casts</li> <li>Ray Layer Mask The LayerMask   passed to the raycast or spherecast. This can be used to ignore certain types   of objects when casting.</li> <li>Observation Stacks The number of previous results to \"stack\" with the cast   results. Note that this can be independent of the \"Stacked Vectors\" setting in   <code>Behavior Parameters</code>.</li> <li>Start Vertical Offset (3D only) The vertical offset of the ray start point.</li> <li>End Vertical Offset (3D only) The vertical offset of the ray end point.</li> <li>Alternating Ray Order Alternating is the default, it gives an order of (0,   -delta, delta, -2delta, 2delta, ..., -ndelta, ndelta). If alternating is   disabled the order is left to right (-ndelta, -(n-1)delta, ..., -delta, 0,   delta, ..., (n-1)delta, ndelta). For general usage there is no difference   but if using custom models the left-to-right layout that matches the spatial   structuring can be preferred (e.g. for processing with conv nets).</li> <li>Use Batched Raycasts (3D only) Whether to use batched raycasts. Enable to use batched raycasts and the jobs system.</li> </ul> <p>In the example image above, the Agent has two <code>RayPerceptionSensorComponent3D</code>s. Both use 3 Rays Per Direction and 90 Max Ray Degrees. One of the components had a vertical offset, so the Agent can tell whether it's clear to jump over the wall.</p> <p>The total size of the created observations is</p> <pre><code>(Observation Stacks) * (1 + 2 * Rays Per Direction) * (Num Detectable Tags + 2)\n</code></pre> <p>so the number of rays and tags should be kept as small as possible to reduce the amount of data used. Note that this is separate from the State Size defined in <code>Behavior Parameters</code>, so you don't need to worry about the formula above when setting the State Size.</p>"},{"location":"Learning-Environment-Design-Agents/#raycast-observation-summary-best-practices","title":"RayCast Observation Summary &amp; Best Practices","text":"<ul> <li>Attach <code>RayPerceptionSensorComponent3D</code> or <code>RayPerceptionSensorComponent2D</code> to   use.</li> <li>This observation type is best used when there is relevant spatial information   for the agent that doesn't require a fully rendered image to convey.</li> <li>Use as few rays and tags as necessary to solve the problem in order to improve   learning stability and agent performance.</li> <li>If you run into performance issues, try using batched raycasts by enabling the Use Batched Raycast setting.   (Only available for 3D ray perception sensors.)</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#grid-observations","title":"Grid Observations","text":"<p>Grid-base observations combine the advantages of 2D spatial representation in visual observations, and the flexibility of defining detectable objects in RayCast observations. The sensor uses a set of box queries in a grid shape and gives a top-down 2D view around the agent. This can be implemented by adding a <code>GridSensorComponent</code> to the Agent GameObject.</p> <p>During observations, the sensor detects the presence of detectable objects in each cell and encode that into one-hot representation. The collected information from each cell forms a 3D tensor observation and will be fed into the convolutional neural network (CNN) of the agent policy just like visual observations.</p> <p></p> <p>The sensor component has the following settings: - Cell Scale The scale of each cell in the grid. - Grid Size Number of cells on each side of the grid. - Agent Game Object The Agent that holds the grid sensor. This is used to   disambiguate objects with the same tag as the agent so that the agent doesn't   detect itself. - Rotate With Agent Whether the grid rotates with the Agent. - Detectable Tags A list of strings corresponding to the types of objects that   the Agent should be able to distinguish between. - Collider Mask The LayerMask   passed to the collider detection. This can be used to ignore certain types   of objects. - Initial Collider Buffer Size The initial size of the Collider buffer used   in the non-allocating Physics calls for each cell. - Max Collider Buffer Size The max size of the Collider buffer used in the   non-allocating Physics calls for each cell.</p> <p>The observation for each grid cell is a one-hot encoding of the detected object. The total size of the created observations is</p> <pre><code>GridSize.x * GridSize.z * Num Detectable Tags\n</code></pre> <p>so the number of detectable tags and size of the grid should be kept as small as possible to reduce the amount of data used. This makes a trade-off between the granularity of the observation and training speed.</p> <p>To allow more variety of observations that grid sensor can capture, the <code>GridSensorComponent</code> and the underlying <code>GridSensorBase</code> also provides interfaces that can be overridden to collect customized observation from detected objects. See the doc on extending grid Sensors for more details on custom grid sensors.</p> <p>Note: The <code>GridSensor</code> only works in 3D environments and will not behave properly in 2D environments.</p>"},{"location":"Learning-Environment-Design-Agents/#grid-observation-summary-best-practices","title":"Grid Observation Summary &amp; Best Practices","text":"<ul> <li>Attach <code>GridSensorComponent</code> to use.</li> <li>This observation type is best used when there is relevant non-visual spatial information that   can be best captured in 2D representations.</li> <li>Use as small grid size and as few tags as necessary to solve the problem in order to improve   learning stability and agent performance.</li> <li>Do not use <code>GridSensor</code> in a 2D game.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#variable-length-observations","title":"Variable Length Observations","text":"<p>It is possible for agents to collect observations from a varying number of GameObjects by using a <code>BufferSensor</code>. You can add a <code>BufferSensor</code> to your Agent by adding a <code>BufferSensorComponent</code> to its GameObject. The <code>BufferSensor</code> can be useful in situations in which the Agent must pay attention to a varying number of entities (for example, a varying number of enemies or projectiles). On the trainer side, the <code>BufferSensor</code> is processed using an attention module. More information about attention mechanisms can be found here. Training or doing inference with variable length observations can be slower than using a flat vector observation. However, attention mechanisms enable solving problems that require comparative reasoning between entities in a scene such as our Sorter environment. Note that even though the <code>BufferSensor</code> can process a variable number of entities, you still need to define a maximum number of entities. This is because our network architecture requires to know what the shape of the observations will be. If fewer entities are observed than the maximum, the observation will be padded with zeros and the trainer will ignore the padded observations. Note that attention layers are invariant to the order of the entities, so there is no need to properly \"order\" the entities before feeding them into the <code>BufferSensor</code>.</p> <p>The <code>BufferSensorComponent</code> Editor inspector has two arguments:  - <code>Observation Size</code> : This is how many floats each entities will be  represented with. This number is fixed and all entities must  have the same representation. For example, if the entities you want to  put into the <code>BufferSensor</code> have for relevant information position and  speed, then the <code>Observation Size</code> should be 6 floats.  - <code>Maximum Number of Entities</code> : This is the maximum number of entities  the <code>BufferSensor</code> will be able to collect.</p> <p>To add an entity's observations to a <code>BufferSensorComponent</code>, you need to call <code>BufferSensorComponent.AppendObservation()</code> in the Agent.CollectObservations() method with a float array of size <code>Observation Size</code> as argument.</p> <p>Note: Currently, the observations put into the <code>BufferSensor</code> are not normalized, you will need to normalize your observations manually between -1 and 1.</p>"},{"location":"Learning-Environment-Design-Agents/#variable-length-observation-summary-best-practices","title":"Variable Length Observation Summary &amp; Best Practices","text":"<ul> <li>Attach <code>BufferSensorComponent</code> to use.</li> <li>Call <code>BufferSensorComponent.AppendObservation()</code> in the  Agent.CollectObservations() methodto add the observations  of an entity to the <code>BufferSensor</code>.</li> <li>Normalize the entities observations before feeding them into the <code>BufferSensor</code>.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#goal-signal","title":"Goal Signal","text":"<p>It is possible for agents to collect observations that will be treated as \"goal signal\". A goal signal is used to condition the policy of the agent, meaning that if the goal changes, the policy (i.e. the mapping from observations to actions) will change as well. Note that this is true for any observation since all observations influence the policy of the Agent to some degree. But by specifying a goal signal explicitly, we can make this conditioning more important to the agent. This feature can be used in settings where an agent must learn to solve different tasks that are similar by some aspects because the agent will learn to reuse learnings from different tasks to generalize better. In Unity, you can specify that a <code>VectorSensor</code> or a <code>CameraSensor</code> is a goal by attaching a <code>VectorSensorComponent</code> or a <code>CameraSensorComponent</code> to the Agent and selecting <code>Goal Signal</code> as <code>Observation Type</code>. On the trainer side, there are two different ways to condition the policy. This setting is determined by the conditioning_type parameter. If set to <code>hyper</code> (default) a HyperNetwork will be used to generate some of the weights of the policy using the goal observations as input. Note that using a HyperNetwork requires a lot of computations, it is recommended to use a smaller number of hidden units in the policy to alleviate this. If set to <code>none</code> the goal signal will be considered as regular observations. For an example on how to use a goal signal, see the GridWorld example.</p>"},{"location":"Learning-Environment-Design-Agents/#goal-signal-summary-best-practices","title":"Goal Signal Summary &amp; Best Practices","text":"<ul> <li>Attach a <code>VectorSensorComponent</code> or <code>CameraSensorComponent</code> to an agent and  set the observation type to goal to use the feature.</li> <li>Set the conditioning_type parameter in the training configuration.</li> <li>Reduce the number of hidden units in the network when using the HyperNetwork  conditioning type.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#actions-and-actuators","title":"Actions and Actuators","text":"<p>An action is an instruction from the Policy that the agent carries out. The action is passed to the an <code>IActionReceiver</code> (either an <code>Agent</code> or an <code>IActuator</code>) as the <code>ActionBuffers</code> parameter when the Academy invokes the <code>IActionReciever.OnActionReceived()</code> function. There are two types of actions supported: Continuous and Discrete.</p> <p>Neither the Policy nor the training algorithm know anything about what the action values themselves mean. The training algorithm simply tries different values for the action list and observes the affect on the accumulated rewards over time and many training episodes. Thus, the only place actions are defined for an Agent is in the <code>OnActionReceived()</code> function.</p> <p>For example, if you designed an agent to move in two dimensions, you could use either continuous or the discrete actions. In the continuous case, you would set the action size to two (one for each dimension), and the agent's Policy would output an action with two floating point values. In the discrete case, you would use one Branch with a size of four (one for each direction), and the Policy would create an action array containing a single element with a value ranging from zero to three. Alternatively, you could create two branches of size two (one for horizontal movement and one for vertical movement), and the Policy would output an action array containing two elements with values ranging from zero to one. You could alternatively use a combination of continuous and discrete actions e.g., using one continuous action for horizontal movement and a discrete branch of size two for the vertical movement.</p> <p>Note that when you are programming actions for an agent, it is often helpful to test your action logic using the <code>Heuristic()</code> method of the Agent, which lets you map keyboard commands to actions.</p>"},{"location":"Learning-Environment-Design-Agents/#continuous-actions","title":"Continuous Actions","text":"<p>When an Agent's Policy has Continuous actions, the <code>ActionBuffers.ContinuousActions</code> passed to the Agent's <code>OnActionReceived()</code> function is an array with length equal to the <code>Continuous Action Size</code> property value. The individual values in the array have whatever meanings that you ascribe to them. If you assign an element in the array as the speed of an Agent, for example, the training process learns to control the speed of the Agent through this parameter.</p> <p>The 3DBall example uses continuous actions with two control values.</p> <p></p> <p>These control values are applied as rotation to the cube:</p> <pre><code>    public override void OnActionReceived(ActionBuffers actionBuffers)\n    {\n        var actionZ = 2f * Mathf.Clamp(actionBuffers.ContinuousActions[0], -1f, 1f);\n        var actionX = 2f * Mathf.Clamp(actionBuffers.ContinuousActions[1], -1f, 1f);\n\n        gameObject.transform.Rotate(new Vector3(0, 0, 1), actionZ);\n        gameObject.transform.Rotate(new Vector3(1, 0, 0), actionX);\n    }\n</code></pre> <p>By default the output from our provided PPO algorithm pre-clamps the values of <code>ActionBuffers.ContinuousActions</code> into the [-1, 1] range. It is a best practice to manually clip these as well, if you plan to use a 3rd party algorithm with your environment. As shown above, you can scale the control values as needed after clamping them.</p>"},{"location":"Learning-Environment-Design-Agents/#discrete-actions","title":"Discrete Actions","text":"<p>When an Agent's Policy uses discrete actions, the <code>ActionBuffers.DiscreteActions</code> passed to the Agent's <code>OnActionReceived()</code> function is an array of integers with length equal to <code>Discrete Branch Size</code>. When defining the discrete actions, <code>Branches</code> is an array of integers, each value corresponds to the number of possibilities for each branch.</p> <p>For example, if we wanted an Agent that can move in a plane and jump, we could define two branches (one for motion and one for jumping) because we want our agent be able to move and jump concurrently. We define the first branch to have 5 possible actions (don't move, go left, go right, go backward, go forward) and the second one to have 2 possible actions (don't jump, jump). The <code>OnActionReceived()</code> method would look something like:</p> <pre><code>// Get the action index for movement\nint movement = actionBuffers.DiscreteActions[0];\n// Get the action index for jumping\nint jump = actionBuffers.DiscreteActions[1];\n\n// Look up the index in the movement action list:\nif (movement == 1) { directionX = -1; }\nif (movement == 2) { directionX = 1; }\nif (movement == 3) { directionZ = -1; }\nif (movement == 4) { directionZ = 1; }\n// Look up the index in the jump action list:\nif (jump == 1 &amp;&amp; IsGrounded()) { directionY = 1; }\n\n// Apply the action results to move the Agent\ngameObject.GetComponent&lt;Rigidbody&gt;().AddForce(\n    new Vector3(\n        directionX * 40f, directionY * 300f, directionZ * 40f));\n</code></pre>"},{"location":"Learning-Environment-Design-Agents/#masking-discrete-actions","title":"Masking Discrete Actions","text":"<p>When using Discrete Actions, it is possible to specify that some actions are impossible for the next decision. When the Agent is controlled by a neural network, the Agent will be unable to perform the specified action. Note that when the Agent is controlled by its Heuristic, the Agent will still be able to decide to perform the masked action. In order to disallow an action, override the <code>Agent.WriteDiscreteActionMask()</code> virtual method, and call <code>SetActionEnabled()</code> on the provided <code>IDiscreteActionMask</code>:</p> <pre><code>public override void WriteDiscreteActionMask(IDiscreteActionMask actionMask)\n{\n    actionMask.SetActionEnabled(branch, actionIndex, isEnabled);\n}\n</code></pre> <p>Where:</p> <ul> <li><code>branch</code> is the index (starting at 0) of the branch on which you want to allow or disallow the action</li> <li><code>actionIndex</code> is the index of the action that you want to allow or disallow.</li> <li><code>isEnabled</code> is a bool indicating whether the action should be allowed or now.</li> </ul> <p>For example, if you have an Agent with 2 branches and on the first branch (branch 0) there are 4 possible actions : \"do nothing\", \"jump\", \"shoot\" and \"change weapon\". Then with the code bellow, the Agent will either \"do nothing\" or \"change weapon\" for their next decision (since action index 1 and 2 are masked)</p> <pre><code>actionMask.SetActionEnabled(0, 1, false);\nactionMask.SetActionEnabled(0, 2, false);\n</code></pre> <p>Notes:</p> <ul> <li>You can call <code>SetActionEnabled</code> multiple times if you want to put masks on multiple   branches.</li> <li>At each step, the state of an action is reset and enabled by default.</li> <li>You cannot mask all the actions of a branch.</li> <li>You cannot mask actions in continuous control.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#iactuator-interface-and-actuatorcomponents","title":"IActuator interface and ActuatorComponents","text":"<p>The Actuator API allows users to abstract behavior out of Agents and in to components (similar to the ISensor API).  The <code>IActuator</code> interface and <code>Agent</code> class both implement the <code>IActionReceiver</code> interface to allow for backward compatibility with the current <code>Agent.OnActionReceived</code>. This means you will not have to change your code until you decide to use the <code>IActuator</code> API.</p> <p>Like the <code>ISensor</code> interface, the <code>IActuator</code> interface is intended for advanced users.</p> <p>The <code>ActuatorComponent</code> abstract class is used to create the actual <code>IActuator</code> at runtime. It must be attached to the same <code>GameObject</code> as the <code>Agent</code>, or to a child <code>GameObject</code>.  Actuators and all of their data structures are initialized during <code>Agent.Initialize</code>.  This was done to prevent an unexpected allocations at runtime.</p> <p>You can find an example of an <code>IActuator</code> implementation in the <code>Basic</code> example scene. NOTE: you do not need to adjust the Actions in the Agent's <code>Behavior Parameters</code> when using an <code>IActuator</code> and <code>ActuatorComponents</code>.</p> <p>Internally, <code>Agent.OnActionReceived</code> uses an <code>IActuator</code> to send actions to the Agent, although this is mostly abstracted from the user.</p>"},{"location":"Learning-Environment-Design-Agents/#actions-summary-best-practices","title":"Actions Summary &amp; Best Practices","text":"<ul> <li>Agents can use <code>Discrete</code> and/or <code>Continuous</code> actions.</li> <li>Discrete actions can have multiple action branches, and it's possible to mask   certain actions so that they won't be taken.</li> <li>In general, fewer actions will make for easier learning.</li> <li>Be sure to set the Continuous Action Size and Discrete Branch Size to the desired   number for each type of action, and not greater, as doing the latter can interfere with the   efficiency of the training process.</li> <li>Continuous action values should be clipped to an   appropriate range. The provided PPO model automatically clips these values   between -1 and 1, but third party training systems may not do so.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#rewards","title":"Rewards","text":"<p>In reinforcement learning, the reward is a signal that the agent has done something right. The PPO reinforcement learning algorithm works by optimizing the choices an agent makes such that the agent earns the highest cumulative reward over time. The better your reward mechanism, the better your agent will learn.</p> <p>Note: Rewards are not used during inference by an Agent using a trained model and is also not used during imitation learning.</p> <p>Perhaps the best advice is to start simple and only add complexity as needed. In general, you should reward results rather than actions you think will lead to the desired results. You can even use the Agent's Heuristic to control the Agent while watching how it accumulates rewards.</p> <p>Allocate rewards to an Agent by calling the <code>AddReward()</code> or <code>SetReward()</code> methods on the agent. The reward assigned between each decision should be in the range [-1,1]. Values outside this range can lead to unstable training. The <code>reward</code> value is reset to zero when the agent receives a new decision. If there are multiple calls to <code>AddReward()</code> for a single agent decision, the rewards will be summed together to evaluate how good the previous decision was. The <code>SetReward()</code> will override all previous rewards given to an agent since the previous decision.</p>"},{"location":"Learning-Environment-Design-Agents/#examples","title":"Examples","text":"<p>You can examine the <code>OnActionReceived()</code> functions defined in the example environments to see how those projects allocate rewards.</p> <p>The <code>GridAgent</code> class in the GridWorld example uses a very simple reward system:</p> <pre><code>Collider[] hitObjects = Physics.OverlapBox(trueAgent.transform.position,\n                                           new Vector3(0.3f, 0.3f, 0.3f));\nif (hitObjects.Where(col =&gt; col.gameObject.tag == \"goal\").ToArray().Length == 1)\n{\n    AddReward(1.0f);\n    EndEpisode();\n}\nelse if (hitObjects.Where(col =&gt; col.gameObject.tag == \"pit\").ToArray().Length == 1)\n{\n    AddReward(-1f);\n    EndEpisode();\n}\n</code></pre> <p>The agent receives a positive reward when it reaches the goal and a negative reward when it falls into the pit. Otherwise, it gets no rewards. This is an example of a sparse reward system. The agent must explore a lot to find the infrequent reward.</p> <p>In contrast, the <code>AreaAgent</code> in the Area example gets a small negative reward every step. In order to get the maximum reward, the agent must finish its task of reaching the goal square as quickly as possible:</p> <pre><code>AddReward( -0.005f);\nMoveAgent(act);\n\nif (gameObject.transform.position.y &lt; 0.0f ||\n    Mathf.Abs(gameObject.transform.position.x - area.transform.position.x) &gt; 8f ||\n    Mathf.Abs(gameObject.transform.position.z + 5 - area.transform.position.z) &gt; 8)\n{\n    AddReward(-1f);\n    EndEpisode();\n}\n</code></pre> <p>The agent also gets a larger negative penalty if it falls off the playing surface.</p> <p>The <code>Ball3DAgent</code> in the 3DBall takes a similar approach, but allocates a small positive reward as long as the agent balances the ball. The agent can maximize its rewards by keeping the ball on the platform:</p> <pre><code>\nSetReward(0.1f);\n\n// When ball falls mark Agent as finished and give a negative penalty\nif ((ball.transform.position.y - gameObject.transform.position.y) &lt; -2f ||\n    Mathf.Abs(ball.transform.position.x - gameObject.transform.position.x) &gt; 3f ||\n    Mathf.Abs(ball.transform.position.z - gameObject.transform.position.z) &gt; 3f)\n{\n    SetReward(-1f);\n    EndEpisode();\n\n}\n</code></pre> <p>The <code>Ball3DAgent</code> also assigns a negative penalty when the ball falls off the platform.</p> <p>Note that all of these environments make use of the <code>EndEpisode()</code> method, which manually terminates an episode when a termination condition is reached. This can be called independently of the <code>Max Step</code> property.</p>"},{"location":"Learning-Environment-Design-Agents/#rewards-summary-best-practices","title":"Rewards Summary &amp; Best Practices","text":"<ul> <li>Use <code>AddReward()</code> to accumulate rewards between decisions. Use <code>SetReward()</code>   to overwrite any previous rewards accumulate between decisions.</li> <li>The magnitude of any given reward should typically not be greater than 1.0 in   order to ensure a more stable learning process.</li> <li>Positive rewards are often more helpful to shaping the desired behavior of an   agent than negative rewards. Excessive negative rewards can result in the   agent failing to learn any meaningful behavior.</li> <li>For locomotion tasks, a small positive reward (+0.1) for forward velocity is   typically used.</li> <li>If you want the agent to finish a task quickly, it is often helpful to provide   a small penalty every step (-0.05) that the agent does not complete the task.   In this case completion of the task should also coincide with the end of the   episode by calling <code>EndEpisode()</code> on the agent when it has accomplished its   goal.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#agent-properties","title":"Agent Properties","text":"<ul> <li><code>Behavior Parameters</code> - The parameters dictating what Policy the Agent will   receive.</li> <li><code>Behavior Name</code> - The identifier for the behavior. Agents with the same     behavior name will learn the same policy.</li> <li><code>Vector Observation</code><ul> <li><code>Space Size</code> - Length of vector observation for the Agent.</li> <li><code>Stacked Vectors</code> - The number of previous vector observations that will   be stacked and used collectively for decision making. This results in the   effective size of the vector observation being passed to the Policy being:   Space Size x Stacked Vectors.</li> </ul> </li> <li><code>Actions</code><ul> <li><code>Continuous Actions</code> - The number of concurrent continuous actions that  the Agent can take.</li> <li><code>Discrete Branches</code> - An array of integers, defines multiple concurrent   discrete actions. The values in the <code>Discrete Branches</code> array correspond   to the number of possible discrete values for each action branch.</li> </ul> </li> <li><code>Model</code> - The neural network model used for inference (obtained after     training)</li> <li><code>Inference Device</code> - Whether to use CPU or GPU to run the model during     inference</li> <li><code>Behavior Type</code> - Determines whether the Agent will do training, inference,     or use its Heuristic() method:<ul> <li><code>Default</code> - the Agent will train if they connect to a python trainer,   otherwise they will perform inference.</li> <li><code>Heuristic Only</code> - the Agent will always use the <code>Heuristic()</code> method.</li> <li><code>Inference Only</code> - the Agent will always perform inference.</li> </ul> </li> <li><code>Team ID</code> - Used to define the team for self-play</li> <li><code>Use Child Sensors</code> - Whether to use all Sensor components attached to child     GameObjects of this Agent.</li> <li><code>Max Step</code> - The per-agent maximum number of steps. Once this number is   reached, the Agent will be reset.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#destroying-an-agent","title":"Destroying an Agent","text":"<p>You can destroy an Agent GameObject during the simulation. Make sure that there is always at least one Agent training at all times by either spawning a new Agent every time one is destroyed or by re-spawning new Agents when the whole environment resets.</p>"},{"location":"Learning-Environment-Design-Agents/#defining-multi-agent-scenarios","title":"Defining Multi-agent Scenarios","text":""},{"location":"Learning-Environment-Design-Agents/#teams-for-adversarial-scenarios","title":"Teams for Adversarial Scenarios","text":"<p>Self-play is triggered by including the self-play hyperparameter hierarchy in the trainer configuration. To distinguish opposing agents, set the team ID to different integer values in the behavior parameters script on the agent prefab.</p> <p> </p> <p>Team ID must be 0 or an integer greater than 0.</p> <p>In symmetric games, since all agents (even on opposing teams) will share the same policy, they should have the same 'Behavior Name' in their Behavior Parameters Script. In asymmetric games, they should have a different Behavior Name in their Behavior Parameters script. Note, in asymmetric games, the agents must have both different Behavior Names and different team IDs!</p> <p>For examples of how to use this feature, you can see the trainer configurations and agent prefabs for our Tennis and Soccer environments. Tennis and Soccer provide examples of symmetric games. To train an asymmetric game, specify trainer configurations for each of your behavior names and include the self-play hyperparameter hierarchy in both.</p>"},{"location":"Learning-Environment-Design-Agents/#groups-for-cooperative-scenarios","title":"Groups for Cooperative Scenarios","text":"<p>Cooperative behavior in ML-Agents can be enabled by instantiating a <code>SimpleMultiAgentGroup</code>, typically in an environment controller or similar script, and adding agents to it using the <code>RegisterAgent(Agent agent)</code> method. Note that all agents added to the same <code>SimpleMultiAgentGroup</code> must have the same behavior name and Behavior Parameters. Using <code>SimpleMultiAgentGroup</code> enables the agents within a group to learn how to work together to achieve a common goal (i.e., maximize a group-given reward), even if one or more of the group members are removed before the episode ends. You can then use this group to add/set rewards, end or interrupt episodes at a group level using the <code>AddGroupReward()</code>, <code>SetGroupReward()</code>, <code>EndGroupEpisode()</code>, and <code>GroupEpisodeInterrupted()</code> methods. For example:</p> <pre><code>// Create a Multi Agent Group in Start() or Initialize()\nm_AgentGroup = new SimpleMultiAgentGroup();\n\n// Register agents in group at the beginning of an episode\nfor (var agent in AgentList)\n{\n  m_AgentGroup.RegisterAgent(agent);\n}\n\n// if the team scores a goal\nm_AgentGroup.AddGroupReward(rewardForGoal);\n\n// If the goal is reached and the episode is over\nm_AgentGroup.EndGroupEpisode();\nResetScene();\n\n// If time ran out and we need to interrupt the episode\nm_AgentGroup.GroupEpisodeInterrupted();\nResetScene();\n</code></pre> <p>Multi Agent Groups should be used with the MA-POCA trainer, which is explicitly designed to train cooperative environments. This can be enabled by using the <code>poca</code> trainer - see the training configurations doc for more information on configuring MA-POCA. When using MA-POCA, agents which are deactivated or removed from the Scene during the episode will still learn to contribute to the group's long term rewards, even if they are not active in the scene to experience them.</p> <p>See the Cooperative Push Block environment for an example of how to use Multi Agent Groups, and the Dungeon Escape environment for an example of how the Multi Agent Group can be used with agents that are removed from the scene mid-episode.</p> <p>NOTE: Groups differ from Teams (for competitive settings) in the following way - Agents working together should be added to the same Group, while agents playing against each other should be given different Team Ids. If in the Scene there is one playing field and two teams, there should be two Groups, one for each team, and each team should be assigned a different Team Id. If this playing field is duplicated many times in the Scene (e.g. for training speedup), there should be two Groups per playing field, and two unique Team Ids for the entire Scene. In environments with both Groups and Team Ids configured, MA-POCA and self-play can be used together for training. In the diagram below, there are two agents on each team, and two playing fields where teams are pitted against each other. All the blue agents should share a Team Id (and the orange ones a different ID), and there should be four group managers, one per pair of agents.</p> <p> </p> <p>Please see the SoccerTwos environment for an example.</p>"},{"location":"Learning-Environment-Design-Agents/#cooperative-behaviors-notes-and-best-practices","title":"Cooperative Behaviors Notes and Best Practices","text":"<ul> <li> <p>An agent can only be registered to one MultiAgentGroup at a time. If you want to re-assign an agent from one group to another, you have to unregister it from the current group first.</p> </li> <li> <p>Agents with different behavior names in the same group are not supported.</p> </li> <li> <p>Agents within groups should always set the <code>Max Steps</code> parameter in the Agent script to 0. Instead, handle Max Steps using the MultiAgentGroup by ending the episode for the entire Group using <code>GroupEpisodeInterrupted()</code>.</p> </li> <li> <p><code>EndGroupEpisode</code> and <code>GroupEpisodeInterrupted</code> do the same job in the game, but has slightly different effect on the training. If the episode is completed, you would want to call <code>EndGroupEpisode</code>. But if the episode is not over but it has been running for enough steps, i.e. reaching max step, you would call <code>GroupEpisodeInterrupted</code>.</p> </li> <li> <p>If an agent finished earlier, e.g. completed tasks/be removed/be killed in the game, do not call <code>EndEpisode()</code> on the Agent. Instead, disable the agent and re-enable it when the next episode starts, or destroy the agent entirely. This is because calling <code>EndEpisode()</code> will call <code>OnEpisodeBegin()</code>, which will reset the agent immediately. While it is possible to call <code>EndEpisode()</code> in this way, it is usually not the desired behavior when training groups of agents.</p> </li> <li> <p>If an agent that was disabled in a scene needs to be re-enabled, it must be re-registered to the MultiAgentGroup.</p> </li> <li> <p>Group rewards are meant to reinforce agents to act in the group's best interest instead of individual ones, and are treated differently than individual agent rewards during training. So calling <code>AddGroupReward()</code> is not equivalent to calling agent.AddReward() on each agent in the group.</p> </li> <li> <p>You can still add incremental rewards to agents using <code>Agent.AddReward()</code> if they are in a Group. These rewards will only be given to those agents and are received when the Agent is active.</p> </li> <li> <p>Environments which use Multi Agent Groups can be trained using PPO or SAC, but agents will not be able to learn from group rewards after deactivation/removal, nor will they behave as cooperatively.</p> </li> </ul>"},{"location":"Learning-Environment-Design-Agents/#recording-demonstrations","title":"Recording Demonstrations","text":"<p>In order to record demonstrations from an agent, add the <code>Demonstration Recorder</code> component to a GameObject in the scene which contains an <code>Agent</code> component. Once added, it is possible to name the demonstration that will be recorded from the agent.</p> <p> </p> <p>When <code>Record</code> is checked, a demonstration will be created whenever the scene is played from the Editor. Depending on the complexity of the task, anywhere from a few minutes or a few hours of demonstration data may be necessary to be useful for imitation learning. To specify an exact number of steps you want to record use the <code>Num Steps To Record</code> field and the editor will end your play session automatically once that many steps are recorded. If you set <code>Num Steps To Record</code> to <code>0</code> then recording will continue until you manually end the play session. Once the play session ends a <code>.demo</code> file will be created in the <code>Assets/Demonstrations</code> folder (by default). This file contains the demonstrations. Clicking on the file will provide metadata about the demonstration in the inspector.</p> <p> </p> <p>You can then specify the path to this file in your training configurations.</p>"},{"location":"Learning-Environment-Design/","title":"Designing a Learning Environment","text":"<p>This page contains general advice on how to design your learning environment, in addition to overviewing aspects of the ML-Agents Unity SDK that pertain to setting up your scene and simulation as opposed to designing your agents within the scene. We have a dedicated page on Designing Agents which includes how to instrument observations, actions and rewards, define teams for multi-agent scenarios and record agent demonstrations for imitation learning.</p> <p>To help on-board to the entire set of functionality provided by the ML-Agents Toolkit, we recommend exploring our API documentation. Additionally, our example environments are a great resource as they provide sample usage of almost all of our features.</p>"},{"location":"Learning-Environment-Design/#the-simulation-and-training-process","title":"The Simulation and Training Process","text":"<p>Training and simulation proceed in steps orchestrated by the ML-Agents Academy class. The Academy works with Agent objects in the scene to step through the simulation.</p> <p>During training, the external Python training process communicates with the Academy to run a series of episodes while it collects data and optimizes its neural network model. When training is completed successfully, you can add the trained model file to your Unity project for later use.</p> <p>The ML-Agents Academy class orchestrates the agent simulation loop as follows:</p> <ol> <li>Calls your Academy's <code>OnEnvironmentReset</code> delegate.</li> <li>Calls the <code>OnEpisodeBegin()</code> function for each Agent in the scene.</li> <li>Gathers information about the scene. This is done by calling the   <code>CollectObservations(VectorSensor sensor)</code> function for each Agent in the   scene, as well as updating their sensor and collecting the resulting   observations.</li> <li>Uses each Agent's Policy to decide on the Agent's next action.</li> <li>Calls the <code>OnActionReceived()</code> function for each Agent in the scene, passing    in the action chosen by the Agent's Policy.</li> <li>Calls the Agent's <code>OnEpisodeBegin()</code> function if the Agent has reached its    <code>Max Step</code> count or has otherwise marked itself as <code>EndEpisode()</code>.</li> </ol> <p>To create a training environment, extend the Agent class to implement the above methods whether you need to implement them or not depends on your specific scenario.</p>"},{"location":"Learning-Environment-Design/#organizing-the-unity-scene","title":"Organizing the Unity Scene","text":"<p>To train and use the ML-Agents Toolkit in a Unity scene, the scene as many Agent subclasses as you need. Agent instances should be attached to the GameObject representing that Agent.</p>"},{"location":"Learning-Environment-Design/#academy","title":"Academy","text":"<p>The Academy is a singleton which orchestrates Agents and their decision making processes. Only a single Academy exists at a time.</p>"},{"location":"Learning-Environment-Design/#academy-resetting","title":"Academy resetting","text":"<p>To alter the environment at the start of each episode, add your method to the Academy's OnEnvironmentReset action.</p> <pre><code>public class MySceneBehavior : MonoBehaviour\n{\n    public void Awake()\n    {\n        Academy.Instance.OnEnvironmentReset += EnvironmentReset;\n    }\n\n    void EnvironmentReset()\n    {\n        // Reset the scene here\n    }\n}\n</code></pre> <p>For example, you might want to reset an Agent to its starting position or move a goal to a random position. An environment resets when the <code>reset()</code> method is called on the Python <code>UnityEnvironment</code>.</p> <p>When you reset an environment, consider the factors that should change so that training is generalizable to different conditions. For example, if you were training a maze-solving agent, you would probably want to change the maze itself for each training episode. Otherwise, the agent would probably on learn to solve one, particular maze, not mazes in general.</p>"},{"location":"Learning-Environment-Design/#multiple-areas","title":"Multiple Areas","text":"<p>In many of the example environments, many copies of the training area are instantiated in the scene. This generally speeds up training, allowing the environment to gather many experiences in parallel. This can be achieved simply by instantiating many Agents with the same Behavior Name. If possible, consider designing your scene to support multiple areas.</p> <p>Check out our example environments to see examples of multiple areas. Additionally, the Making a New Learning Environment guide demonstrates this option.</p>"},{"location":"Learning-Environment-Design/#environments","title":"Environments","text":"<p>When you create a training environment in Unity, you must set up the scene so that it can be controlled by the external training process. Considerations include:</p> <ul> <li>The training scene must start automatically when your Unity application is   launched by the training process.</li> <li>The Academy must reset the scene to a valid starting point for each episode of   training.</li> <li>A training episode must have a definite end \u2014 either using <code>Max Steps</code> or by   each Agent ending its episode manually with <code>EndEpisode()</code>.</li> </ul>"},{"location":"Learning-Environment-Design/#environment-parameters","title":"Environment Parameters","text":"<p>Curriculum learning and environment parameter randomization are two training methods that control specific parameters in your environment. As such, it is important to ensure that your environment parameters are updated at each step to the correct values. To enable this, we expose a <code>EnvironmentParameters</code> C# class that you can use to retrieve the values of the parameters defined in the training configurations for both of those features. Please see our documentation for curriculum learning and environment parameter randomization for details.</p> <p>We recommend modifying the environment from the Agent's <code>OnEpisodeBegin()</code> function by leveraging <code>Academy.Instance.EnvironmentParameters</code>. See the WallJump example environment for a sample usage (specifically, WallJumpAgent.cs ).</p>"},{"location":"Learning-Environment-Design/#agent","title":"Agent","text":"<p>The Agent class represents an actor in the scene that collects observations and carries out actions. The Agent class is typically attached to the GameObject in the scene that otherwise represents the actor \u2014 for example, to a player object in a football game or a car object in a vehicle simulation. Every Agent must have appropriate <code>Behavior Parameters</code>.</p> <p>Generally, when creating an Agent, you should extend the Agent class and implement the <code>CollectObservations(VectorSensor sensor)</code> and <code>OnActionReceived()</code> methods:</p> <ul> <li><code>CollectObservations(VectorSensor sensor)</code> \u2014 Collects the Agent's observation   of its environment.</li> <li><code>OnActionReceived()</code> \u2014 Carries out the action chosen by the Agent's Policy and   assigns a reward to the current state.</li> </ul> <p>Your implementations of these functions determine how the Behavior Parameters assigned to this Agent must be set.</p> <p>You must also determine how an Agent finishes its task or times out. You can manually terminate an Agent episode in your <code>OnActionReceived()</code> function when the Agent has finished (or irrevocably failed) its task by calling the <code>EndEpisode()</code> function. You can also set the Agent's <code>Max Steps</code> property to a positive value and the Agent will consider the episode over after it has taken that many steps. You can use the <code>Agent.OnEpisodeBegin()</code> function to prepare the Agent to start again.</p> <p>See Agents for detailed information about programming your own Agents.</p>"},{"location":"Learning-Environment-Design/#recording-statistics","title":"Recording Statistics","text":"<p>We offer developers a mechanism to record statistics from within their Unity environments. These statistics are aggregated and generated during the training process. To record statistics, see the <code>StatsRecorder</code> C# class.</p> <p>See the FoodCollector example environment for a sample usage (specifically, FoodCollectorSettings.cs ).</p>"},{"location":"Learning-Environment-Examples/","title":"Example Learning Environments","text":"<p>The Unity ML-Agents Toolkit includes an expanding set of example environments that highlight the various features of the toolkit. These environments can also serve as templates for new environments or as ways to test new ML algorithms. Environments are located in <code>Project/Assets/ML-Agents/Examples</code> and summarized below.</p> <p>For the environments that highlight specific features of the toolkit, we provide the pre-trained model files and the training config file that enables you to train the scene yourself. The environments that are designed to serve as challenges for researchers do not have accompanying pre-trained model files or training configs and are marked as Optional below.</p> <p>This page only overviews the example environments we provide. To learn more on how to design and build your own environments see our Making a New Learning Environment page. If you would like to contribute environments, please see our contribution guidelines page.</p>"},{"location":"Learning-Environment-Examples/#basic","title":"Basic","text":"<ul> <li>Set-up: A linear movement task where the agent must move left or right to   rewarding states.</li> <li>Goal: Move to the most reward state.</li> <li>Agents: The environment contains one agent.</li> <li>Agent Reward Function:</li> <li>-0.01 at each step</li> <li>+0.1 for arriving at suboptimal state.</li> <li>+1.0 for arriving at optimal state.</li> <li>Behavior Parameters:</li> <li>Vector Observation space: One variable corresponding to current state.</li> <li>Actions: 1 discrete action branch with 3 actions (Move left, do nothing, move     right).</li> <li>Visual Observations: None</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward: 0.93</li> </ul>"},{"location":"Learning-Environment-Examples/#3dball-3d-balance-ball","title":"3DBall: 3D Balance Ball","text":"<ul> <li>Set-up: A balance-ball task, where the agent balances the ball on it's head.</li> <li>Goal: The agent must balance the ball on it's head for as long as possible.</li> <li>Agents: The environment contains 12 agents of the same kind, all using the   same Behavior Parameters.</li> <li>Agent Reward Function:</li> <li>+0.1 for every step the ball remains on it's head.</li> <li>-1.0 if the ball falls off.</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 8 variables corresponding to rotation of the agent     cube, and position and velocity of ball.</li> <li>Vector Observation space (Hard Version): 5 variables corresponding to     rotation of the agent cube and position of ball.</li> <li>Actions: 2 continuous actions, with one value corresponding to     X-rotation, and the other to Z-rotation.</li> <li>Visual Observations: Third-person view from the upper-front of the agent. Use     <code>Visual3DBall</code> scene.</li> <li>Float Properties: Three</li> <li>scale: Specifies the scale of the ball in the 3 dimensions (equal across the     three dimensions)<ul> <li>Default: 1</li> <li>Recommended Minimum: 0.2</li> <li>Recommended Maximum: 5</li> </ul> </li> <li>gravity: Magnitude of gravity<ul> <li>Default: 9.81</li> <li>Recommended Minimum: 4</li> <li>Recommended Maximum: 105</li> </ul> </li> <li>mass: Specifies mass of the ball<ul> <li>Default: 1</li> <li>Recommended Minimum: 0.1</li> <li>Recommended Maximum: 20</li> </ul> </li> <li>Benchmark Mean Reward: 100</li> </ul>"},{"location":"Learning-Environment-Examples/#gridworld","title":"GridWorld","text":"<ul> <li>Set-up: A multi-goal version of the grid-world task. Scene contains agent, goal,   and obstacles.</li> <li>Goal: The agent must navigate the grid to the appropriate goal while   avoiding the obstacles.</li> <li>Agents: The environment contains nine agents with the same Behavior   Parameters.</li> <li>Agent Reward Function:</li> <li>-0.01 for every step.</li> <li>+1.0 if the agent navigates to the correct goal (episode ends).</li> <li>-1.0 if the agent navigates to an incorrect goal (episode ends).</li> <li>Behavior Parameters:</li> <li>Vector Observation space: None</li> <li>Actions: 1 discrete action branch with 5 actions, corresponding to movement in     cardinal directions or not moving. Note that for this environment,     action masking     is turned on by default (this option can be toggled using the <code>Mask Actions</code>     checkbox within the <code>trueAgent</code> GameObject). The trained model file provided     was generated with action masking turned on.</li> <li>Visual Observations: One corresponding to top-down view of GridWorld.</li> <li>Goal Signal : A one hot vector corresponding to which color is the correct goal   for the Agent</li> <li>Float Properties: Three, corresponding to grid size, number of green goals, and   number of red goals.</li> <li>Benchmark Mean Reward: 0.8</li> </ul>"},{"location":"Learning-Environment-Examples/#push-block","title":"Push Block","text":"<ul> <li>Set-up: A platforming environment where the agent can push a block around.</li> <li>Goal: The agent must push the block to the goal.</li> <li>Agents: The environment contains one agent.</li> <li>Agent Reward Function:</li> <li>-0.0025 for every step.</li> <li>+1.0 if the block touches the goal.</li> <li>Behavior Parameters:</li> <li>Vector Observation space: (Continuous) 70 variables corresponding to 14     ray-casts each detecting one of three possible objects (wall, goal, or     block).</li> <li>Actions: 1 discrete action branch with 7 actions, corresponding to turn clockwise     and counterclockwise, move along four different face directions, or do nothing.</li> <li>Float Properties: Four</li> <li>block_scale: Scale of the block along the x and z dimensions<ul> <li>Default: 2</li> <li>Recommended Minimum: 0.5</li> <li>Recommended Maximum: 4</li> </ul> </li> <li>dynamic_friction: Coefficient of friction for the ground material acting on     moving objects<ul> <li>Default: 0</li> <li>Recommended Minimum: 0</li> <li>Recommended Maximum: 1</li> </ul> </li> <li>static_friction: Coefficient of friction for the ground material acting on     stationary objects<ul> <li>Default: 0</li> <li>Recommended Minimum: 0</li> <li>Recommended Maximum: 1</li> </ul> </li> <li>block_drag: Effect of air resistance on block<ul> <li>Default: 0.5</li> <li>Recommended Minimum: 0</li> <li>Recommended Maximum: 2000</li> </ul> </li> <li>Benchmark Mean Reward: 4.5</li> </ul>"},{"location":"Learning-Environment-Examples/#wall-jump","title":"Wall Jump","text":"<ul> <li>Set-up: A platforming environment where the agent can jump over a wall.</li> <li>Goal: The agent must use the block to scale the wall and reach the goal.</li> <li>Agents: The environment contains one agent linked to two different Models. The   Policy the agent is linked to changes depending on the height of the wall. The   change of Policy is done in the WallJumpAgent class.</li> <li>Agent Reward Function:</li> <li>-0.0005 for every step.</li> <li>+1.0 if the agent touches the goal.</li> <li>-1.0 if the agent falls off the platform.</li> <li>Behavior Parameters:</li> <li>Vector Observation space: Size of 74, corresponding to 14 ray casts each     detecting 4 possible objects. plus the global position of the agent and     whether or not the agent is grounded.</li> <li>Actions: 4 discrete action branches:<ul> <li>Forward Motion (3 possible actions: Forward, Backwards, No Action)</li> <li>Rotation (3 possible actions: Rotate Left, Rotate Right, No Action)</li> <li>Side Motion (3 possible actions: Left, Right, No Action)</li> <li>Jump (2 possible actions: Jump, No Action)</li> </ul> </li> <li>Visual Observations: None</li> <li>Float Properties: Four</li> <li>Benchmark Mean Reward (Big &amp; Small Wall): 0.8</li> </ul>"},{"location":"Learning-Environment-Examples/#crawler","title":"Crawler","text":"<ul> <li>Set-up: A creature with 4 arms and 4 forearms.</li> <li>Goal: The agents must move its body toward the goal direction without falling.</li> <li>Agents: The environment contains 10 agents with same Behavior Parameters.</li> <li>Agent Reward Function (independent):   The reward function is now geometric meaning the reward each step is a product   of all the rewards instead of a sum, this helps the agent try to maximize all   rewards instead of the easiest rewards.</li> <li>Body velocity matches goal velocity. (normalized between (0,1))</li> <li>Head direction alignment with goal direction. (normalized between (0,1))</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 172 variables corresponding to position, rotation,     velocity, and angular velocities of each limb plus the acceleration and     angular acceleration of the body.</li> <li>Actions: 20 continuous actions, corresponding to target     rotations for joints.</li> <li>Visual Observations: None</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward: 3000</li> </ul>"},{"location":"Learning-Environment-Examples/#worm","title":"Worm","text":"<ul> <li>Set-up: A worm with a head and 3 body segments.</li> <li>Goal: The agents must move its body toward the goal direction.</li> <li>Agents: The environment contains 10 agents with same Behavior Parameters.</li> <li>Agent Reward Function (independent):   The reward function is now geometric meaning the reward each step is a product   of all the rewards instead of a sum, this helps the agent try to maximize all   rewards instead of the easiest rewards.</li> <li>Body velocity matches goal velocity. (normalized between (0,1))</li> <li>Body direction alignment with goal direction. (normalized between (0,1))</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 64 variables corresponding to position, rotation,     velocity, and angular velocities of each limb plus the acceleration and     angular acceleration of the body.</li> <li>Actions: 9 continuous actions, corresponding to target     rotations for joints.</li> <li>Visual Observations: None</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward: 800</li> </ul>"},{"location":"Learning-Environment-Examples/#food-collector","title":"Food Collector","text":"<ul> <li>Set-up: A multi-agent environment where agents compete to collect food.</li> <li>Goal: The agents must learn to collect as many green food spheres as possible   while avoiding red spheres.</li> <li>Agents: The environment contains 5 agents with same Behavior Parameters.</li> <li>Agent Reward Function (independent):</li> <li>+1 for interaction with green spheres</li> <li>-1 for interaction with red spheres</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 53 corresponding to velocity of agent (2), whether     agent is frozen and/or shot its laser (2), plus grid based perception of     objects around agent's forward direction (40 by 40 with 6 different categories).</li> <li>Actions:<ul> <li>3 continuous actions correspond to Forward Motion, Side Motion and Rotation</li> <li>1 discrete acion branch for Laser with 2 possible actions corresponding to   Shoot Laser or No Action</li> </ul> </li> <li>Visual Observations (Optional): First-person camera per-agent, plus one vector     flag representing the frozen state of the agent. This scene uses a combination     of vector and visual observations and the training will not succeed without     the frozen vector flag. Use <code>VisualFoodCollector</code> scene.</li> <li>Float Properties: Two</li> <li>laser_length: Length of the laser used by the agent<ul> <li>Default: 1</li> <li>Recommended Minimum: 0.2</li> <li>Recommended Maximum: 7</li> </ul> </li> <li>agent_scale: Specifies the scale of the agent in the 3 dimensions (equal     across the three dimensions)<ul> <li>Default: 1</li> <li>Recommended Minimum: 0.5</li> <li>Recommended Maximum: 5</li> </ul> </li> <li>Benchmark Mean Reward: 10</li> </ul>"},{"location":"Learning-Environment-Examples/#hallway","title":"Hallway","text":"<ul> <li>Set-up: Environment where the agent needs to find information in a room,   remember it, and use it to move to the correct goal.</li> <li>Goal: Move to the goal which corresponds to the color of the block in the   room.</li> <li>Agents: The environment contains one agent.</li> <li>Agent Reward Function (independent):</li> <li>+1 For moving to correct goal.</li> <li>-0.1 For moving to incorrect goal.</li> <li>-0.0003 Existential penalty.</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 30 corresponding to local ray-casts detecting     objects, goals, and walls.</li> <li>Actions: 1 discrete action Branch, with 4 actions corresponding to agent     rotation and forward/backward movement.</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward: 0.7</li> <li>To train this environment, you can enable curiosity by adding the <code>curiosity</code> reward signal     in <code>config/ppo/Hallway.yaml</code></li> </ul>"},{"location":"Learning-Environment-Examples/#soccer-twos","title":"Soccer Twos","text":"<ul> <li>Set-up: Environment where four agents compete in a 2 vs 2 toy soccer game.</li> <li>Goal:</li> <li>Get the ball into the opponent's goal while preventing the ball from     entering own goal.</li> <li>Agents: The environment contains two different Multi Agent Groups with two agents in each.   Parameters : SoccerTwos.</li> <li>Agent Reward Function (dependent):</li> <li>(1 - <code>accumulated time penalty</code>) When ball enters opponent's goal     <code>accumulated time penalty</code> is incremented by (1 / <code>MaxStep</code>) every fixed     update and is reset to 0 at the beginning of an episode.</li> <li>-1 When ball enters team's goal.</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 336 corresponding to 11 ray-casts forward     distributed over 120 degrees and 3 ray-casts backward distributed over 90     degrees each detecting 6 possible object types, along with the object's     distance. The forward ray-casts contribute 264 state dimensions and backward     72 state dimensions over three observation stacks.</li> <li>Actions: 3 discrete branched actions corresponding to     forward, backward, sideways movement, as well as rotation.</li> <li>Visual Observations: None</li> <li>Float Properties: Two</li> <li>ball_scale: Specifies the scale of the ball in the 3 dimensions (equal     across the three dimensions)<ul> <li>Default: 7.5</li> <li>Recommended minimum: 4</li> <li>Recommended maximum: 10</li> </ul> </li> <li>gravity: Magnitude of the gravity<ul> <li>Default: 9.81</li> <li>Recommended minimum: 6</li> <li>Recommended maximum: 20</li> </ul> </li> </ul>"},{"location":"Learning-Environment-Examples/#strikers-vs-goalie","title":"Strikers Vs. Goalie","text":"<ul> <li>Set-up: Environment where two agents compete in a 2 vs 1 soccer variant.</li> <li>Goal:</li> <li>Striker: Get the ball into the opponent's goal.</li> <li>Goalie: Keep the ball out of the goal.</li> <li>Agents: The environment contains two different Multi Agent Groups. One with two Strikers and the other one Goalie.   Behavior Parameters : Striker, Goalie.</li> <li>Striker Agent Reward Function (dependent):</li> <li>+1 When ball enters opponent's goal.</li> <li>-0.001 Existential penalty.</li> <li>Goalie Agent Reward Function (dependent):</li> <li>-1 When ball enters goal.</li> <li>0.001 Existential bonus.</li> <li>Behavior Parameters:</li> <li>Striker Vector Observation space: 294 corresponding to 11 ray-casts forward     distributed over 120 degrees and 3 ray-casts backward distributed over 90     degrees each detecting 5 possible object types, along with the object's     distance. The forward ray-casts contribute 231 state dimensions and backward     63 state dimensions over three observation stacks.</li> <li>Striker Actions: 3 discrete branched actions corresponding     to forward, backward, sideways movement, as well as rotation.</li> <li>Goalie Vector Observation space: 738 corresponding to 41 ray-casts     distributed over 360 degrees each detecting 4 possible object types, along     with the object's distance and 3 observation stacks.</li> <li>Goalie Actions: 3 discrete branched actions corresponding     to forward, backward, sideways movement, as well as rotation.</li> <li>Visual Observations: None</li> <li>Float Properties: Two</li> <li>ball_scale: Specifies the scale of the ball in the 3 dimensions (equal     across the three dimensions)<ul> <li>Default: 7.5</li> <li>Recommended minimum: 4</li> <li>Recommended maximum: 10</li> </ul> </li> <li>gravity: Magnitude of the gravity<ul> <li>Default: 9.81</li> <li>Recommended minimum: 6</li> <li>Recommended maximum: 20</li> </ul> </li> </ul>"},{"location":"Learning-Environment-Examples/#walker","title":"Walker","text":"<ul> <li>Set-up: Physics-based Humanoid agents with 26 degrees of freedom. These DOFs   correspond to articulation of the following body-parts: hips, chest, spine,   head, thighs, shins, feet, arms, forearms and hands.</li> <li>Goal: The agents must move its body toward the goal direction without falling.</li> <li>Agents: The environment contains 10 independent agents with same Behavior   Parameters.</li> <li>Agent Reward Function (independent):   The reward function is now geometric meaning the reward each step is a product   of all the rewards instead of a sum, this helps the agent try to maximize all   rewards instead of the easiest rewards.</li> <li>Body velocity matches goal velocity. (normalized between (0,1))</li> <li>Head direction alignment with goal direction. (normalized between (0,1))</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 243 variables corresponding to position, rotation,     velocity, and angular velocities of each limb, along with goal direction.</li> <li>Actions: 39 continuous actions, corresponding to target     rotations and strength applicable to the joints.</li> <li>Visual Observations: None</li> <li>Float Properties: Four</li> <li>gravity: Magnitude of gravity<ul> <li>Default: 9.81</li> <li>Recommended Minimum:</li> <li>Recommended Maximum:</li> </ul> </li> <li>hip_mass: Mass of the hip component of the walker<ul> <li>Default: 8</li> <li>Recommended Minimum: 7</li> <li>Recommended Maximum: 28</li> </ul> </li> <li>chest_mass: Mass of the chest component of the walker<ul> <li>Default: 8</li> <li>Recommended Minimum: 3</li> <li>Recommended Maximum: 20</li> </ul> </li> <li>spine_mass: Mass of the spine component of the walker<ul> <li>Default: 8</li> <li>Recommended Minimum: 3</li> <li>Recommended Maximum: 20</li> </ul> </li> <li>Benchmark Mean Reward : 2500</li> </ul>"},{"location":"Learning-Environment-Examples/#pyramids","title":"Pyramids","text":"<ul> <li>Set-up: Environment where the agent needs to press a button to spawn a   pyramid, then navigate to the pyramid, knock it over, and move to the gold   brick at the top.</li> <li>Goal: Move to the golden brick on top of the spawned pyramid.</li> <li>Agents: The environment contains one agent.</li> <li>Agent Reward Function (independent):</li> <li>+2 For moving to golden brick (minus 0.001 per step).</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 148 corresponding to local ray-casts detecting     switch, bricks, golden brick, and walls, plus variable indicating switch     state.</li> <li>Actions: 1 discrete action branch, with 4 actions corresponding to agent rotation and     forward/backward movement.</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward: 1.75</li> </ul>"},{"location":"Learning-Environment-Examples/#match-3","title":"Match 3","text":"<ul> <li>Set-up: Simple match-3 game. Matched pieces are removed, and remaining pieces drop down. New pieces are spawned randomly at the top, with a chance of being \"special\".</li> <li>Goal: Maximize score from matching pieces.</li> <li>Agents: The environment contains several independent Agents.</li> <li>Agent Reward Function (independent):</li> <li>.01 for each normal piece cleared. Special pieces are worth 2x or 3x.</li> <li>Behavior Parameters:</li> <li>None</li> <li>Observations and actions are defined with a sensor and actuator respectively.</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward:</li> <li>39.5 for visual observations</li> <li>38.5 for vector observations</li> <li>34.2 for simple heuristic (pick a random valid move)</li> <li>37.0 for greedy heuristic (pick the highest-scoring valid move)</li> </ul>"},{"location":"Learning-Environment-Examples/#sorter","title":"Sorter","text":"<ul> <li>Set-up: The Agent is in a circular room with numbered tiles. The values of the  tiles are random between 1 and 20. The tiles present in the room are randomized  at each episode. When the Agent visits a tile, it turns green.</li> <li>Goal: Visit all the tiles in ascending order.</li> <li>Agents: The environment contains a single Agent</li> <li>Agent Reward Function:</li> <li>-.0002 Existential penalty.</li> <li>+1 For visiting the right tile</li> <li>-1 For visiting the wrong tile</li> <li>BehaviorParameters:</li> <li>Vector Observations : 4 : 2 floats for Position and 2 floats for orientation</li> <li>Variable Length Observations : Between 1 and 20 entities (one for each tile)   each with 22 observations, the first 20 are one hot encoding of the value of the tile,   the 21st and 22nd represent the position of the tile relative to the Agent and the 23rd   is <code>1</code> if the tile was visited and <code>0</code> otherwise.</li> <li>Actions: 3 discrete branched actions corresponding to forward, backward,   sideways movement, as well as rotation.</li> <li>Float Properties: One<ul> <li>num_tiles: The maximum number of tiles to sample.</li> <li>Default: 2</li> <li>Recommended Minimum: 1</li> <li>Recommended Maximum: 20</li> </ul> </li> <li>Benchmark Mean Reward: Depends on the number of tiles.</li> </ul>"},{"location":"Learning-Environment-Examples/#cooperative-push-block","title":"Cooperative Push Block","text":"<ul> <li>Set-up: Similar to Push Block, the agents are in an area with blocks that need to be pushed into a goal. Small blocks can be pushed by one agents and are worth +1 value, medium blocks require two agents to push in and are worth +2, and large blocks require all 3 agents to push and are worth +3.</li> <li>Goal: Push all blocks into the goal.</li> <li>Agents: The environment contains three Agents in a Multi Agent Group.</li> <li>Agent Reward Function:</li> <li>-0.0001 Existential penalty, as a group reward.</li> <li>+1, +2, or +3 for pushing in a block, added as a group reward.</li> <li>Behavior Parameters:</li> <li>Observation space: A single Grid Sensor with separate tags for each block size,     the goal, the walls, and other agents.</li> <li>Actions: 1 discrete action branch with 7 actions, corresponding to turn clockwise     and counterclockwise, move along four different face directions, or do nothing.</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward: 11 (Group Reward)</li> </ul>"},{"location":"Learning-Environment-Examples/#dungeon-escape","title":"Dungeon Escape","text":"<ul> <li>Set-up: Agents are trapped in a dungeon with a dragon, and must work together to escape.   To retrieve the key, one of the agents must find and slay the dragon, sacrificing itself   to do so. The dragon will drop a key for the others to use. The other agents can then pick   up this key and unlock the dungeon door. If the agents take too long, the dragon will escape   through a portal and the environment resets.</li> <li>Goal: Unlock the dungeon door and leave.</li> <li>Agents: The environment contains three Agents in a Multi Agent Group and one Dragon, which   moves in a predetermined pattern.</li> <li>Agent Reward Function:</li> <li>+1 group reward if any agent successfully unlocks the door and leaves the dungeon.</li> <li>Behavior Parameters:</li> <li>Observation space: A Ray Perception Sensor with separate tags for the walls, other agents,     the door, key, the dragon, and the dragon's portal. A single Vector Observation which indicates     whether the agent is holding a key.</li> <li>Actions: 1 discrete action branch with 7 actions, corresponding to turn clockwise     and counterclockwise, move along four different face directions, or do nothing.</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward: 1.0 (Group Reward)</li> </ul>"},{"location":"Learning-Environment-Executable/","title":"Using an Environment Executable","text":"<p>This section will help you create and use built environments rather than the Editor to interact with an environment. Using an executable has some advantages over using the Editor:</p> <ul> <li>You can exchange executable with other people without having to share your   entire repository.</li> <li>You can put your executable on a remote machine for faster training.</li> <li>You can use <code>Server Build</code> (<code>Headless</code>) mode for faster training (as long as the executable does not need rendering).</li> <li>You can keep using the Unity Editor for other tasks while the agents are   training.</li> </ul>"},{"location":"Learning-Environment-Executable/#building-the-3dball-environment","title":"Building the 3DBall environment","text":"<p>The first step is to open the Unity scene containing the 3D Balance Ball environment:</p> <ol> <li>Launch Unity.</li> <li>On the Projects dialog, choose the Open option at the top of the window.</li> <li>Using the file dialog that opens, locate the <code>Project</code> folder within the    ML-Agents project and click Open.</li> <li>In the Project window, navigate to the folder    <code>Assets/ML-Agents/Examples/3DBall/Scenes/</code>.</li> <li>Double-click the <code>3DBall</code> file to load the scene containing the Balance Ball    environment.</li> </ol> <p></p> <p>Next, we want the set up scene to play correctly when the training process launches our environment executable. This means:</p> <ul> <li>The environment application runs in the background.</li> <li>No dialogs require interaction.</li> <li> <p>The correct scene loads automatically.</p> </li> <li> <p>Open Player Settings (menu: Edit &gt; Project Settings &gt; Player).</p> </li> <li>Under Resolution and Presentation:</li> <li>Ensure that Run in Background is Checked.</li> <li>Ensure that Display Resolution Dialog is set to Disabled. (Note: this     setting may not be available in newer versions of the editor.)</li> <li>Open the Build Settings window (menu:File &gt; Build Settings).</li> <li>Choose your target platform.</li> <li>(optional) Select \u201cDevelopment Build\u201d to      log debug messages.</li> <li>If any scenes are shown in the Scenes in Build list, make sure that the    3DBall Scene is the only one checked. (If the list is empty, then only the    current scene is included in the build).</li> <li>Click Build:</li> <li>In the File dialog, navigate to your ML-Agents directory.</li> <li>Assign a file name and click Save.</li> <li>(For Windows\uff09With Unity 2018.1, it will ask you to select a folder instead      of a file name. Create a subfolder within the root directory and select      that folder to build. In the following steps you will refer to this      subfolder's name as <code>env_name</code>. You cannot create builds in the Assets      folder</li> </ul> <p></p> <p>Now that we have a Unity executable containing the simulation environment, we can interact with it.</p>"},{"location":"Learning-Environment-Executable/#interacting-with-the-environment","title":"Interacting with the Environment","text":"<p>If you want to use the Python API to interact with your executable, you can pass the name of the executable with the argument 'file_name' of the <code>UnityEnvironment</code>. For instance:</p> <pre><code>from mlagents_envs.environment import UnityEnvironment\nenv = UnityEnvironment(file_name=&lt;env_name&gt;)\n</code></pre>"},{"location":"Learning-Environment-Executable/#training-the-environment","title":"Training the Environment","text":"<ol> <li>Open a command or terminal window.</li> <li>Navigate to the folder where you installed the ML-Agents Toolkit. If you    followed the default installation, then navigate to the    <code>ml-agents/</code> folder.</li> <li>Run    <code>mlagents-learn &lt;trainer-config-file&gt; --env=&lt;env_name&gt; --run-id=&lt;run-identifier&gt;</code>    Where:</li> <li><code>&lt;trainer-config-file&gt;</code> is the file path of the trainer configuration yaml</li> <li><code>&lt;env_name&gt;</code> is the name and path to the executable you exported from Unity      (without extension)</li> <li><code>&lt;run-identifier&gt;</code> is a string used to separate the results of different      training runs</li> </ol> <p>For example, if you are training with a 3DBall executable, and you saved it to the directory where you installed the ML-Agents Toolkit, run:</p> <pre><code>mlagents-learn config/ppo/3DBall.yaml --env=3DBall --run-id=firstRun\n</code></pre> <p>And you should see something like</p> <pre><code>ml-agents$ mlagents-learn config/ppo/3DBall.yaml --env=3DBall --run-id=first-run\n\n\n                        \u2584\u2584\u2584\u2593\u2593\u2593\u2593\n                   \u2553\u2593\u2593\u2593\u2593\u2593\u2593\u2588\u2593\u2593\u2593\u2593\u2593\n              ,\u2584\u2584\u2584m\u2580\u2580\u2580'  ,\u2593\u2593\u2593\u2580\u2593\u2593\u2584                           \u2593\u2593\u2593  \u2593\u2593\u258c\n            \u2584\u2593\u2593\u2593\u2580'      \u2584\u2593\u2593\u2580  \u2593\u2593\u2593      \u2584\u2584     \u2584\u2584 ,\u2584\u2584 \u2584\u2584\u2584\u2584   ,\u2584\u2584 \u2584\u2593\u2593\u258c\u2584 \u2584\u2584\u2584    ,\u2584\u2584\n          \u2584\u2593\u2593\u2593\u2580        \u2584\u2593\u2593\u2580   \u2590\u2593\u2593\u258c     \u2593\u2593\u258c   \u2590\u2593\u2593 \u2590\u2593\u2593\u2593\u2580\u2580\u2580\u2593\u2593\u258c \u2593\u2593\u2593 \u2580\u2593\u2593\u258c\u2580 ^\u2593\u2593\u258c  \u2552\u2593\u2593\u258c\n        \u2584\u2593\u2593\u2593\u2593\u2593\u2584\u2584\u2584\u2584\u2584\u2584\u2584\u2584\u2593\u2593\u2593      \u2593\u2580      \u2593\u2593\u258c   \u2590\u2593\u2593 \u2590\u2593\u2593    \u2593\u2593\u2593 \u2593\u2593\u2593  \u2593\u2593\u258c   \u2590\u2593\u2593\u2584 \u2593\u2593\u258c\n        \u2580\u2593\u2593\u2593\u2593\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2593\u2593\u2584     \u2593\u2593      \u2593\u2593\u258c   \u2590\u2593\u2593 \u2590\u2593\u2593    \u2593\u2593\u2593 \u2593\u2593\u2593  \u2593\u2593\u258c    \u2590\u2593\u2593\u2590\u2593\u2593\n          ^\u2588\u2593\u2593\u2593        \u2580\u2593\u2593\u2584   \u2590\u2593\u2593\u258c     \u2593\u2593\u2593\u2593\u2584\u2593\u2593\u2593\u2593 \u2590\u2593\u2593    \u2593\u2593\u2593 \u2593\u2593\u2593  \u2593\u2593\u2593\u2584    \u2593\u2593\u2593\u2593`\n            '\u2580\u2593\u2593\u2593\u2584      ^\u2593\u2593\u2593  \u2593\u2593\u2593       \u2514\u2580\u2580\u2580\u2580 \u2580\u2580 ^\u2580\u2580    `\u2580\u2580 `\u2580\u2580   '\u2580\u2580    \u2590\u2593\u2593\u258c\n               \u2580\u2580\u2580\u2580\u2593\u2584\u2584\u2584   \u2593\u2593\u2593\u2593\u2593\u2593,                                      \u2593\u2593\u2593\u2593\u2580\n                   `\u2580\u2588\u2593\u2593\u2593\u2593\u2593\u2593\u2593\u2593\u2593\u258c\n                        \u00ac`\u2580\u2580\u2580\u2588\u2593\n\n</code></pre> <p>Note: If you're using Anaconda, don't forget to activate the ml-agents environment first.</p> <p>If <code>mlagents-learn</code> runs correctly and starts training, you should see something like this:</p> <pre><code>CrashReporter: initialized\nMono path[0] = '/Users/dericp/workspace/ml-agents/3DBall.app/Contents/Resources/Data/Managed'\nMono config path = '/Users/dericp/workspace/ml-agents/3DBall.app/Contents/MonoBleedingEdge/etc'\nINFO:mlagents_envs:\n'Ball3DAcademy' started successfully!\nUnity Academy name: Ball3DAcademy\n\nINFO:mlagents_envs:Connected new brain:\nUnity brain name: Ball3DLearning\n        Number of Visual Observations (per agent): 0\n        Vector Observation space size (per agent): 8\n        Number of stacked Vector Observation: 1\nINFO:mlagents_envs:Hyperparameters for the PPO Trainer of brain Ball3DLearning:\n        batch_size:          64\n        beta:                0.001\n        buffer_size:         12000\n        epsilon:             0.2\n        gamma:               0.995\n        hidden_units:        128\n        lambd:               0.99\n        learning_rate:       0.0003\n        max_steps:           5.0e4\n        normalize:           True\n        num_epoch:           3\n        num_layers:          2\n        time_horizon:        1000\n        sequence_length:     64\n        summary_freq:        1000\n        use_recurrent:       False\n        memory_size:         256\n        use_curiosity:       False\n        curiosity_strength:  0.01\n        curiosity_enc_size:  128\n        output_path: ./results/first-run-0/Ball3DLearning\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 1000. Mean Reward: 1.242. Std of Reward: 0.746. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 2000. Mean Reward: 1.319. Std of Reward: 0.693. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 3000. Mean Reward: 1.804. Std of Reward: 1.056. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 4000. Mean Reward: 2.151. Std of Reward: 1.432. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 5000. Mean Reward: 3.175. Std of Reward: 2.250. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 6000. Mean Reward: 4.898. Std of Reward: 4.019. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 7000. Mean Reward: 6.716. Std of Reward: 5.125. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 8000. Mean Reward: 12.124. Std of Reward: 11.929. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 9000. Mean Reward: 18.151. Std of Reward: 16.871. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 10000. Mean Reward: 27.284. Std of Reward: 28.667. Training.\n</code></pre> <p>You can press Ctrl+C to stop the training, and your trained model will be at <code>results/&lt;run-identifier&gt;/&lt;behavior_name&gt;.onnx</code>, which corresponds to your model's latest checkpoint. (Note: There is a known bug on Windows that causes the saving of the model to fail when you early terminate the training, it's recommended to wait until Step has reached the max_steps parameter you set in your config YAML.) You can now embed this trained model into your Agent by following the steps below:</p> <ol> <li>Move your model file into    <code>Project/Assets/ML-Agents/Examples/3DBall/TFModels/</code>.</li> <li>Open the Unity Editor, and select the 3DBall scene as described above.</li> <li>Select the 3DBall prefab from the Project window and select Agent.</li> <li>Drag the <code>&lt;behavior_name&gt;.onnx</code> file from the Project window of the Editor to    the Model placeholder in the Ball3DAgent inspector window.</li> <li>Press the Play button at the top of the Editor.</li> </ol>"},{"location":"Learning-Environment-Executable/#training-on-headless-server","title":"Training on Headless Server","text":"<p>To run training on headless server with no graphics rendering support, you need to turn off graphics display in the Unity executable. There are two ways to achieve this: 1. Pass <code>--no-graphics</code> option to mlagents-learn training command. This is equivalent to    adding <code>-nographics -batchmode</code> to the Unity executable's commandline. 2. Build your Unity executable with Server Build. You can find this setting in Build Settings    in the Unity Editor.</p> <p>If you want to train with graphics (for example, using camera and visual observations), you'll need to set up display rendering support (e.g. xvfb) on you server machine. In our Colab Notebook Tutorials, the Setup section has examples of setting up xvfb on servers.</p>"},{"location":"Limitations/","title":"Limitations","text":"<p>See the package-specific Limitations pages:</p> <ul> <li><code>com.unity.mlagents</code> Unity package</li> <li><code>mlagents</code> Python package</li> <li><code>mlagents_envs</code> Python package</li> </ul>"},{"location":"ML-Agents-Envs-README/","title":"Unity ML-Agents Python Interface","text":"<p>The <code>mlagents_envs</code> Python package is part of the ML-Agents Toolkit. <code>mlagents_envs</code> provides three Python APIs that allows direct interaction with the Unity game engine: - A single agent API (Gym API) - A gym-like multi-agent API (PettingZoo API) - A low-level API (LLAPI)</p> <p>The LLAPI is used by the trainer implementation in <code>mlagents</code>. <code>mlagents_envs</code> can be used independently of <code>mlagents</code> for Python communication.</p>"},{"location":"ML-Agents-Envs-README/#installation","title":"Installation","text":"<p>Install the <code>mlagents_envs</code> package with:</p> <pre><code>python -m pip install mlagents_envs==1.0.0\n</code></pre>"},{"location":"ML-Agents-Envs-README/#usage-more-information","title":"Usage &amp; More Information","text":"<p>See - Gym API Guide - PettingZoo API Guide - Python API Guide</p> <p>for more information on how to use the API to interact with a Unity environment.</p> <p>For more information on the ML-Agents Toolkit and how to instrument a Unity scene with the ML-Agents SDK, check out the main ML-Agents Toolkit documentation.</p>"},{"location":"ML-Agents-Envs-README/#limitations","title":"Limitations","text":"<ul> <li><code>mlagents_envs</code> uses localhost ports to exchange data between Unity and   Python. As such, multiple instances can have their ports collide, leading to   errors. Make sure to use a different port if you are using multiple instances   of <code>UnityEnvironment</code>.</li> <li>Communication between Unity and the Python <code>UnityEnvironment</code> is not secure.</li> <li>On Linux, ports are not released immediately after the communication closes.   As such, you cannot reuse ports right after closing a <code>UnityEnvironment</code>.</li> </ul>"},{"location":"ML-Agents-Overview/","title":"ML-Agents Toolkit Overview","text":"<p>Table of Contents</p> <ul> <li>Running Example: Training NPC Behaviors</li> <li>Key Components</li> <li>Training Modes</li> <li>Built-in Training and Inference<ul> <li>Cross-Platform Inference</li> </ul> </li> <li>Custom Training and Inference</li> <li>Flexible Training Scenarios</li> <li>Training Methods: Environment-agnostic</li> <li>A Quick Note on Reward Signals</li> <li>Deep Reinforcement Learning<ul> <li>Curiosity for Sparse-reward Environments</li> <li>RND for Sparse-reward Environments</li> </ul> </li> <li>Imitation Learning<ul> <li>GAIL (Generative Adversarial Imitation Learning)</li> <li>Behavioral Cloning (BC)</li> <li>Recording Demonstrations</li> </ul> </li> <li>Summary</li> <li>Training Methods: Environment-specific</li> <li>Training in Competitive Multi-Agent Environments with Self-Play</li> <li>Training in Cooperative Multi-Agent Environments with MA-POCA</li> <li>Solving Complex Tasks using Curriculum Learning</li> <li>Training Robust Agents using Environment Parameter Randomization</li> <li>Model Types</li> <li>Learning from Vector Observations</li> <li>Learning from Cameras using Convolutional Neural Networks</li> <li>Learning from Variable Length Observations using Attention</li> <li>Memory-enhanced Agents using Recurrent Neural Networks</li> <li>Additional Features</li> <li>Summary and Next Steps</li> </ul> <p>The Unity Machine Learning Agents Toolkit (ML-Agents Toolkit) is an open-source project that enables games and simulations to serve as environments for training intelligent agents. Agents can be trained using reinforcement learning, imitation learning, neuroevolution, or other machine learning methods through a simple-to-use Python API. We also provide implementations (based on PyTorch) of state-of-the-art algorithms to enable game developers and hobbyists to easily train intelligent agents for 2D, 3D and VR/AR games. These trained agents can be used for multiple purposes, including controlling NPC behavior (in a variety of settings such as multi-agent and adversarial), automated testing of game builds and evaluating different game design decisions pre-release. The ML-Agents Toolkit is mutually beneficial for both game developers and AI researchers as it provides a central platform where advances in AI can be evaluated on Unity\u2019s rich environments and then made accessible to the wider research and game developer communities.</p> <p>Depending on your background (i.e. researcher, game developer, hobbyist), you may have very different questions on your mind at the moment. To make your transition to the ML-Agents Toolkit easier, we provide several background pages that include overviews and helpful resources on the Unity Engine, machine learning and PyTorch. We strongly recommend browsing the relevant background pages if you're not familiar with a Unity scene, basic machine learning concepts or have not previously heard of PyTorch.</p> <p>The remainder of this page contains a deep dive into ML-Agents, its key components, different training modes and scenarios. By the end of it, you should have a good sense of what the ML-Agents Toolkit allows you to do. The subsequent documentation pages provide examples of how to use ML-Agents. To get started, watch this demo video of ML-Agents in action.</p>"},{"location":"ML-Agents-Overview/#running-example-training-npc-behaviors","title":"Running Example: Training NPC Behaviors","text":"<p>To help explain the material and terminology in this page, we'll use a hypothetical, running example throughout. We will explore the problem of training the behavior of a non-playable character (NPC) in a game. (An NPC is a game character that is never controlled by a human player and its behavior is pre-defined by the game developer.) More specifically, let's assume we're building a multi-player, war-themed game in which players control the soldiers. In this game, we have a single NPC who serves as a medic, finding and reviving wounded players. Lastly, let us assume that there are two teams, each with five players and one NPC medic.</p> <p>The behavior of a medic is quite complex. It first needs to avoid getting injured, which requires detecting when it is in danger and moving to a safe location. Second, it needs to be aware of which of its team members are injured and require assistance. In the case of multiple injuries, it needs to assess the degree of injury and decide who to help first. Lastly, a good medic will always place itself in a position where it can quickly help its team members. Factoring in all of these traits means that at every instance, the medic needs to measure several attributes of the environment (e.g. position of team members, position of enemies, which of its team members are injured and to what degree) and then decide on an action (e.g. hide from enemy fire, move to help one of its members). Given the large number of settings of the environment and the large number of actions that the medic can take, defining and implementing such complex behaviors by hand is challenging and prone to errors.</p> <p>With ML-Agents, it is possible to train the behaviors of such NPCs (called Agents) using a variety of methods. The basic idea is quite simple. We need to define three entities at every moment of the game (called environment):</p> <ul> <li>Observations - what the medic perceives about the environment.   Observations can be numeric and/or visual. Numeric observations measure   attributes of the environment from the point of view of the agent. For our   medic this would be attributes of the battlefield that are visible to it. For   most interesting environments, an agent will require several continuous   numeric observations. Visual observations, on the other hand, are images   generated from the cameras attached to the agent and represent what the agent   is seeing at that point in time. It is common to confuse an agent's   observation with the environment (or game) state. The environment state   represents information about the entire scene containing all the game   characters. The agents observation, however, only contains information that   the agent is aware of and is typically a subset of the environment state. For   example, the medic observation cannot include information about an enemy in   hiding that the medic is unaware of.</li> <li>Actions - what actions the medic can take. Similar to observations,   actions can either be continuous or discrete depending on the complexity of   the environment and agent. In the case of the medic, if the environment is a   simple grid world where only their location matters, then a discrete action   taking on one of four values (north, south, east, west) suffices. However, if   the environment is more complex and the medic can move freely then using two   continuous actions (one for direction and another for speed) is more   appropriate.</li> <li>Reward signals - a scalar value indicating how well the medic is doing.   Note that the reward signal need not be provided at every moment, but only   when the medic performs an action that is good or bad. For example, it can   receive a large negative reward if it dies, a modest positive reward whenever   it revives a wounded team member, and a modest negative reward when a wounded   team member dies due to lack of assistance. Note that the reward signal is how   the objectives of the task are communicated to the agent, so they need to be   set up in a manner where maximizing reward generates the desired optimal   behavior.</li> </ul> <p>After defining these three entities (the building blocks of a reinforcement learning task), we can now train the medic's behavior. This is achieved by simulating the environment for many trials where the medic, over time, learns what is the optimal action to take for every observation it measures by maximizing its future reward. The key is that by learning the actions that maximize its reward, the medic is learning the behaviors that make it a good medic (i.e. one who saves the most number of lives). In reinforcement learning terminology, the behavior that is learned is called a policy, which is essentially a (optimal) mapping from observations to actions. Note that the process of learning a policy through running simulations is called the training phase, while playing the game with an NPC that is using its learned policy is called the inference phase.</p> <p>The ML-Agents Toolkit provides all the necessary tools for using Unity as the simulation engine for learning the policies of different objects in a Unity environment. In the next few sections, we discuss how the ML-Agents Toolkit achieves this and what features it provides.</p>"},{"location":"ML-Agents-Overview/#key-components","title":"Key Components","text":"<p>The ML-Agents Toolkit contains five high-level components:</p> <ul> <li>Learning Environment - which contains the Unity scene and all the game   characters. The Unity scene provides the environment in which agents observe,   act, and learn. How you set up the Unity scene to serve as a learning   environment really depends on your goal. You may be trying to solve a specific   reinforcement learning problem of limited scope, in which case you can use the   same scene for both training and for testing trained agents. Or, you may be   training agents to operate in a complex game or simulation. In this case, it   might be more efficient and practical to create a purpose-built training   scene. The ML-Agents Toolkit includes an ML-Agents Unity SDK   (<code>com.unity.ml-agents</code> package) that enables you to transform any Unity scene   into a learning environment by defining the agents and their behaviors.</li> <li>Python Low-Level API - which contains a low-level Python interface for   interacting and manipulating a learning environment. Note that, unlike the   Learning Environment, the Python API is not part of Unity, but lives outside   and communicates with Unity through the Communicator. This API is contained in   a dedicated <code>mlagents_envs</code> Python package and is used by the Python training   process to communicate with and control the Academy during training. However,   it can be used for other purposes as well. For example, you could use the API   to use Unity as the simulation engine for your own machine learning   algorithms. See Python API for more information.</li> <li>External Communicator - which connects the Learning Environment with the   Python Low-Level API. It lives within the Learning Environment.</li> <li>Python Trainers which contains all the machine learning algorithms that   enable training agents. The algorithms are implemented in Python and are part   of their own <code>mlagents</code> Python package. The package exposes a single   command-line utility <code>mlagents-learn</code> that supports all the training methods   and options outlined in this document. The Python Trainers interface solely   with the Python Low-Level API.</li> <li>Gym Wrapper (not pictured). A common way in which machine learning   researchers interact with simulation environments is via a wrapper provided by   OpenAI called gym. We provide a gym wrapper   in the <code>ml-agents-envs</code> package and instructions for using   it with existing machine learning algorithms which utilize gym.</li> <li>PettingZoo Wrapper (not pictured) PettingZoo is python API for   interacting with multi-agent simulation environments that provides a   gym-like interface. We provide a PettingZoo wrapper for Unity ML-Agents   environments in the <code>ml-agents-envs</code> package and   instructions for using it with machine learning   algorithms.</li> </ul> <p> </p> <p>Simplified block diagram of ML-Agents.</p> <p>The Learning Environment contains two Unity Components that help organize the Unity scene:</p> <ul> <li>Agents - which is attached to a Unity GameObject (any character within a   scene) and handles generating its observations, performing the actions it   receives and assigning a reward (positive / negative) when appropriate. Each   Agent is linked to a Behavior.</li> <li>Behavior - defines specific attributes of the agent such as the number of   actions that agent can take. Each Behavior is uniquely identified by a   <code>Behavior Name</code> field. A Behavior can be thought as a function that receives   observations and rewards from the Agent and returns actions. A Behavior can be   of one of three types: Learning, Heuristic or Inference. A Learning Behavior   is one that is not, yet, defined but about to be trained. A Heuristic Behavior   is one that is defined by a hard-coded set of rules implemented in code. An   Inference Behavior is one that includes a trained Neural Network file. In   essence, after a Learning Behavior is trained, it becomes an Inference   Behavior.</li> </ul> <p>Every Learning Environment will always have one Agent for every character in the scene. While each Agent must be linked to a Behavior, it is possible for Agents that have similar observations and actions to have the same Behavior. In our sample game, we have two teams each with their own medic. Thus we will have two Agents in our Learning Environment, one for each medic, but both of these medics can have the same Behavior. This does not mean that at each instance they will have identical observation and action values.</p> <p> </p> <p>Example block diagram of ML-Agents Toolkit for our sample game.</p> <p>Note that in a single environment, there can be multiple Agents and multiple Behaviors at the same time. For example, if we expanded our game to include tank driver NPCs, then the Agent attached to those characters cannot share its Behavior with the Agent linked to the medics (medics and drivers have different actions). The Learning Environment through the Academy (not represented in the diagram) ensures that all the Agents are in sync in addition to controlling environment-wide settings.</p> <p>Lastly, it is possible to exchange data between Unity and Python outside of the machine learning loop through Side Channels. One example of using Side Channels is to exchange data with Python about Environment Parameters. The following diagram illustrates the above.</p> <p> </p>"},{"location":"ML-Agents-Overview/#training-modes","title":"Training Modes","text":"<p>Given the flexibility of ML-Agents, there are a few ways in which training and inference can proceed.</p>"},{"location":"ML-Agents-Overview/#built-in-training-and-inference","title":"Built-in Training and Inference","text":"<p>As mentioned previously, the ML-Agents Toolkit ships with several implementations of state-of-the-art algorithms for training intelligent agents. More specifically, during training, all the medics in the scene send their observations to the Python API through the External Communicator. The Python API processes these observations and sends back actions for each medic to take. During training these actions are mostly exploratory to help the Python API learn the best policy for each medic. Once training concludes, the learned policy for each medic can be exported as a model file. Then during the inference phase, the medics still continue to generate their observations, but instead of being sent to the Python API, they will be fed into their (internal, embedded) model to generate the optimal action for each medic to take at every point in time.</p> <p>The Getting Started Guide tutorial covers this training mode with the 3D Balance Ball sample environment.</p>"},{"location":"ML-Agents-Overview/#cross-platform-inference","title":"Cross-Platform Inference","text":"<p>It is important to note that the ML-Agents Toolkit leverages the Sentis to run the models within a Unity scene such that an agent can take the optimal action at each step. Given that Sentis support most platforms that Unity does, this means that any model you train with the ML-Agents Toolkit can be embedded into your Unity application that runs on any platform. See our dedicated blog post for additional information.</p>"},{"location":"ML-Agents-Overview/#custom-training-and-inference","title":"Custom Training and Inference","text":"<p>In the previous mode, the Agents were used for training to generate a PyTorch model that the Agents can later use. However, any user of the ML-Agents Toolkit can leverage their own algorithms for training. In this case, the behaviors of all the Agents in the scene will be controlled within Python. You can even turn your environment into a gym.</p> <p>We do not currently have a tutorial highlighting this mode, but you can learn more about the Python API here.</p>"},{"location":"ML-Agents-Overview/#flexible-training-scenarios","title":"Flexible Training Scenarios","text":"<p>While the discussion so-far has mostly focused on training a single agent, with ML-Agents, several training scenarios are possible. We are excited to see what kinds of novel and fun environments the community creates. For those new to training intelligent agents, below are a few examples that can serve as inspiration:</p> <ul> <li>Single-Agent. A single agent, with its own reward signal. The traditional way   of training an agent. An example is any single-player game, such as Chicken.</li> <li>Simultaneous Single-Agent. Multiple independent agents with independent reward   signals with same <code>Behavior Parameters</code>. A parallelized version of the   traditional training scenario, which can speed-up and stabilize the training   process. Helpful when you have multiple versions of the same character in an   environment who should learn similar behaviors. An example might be training a   dozen robot-arms to each open a door simultaneously.</li> <li>Adversarial Self-Play. Two interacting agents with inverse reward signals. In   two-player games, adversarial self-play can allow an agent to become   increasingly more skilled, while always having the perfectly matched opponent:   itself. This was the strategy employed when training AlphaGo, and more   recently used by OpenAI to train a human-beating 1-vs-1 Dota 2 agent.</li> <li>Cooperative Multi-Agent. Multiple interacting agents with a shared reward   signal with same or different <code>Behavior Parameters</code>. In this scenario, all   agents must work together to accomplish a task that cannot be done alone.   Examples include environments where each agent only has access to partial   information, which needs to be shared in order to accomplish the task or   collaboratively solve a puzzle.</li> <li>Competitive Multi-Agent. Multiple interacting agents with inverse reward   signals with same or different <code>Behavior Parameters</code>. In this scenario, agents   must compete with one another to either win a competition, or obtain some   limited set of resources. All team sports fall into this scenario.</li> <li>Ecosystem. Multiple interacting agents with independent reward signals with   same or different <code>Behavior Parameters</code>. This scenario can be thought of as   creating a small world in which animals with different goals all interact,   such as a savanna in which there might be zebras, elephants and giraffes, or   an autonomous driving simulation within an urban environment.</li> </ul>"},{"location":"ML-Agents-Overview/#training-methods-environment-agnostic","title":"Training Methods: Environment-agnostic","text":"<p>The remaining sections overview the various state-of-the-art machine learning algorithms that are part of the ML-Agents Toolkit. If you aren't studying machine and reinforcement learning as a subject and just want to train agents to accomplish tasks, you can treat these algorithms as black boxes. There are a few training-related parameters to adjust inside Unity as well as on the Python training side, but you do not need in-depth knowledge of the algorithms themselves to successfully create and train agents. Step-by-step procedures for running the training process are provided in the Training ML-Agents page.</p> <p>This section specifically focuses on the training methods that are available regardless of the specifics of your learning environment.</p>"},{"location":"ML-Agents-Overview/#a-quick-note-on-reward-signals","title":"A Quick Note on Reward Signals","text":"<p>In this section we introduce the concepts of intrinsic and extrinsic rewards, which helps explain some of the training methods.</p> <p>In reinforcement learning, the end goal for the Agent is to discover a behavior (a Policy) that maximizes a reward. You will need to provide the agent one or more reward signals to use during training. Typically, a reward is defined by your environment, and corresponds to reaching some goal. These are what we refer to as extrinsic rewards, as they are defined external of the learning algorithm.</p> <p>Rewards, however, can be defined outside of the environment as well, to encourage the agent to behave in certain ways, or to aid the learning of the true extrinsic reward. We refer to these rewards as intrinsic reward signals. The total reward that the agent will learn to maximize can be a mix of extrinsic and intrinsic reward signals.</p> <p>The ML-Agents Toolkit allows reward signals to be defined in a modular way, and we provide four reward signals that can the mixed and matched to help shape your agent's behavior:</p> <ul> <li><code>extrinsic</code>: represents the rewards defined in your environment, and is   enabled by default</li> <li><code>gail</code>: represents an intrinsic reward signal that is defined by GAIL (see   below)</li> <li><code>curiosity</code>: represents an intrinsic reward signal that encourages exploration   in sparse-reward environments that is defined by the Curiosity module (see   below).</li> <li><code>rnd</code>: represents an intrinsic reward signal that encourages exploration   in sparse-reward environments that is defined by the Curiosity module (see   below).</li> </ul>"},{"location":"ML-Agents-Overview/#deep-reinforcement-learning","title":"Deep Reinforcement Learning","text":"<p>ML-Agents provide an implementation of two reinforcement learning algorithms:</p> <ul> <li>Proximal Policy Optimization (PPO)</li> <li>Soft Actor-Critic (SAC)</li> </ul> <p>The default algorithm is PPO. This is a method that has been shown to be more general purpose and stable than many other RL algorithms.</p> <p>In contrast with PPO, SAC is off-policy, which means it can learn from experiences collected at any time during the past. As experiences are collected, they are placed in an experience replay buffer and randomly drawn during training. This makes SAC significantly more sample-efficient, often requiring 5-10 times less samples to learn the same task as PPO. However, SAC tends to require more model updates. SAC is a good choice for heavier or slower environments (about 0.1 seconds per step or more). SAC is also a \"maximum entropy\" algorithm, and enables exploration in an intrinsic way. Read more about maximum entropy RL here.</p>"},{"location":"ML-Agents-Overview/#curiosity-for-sparse-reward-environments","title":"Curiosity for Sparse-reward Environments","text":"<p>In environments where the agent receives rare or infrequent rewards (i.e. sparse-reward), an agent may never receive a reward signal on which to bootstrap its training process. This is a scenario where the use of an intrinsic reward signals can be valuable. Curiosity is one such signal which can help the agent explore when extrinsic rewards are sparse.</p> <p>The <code>curiosity</code> Reward Signal enables the Intrinsic Curiosity Module. This is an implementation of the approach described in Curiosity-driven Exploration by Self-supervised Prediction by Pathak, et al. It trains two networks:</p> <ul> <li>an inverse model, which takes the current and next observation of the agent,   encodes them, and uses the encoding to predict the action that was taken   between the observations</li> <li>a forward model, which takes the encoded current observation and action, and   predicts the next encoded observation.</li> </ul> <p>The loss of the forward model (the difference between the predicted and actual encoded observations) is used as the intrinsic reward, so the more surprised the model is, the larger the reward will be.</p> <p>For more information, see our dedicated blog post on the Curiosity module.</p>"},{"location":"ML-Agents-Overview/#rnd-for-sparse-reward-environments","title":"RND for Sparse-reward Environments","text":"<p>Similarly to Curiosity, Random Network Distillation (RND) is useful in sparse or rare reward environments as it helps the Agent explore. The RND Module is implemented following the paper Exploration by Random Network Distillation. RND uses two networks:  - The first is a network with fixed random weights that takes observations as inputs and  generates an encoding  - The second is a network with similar architecture that is trained to predict the  outputs of the first network and uses the observations the Agent collects as training data.</p> <p>The loss (the squared difference between the predicted and actual encoded observations) of the trained model is used as intrinsic reward. The more an Agent visits a state, the more accurate the predictions and the lower the rewards which encourages the Agent to explore new states with higher prediction errors.</p>"},{"location":"ML-Agents-Overview/#imitation-learning","title":"Imitation Learning","text":"<p>It is often more intuitive to simply demonstrate the behavior we want an agent to perform, rather than attempting to have it learn via trial-and-error methods. For example, instead of indirectly training a medic with the help of a reward function, we can give the medic real world examples of observations from the game and actions from a game controller to guide the medic's behavior. Imitation Learning uses pairs of observations and actions from a demonstration to learn a policy. See this video demo of imitation learning .</p> <p>Imitation learning can either be used alone or in conjunction with reinforcement learning. If used alone it can provide a mechanism for learning a specific type of behavior (i.e. a specific style of solving the task). If used in conjunction with reinforcement learning it can dramatically reduce the time the agent takes to solve the environment. This can be especially pronounced in sparse-reward environments. For instance, on the Pyramids environment, using 6 episodes of demonstrations can reduce training steps by more than 4 times. See Behavioral Cloning + GAIL + Curiosity + RL below.</p> <p> </p> <p>The ML-Agents Toolkit provides a way to learn directly from demonstrations, as well as use them to help speed up reward-based training (RL). We include two algorithms called Behavioral Cloning (BC) and Generative Adversarial Imitation Learning (GAIL). In most scenarios, you can combine these two features:</p> <ul> <li>If you want to help your agents learn (especially with environments that have   sparse rewards) using pre-recorded demonstrations, you can generally enable   both GAIL and Behavioral Cloning at low strengths in addition to having an   extrinsic reward. An example of this is provided for the PushBlock example   environment in <code>config/imitation/PushBlock.yaml</code>.</li> <li>If you want to train purely from demonstrations with GAIL and BC without an   extrinsic reward signal, please see the CrawlerStatic example environment under   in <code>config/imitation/CrawlerStatic.yaml</code>.</li> </ul> <p>Note: GAIL introduces a survivor bias to the learning process. That is, by giving positive rewards based on similarity to the expert, the agent is incentivized to remain alive for as long as possible. This can directly conflict with goal-oriented tasks like our PushBlock or Pyramids example environments where an agent must reach a goal state thus ending the episode as quickly as possible. In these cases, we strongly recommend that you use a low strength GAIL reward signal and a sparse extrinisic signal when the agent achieves the task. This way, the GAIL reward signal will guide the agent until it discovers the extrnisic signal and will not overpower it. If the agent appears to be ignoring the extrinsic reward signal, you should reduce the strength of GAIL.</p>"},{"location":"ML-Agents-Overview/#gail-generative-adversarial-imitation-learning","title":"GAIL (Generative Adversarial Imitation Learning)","text":"<p>GAIL, or Generative Adversarial Imitation Learning, uses an adversarial approach to reward your Agent for behaving similar to a set of demonstrations. GAIL can be used with or without environment rewards, and works well when there are a limited number of demonstrations. In this framework, a second neural network, the discriminator, is taught to distinguish whether an observation/action is from a demonstration or produced by the agent. This discriminator can then examine a new observation/action and provide it a reward based on how close it believes this new observation/action is to the provided demonstrations.</p> <p>At each training step, the agent tries to learn how to maximize this reward. Then, the discriminator is trained to better distinguish between demonstrations and agent state/actions. In this way, while the agent gets better and better at mimicking the demonstrations, the discriminator keeps getting stricter and stricter and the agent must try harder to \"fool\" it.</p> <p>This approach learns a policy that produces states and actions similar to the demonstrations, requiring fewer demonstrations than direct cloning of the actions. In addition to learning purely from demonstrations, the GAIL reward signal can be mixed with an extrinsic reward signal to guide the learning process.</p>"},{"location":"ML-Agents-Overview/#behavioral-cloning-bc","title":"Behavioral Cloning (BC)","text":"<p>BC trains the Agent's policy to exactly mimic the actions shown in a set of demonstrations. The BC feature can be enabled on the PPO or SAC trainers. As BC cannot generalize past the examples shown in the demonstrations, BC tends to work best when there exists demonstrations for nearly all of the states that the agent can experience, or in conjunction with GAIL and/or an extrinsic reward.</p>"},{"location":"ML-Agents-Overview/#recording-demonstrations","title":"Recording Demonstrations","text":"<p>Demonstrations of agent behavior can be recorded from the Unity Editor or build, and saved as assets. These demonstrations contain information on the observations, actions, and rewards for a given agent during the recording session. They can be managed in the Editor, as well as used for training with BC and GAIL. See the Designing Agents page for more information on how to record demonstrations for your agent.</p>"},{"location":"ML-Agents-Overview/#summary","title":"Summary","text":"<p>To summarize, we provide 3 training methods: BC, GAIL and RL (PPO or SAC) that can be used independently or together:</p> <ul> <li>BC can be used on its own or as a pre-training step before GAIL and/or RL</li> <li>GAIL can be used with or without extrinsic rewards</li> <li>RL can be used on its own (either PPO or SAC) or in conjunction with BC and/or   GAIL.</li> </ul> <p>Leveraging either BC or GAIL requires recording demonstrations to be provided as input to the training algorithms.</p>"},{"location":"ML-Agents-Overview/#training-methods-environment-specific","title":"Training Methods: Environment-specific","text":"<p>In addition to the three environment-agnostic training methods introduced in the previous section, the ML-Agents Toolkit provides additional methods that can aid in training behaviors for specific types of environments.</p>"},{"location":"ML-Agents-Overview/#training-in-competitive-multi-agent-environments-with-self-play","title":"Training in Competitive Multi-Agent Environments with Self-Play","text":"<p>ML-Agents provides the functionality to train both symmetric and asymmetric adversarial games with Self-Play. A symmetric game is one in which opposing agents are equal in form, function and objective. Examples of symmetric games are our Tennis and Soccer example environments. In reinforcement learning, this means both agents have the same observation and actions and learn from the same reward function and so they can share the same policy. In asymmetric games, this is not the case. An example of an asymmetric games are Hide and Seek. Agents in these types of games do not always have the same observation or actions and so sharing policy networks is not necessarily ideal.</p> <p>With self-play, an agent learns in adversarial games by competing against fixed, past versions of its opponent (which could be itself as in symmetric games) to provide a more stable, stationary learning environment. This is compared to competing against the current, best opponent in every episode, which is constantly changing (because it's learning).</p> <p>Self-play can be used with our implementations of both Proximal Policy Optimization (PPO) and Soft Actor-Critic (SAC). However, from the perspective of an individual agent, these scenarios appear to have non-stationary dynamics because the opponent is often changing. This can cause significant issues in the experience replay mechanism used by SAC. Thus, we recommend that users use PPO. For further reading on this issue in particular, see the paper Stabilising Experience Replay for Deep Multi-Agent Reinforcement Learning.</p> <p>See our Designing Agents page for more information on setting up teams in your Unity scene. Also, read our blog post on self-play for additional information. Additionally, check ELO Rating System the method we use to calculate the relative skill level between two players.</p>"},{"location":"ML-Agents-Overview/#training-in-cooperative-multi-agent-environments-with-ma-poca","title":"Training In Cooperative Multi-Agent Environments with MA-POCA","text":"<p>ML-Agents provides the functionality for training cooperative behaviors - i.e., groups of agents working towards a common goal, where the success of the individual is linked to the success of the whole group. In such a scenario, agents typically receive rewards as a group. For instance, if a team of agents wins a game against an opposing team, everyone is rewarded - even agents who did not directly contribute to the win. This makes learning what to do as an individual difficult - you may get a win for doing nothing, and a loss for doing your best.</p> <p>In ML-Agents, we provide MA-POCA (MultiAgent POsthumous Credit Assignment), which is a novel multi-agent trainer that trains a centralized critic, a neural network that acts as a \"coach\" for a whole group of agents. You can then give rewards to the team as a whole, and the agents will learn how best to contribute to achieving that reward. Agents can also be given rewards individually, and the team will work together to help the individual achieve those goals. During an episode, agents can be added or removed from the group, such as when agents spawn or die in a game. If agents are removed mid-episode (e.g., if teammates die or are removed from the game), they will still learn whether their actions contributed to the team winning later, enabling agents to take group-beneficial actions even if they result in the individual being removed from the game (i.e., self-sacrifice). MA-POCA can also be combined with self-play to train teams of agents to play against each other.</p> <p>To learn more about enabling cooperative behaviors for agents in an ML-Agents environment, check out this page.</p> <p>To learn more about MA-POCA, please see our paper On the Use and Misuse of Absorbing States in Multi-Agent Reinforcement Learning. For further reading, MA-POCA builds on previous work in multi-agent cooperative learning (Lowe et al., Foerster et al., among others) to enable the above use-cases.</p>"},{"location":"ML-Agents-Overview/#solving-complex-tasks-using-curriculum-learning","title":"Solving Complex Tasks using Curriculum Learning","text":"<p>Curriculum learning is a way of training a machine learning model where more difficult aspects of a problem are gradually introduced in such a way that the model is always optimally challenged. This idea has been around for a long time, and it is how we humans typically learn. If you imagine any childhood primary school education, there is an ordering of classes and topics. Arithmetic is taught before algebra, for example. Likewise, algebra is taught before calculus. The skills and knowledge learned in the earlier subjects provide a scaffolding for later lessons. The same principle can be applied to machine learning, where training on easier tasks can provide a scaffolding for harder tasks in the future.</p> <p>Imagine training the medic to scale a wall to arrive at a wounded team member. The starting point when training a medic to accomplish this task will be a random policy. That starting policy will have the medic running in circles, and will likely never, or very rarely scale the wall properly to revive their team member (and achieve the reward). If we start with a simpler task, such as moving toward an unobstructed team member, then the medic can easily learn to accomplish the task. From there, we can slowly add to the difficulty of the task by increasing the size of the wall until the medic can complete the initially near-impossible task of scaling the wall. We have included an environment to demonstrate this with ML-Agents, called Wall Jump.</p> <p></p> <p>Demonstration of a hypothetical curriculum training scenario in which a progressively taller wall obstructs the path to the goal.</p> <p>[Note: The example provided above is for instructional purposes, and was based on an early version of the Wall Jump example environment. As such, it is not possible to directly replicate the results here using that environment.]</p> <p>The ML-Agents Toolkit supports modifying custom environment parameters during the training process to aid in learning. This allows elements of the environment related to difficulty or complexity to be dynamically adjusted based on training progress. The Training ML-Agents page has more information on defining training curriculums.</p>"},{"location":"ML-Agents-Overview/#training-robust-agents-using-environment-parameter-randomization","title":"Training Robust Agents using Environment Parameter Randomization","text":"<p>An agent trained on a specific environment, may be unable to generalize to any tweaks or variations in the environment (in machine learning this is referred to as overfitting). This becomes problematic in cases where environments are instantiated with varying objects or properties. One mechanism to alleviate this and train more robust agents that can generalize to unseen variations of the environment is to expose them to these variations during training. Similar to Curriculum Learning, where environments become more difficult as the agent learns, the ML-Agents Toolkit provides a way to randomly sample parameters of the environment during training. We refer to this approach as Environment Parameter Randomization. For those familiar with Reinforcement Learning research, this approach is based on the concept of Domain Randomization. By using parameter randomization during training, the agent can be better suited to adapt (with higher performance) to future unseen variations of the environment.</p> Ball scale of 0.5 Ball scale of 4 <p>Example of variations of the 3D Ball environment. The environment parameters are <code>gravity</code>, <code>ball_mass</code> and <code>ball_scale</code>.</p>"},{"location":"ML-Agents-Overview/#model-types","title":"Model Types","text":"<p>Regardless of the training method deployed, there are a few model types that users can train using the ML-Agents Toolkit. This is due to the flexibility in defining agent observations, which include vector, ray cast and visual observations. You can learn more about how to instrument an agent's observation in the Designing Agents guide.</p>"},{"location":"ML-Agents-Overview/#learning-from-vector-observations","title":"Learning from Vector Observations","text":"<p>Whether an agent's observations are ray cast or vector, the ML-Agents Toolkit provides a fully connected neural network model to learn from those observations. At training time you can configure different aspects of this model such as the number of hidden units and number of layers.</p>"},{"location":"ML-Agents-Overview/#learning-from-cameras-using-convolutional-neural-networks","title":"Learning from Cameras using Convolutional Neural Networks","text":"<p>Unlike other platforms, where the agent\u2019s observation might be limited to a single vector or image, the ML-Agents Toolkit allows multiple cameras to be used for observations per agent. This enables agents to learn to integrate information from multiple visual streams. This can be helpful in several scenarios such as training a self-driving car which requires multiple cameras with different viewpoints, or a navigational agent which might need to integrate aerial and first-person visuals. You can learn more about adding visual observations to an agent here.</p> <p>When visual observations are utilized, the ML-Agents Toolkit leverages convolutional neural networks (CNN) to learn from the input images. We offer three network architectures:</p> <ul> <li>a simple encoder which consists of two convolutional layers</li> <li>the implementation proposed by   Mnih et al., consisting of   three convolutional layers,</li> <li>the IMPALA Resnet consisting of three   stacked layers, each with two residual blocks, making a much larger network   than the other two.</li> </ul> <p>The choice of the architecture depends on the visual complexity of the scene and the available computational resources.</p>"},{"location":"ML-Agents-Overview/#learning-from-variable-length-observations-using-attention","title":"Learning from Variable Length Observations using Attention","text":"<p>Using the ML-Agents Toolkit, it is possible to have agents learn from a varying number of inputs. To do so, each agent can keep track of a buffer of vector observations. At each step, the agent will go through all the elements in the buffer and extract information but the elements in the buffer can change at every step. This can be useful in scenarios in which the agents must keep track of a varying number of elements throughout the episode. For example in a game where an agent must learn to avoid projectiles, but the projectiles can vary in numbers.</p> <p></p> <p>You can learn more about variable length observations here. When variable length observations are utilized, the ML-Agents Toolkit leverages attention networks to learn from a varying number of entities. Agents using attention will ignore entities that are deemed not relevant and pay special attention to entities relevant to the current situation based on context.</p>"},{"location":"ML-Agents-Overview/#memory-enhanced-agents-using-recurrent-neural-networks","title":"Memory-enhanced Agents using Recurrent Neural Networks","text":"<p>Have you ever entered a room to get something and immediately forgot what you were looking for? Don't let that happen to your agents.</p> <p></p> <p>In some scenarios, agents must learn to remember the past in order to take the best decision. When an agent only has partial observability of the environment, keeping track of past observations can help the agent learn. Deciding what the agents should remember in order to solve a task is not easy to do by hand, but our training algorithms can learn to keep track of what is important to remember with LSTM.</p>"},{"location":"ML-Agents-Overview/#additional-features","title":"Additional Features","text":"<p>Beyond the flexible training scenarios available, the ML-Agents Toolkit includes additional features which improve the flexibility and interpretability of the training process.</p> <ul> <li>Concurrent Unity Instances - We enable developers to run concurrent,   parallel instances of the Unity executable during training. For certain   scenarios, this should speed up training. Check out our dedicated page on   creating a Unity executable and the   Training ML-Agents   page for instructions on how to set the number of concurrent instances.</li> <li>Recording Statistics from Unity - We enable developers to   record statistics from   within their Unity environments. These statistics are aggregated and generated   during the training process.</li> <li>Custom Side Channels - We enable developers to   create custom side channels to manage data transfer   between Unity and Python that is unique to their training workflow and/or   environment.</li> <li>Custom Samplers - We enable developers to   create custom sampling methods   for Environment Parameter Randomization. This enables users to customize this   training method for their particular environment.</li> </ul>"},{"location":"ML-Agents-Overview/#summary-and-next-steps","title":"Summary and Next Steps","text":"<p>To briefly summarize: The ML-Agents Toolkit enables games and simulations built in Unity to serve as the platform for training intelligent agents. It is designed to enable a large variety of training modes and scenarios and comes packed with several features to enable researchers and developers to leverage (and enhance) machine learning within Unity.</p> <p>In terms of next steps:</p> <ul> <li>For a walkthrough of running ML-Agents with a simple scene, check out the   Getting Started guide.</li> <li>For a \"Hello World\" introduction to creating your own Learning Environment,   check out the   Making a New Learning Environment page.</li> <li>For an overview on the more complex example environments that are provided in   this toolkit, check out the   Example Environments page.</li> <li>For more information on the various training options available, check out the   Training ML-Agents page.</li> </ul>"},{"location":"ML-Agents-README/","title":"Unity ML-Agents Trainers","text":"<p>The <code>mlagents</code> Python package is part of the ML-Agents Toolkit. <code>mlagents</code> provides a set of reinforcement and imitation learning algorithms designed to be used with Unity environments. The algorithms interface with the Python API provided by the <code>mlagents_envs</code> package. See here for more information on <code>mlagents_envs</code>.</p> <p>The algorithms can be accessed using the: <code>mlagents-learn</code> access point. See here for more information on using this package.</p>"},{"location":"ML-Agents-README/#installation","title":"Installation","text":"<p>Install the <code>mlagents</code> package with:</p> <pre><code>python -m pip install mlagents==1.0.0\n</code></pre>"},{"location":"ML-Agents-README/#usage-more-information","title":"Usage &amp; More Information","text":"<p>For more information on the ML-Agents Toolkit and how to instrument a Unity scene with the ML-Agents SDK, check out the main ML-Agents Toolkit documentation.</p>"},{"location":"ML-Agents-README/#limitations","title":"Limitations","text":"<ul> <li>Resuming self-play from a checkpoint resets the reported ELO to the default   value.</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/","title":"Unity ML-Agents Toolkit Documentation","text":""},{"location":"ML-Agents-Toolkit-Documentation/#installation-set-up","title":"Installation &amp; Set-up","text":"<ul> <li>Installation</li> <li>Using Virtual Environment</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#getting-started","title":"Getting Started","text":"<ul> <li>Getting Started Guide</li> <li>ML-Agents Toolkit Overview</li> <li>Background: Unity</li> <li>Background: Machine Learning</li> <li>Background: PyTorch</li> <li>Example Environments</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#creating-learning-environments","title":"Creating Learning Environments","text":"<ul> <li>Making a New Learning Environment</li> <li>Designing a Learning Environment</li> <li>Designing Agents</li> <li>Using an Executable Environment</li> <li>ML-Agents Package Settings</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#training-inference","title":"Training &amp; Inference","text":"<ul> <li>Training ML-Agents</li> <li>Training Configuration File</li> <li>Using TensorBoard to Observe Training</li> <li>Profiling Trainers</li> <li>Sentis</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#extending-ml-agents","title":"Extending ML-Agents","text":"<ul> <li>Creating Custom Side Channels</li> <li>Creating Custom Samplers for Environment Parameter Randomization</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#hugging-face-integration","title":"Hugging Face Integration","text":"<ul> <li>Using Hugging Face to download and upload trained models</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#python-tutorial-with-google-colab","title":"Python Tutorial with Google Colab","text":"<ul> <li>Using a UnityEnvironment</li> <li>Q-Learning with a UnityEnvironment</li> <li>Using Side Channels on a UnityEnvironment</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#help","title":"Help","text":"<ul> <li>Migrating from earlier versions of ML-Agents</li> <li>Frequently Asked Questions</li> <li>ML-Agents Glossary</li> <li>Limitations</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#api-docs","title":"API Docs","text":"<ul> <li>API Reference</li> <li>Python API Documentation</li> <li>How to use the Python API</li> <li>How to use the Unity Environment Registry</li> <li>Wrapping Learning Environment as a Gym (+Baselines/Dopamine Integration)</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#translations","title":"Translations","text":"<p>To make the Unity ML-Agents Toolkit accessible to the global research and Unity developer communities, we're attempting to create and maintain translations of our documentation. We've started with translating a subset of the documentation to one language (Chinese), but we hope to continue translating more pages and to other languages. Consequently, we welcome any enhancements and improvements from the community.</p> <ul> <li>Chinese</li> <li>Korean</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#deprecated-docs","title":"Deprecated Docs","text":"<p>We no longer use them ourselves and so they may not be up-to-date. We've decided to keep them up just in case they are helpful to you.</p> <ul> <li>Windows Anaconda Installation</li> <li>Using Docker</li> <li>Training on the Cloud with Amazon Web Services</li> <li>Training on the Cloud with Microsoft Azure</li> <li>Using the Video Recorder</li> </ul>"},{"location":"Migrating/","title":"Upgrading","text":""},{"location":"Migrating/#migrating","title":"Migrating","text":""},{"location":"Migrating/#migrating-to-the-ml-agents-envs-0300-package","title":"Migrating to the ml-agents-envs 0.30.0 package","text":"<ul> <li>Python 3.10.12 is now the minimum version of python supported due to python3.6 EOL.   Please update your python installation to 3.10.12 or higher.</li> <li>The <code>gym-unity</code> package has been refactored into the <code>ml-agents-envs</code> package. Please update your imports accordingly.</li> <li>Example:</li> <li>Before</li> </ul> <pre><code>from gym_unity.unity_gym_env import UnityToGymWrapper\n</code></pre> <ul> <li>After:</li> </ul> <pre><code>from mlagents_envs.envs.unity_gym_env import UnityToGymWrapper\n</code></pre>"},{"location":"Migrating/#migrating-the-package-to-version-2x","title":"Migrating the package to version 2.x","text":"<ul> <li>The official version of Unity ML-Agents supports is now 2022.3 LTS. If you run   into issues, please consider deleting your project's Library folder and reponening your   project.</li> <li>If you used any of the APIs that were deprecated before version 2.0, you need to use their replacement. These deprecated APIs have been removed. See the migration steps bellow for specific API replacements.</li> </ul>"},{"location":"Migrating/#deprecated-methods-removed","title":"Deprecated methods removed","text":"Deprecated API Suggested Replacement <code>IActuator ActuatorComponent.CreateActuator()</code> <code>IActuator[] ActuatorComponent.CreateActuators()</code> <code>IActionReceiver.PackActions(in float[] destination)</code> none <code>Agent.CollectDiscreteActionMasks(DiscreteActionMasker actionMasker)</code> <code>Agent.WriteDiscreteActionMask(IDiscreteActionMask actionMask)</code> <code>Agent.Heuristic(float[] actionsOut)</code> <code>Agent.Heuristic(in ActionBuffers actionsOut)</code> <code>Agent.OnActionReceived(float[] vectorAction)</code> <code>Agent.OnActionReceived(ActionBuffers actions)</code> <code>Agent.GetAction()</code> <code>Agent.GetStoredActionBuffers()</code> <code>BrainParameters.SpaceType</code>, <code>VectorActionSize</code>, <code>VectorActionSpaceType</code>, and <code>NumActions</code> <code>BrainParameters.ActionSpec</code> <code>ObservationWriter.AddRange(IEnumerable&lt;float&gt; data, int writeOffset = 0)</code> <code>ObservationWriter. AddList(IList&lt;float&gt; data, int writeOffset = 0</code> <code>SensorComponent.IsVisual()</code> and <code>IsVector()</code> none <code>VectorSensor.AddObservation(IEnumerable&lt;float&gt; observation)</code> <code>VectorSensor.AddObservation(IList&lt;float&gt; observation)</code> <code>SideChannelsManager</code> <code>SideChannelManager</code>"},{"location":"Migrating/#idiscreteactionmask-changes","title":"IDiscreteActionMask changes","text":"<ul> <li>The interface for disabling specific discrete actions has changed. <code>IDiscreteActionMask.WriteMask()</code> was removed, and replaced with <code>SetActionEnabled()</code>. Instead of returning an IEnumerable with indices to disable, you can now call <code>SetActionEnabled</code> for each index to disable (or enable). As an example, if you overrode <code>Agent.WriteDiscreteActionMask()</code> with something that looked like:</li> </ul> <pre><code>public override void WriteDiscreteActionMask(IDiscreteActionMask actionMask)\n{\n    var branch = 2;\n    var actionsToDisable = new[] {1, 3};\n    actionMask.WriteMask(branch, actionsToDisable);\n}\n</code></pre> <p>the equivalent code would now be</p> <pre><code>public override void WriteDiscreteActionMask(IDiscreteActionMask actionMask)\n{\n    var branch = 2;\n    actionMask.SetActionEnabled(branch, 1, false);\n    actionMask.SetActionEnabled(branch, 3, false);\n}\n</code></pre>"},{"location":"Migrating/#iactuator-changes","title":"IActuator changes","text":"<ul> <li>The <code>IActuator</code> interface now implements <code>IHeuristicProvider</code>.  Please add the corresponding <code>Heuristic(in ActionBuffers)</code> method to your custom Actuator classes.</li> </ul>"},{"location":"Migrating/#isensor-and-sensorcomponent-changes","title":"ISensor and SensorComponent changes","text":"<ul> <li>The <code>ISensor.GetObservationShape()</code> method and <code>ITypedSensor</code> and <code>IDimensionPropertiesSensor</code> interfaces were removed, and <code>GetObservationSpec()</code> was added. You can use <code>ObservationSpec.Vector()</code> or <code>ObservationSpec.Visual()</code> to generate <code>ObservationSpec</code>s that are equivalent to the previous shape. For example, if your old ISensor looked like:</li> </ul> <pre><code>public override int[] GetObservationShape()\n{\n    return new[] { m_Height, m_Width, m_NumChannels };\n}\n</code></pre> <p>the equivalent code would now be</p> <pre><code>public override ObservationSpec GetObservationSpec()\n{\n    return ObservationSpec.Visual(m_Height, m_Width, m_NumChannels);\n}\n</code></pre> <ul> <li>The <code>ISensor.GetCompressionType()</code> method and <code>ISparseChannelSensor</code> interface was removed, and <code>GetCompressionSpec()</code> was added. You can use <code>CompressionSpec.Default()</code> or <code>CompressionSpec.Compressed()</code> to generate <code>CompressionSpec</code>s that are  equivalent to  the previous values. For example, if your old ISensor looked like:  ```csharp public virtual SensorCompressionType GetCompressionType() {     return SensorCompressionType.None; }</li> </ul> <pre><code>\nthe equivalent code would now be\n\n```csharp\npublic CompressionSpec GetCompressionSpec()\n{\n    return CompressionSpec.Default();\n}\n</code></pre> <ul> <li>The abstract method <code>SensorComponent.GetObservationShape()</code> was removed.</li> <li>The abstract method <code>SensorComponent.CreateSensor()</code> was replaced with <code>CreateSensors()</code>, which returns an <code>ISensor[]</code>.</li> </ul>"},{"location":"Migrating/#match3-integration-changes","title":"Match3 integration changes","text":"<p>The Match-3 integration utilities were moved from <code>com.unity.ml-agents.extensions</code> to <code>com.unity.ml-agents</code>.</p> <p>The <code>AbstractBoard</code> interface was changed: * <code>AbstractBoard</code> no longer contains <code>Rows</code>, <code>Columns</code>, <code>NumCellTypes</code>, and <code>NumSpecialTypes</code> fields. * <code>public abstract BoardSize GetMaxBoardSize()</code> was added as an abstract method. <code>BoardSize</code> is a new struct that contains <code>Rows</code>, <code>Columns</code>, <code>NumCellTypes</code>, and <code>NumSpecialTypes</code> fields, with the same meanings as the old <code>AbstractBoard</code> fields. * <code>public virtual BoardSize GetCurrentBoardSize()</code> is an optional method; by default it returns <code>GetMaxBoardSize()</code>. If you wish to use a single behavior to work with multiple board sizes, override <code>GetCurrentBoardSize()</code> to return the current <code>BoardSize</code>. The values returned by <code>GetCurrentBoardSize()</code> must be less than or equal to the corresponding values from <code>GetMaxBoardSize()</code>.</p>"},{"location":"Migrating/#gridsensor-changes","title":"GridSensor changes","text":"<p>The sensor configuration has changed: * The sensor implementation has been refactored and exsisting GridSensor created from extension package will not work in newer version. Some errors might show up when loading the old sensor in the scene. You'll need to remove the old sensor and create a new GridSensor. * These parameters names have changed but still refer to the same concept in the sensor: <code>GridNumSide</code> -&gt; <code>GridSize</code>, <code>RotateToAgent</code> -&gt; <code>RotateWithAgent</code>, <code>ObserveMask</code> -&gt; <code>ColliderMask</code>, <code>DetectableObjects</code> -&gt; <code>DetectableTags</code> * <code>DepthType</code> (<code>ChanelBase</code>/<code>ChannelHot</code>) option and <code>ChannelDepth</code> are removed. Now the default is one-hot encoding for detected tag. If you were using original GridSensor without overriding any method, switching to new GridSensor will produce similar effect for training although the actual observations will be slightly different.</p> <p>For creating your GridSensor implementation with custom data: * To create custom GridSensor, derive from <code>GridSensorBase</code> instead of <code>GridSensor</code>. Besides overriding <code>GetObjectData()</code>, you will also need to consider override <code>GetCellObservationSize()</code>, <code>IsDataNormalized()</code> and <code>GetProcessCollidersMethod()</code> according to the data you collect. Also you'll need to override <code>GridSensorComponent.GetGridSensors()</code> and return your custom GridSensor. * The input argument <code>tagIndex</code> in <code>GetObjectData()</code> has changed from 1-indexed to 0-indexed and the data type changed from <code>float</code> to <code>int</code>. The index of first detectable tag will be 0 instead of 1. <code>normalizedDistance</code> was removed from input. * The observation data should be written to the input <code>dataBuffer</code> instead of creating and returning a new array. * Removed the constraint of all data required to be normalized. You should specify it in <code>IsDataNormalized()</code>. Sensors with non-normalized data cannot use PNG compression type. * The sensor will not further encode the data recieved from <code>GetObjectData()</code> anymore. The values recieved from <code>GetObjectData()</code> will be the observation sent to the trainer.</p>"},{"location":"Migrating/#lstm-models-from-previous-releases-no-longer-supported","title":"LSTM models from previous releases no longer supported","text":"<p>The way that Sentis processes LSTM (recurrent neural networks) has changed. As a result, models trained with previous versions of ML-Agents will not be usable at inference if they were trained with a <code>memory</code> setting in the <code>.yaml</code> config file. If you want to use a model that has a recurrent neural network in this release of ML-Agents, you need to train the model using the python trainer from this release.</p>"},{"location":"Migrating/#migrating-to-release-13","title":"Migrating to Release 13","text":""},{"location":"Migrating/#implementing-iheuristic-in-your-iactuator-implementations","title":"Implementing IHeuristic in your IActuator implementations","text":"<ul> <li>If you have any custom actuators, you can now implement the <code>IHeuristicProvider</code> interface to have your actuator   handle the generation of actions when an Agent is running in heuristic mode.</li> <li><code>VectorSensor.AddObservation(IEnumerable&lt;float&gt;)</code> is deprecated. Use <code>VectorSensor.AddObservation(IList&lt;float&gt;)</code>   instead.</li> <li><code>ObservationWriter.AddRange()</code> is deprecated. Use <code>ObservationWriter.AddList()</code> instead.</li> <li><code>ActuatorComponent.CreateAcuator()</code> is deprecated.  Please use override <code>ActuatorComponent.CreateActuators</code>   instead.  Since <code>ActuatorComponent.CreateActuator()</code> is abstract, you will still need to override it in your   class until it is removed.  It is only ever called if you don't override <code>ActuatorComponent.CreateActuators</code>.   You can suppress the warnings by surrounding the method with the following pragma:     <code>c#     #pragma warning disable 672     public IActuator CreateActuator() { ... }     #pragma warning restore 672</code></li> </ul>"},{"location":"Migrating/#migrating_1","title":"Migrating","text":""},{"location":"Migrating/#migrating-to-release-11","title":"Migrating to Release 11","text":""},{"location":"Migrating/#agent-virtual-method-deprecation","title":"Agent virtual method deprecation","text":"<ul> <li><code>Agent.CollectDiscreteActionMasks()</code> was deprecated and should be replaced with <code>Agent.WriteDiscreteActionMask()</code></li> <li><code>Agent.Heuristic(float[])</code> was deprecated and should be replaced with <code>Agent.Heuristic(ActionBuffers)</code>.</li> <li><code>Agent.OnActionReceived(float[])</code> was deprecated and should be replaced with <code>Agent.OnActionReceived(ActionBuffers)</code>.</li> <li><code>Agent.GetAction()</code> was deprecated and should be replaced with <code>Agent.GetStoredActionBuffers()</code>.</li> </ul> <p>The default implementation of these will continue to call the deprecated versions where appropriate. However, the deprecated versions may not be compatible with continuous and discrete actions on the same Agent.</p>"},{"location":"Migrating/#brainparameters-field-and-method-deprecation","title":"BrainParameters field and method deprecation","text":"<ul> <li><code>BrainParameters.VectorActionSize</code> was deprecated; you can now set <code>BrainParameters.ActionSpec.NumContinuousActions</code>  or <code>BrainParameters.ActionSpec.BranchSizes</code> instead.</li> <li><code>BrainParameters.VectorActionSpaceType</code> was deprecated, since both continuous and discrete actions can now be used.</li> <li><code>BrainParameters.NumActions()</code> was deprecated. Use  <code>BrainParameters.ActionSpec.NumContinuousActions</code> and  <code>BrainParameters.ActionSpec.NumDiscreteActions</code> instead.</li> </ul>"},{"location":"Migrating/#migrating-from-release-7-to-latest","title":"Migrating from Release 7 to latest","text":""},{"location":"Migrating/#important-changes","title":"Important changes","text":"<ul> <li>Some trainer files were moved. If you were using the <code>TrainerFactory</code> class, it was moved to the <code>trainers/trainer</code> folder.</li> <li>The <code>components</code> folder containing <code>bc</code> and <code>reward_signals</code> code was moved to the <code>trainers/tf</code> folder</li> </ul>"},{"location":"Migrating/#steps-to-migrate","title":"Steps to Migrate","text":"<ul> <li>Replace calls to <code>from mlagents.trainers.trainer_util import TrainerFactory</code> to <code>from mlagents.trainers.trainer import TrainerFactory</code></li> <li>Replace calls to <code>from mlagents.trainers.trainer_util import handle_existing_directories</code> to <code>from mlagents.trainers.directory_utils import validate_existing_directories</code></li> <li>Replace <code>mlagents.trainers.components</code> with <code>mlagents.trainers.tf.components</code> in your import statements.</li> </ul>"},{"location":"Migrating/#migrating-from-release-3-to-release-7","title":"Migrating from Release 3 to Release 7","text":""},{"location":"Migrating/#important-changes_1","title":"Important changes","text":"<ul> <li>The Parameter Randomization feature has been merged with the Curriculum feature. It is now possible to specify a sampler in the lesson of a Curriculum. Curriculum has been refactored and is now specified at the level of the parameter, not the behavior. More information here.(#4160)</li> </ul>"},{"location":"Migrating/#steps-to-migrate_1","title":"Steps to Migrate","text":"<ul> <li>The configuration format for curriculum and parameter randomization has changed. To upgrade your configuration files, an upgrade script has been provided. Run <code>python -m mlagents.trainers.upgrade_config -h</code> to see the script usage. Note that you will have had to upgrade to/install the current version of ML-Agents before running the script. To update manually:</li> <li>If your config file used a <code>parameter_randomization</code> section, rename that section to <code>environment_parameters</code></li> <li>If your config file used a <code>curriculum</code> section, you will need to rewrite your curriculum with this format.</li> </ul>"},{"location":"Migrating/#migrating-from-release-1-to-release-3","title":"Migrating from Release 1 to Release 3","text":""},{"location":"Migrating/#important-changes_2","title":"Important changes","text":"<ul> <li>Training artifacts (trained models, summaries) are now found under <code>results/</code>   instead of <code>summaries/</code> and <code>models/</code>.</li> <li>Trainer configuration, curriculum configuration, and parameter randomization   configuration have all been moved to a single YAML file. (#3791)</li> <li>Trainer configuration format has changed, and using a \"default\" behavior name has   been deprecated. (#3936)</li> <li><code>max_step</code> in the <code>TerminalStep</code> and <code>TerminalSteps</code> objects was renamed <code>interrupted</code>.</li> <li>On the UnityEnvironment API, <code>get_behavior_names()</code> and <code>get_behavior_specs()</code> methods were combined into the property <code>behavior_specs</code> that contains a mapping from behavior names to behavior spec.</li> <li><code>use_visual</code> and <code>allow_multiple_visual_obs</code> in the <code>UnityToGymWrapper</code> constructor were replaced by <code>allow_multiple_obs</code> which allows one or more visual observations and vector observations to be used simultaneously.</li> <li><code>--save-freq</code> has been removed from the CLI and is now configurable in the trainer configuration   file.</li> <li><code>--lesson</code> has been removed from the CLI. Lessons will resume when using <code>--resume</code>.   To start at a different lesson, modify your Curriculum configuration.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_2","title":"Steps to Migrate","text":"<ul> <li>To upgrade your configuration files, an upgrade script has been provided. Run   <code>python -m mlagents.trainers.upgrade_config -h</code> to see the script usage. Note that you will have   had to upgrade to/install the current version of ML-Agents before running the script.</li> </ul> <p>To do it manually, copy your <code>&lt;BehaviorName&gt;</code> sections from <code>trainer_config.yaml</code> into a separate trainer configuration file, under a <code>behaviors</code> section.   The <code>default</code> section is no longer needed. This new file should be specific to your environment, and not contain   configurations for multiple environments (unless they have the same Behavior Names).   - You will need to reformat your trainer settings as per the example.   - If your training uses curriculum, move those configurations under a <code>curriculum</code> section.   - If your training uses parameter randomization, move   the contents of the sampler config to <code>parameter_randomization</code> in the main trainer configuration. - If you are using <code>UnityEnvironment</code> directly, replace <code>max_step</code> with <code>interrupted</code>  in the <code>TerminalStep</code> and <code>TerminalSteps</code> objects.  - Replace usage of <code>get_behavior_names()</code> and <code>get_behavior_specs()</code> in UnityEnvironment with <code>behavior_specs</code>.  - If you use the <code>UnityToGymWrapper</code>, remove <code>use_visual</code> and <code>allow_multiple_visual_obs</code>  from the constructor and add <code>allow_multiple_obs = True</code> if the environment contains either  both visual and vector observations or multiple visual observations.  - If you were setting <code>--save-freq</code> in the CLI, add a <code>checkpoint_interval</code> value in your   trainer configuration, and set it equal to <code>save-freq * n_agents_in_scene</code>.</p>"},{"location":"Migrating/#migrating-from-015-to-release-1","title":"Migrating from 0.15 to Release 1","text":""},{"location":"Migrating/#important-changes_3","title":"Important changes","text":"<ul> <li>The <code>MLAgents</code> C# namespace was renamed to <code>Unity.MLAgents</code>, and other nested   namespaces were similarly renamed (#3843).</li> <li>The <code>--load</code> and <code>--train</code> command-line flags have been deprecated and   replaced with <code>--resume</code> and <code>--inference</code>.</li> <li>Running with the same <code>--run-id</code> twice will now throw an error.</li> <li>The <code>play_against_current_self_ratio</code> self-play trainer hyperparameter has   been renamed to <code>play_against_latest_model_ratio</code></li> <li>Removed the multi-agent gym option from the gym wrapper. For multi-agent   scenarios, use the Low Level Python API.</li> <li>The low level Python API has changed. You can look at the document   Low Level Python API documentation for more information. If   you use <code>mlagents-learn</code> for training, this should be a transparent change.</li> <li>The obsolete <code>Agent</code> methods <code>GiveModel</code>, <code>Done</code>, <code>InitializeAgent</code>,   <code>AgentAction</code> and <code>AgentReset</code> have been removed.</li> <li>The signature of <code>Agent.Heuristic()</code> was changed to take a <code>float[]</code> as a   parameter, instead of returning the array. This was done to prevent a common   source of error where users would return arrays of the wrong size.</li> <li>The SideChannel API has changed (#3833, #3660) :</li> <li>Introduced the <code>SideChannelManager</code> to register, unregister and access side     channels.</li> <li><code>EnvironmentParameters</code> replaces the default <code>FloatProperties</code>. You can     access the <code>EnvironmentParameters</code> with     <code>Academy.Instance.EnvironmentParameters</code> on C#. If you were previously     creating a <code>UnityEnvironment</code> in python and passing it a     <code>FloatPropertiesChannel</code>, create an <code>EnvironmentParametersChannel</code> instead.</li> <li><code>SideChannel.OnMessageReceived</code> is now a protected method (was public)</li> <li>SideChannel IncomingMessages methods now take an optional default argument,     which is used when trying to read more data than the message contains.</li> <li>Added a feature to allow sending stats from C# environments to TensorBoard     (and other python StatsWriters). To do this from your code, use     <code>Academy.Instance.StatsRecorder.Add(key, value)</code>(#3660)</li> <li><code>num_updates</code> and <code>train_interval</code> for SAC have been replaced with   <code>steps_per_update</code>.</li> <li>The <code>UnityEnv</code> class from the <code>gym-unity</code> package was renamed   <code>UnityToGymWrapper</code> and no longer creates the <code>UnityEnvironment</code>. Instead, the   <code>UnityEnvironment</code> must be passed as input to the constructor of   <code>UnityToGymWrapper</code></li> <li>Public fields and properties on several classes were renamed to follow Unity's   C# style conventions. All public fields and properties now use \"PascalCase\"   instead of \"camelCase\"; for example, <code>Agent.maxStep</code> was renamed to   <code>Agent.MaxStep</code>. For a full list of changes, see the pull request. (#3828)</li> <li><code>WriteAdapter</code> was renamed to <code>ObservationWriter</code>. (#3834)</li> </ul>"},{"location":"Migrating/#steps-to-migrate_3","title":"Steps to Migrate","text":"<ul> <li>In C# code, replace <code>using MLAgents</code> with <code>using Unity.MLAgents</code>. Replace   other nested namespaces such as <code>using MLAgents.Sensors</code> with   <code>using Unity.MLAgents.Sensors</code></li> <li>Replace the <code>--load</code> flag with <code>--resume</code> when calling <code>mlagents-learn</code>, and   don't use the <code>--train</code> flag as training will happen by default. To run   without training, use <code>--inference</code>.</li> <li>To force-overwrite files from a pre-existing run, add the <code>--force</code>   command-line flag.</li> <li>The Jupyter notebooks have been removed from the repository.</li> <li>If your Agent class overrides <code>Heuristic()</code>, change the signature to   <code>public override void Heuristic(float[] actionsOut)</code> and assign values to   <code>actionsOut</code> instead of returning an array.</li> <li>If you used <code>SideChannels</code> you must:</li> <li>Replace <code>Academy.FloatProperties</code> with     <code>Academy.Instance.EnvironmentParameters</code>.</li> <li><code>Academy.RegisterSideChannel</code> and <code>Academy.UnregisterSideChannel</code> were     removed. Use <code>SideChannelManager.RegisterSideChannel</code> and     <code>SideChannelManager.UnregisterSideChannel</code> instead.</li> <li>Set <code>steps_per_update</code> to be around equal to the number of agents in your   environment, times <code>num_updates</code> and divided by <code>train_interval</code>.</li> <li>Replace <code>UnityEnv</code> with <code>UnityToGymWrapper</code> in your code. The constructor no   longer takes a file name as input but a fully constructed <code>UnityEnvironment</code>   instead.</li> <li>Update uses of \"camelCase\" fields and properties to \"PascalCase\".</li> </ul>"},{"location":"Migrating/#migrating-from-014-to-015","title":"Migrating from 0.14 to 0.15","text":""},{"location":"Migrating/#important-changes_4","title":"Important changes","text":"<ul> <li>The <code>Agent.CollectObservations()</code> virtual method now takes as input a   <code>VectorSensor</code> sensor as argument. The <code>Agent.AddVectorObs()</code> methods were   removed.</li> <li>The <code>SetMask</code> was renamed to <code>SetMask</code> method must now be called on the   <code>DiscreteActionMasker</code> argument of the <code>CollectDiscreteActionMasks</code> virtual   method.</li> <li>We consolidated our API for <code>DiscreteActionMasker</code>. <code>SetMask</code> takes two   arguments : the branch index and the list of masked actions for that branch.</li> <li>The <code>Monitor</code> class has been moved to the Examples Project. (It was prone to   errors during testing)</li> <li>The <code>MLAgents.Sensors</code> namespace has been introduced. All sensors classes are   part of the <code>MLAgents.Sensors</code> namespace.</li> <li>The <code>MLAgents.SideChannels</code> namespace has been introduced. All side channel   classes are part of the <code>MLAgents.SideChannels</code> namespace.</li> <li>The interface for <code>RayPerceptionSensor.PerceiveStatic()</code> was changed to take   an input class and write to an output class, and the method was renamed to   <code>Perceive()</code>.</li> <li>The <code>SetMask</code> method must now be called on the <code>DiscreteActionMasker</code> argument   of the <code>CollectDiscreteActionMasks</code> method.</li> <li>The method <code>GetStepCount()</code> on the Agent class has been replaced with the   property getter <code>StepCount</code></li> <li>The <code>--multi-gpu</code> option has been removed temporarily.</li> <li><code>AgentInfo.actionMasks</code> has been renamed to <code>AgentInfo.discreteActionMasks</code>.</li> <li><code>BrainParameters</code> and <code>SpaceType</code> have been removed from the public API</li> <li><code>BehaviorParameters</code> have been removed from the public API.</li> <li><code>DecisionRequester</code> has been made internal (you can still use the   DecisionRequesterComponent from the inspector). <code>RepeatAction</code> was renamed   <code>TakeActionsBetweenDecisions</code> for clarity.</li> <li>The following methods in the <code>Agent</code> class have been renamed. The original   method names will be removed in a later release:</li> <li><code>InitializeAgent()</code> was renamed to <code>Initialize()</code></li> <li><code>AgentAction()</code> was renamed to <code>OnActionReceived()</code></li> <li><code>AgentReset()</code> was renamed to <code>OnEpsiodeBegin()</code></li> <li><code>Done()</code> was renamed to <code>EndEpisode()</code></li> <li><code>GiveModel()</code> was renamed to <code>SetModel()</code></li> <li>The <code>IFloatProperties</code> interface has been removed.</li> <li>The interface for SideChannels was changed:</li> <li>In C#, <code>OnMessageReceived</code> now takes a <code>IncomingMessage</code> argument, and     <code>QueueMessageToSend</code> takes an <code>OutgoingMessage</code> argument.</li> <li>In python, <code>on_message_received</code> now takes a <code>IncomingMessage</code> argument, and     <code>queue_message_to_send</code> takes an <code>OutgoingMessage</code> argument.</li> <li>Automatic stepping for Academy is now controlled from the     AutomaticSteppingEnabled property.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_4","title":"Steps to Migrate","text":"<ul> <li>Add the <code>using MLAgents.Sensors;</code> in addition to <code>using MLAgents;</code> on top of   your Agent's script.</li> <li>Replace your Agent's implementation of <code>CollectObservations()</code> with   <code>CollectObservations(VectorSensor sensor)</code>. In addition, replace all calls to   <code>AddVectorObs()</code> with <code>sensor.AddObservation()</code> or   <code>sensor.AddOneHotObservation()</code> on the <code>VectorSensor</code> passed as argument.</li> <li>Replace your calls to <code>SetActionMask</code> on your Agent to   <code>DiscreteActionMasker.SetActionMask</code> in <code>CollectDiscreteActionMasks</code>.</li> <li>If you call <code>RayPerceptionSensor.PerceiveStatic()</code> manually, add your inputs   to a <code>RayPerceptionInput</code>. To get the previous float array output, iterate   through <code>RayPerceptionOutput.rayOutputs</code> and call   <code>RayPerceptionOutput.RayOutput.ToFloatArray()</code>.</li> <li>Replace all calls to <code>Agent.GetStepCount()</code> with <code>Agent.StepCount</code></li> <li>We strongly recommend replacing the following methods with their new   equivalent as they will be removed in a later release:</li> <li><code>InitializeAgent()</code> to <code>Initialize()</code></li> <li><code>AgentAction()</code> to <code>OnActionReceived()</code></li> <li><code>AgentReset()</code> to <code>OnEpisodeBegin()</code></li> <li><code>Done()</code> to <code>EndEpisode()</code></li> <li><code>GiveModel()</code> to <code>SetModel()</code></li> <li>Replace <code>IFloatProperties</code> variables with <code>FloatPropertiesChannel</code> variables.</li> <li>If you implemented custom <code>SideChannels</code>, update the signatures of your   methods, and add your data to the <code>OutgoingMessage</code> or read it from the   <code>IncomingMessage</code>.</li> <li>Replace calls to Academy.EnableAutomaticStepping()/DisableAutomaticStepping()   with Academy.AutomaticSteppingEnabled = true/false.</li> </ul>"},{"location":"Migrating/#migrating-from-013-to-014","title":"Migrating from 0.13 to 0.14","text":""},{"location":"Migrating/#important-changes_5","title":"Important changes","text":"<ul> <li>The <code>UnitySDK</code> folder has been split into a Unity Package   (<code>com.unity.ml-agents</code>) and an examples project (<code>Project</code>). Please follow the   Installation Guide to get up and running with this new repo   structure.</li> <li>Several changes were made to how agents are reset and marked as done:</li> <li>Calling <code>Done()</code> on the Agent will now reset it immediately and call the     <code>AgentReset</code> virtual method. (This is to simplify the previous logic in     which the Agent had to wait for the next <code>EnvironmentStep</code> to reset)</li> <li>The \"Reset on Done\" setting in AgentParameters was removed; this is now     effectively always true. <code>AgentOnDone</code> virtual method on the Agent has been     removed.</li> <li>The <code>Decision Period</code> and <code>On Demand decision</code> checkbox have been removed from   the Agent. On demand decision is now the default (calling <code>RequestDecision</code> on   the Agent manually.)</li> <li>The Academy class was changed to a singleton, and its virtual methods were   removed.</li> <li>Trainer steps are now counted per-Agent, not per-environment as in previous   versions. For instance, if you have 10 Agents in the scene, 20 environment   steps now corresponds to 200 steps as printed in the terminal and in   Tensorboard.</li> <li>Curriculum config files are now YAML formatted and all curricula for a   training run are combined into a single file.</li> <li>The <code>--num-runs</code> command-line option has been removed from <code>mlagents-learn</code>.</li> <li>Several fields on the Agent were removed or made private in order to simplify   the interface.</li> <li>The <code>agentParameters</code> field of the Agent has been removed. (Contained only     <code>maxStep</code> information)</li> <li><code>maxStep</code> is now a public field on the Agent. (Was moved from     <code>agentParameters</code>)</li> <li>The <code>Info</code> field of the Agent has been made private. (Was only used     internally and not meant to be modified outside of the Agent)</li> <li>The <code>GetReward()</code> method on the Agent has been removed. (It was being     confused with <code>GetCumulativeReward()</code>)</li> <li>The <code>AgentAction</code> struct no longer contains a <code>value</code> field. (Value     estimates were not set during inference)</li> <li>The <code>GetValueEstimate()</code> method on the Agent has been removed.</li> <li>The <code>UpdateValueAction()</code> method on the Agent has been removed.</li> <li>The deprecated <code>RayPerception3D</code> and <code>RayPerception2D</code> classes were removed,   and the <code>legacyHitFractionBehavior</code> argument was removed from   <code>RayPerceptionSensor.PerceiveStatic()</code>.</li> <li>RayPerceptionSensor was inconsistent in how it handle scale on the Agent's   transform. It now scales the ray length and sphere size for casting as the   transform's scale changes.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_5","title":"Steps to Migrate","text":"<ul> <li>Follow the instructions on how to install the <code>com.unity.ml-agents</code> package   into your project in the Installation Guide.</li> <li>If your Agent implemented <code>AgentOnDone</code> and did not have the checkbox   <code>Reset On Done</code> checked in the inspector, you must call the code that was in   <code>AgentOnDone</code> manually.</li> <li>If you give your Agent a reward or penalty at the end of an episode (e.g. for   reaching a goal or falling off of a platform), make sure you call   <code>AddReward()</code> or <code>SetReward()</code> before calling <code>Done()</code>. Previously, the   order didn't matter.</li> <li>If you were not using <code>On Demand Decision</code> for your Agent, you must add a   <code>DecisionRequester</code> component to your Agent GameObject and set its   <code>Decision Period</code> field to the old <code>Decision Period</code> of the Agent.</li> <li>If you have a class that inherits from Academy:</li> <li>If the class didn't override any of the virtual methods and didn't store any     additional data, you can just remove the old script from the scene.</li> <li>If the class had additional data, create a new MonoBehaviour and store the     data in the new MonoBehaviour instead.</li> <li>If the class overrode the virtual methods, create a new MonoBehaviour and     move the logic to it:<ul> <li>Move the InitializeAcademy code to MonoBehaviour.Awake</li> <li>Move the AcademyStep code to MonoBehaviour.FixedUpdate</li> <li>Move the OnDestroy code to MonoBehaviour.OnDestroy.</li> <li>Move the AcademyReset code to a new method and add it to the   Academy.OnEnvironmentReset action.</li> </ul> </li> <li>Multiply <code>max_steps</code> and <code>summary_freq</code> in your <code>trainer_config.yaml</code> by the   number of Agents in the scene.</li> <li>Combine curriculum configs into a single file. See   the WallJump curricula for an example of   the new curriculum config format. A tool like https://www.json2yaml.com may be   useful to help with the conversion.</li> <li>If you have a model trained which uses RayPerceptionSensor and has non-1.0   scale in the Agent's transform, it must be retrained.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v0120-to-v0130","title":"Migrating from ML-Agents Toolkit v0.12.0 to v0.13.0","text":""},{"location":"Migrating/#important-changes_6","title":"Important changes","text":"<ul> <li>The low level Python API has changed. You can look at the document   Low Level Python API documentation for more information. This   should only affect you if you're writing a custom trainer; if you use   <code>mlagents-learn</code> for training, this should be a transparent change.</li> <li><code>reset()</code> on the Low-Level Python API no longer takes a <code>train_mode</code>     argument. To modify the performance/speed of the engine, you must use an     <code>EngineConfigurationChannel</code></li> <li><code>reset()</code> on the Low-Level Python API no longer takes a <code>config</code> argument.     <code>UnityEnvironment</code> no longer has a <code>reset_parameters</code> field. To modify float     properties in the environment, you must use a <code>FloatPropertiesChannel</code>. For     more information, refer to the     Low Level Python API documentation</li> <li><code>CustomResetParameters</code> are now removed.</li> <li>The Academy no longer has a <code>Training Configuration</code> nor   <code>Inference Configuration</code> field in the inspector. To modify the configuration   from the Low-Level Python API, use an <code>EngineConfigurationChannel</code>. To modify   it during training, use the new command line arguments <code>--width</code>, <code>--height</code>,   <code>--quality-level</code>, <code>--time-scale</code> and <code>--target-frame-rate</code> in   <code>mlagents-learn</code>.</li> <li>The Academy no longer has a <code>Default Reset Parameters</code> field in the inspector.   The Academy class no longer has a <code>ResetParameters</code>. To access shared float   properties with Python, use the new <code>FloatProperties</code> field on the Academy.</li> <li>Offline Behavioral Cloning has been removed. To learn from demonstrations, use   the GAIL and Behavioral Cloning features with either PPO or SAC.</li> <li><code>mlagents.envs</code> was renamed to <code>mlagents_envs</code>. The previous repo layout   depended on PEP420, which caused   problems with some of our tooling such as mypy and pylint.</li> <li>The official version of Unity ML-Agents supports is now 2022.3 LTS. If you run   into issues, please consider deleting your library folder and reponening your   projects. You will need to install the Sentis package into your project in   order to ML-Agents to compile correctly.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_6","title":"Steps to Migrate","text":"<ul> <li>If you had a custom <code>Training Configuration</code> in the Academy inspector, you   will need to pass your custom configuration at every training run using the   new command line arguments <code>--width</code>, <code>--height</code>, <code>--quality-level</code>,   <code>--time-scale</code> and <code>--target-frame-rate</code>.</li> <li>If you were using <code>--slow</code> in <code>mlagents-learn</code>, you will need to pass your old   <code>Inference Configuration</code> of the Academy inspector with the new command line   arguments <code>--width</code>, <code>--height</code>, <code>--quality-level</code>, <code>--time-scale</code> and   <code>--target-frame-rate</code> instead.</li> <li>Any imports from <code>mlagents.envs</code> should be replaced with <code>mlagents_envs</code>.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v0110-to-v0120","title":"Migrating from ML-Agents Toolkit v0.11.0 to v0.12.0","text":""},{"location":"Migrating/#important-changes_7","title":"Important Changes","text":"<ul> <li>Text actions and observations, and custom action and observation protos have   been removed.</li> <li>RayPerception3D and RayPerception2D are marked deprecated, and will be removed   in a future release. They can be replaced by RayPerceptionSensorComponent3D   and RayPerceptionSensorComponent2D.</li> <li>The <code>Use Heuristic</code> checkbox in Behavior Parameters has been replaced with a   <code>Behavior Type</code> dropdown menu. This has the following options:</li> <li><code>Default</code> corresponds to the previous unchecked behavior, meaning that     Agents will train if they connect to a python trainer, otherwise they will     perform inference.</li> <li><code>Heuristic Only</code> means the Agent will always use the <code>Heuristic()</code> method.     This corresponds to having \"Use Heuristic\" selected in 0.11.0.</li> <li><code>Inference Only</code> means the Agent will always perform inference.</li> <li>ML-Agents was upgraded to use Sentis 1.2.0-exp.2 and is installed via the package manager.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_7","title":"Steps to Migrate","text":"<ul> <li>We fixed a bug in   <code>RayPerception3d.Perceive()</code> that was causing the <code>endOffset</code> to be used   incorrectly. However this may produce different behavior from previous   versions if you use a non-zero <code>startOffset</code>. To reproduce the old behavior,   you should increase the value of <code>endOffset</code> by <code>startOffset</code>. You can   verify your raycasts are performing as expected in scene view using the debug   rays.</li> <li>If you use RayPerception3D, replace it with RayPerceptionSensorComponent3D   (and similarly for 2D). The settings, such as ray angles and detectable tags,   are configured on the component now. RayPerception3D would contribute   <code>(# of rays) * (# of tags + 2)</code> to the State Size in Behavior Parameters, but   this is no longer necessary, so you should reduce the State Size by this   amount. Making this change will require retraining your model, since the   observations that RayPerceptionSensorComponent3D produces are different from   the old behavior.</li> <li>If you see messages such as   <code>The type or namespace 'Sentis' could not be found</code> or   <code>The type or namespace 'Google' could not be found</code>, you will need to   install the Sentis preview package.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v010-to-v0110","title":"Migrating from ML-Agents Toolkit v0.10 to v0.11.0","text":""},{"location":"Migrating/#important-changes_8","title":"Important Changes","text":"<ul> <li>The definition of the gRPC service has changed.</li> <li>The online BC training feature has been removed.</li> <li>The BroadcastHub has been deprecated. If there is a training Python process,   all LearningBrains in the scene will automatically be trained. If there is no   Python process, inference will be used.</li> <li>The Brain ScriptableObjects have been deprecated. The Brain Parameters are now   on the Agent and are referred to as Behavior Parameters. Make sure the   Behavior Parameters is attached to the Agent GameObject.</li> <li>To use a heuristic behavior, implement the <code>Heuristic()</code> method in the Agent   class and check the <code>use heuristic</code> checkbox in the Behavior Parameters.</li> <li>Several changes were made to the setup for visual observations (i.e. using   Cameras or RenderTextures):</li> <li>Camera resolutions are no longer stored in the Brain Parameters.</li> <li>AgentParameters no longer stores lists of Cameras and RenderTextures</li> <li>To add visual observations to an Agent, you must now attach a     CameraSensorComponent or RenderTextureComponent to the agent. The     corresponding Camera or RenderTexture can be added to these in the editor,     and the resolution and color/grayscale is configured on the component     itself.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_8","title":"Steps to Migrate","text":"<ul> <li>In order to be able to train, make sure both your ML-Agents Python package and   UnitySDK code come from the v0.11 release. Training will not work, for   example, if you update the ML-Agents Python package, and only update the API   Version in UnitySDK.</li> <li>If your Agents used visual observations, you must add a CameraSensorComponent   corresponding to each old Camera in the Agent's camera list (and similarly for   RenderTextures).</li> <li>Since Brain ScriptableObjects have been removed, you will need to delete all   the Brain ScriptableObjects from your <code>Assets</code> folder. Then, add a   <code>Behavior Parameters</code> component to each <code>Agent</code> GameObject. You will then need   to complete the fields on the new <code>Behavior Parameters</code> component with the   BrainParameters of the old Brain.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v09-to-v010","title":"Migrating from ML-Agents Toolkit v0.9 to v0.10","text":""},{"location":"Migrating/#important-changes_9","title":"Important Changes","text":"<ul> <li>We have updated the C# code in our repository to be in line with Unity Coding   Conventions. This has changed the name of some public facing classes and   enums.</li> <li>The example environments have been updated. If you were using these   environments to benchmark your training, please note that the resulting   rewards may be slightly different in v0.10.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_9","title":"Steps to Migrate","text":"<ul> <li><code>UnitySDK/Assets/ML-Agents/Scripts/Communicator.cs</code> and its class   <code>Communicator</code> have been renamed to   <code>UnitySDK/Assets/ML-Agents/Scripts/ICommunicator.cs</code> and <code>ICommunicator</code>   respectively.</li> <li>The <code>SpaceType</code> Enums <code>discrete</code>, and <code>continuous</code> have been renamed to   <code>Discrete</code> and <code>Continuous</code>.</li> <li>We have removed the <code>Done</code> call as well as the capacity to set <code>Max Steps</code> on   the Academy. Therefore an AcademyReset will never be triggered from C# (only   from Python). If you want to reset the simulation after a fixed number of   steps, or when an event in the simulation occurs, we recommend looking at our   multi-agent example environments (such as FoodCollector). In our examples,   groups of Agents can be reset through an \"Area\" that can reset groups of   Agents.</li> <li>The import for <code>mlagents.envs.UnityEnvironment</code> was removed. If you are using   the Python API, change <code>from mlagents_envs import UnityEnvironment</code> to   <code>from mlagents_envs.environment import UnityEnvironment</code>.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v08-to-v09","title":"Migrating from ML-Agents Toolkit v0.8 to v0.9","text":""},{"location":"Migrating/#important-changes_10","title":"Important Changes","text":"<ul> <li>We have changed the way reward signals (including Curiosity) are defined in   the <code>trainer_config.yaml</code>.</li> <li>When using multiple environments, every \"step\" is recorded in TensorBoard.</li> <li>The steps in the command line console corresponds to a single step of a single   environment. Previously, each step corresponded to one step for all   environments (i.e., <code>num_envs</code> steps).</li> </ul>"},{"location":"Migrating/#steps-to-migrate_10","title":"Steps to Migrate","text":"<ul> <li>If you were overriding any of these following parameters in your config file,   remove them from the top-level config and follow the steps below:</li> <li><code>gamma</code>: Define a new <code>extrinsic</code> reward signal and set it's <code>gamma</code> to your     new gamma.</li> <li><code>use_curiosity</code>, <code>curiosity_strength</code>, <code>curiosity_enc_size</code>: Define a     <code>curiosity</code> reward signal and set its <code>strength</code> to <code>curiosity_strength</code>,     and <code>encoding_size</code> to <code>curiosity_enc_size</code>. Give it the same <code>gamma</code> as     your <code>extrinsic</code> signal to mimic previous behavior.</li> <li>TensorBoards generated when running multiple environments in v0.8 are not   comparable to those generated in v0.9 in terms of step count. Multiply your   v0.8 step count by <code>num_envs</code> for an approximate comparison. You may need to   change <code>max_steps</code> in your config as appropriate as well.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v07-to-v08","title":"Migrating from ML-Agents Toolkit v0.7 to v0.8","text":""},{"location":"Migrating/#important-changes_11","title":"Important Changes","text":"<ul> <li>We have split the Python packages into two separate packages <code>ml-agents</code> and   <code>ml-agents-envs</code>.</li> <li><code>--worker-id</code> option of <code>learn.py</code> has been removed, use <code>--base-port</code> instead   if you'd like to run multiple instances of <code>learn.py</code>.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_11","title":"Steps to Migrate","text":"<ul> <li>If you are installing via PyPI, there is no change.</li> <li>If you intend to make modifications to <code>ml-agents</code> or <code>ml-agents-envs</code> please   check the Installing for Development in the   Installation documentation.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v06-to-v07","title":"Migrating from ML-Agents Toolkit v0.6 to v0.7","text":""},{"location":"Migrating/#important-changes_12","title":"Important Changes","text":"<ul> <li>We no longer support TFS and are now using the   Sentis</li> </ul>"},{"location":"Migrating/#steps-to-migrate_12","title":"Steps to Migrate","text":"<ul> <li>Make sure to remove the <code>ENABLE_TENSORFLOW</code> flag in your Unity Project   settings</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v05-to-v06","title":"Migrating from ML-Agents Toolkit v0.5 to v0.6","text":""},{"location":"Migrating/#important-changes_13","title":"Important Changes","text":"<ul> <li>Brains are now Scriptable Objects instead of MonoBehaviors.</li> <li> <p>You can no longer modify the type of a Brain. If you want to switch between   <code>PlayerBrain</code> and <code>LearningBrain</code> for multiple agents, you will need to assign   a new Brain to each agent separately. Note: You can pass the same Brain to   multiple agents in a scene by leveraging Unity's prefab system or look for all   the agents in a scene using the search bar of the <code>Hierarchy</code> window with the   word <code>Agent</code>.</p> </li> <li> <p>We replaced the Internal and External Brain with Learning Brain.   When you need to train a model, you need to drag it into the <code>Broadcast Hub</code>   inside the <code>Academy</code> and check the <code>Control</code> checkbox.</p> </li> <li>We removed the <code>Broadcast</code> checkbox of the Brain, to use the broadcast   functionality, you need to drag the Brain into the <code>Broadcast Hub</code>.</li> <li>When training multiple Brains at the same time, each model is now stored into   a separate model file rather than in the same file under different graph   scopes.</li> <li>The Learning Brain graph scope, placeholder names, output names and custom   placeholders can no longer be modified.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_13","title":"Steps to Migrate","text":"<ul> <li>To update a scene from v0.5 to v0.6, you must:</li> <li>Remove the <code>Brain</code> GameObjects in the scene. (Delete all of the Brain     GameObjects under Academy in the scene.)</li> <li>Create new <code>Brain</code> Scriptable Objects using <code>Assets -&gt; Create -&gt; ML-Agents</code>     for each type of the Brain you plan to use, and put the created files under     a folder called Brains within your project.</li> <li>Edit their <code>Brain Parameters</code> to be the same as the parameters used in the     <code>Brain</code> GameObjects.</li> <li>Agents have a <code>Brain</code> field in the Inspector, you need to drag the     appropriate Brain ScriptableObject in it.</li> <li>The Academy has a <code>Broadcast Hub</code> field in the inspector, which is list of     brains used in the scene. To train or control your Brain from the     <code>mlagents-learn</code> Python script, you need to drag the relevant     <code>LearningBrain</code> ScriptableObjects used in your scene into entries into this     list.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v04-to-v05","title":"Migrating from ML-Agents Toolkit v0.4 to v0.5","text":""},{"location":"Migrating/#important","title":"Important","text":"<ul> <li>The Unity project <code>unity-environment</code> has been renamed <code>UnitySDK</code>.</li> <li>The <code>python</code> folder has been renamed to <code>ml-agents</code>. It now contains two   packages, <code>mlagents.env</code> and <code>mlagents.trainers</code>. <code>mlagents.env</code> can be used   to interact directly with a Unity environment, while <code>mlagents.trainers</code>   contains the classes for training agents.</li> <li>The supported Unity version has changed from <code>2017.1 or later</code> to   <code>2017.4 or later</code>. 2017.4 is an LTS (Long Term Support) version that helps us   maintain good quality and support. Earlier versions of Unity might still work,   but you may encounter an   error listed here.</li> </ul>"},{"location":"Migrating/#unity-api","title":"Unity API","text":"<ul> <li>Discrete Actions now use branches. You can   now specify concurrent discrete actions. You will need to update the Brain   Parameters in the Brain Inspector in all your environments that use discrete   actions. Refer to the   discrete action documentation   for more information.</li> </ul>"},{"location":"Migrating/#python-api","title":"Python API","text":"<ul> <li>In order to run a training session, you can now use the command   <code>mlagents-learn</code> instead of <code>python3 learn.py</code> after installing the <code>mlagents</code>   packages. This change is documented   here. For example, if we   previously ran</li> </ul> <p><code>sh   python3 learn.py 3DBall --train</code></p> <p>from the <code>python</code> subdirectory (which is changed to <code>ml-agents</code> subdirectory   in v0.5), we now run</p> <p><code>sh   mlagents-learn config/trainer_config.yaml --env=3DBall --train</code></p> <p>from the root directory where we installed the ML-Agents Toolkit.</p> <ul> <li>It is now required to specify the path to the yaml trainer configuration file   when running <code>mlagents-learn</code>. For an example trainer configuration file, see   trainer_config.yaml. An example of passing a   trainer configuration to <code>mlagents-learn</code> is shown above.</li> <li>The environment name is now passed through the <code>--env</code> option.</li> <li>Curriculum learning has been changed. In summary:</li> <li>Curriculum files for the same environment must now be placed into a folder.     Each curriculum file should be named after the Brain whose curriculum it     specifies.</li> <li><code>min_lesson_length</code> now specifies the minimum number of episodes in a lesson     and affects reward thresholding.</li> <li>It is no longer necessary to specify the <code>Max Steps</code> of the Academy to use     curriculum learning.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v03-to-v04","title":"Migrating from ML-Agents Toolkit v0.3 to v0.4","text":""},{"location":"Migrating/#unity-api_1","title":"Unity API","text":"<ul> <li><code>using MLAgents;</code> needs to be added in all of the C# scripts that use   ML-Agents.</li> </ul>"},{"location":"Migrating/#python-api_1","title":"Python API","text":"<ul> <li>We've changed some of the Python packages dependencies in requirement.txt   file. Make sure to run <code>pip3 install -e .</code> within your <code>ml-agents/python</code>   folder to update your Python packages.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v02-to-v03","title":"Migrating from ML-Agents Toolkit v0.2 to v0.3","text":"<p>There are a large number of new features and improvements in the ML-Agents toolkit v0.3 which change both the training process and Unity API in ways which will cause incompatibilities with environments made using older versions. This page is designed to highlight those changes for users familiar with v0.1 or v0.2 in order to ensure a smooth transition.</p>"},{"location":"Migrating/#important_1","title":"Important","text":"<ul> <li>The ML-Agents Toolkit is no longer compatible with Python 2.</li> </ul>"},{"location":"Migrating/#python-training","title":"Python Training","text":"<ul> <li>The training script <code>ppo.py</code> and <code>PPO.ipynb</code> Python notebook have been   replaced with a single <code>learn.py</code> script as the launching point for training   with ML-Agents. For more information on using <code>learn.py</code>, see   here.</li> <li>Hyperparameters for training Brains are now stored in the   <code>trainer_config.yaml</code> file. For more information on using this file, see   here.</li> </ul>"},{"location":"Migrating/#unity-api_2","title":"Unity API","text":"<ul> <li>Modifications to an Agent's rewards must now be done using either   <code>AddReward()</code> or <code>SetReward()</code>.</li> <li>Setting an Agent to done now requires the use of the <code>Done()</code> method.</li> <li><code>CollectStates()</code> has been replaced by <code>CollectObservations()</code>, which now no   longer returns a list of floats.</li> <li>To collect observations, call <code>AddVectorObs()</code> within <code>CollectObservations()</code>.   Note that you can call <code>AddVectorObs()</code> with floats, integers, lists and   arrays of floats, Vector3 and Quaternions.</li> <li><code>AgentStep()</code> has been replaced by <code>AgentAction()</code>.</li> <li><code>WaitTime()</code> has been removed.</li> <li>The <code>Frame Skip</code> field of the Academy is replaced by the Agent's   <code>Decision Frequency</code> field, enabling the Agent to make decisions at different   frequencies.</li> <li>The names of the inputs in the Internal Brain have been changed. You must   replace <code>state</code> with <code>vector_observation</code> and <code>observation</code> with   <code>visual_observation</code>. In addition, you must remove the <code>epsilon</code> placeholder.</li> </ul>"},{"location":"Migrating/#semantics","title":"Semantics","text":"<p>In order to more closely align with the terminology used in the Reinforcement Learning field, and to be more descriptive, we have changed the names of some of the concepts used in ML-Agents. The changes are highlighted in the table below.</p> Old - v0.2 and earlier New - v0.3 and later State Vector Observation Observation Visual Observation Action Vector Action N/A Text Observation N/A Text Action"},{"location":"Package-Settings/","title":"ML-Agents Package Settings","text":"<p>ML-Agents Package Settings contains settings that apply to the whole project. It allows you to configure ML-Agents-specific settings in the Editor.  These settings are available for use in both the Editor and Player.</p> <p>You can find them at <code>Edit</code> &gt; <code>Project Settings...</code> &gt; <code>ML-Agents</code>. It lists out all the available settings and their default values.</p>"},{"location":"Package-Settings/#create-custom-settings","title":"Create Custom Settings","text":"<p>In order to to use your own settings for your project, you'll need to create a settings asset.</p> <p>You can do this by clicking the <code>Create Settings Asset</code> buttom or clicking the gear on the top right and select <code>New Settings Asset...</code>. The asset file can be placed anywhere in the <code>Asset/</code> folder in your project. After Creating the settings asset, you'll be able to modify the settings for your project and your settings will be saved in the asset.</p> <p></p>"},{"location":"Package-Settings/#multiple-custom-settings-for-different-scenarios","title":"Multiple Custom Settings for Different Scenarios","text":"<p>You can create multiple settings assets in one project.</p> <p>By clicking the gear on the top right you'll see all available settings listed in the drop-down menu to choose from.</p> <p>This allows you to create different settings for different scenatios. For example, you can create two separate settings for training and inference, and specify which one you want to use according to what you're currently running.</p> <p></p>"},{"location":"Profiling-Python/","title":"Profiling in Python","text":"<p>As part of the ML-Agents Tookit, we provide a lightweight profiling system, in order to identity hotspots in the training process and help spot regressions from changes.</p> <p>Timers are hierarchical, meaning that the time tracked in a block of code can be further split into other blocks if desired. This also means that a function that is called from multiple places in the code will appear in multiple places in the timing output.</p> <p>All timers operate using a \"global\" instance by default, but this can be overridden if necessary (mainly for testing).</p>"},{"location":"Profiling-Python/#adding-profiling","title":"Adding Profiling","text":"<p>There are two ways to indicate code should be included in profiling. The simplest way is to add the <code>@timed</code> decorator to a function or method of interested.</p> <pre><code>class TrainerController:\n    # ....\n    @timed\n    def advance(self, env: EnvManager) -&gt; int:\n        # do stuff\n</code></pre> <p>You can also used the <code>hierarchical_timer</code> context manager.</p> <pre><code>with hierarchical_timer(\"communicator.exchange\"):\n    outputs = self.communicator.exchange(step_input)\n</code></pre> <p>The context manager may be easier than the <code>@timed</code> decorator for profiling different parts of a large function, or profiling calls to abstract methods that might not use decorator.</p>"},{"location":"Profiling-Python/#output","title":"Output","text":"<p>By default, at the end of training, timers are collected and written in json format to <code>{summaries_dir}/{run_id}_timers.json</code>. The output consists of node objects with the following keys:</p> <ul> <li>total (float): The total time in seconds spent in the block, including child   calls.</li> <li>count (int): The number of times the block was called.</li> <li>self (float): The total time in seconds spent in the block, excluding child   calls.</li> <li>children (dictionary): A dictionary of child nodes, keyed by the node name.</li> <li>is_parallel (bool): Indicates that the block of code was executed in multiple   threads or processes (see below). This is optional and defaults to false.</li> </ul>"},{"location":"Profiling-Python/#parallel-execution","title":"Parallel execution","text":""},{"location":"Profiling-Python/#subprocesses","title":"Subprocesses","text":"<p>For code that executes in multiple processes (for example, SubprocessEnvManager), we periodically send the timer information back to the \"main\" process, aggregate the timers there, and flush them in the subprocess. Note that (depending on the number of processes) this can result in timers where the total time may exceed the parent's total time. This is analogous to the difference between \"real\" and \"user\" values reported from the unix <code>time</code> command. In the timer output, blocks that were run in parallel are indicated by the <code>is_parallel</code> flag.</p>"},{"location":"Profiling-Python/#threads","title":"Threads","text":"<p>Timers currently use <code>time.perf_counter()</code> to track time spent, which may not give accurate results for multiple threads. If this is problematic, set <code>threaded: false</code> in your trainer configuration.</p>"},{"location":"Python-Custom-Trainer-Plugin/","title":"Unity Ml-Agents Custom trainers Plugin","text":"<p>As an attempt to bring a wider variety of reinforcement learning algorithms to our users, we have added custom trainers capabilities. we introduce an extensible plugin system to define new trainers based on the High level trainer API in <code>Ml-agents</code> Package. This will allow rerouting <code>mlagents-learn</code> CLI to custom trainers and extending the config files with hyper-parameters specific to your new trainers. We will expose a high-level extensible trainer (both on-policy, and off-policy trainers) optimizer and hyperparameter classes with documentation for the use of this plugin. For more infromation on how python plugin system works see Plugin interfaces.</p>"},{"location":"Python-Custom-Trainer-Plugin/#overview","title":"Overview","text":"<p>Model-free RL algorithms generally fall into two broad categories: on-policy and off-policy. On-policy algorithms perform updates based on data gathered from the current policy. Off-policy algorithms learn a Q function from a buffer of previous data, then use this Q function to make decisions. Off-policy algorithms have three key benefits in the context of ML-Agents: They tend to use fewer samples than on-policy as they can pull and re-use data from the buffer many times. They allow player demonstrations to be inserted in-line with RL data into the buffer, enabling new ways of doing imitation learning by streaming player data.</p> <p>To add new custom trainers to ML-agents, you would need to create a new python package. To give you an idea of how to structure your package, we have created a mlagents_trainer_plugin package ourselves as an example, with implementation of <code>A2c</code> and <code>DQN</code> algorithms. You would need a <code>setup.py</code> file to list extra requirements and register the new RL algorithm in ml-agents ecosystem and be able to call <code>mlagents-learn</code> CLI with your customized configuration.</p> <pre><code>\u251c\u2500\u2500 mlagents_trainer_plugin\n\u2502    \u251c\u2500\u2500 __init__.py\n\u2502    \u251c\u2500\u2500 a2c\n\u2502    \u2502    \u251c\u2500\u2500 __init__.py\n\u2502    \u2502    \u251c\u2500\u2500 a2c_3DBall.yaml\n\u2502    \u2502    \u251c\u2500\u2500 a2c_optimizer.py\n\u2502    \u2502    \u2514\u2500\u2500 a2c_trainer.py\n\u2502    \u2514\u2500\u2500 dqn\n\u2502        \u251c\u2500\u2500 __init__.py\n\u2502        \u251c\u2500\u2500 dqn_basic.yaml\n\u2502        \u251c\u2500\u2500 dqn_optimizer.py\n\u2502        \u2514\u2500\u2500 dqn_trainer.py\n\u2514\u2500\u2500 setup.py\n</code></pre>"},{"location":"Python-Custom-Trainer-Plugin/#installation-and-execution","title":"Installation and Execution","text":"<p>If you haven't already, follow the installation instructions. Once you have the <code>ml-agents-env</code> and <code>ml-agents</code> packages you can install the plugin package. From the repository's root directory install <code>ml-agents-trainer-plugin</code> (or replace with the name of your plugin folder).</p> <pre><code>pip3 install -e &lt;./ml-agents-trainer-plugin&gt;\n</code></pre> <p>Following the previous installations your package is added as an entrypoint and you can use a config file with new trainers:</p> <pre><code>mlagents-learn ml-agents-trainer-plugin/mlagents_trainer_plugin/a2c/a2c_3DBall.yaml --run-id &lt;run-id-name&gt;\n--env &lt;env-executable&gt;\n</code></pre>"},{"location":"Python-Custom-Trainer-Plugin/#tutorial","title":"Tutorial","text":"<p>Here\u2019s a step-by-step tutorial on how to write a setup file and extend ml-agents trainers, optimizers, and hyperparameter settings.To extend ML-agents classes see references on trainers and Optimizer.</p>"},{"location":"Python-Gym-API-Documentation/","title":"Table of Contents","text":"<ul> <li>mlagents_envs.envs.unity_gym_env</li> <li>UnityGymException</li> <li>UnityToGymWrapper<ul> <li>__init__</li> <li>reset</li> <li>step</li> <li>render</li> <li>close</li> <li>seed</li> </ul> </li> <li>ActionFlattener<ul> <li>__init__</li> <li>lookup_action</li> </ul> </li> </ul>"},{"location":"Python-Gym-API-Documentation/#mlagents_envsenvsunity_gym_env","title":"mlagents_envs.envs.unity_gym_env","text":""},{"location":"Python-Gym-API-Documentation/#unitygymexception-objects","title":"UnityGymException Objects","text":"<pre><code>class UnityGymException(error.Error)\n</code></pre> <p>Any error related to the gym wrapper of ml-agents.</p> <p></p>"},{"location":"Python-Gym-API-Documentation/#unitytogymwrapper-objects","title":"UnityToGymWrapper Objects","text":"<pre><code>class UnityToGymWrapper(gym.Env)\n</code></pre> <p>Provides Gym wrapper for Unity Learning Environments.</p> <p></p>"},{"location":"Python-Gym-API-Documentation/#__init__","title":"__init__","text":"<pre><code> | __init__(unity_env: BaseEnv, uint8_visual: bool = False, flatten_branched: bool = False, allow_multiple_obs: bool = False, action_space_seed: Optional[int] = None)\n</code></pre> <p>Environment initialization</p> <p>Arguments:</p> <ul> <li><code>unity_env</code>: The Unity BaseEnv to be wrapped in the gym. Will be closed when the UnityToGymWrapper closes.</li> <li><code>uint8_visual</code>: Return visual observations as uint8 (0-255) matrices instead of float (0.0-1.0).</li> <li><code>flatten_branched</code>: If True, turn branched discrete action spaces into a Discrete space rather than     MultiDiscrete.</li> <li><code>allow_multiple_obs</code>: If True, return a list of np.ndarrays as observations with the first elements     containing the visual observations and the last element containing the array of vector observations.     If False, returns a single np.ndarray containing either only a single visual observation or the array of     vector observations.</li> <li><code>action_space_seed</code>: If non-None, will be used to set the random seed on created gym.Space instances.</li> </ul> <p></p>"},{"location":"Python-Gym-API-Documentation/#reset","title":"reset","text":"<pre><code> | reset() -&gt; Union[List[np.ndarray], np.ndarray]\n</code></pre> <p>Resets the state of the environment and returns an initial observation. Returns: observation (object/list): the initial observation of the space.</p> <p></p>"},{"location":"Python-Gym-API-Documentation/#step","title":"step","text":"<pre><code> | step(action: List[Any]) -&gt; GymStepResult\n</code></pre> <p>Run one timestep of the environment's dynamics. When end of episode is reached, you are responsible for calling <code>reset()</code> to reset this environment's state. Accepts an action and returns a tuple (observation, reward, done, info).</p> <p>Arguments:</p> <ul> <li><code>action</code> object/list - an action provided by the environment</li> </ul> <p>Returns:</p> <ul> <li><code>observation</code> object/list - agent's observation of the current environment   reward (float/list) : amount of reward returned after previous action</li> <li><code>done</code> boolean/list - whether the episode has ended.</li> <li><code>info</code> dict - contains auxiliary diagnostic information.</li> </ul> <p></p>"},{"location":"Python-Gym-API-Documentation/#render","title":"render","text":"<pre><code> | render(mode=\"rgb_array\")\n</code></pre> <p>Return the latest visual observations. Note that it will not render a new frame of the environment.</p> <p></p>"},{"location":"Python-Gym-API-Documentation/#close","title":"close","text":"<pre><code> | close() -&gt; None\n</code></pre> <p>Override _close in your subclass to perform any necessary cleanup. Environments will automatically close() themselves when garbage collected or when the program exits.</p> <p></p>"},{"location":"Python-Gym-API-Documentation/#seed","title":"seed","text":"<pre><code> | seed(seed: Any = None) -&gt; None\n</code></pre> <p>Sets the seed for this env's random number generator(s). Currently not implemented.</p> <p></p>"},{"location":"Python-Gym-API-Documentation/#actionflattener-objects","title":"ActionFlattener Objects","text":"<pre><code>class ActionFlattener()\n</code></pre> <p>Flattens branched discrete action spaces into single-branch discrete action spaces.</p> <p></p>"},{"location":"Python-Gym-API-Documentation/#__init___1","title":"__init__","text":"<pre><code> | __init__(branched_action_space)\n</code></pre> <p>Initialize the flattener.</p> <p>Arguments:</p> <ul> <li><code>branched_action_space</code>: A List containing the sizes of each branch of the action space, e.g. [2,3,3] for three branches with size 2, 3, and 3 respectively.</li> </ul> <p></p>"},{"location":"Python-Gym-API-Documentation/#lookup_action","title":"lookup_action","text":"<pre><code> | lookup_action(action)\n</code></pre> <p>Convert a scalar discrete action into a unique set of branched actions.</p> <p>Arguments:</p> <ul> <li><code>action</code>: A scalar value representing one of the discrete actions.</li> </ul> <p>Returns:</p> <p>The List containing the branched actions.</p>"},{"location":"Python-Gym-API/","title":"Unity ML-Agents Gym Wrapper","text":"<p>A common way in which machine learning researchers interact with simulation environments is via a wrapper provided by OpenAI called <code>gym</code>. For more information on the gym interface, see here.</p> <p>We provide a gym wrapper and instructions for using it with existing machine learning algorithms which utilize gym. Our wrapper provides interfaces on top of our <code>UnityEnvironment</code> class, which is the default way of interfacing with a Unity environment via Python.</p>"},{"location":"Python-Gym-API/#installation","title":"Installation","text":"<p>The gym wrapper is part of the <code>mlgents_envs</code> package. Please refer to the mlagents_envs installation instructions.</p>"},{"location":"Python-Gym-API/#using-the-gym-wrapper","title":"Using the Gym Wrapper","text":"<p>The gym interface is available from <code>gym_unity.envs</code>. To launch an environment from the root of the project repository use:</p> <pre><code>from mlagents_envs.envs.unity_gym_env import UnityToGymWrapper\n\nenv = UnityToGymWrapper(unity_env, uint8_visual, flatten_branched, allow_multiple_obs)\n</code></pre> <ul> <li> <p><code>unity_env</code> refers to the Unity environment to be wrapped.</p> </li> <li> <p><code>uint8_visual</code> refers to whether to output visual observations as <code>uint8</code>   values (0-255). Many common Gym environments (e.g. Atari) do this. By default   they will be floats (0.0-1.0). Defaults to <code>False</code>.</p> </li> <li> <p><code>flatten_branched</code> will flatten a branched discrete action space into a Gym   Discrete. Otherwise, it will be converted into a MultiDiscrete. Defaults to   <code>False</code>.</p> </li> <li> <p><code>allow_multiple_obs</code> will return a list of observations. The first elements   contain the visual observations and the last element contains the array of   vector observations. If False the environment returns a single array (containing   a single visual observations, if present, otherwise the vector observation).   Defaults to <code>False</code>.</p> </li> <li> <p><code>action_space_seed</code> is the optional seed for action sampling. If non-None, will   be used to set the random seed on created gym.Space instances.</p> </li> </ul> <p>The returned environment <code>env</code> will function as a gym.</p>"},{"location":"Python-Gym-API/#limitations","title":"Limitations","text":"<ul> <li>It is only possible to use an environment with a single Agent.</li> <li>By default, the first visual observation is provided as the <code>observation</code>, if   present. Otherwise, vector observations are provided. You can receive all   visual and vector observations by using the <code>allow_multiple_obs=True</code> option in   the gym parameters. If set to <code>True</code>, you will receive a list of <code>observation</code>   instead of only one.</li> <li>The <code>TerminalSteps</code> or <code>DecisionSteps</code> output from the environment can still   be accessed from the <code>info</code> provided by <code>env.step(action)</code>.</li> <li>Stacked vector observations are not supported.</li> <li>Environment registration for use with <code>gym.make()</code> is currently not supported.</li> <li>Calling env.render() will not render a new frame of the environment. It will   return the latest visual observation if using visual observations.</li> </ul>"},{"location":"Python-Gym-API/#running-openai-baselines-algorithms","title":"Running OpenAI Baselines Algorithms","text":"<p>OpenAI provides a set of open-source maintained and tested Reinforcement Learning algorithms called the Baselines.</p> <p>Using the provided Gym wrapper, it is possible to train ML-Agents environments using these algorithms. This requires the creation of custom training scripts to launch each algorithm. In most cases these scripts can be created by making slight modifications to the ones provided for Atari and Mujoco environments.</p> <p>These examples were tested with baselines version 0.1.6.</p>"},{"location":"Python-Gym-API/#example-dqn-baseline","title":"Example - DQN Baseline","text":"<p>In order to train an agent to play the <code>GridWorld</code> environment using the Baselines DQN algorithm, you first need to install the baselines package using pip:</p> <pre><code>pip install git+git://github.com/openai/baselines\n</code></pre> <p>Next, create a file called <code>train_unity.py</code>. Then create an <code>/envs/</code> directory and build the environment to that directory. For more information on building Unity environments, see here. Note that because of limitations of the DQN baseline, the environment must have a single visual observation, a single discrete action and a single Agent in the scene. Add the following code to the <code>train_unity.py</code> file:</p> <pre><code>import gym\n\nfrom baselines import deepq\nfrom baselines import logger\n\nfrom mlagents_envs.environment import UnityEnvironment\nfrom mlagents_envs.envs.unity_gym_env import UnityToGymWrapper\n\n\ndef main():\n  unity_env = UnityEnvironment( &lt; path - to - environment &gt;)\n  env = UnityToGymWrapper(unity_env, uint8_visual=True)\n  logger.configure('./logs')  # Change to log in a different directory\n  act = deepq.learn(\n    env,\n    \"cnn\",  # For visual inputs\n    lr=2.5e-4,\n    total_timesteps=1000000,\n    buffer_size=50000,\n    exploration_fraction=0.05,\n    exploration_final_eps=0.1,\n    print_freq=20,\n    train_freq=5,\n    learning_starts=20000,\n    target_network_update_freq=50,\n    gamma=0.99,\n    prioritized_replay=False,\n    checkpoint_freq=1000,\n    checkpoint_path='./logs',  # Change to save model in a different directory\n    dueling=True\n  )\n  print(\"Saving model to unity_model.pkl\")\n  act.save(\"unity_model.pkl\")\n\n\nif __name__ == '__main__':\n  main()\n</code></pre> <p>To start the training process, run the following from the directory containing <code>train_unity.py</code>:</p> <pre><code>python -m train_unity\n</code></pre>"},{"location":"Python-Gym-API/#other-algorithms","title":"Other Algorithms","text":"<p>Other algorithms in the Baselines repository can be run using scripts similar to the examples from the baselines package. In most cases, the primary changes needed to use a Unity environment are to import <code>UnityToGymWrapper</code>, and to replace the environment creation code, typically <code>gym.make()</code>, with a call to <code>UnityToGymWrapper(unity_environment)</code> passing the environment as input.</p> <p>A typical rule of thumb is that for vision-based environments, modification should be done to Atari training scripts, and for vector observation environments, modification should be done to Mujoco scripts.</p> <p>Some algorithms will make use of <code>make_env()</code> or <code>make_mujoco_env()</code> functions. You can define a similar function for Unity environments. An example of such a method using the PPO2 baseline:</p> <pre><code>from mlagents_envs.environment import UnityEnvironment\nfrom mlagents_envs.envs import UnityToGymWrapper\nfrom baselines.common.vec_env.subproc_vec_env import SubprocVecEnv\nfrom baselines.common.vec_env.dummy_vec_env import DummyVecEnv\nfrom baselines.bench import Monitor\nfrom baselines import logger\nimport baselines.ppo2.ppo2 as ppo2\n\nimport os\n\ntry:\n  from mpi4py import MPI\nexcept ImportError:\n  MPI = None\n\n\ndef make_unity_env(env_directory, num_env, visual, start_index=0):\n  \"\"\"\n  Create a wrapped, monitored Unity environment.\n  \"\"\"\n\n  def make_env(rank, use_visual=True):  # pylint: disable=C0111\n    def _thunk():\n      unity_env = UnityEnvironment(env_directory, base_port=5000 + rank)\n      env = UnityToGymWrapper(unity_env, uint8_visual=True)\n      env = Monitor(env, logger.get_dir() and os.path.join(logger.get_dir(), str(rank)))\n      return env\n\n    return _thunk\n\n  if visual:\n    return SubprocVecEnv([make_env(i + start_index) for i in range(num_env)])\n  else:\n    rank = MPI.COMM_WORLD.Get_rank() if MPI else 0\n    return DummyVecEnv([make_env(rank, use_visual=False)])\n\n\ndef main():\n  env = make_unity_env( &lt; path - to - environment &gt;, 4, True)\n  ppo2.learn(\n    network=\"mlp\",\n    env=env,\n    total_timesteps=100000,\n    lr=1e-3,\n  )\n\n\nif __name__ == '__main__':\n  main()\n</code></pre>"},{"location":"Python-Gym-API/#run-google-dopamine-algorithms","title":"Run Google Dopamine Algorithms","text":"<p>Google provides a framework Dopamine, and implementations of algorithms, e.g. DQN, Rainbow, and the C51 variant of Rainbow. Using the Gym wrapper, we can run Unity environments using Dopamine.</p> <p>First, after installing the Gym wrapper, clone the Dopamine repository.</p> <pre><code>git clone https://github.com/google/dopamine\n</code></pre> <p>Then, follow the appropriate install instructions as specified on Dopamine's homepage. Note that the Dopamine guide specifies using a virtualenv. If you choose to do so, make sure your unity_env package is also installed within the same virtualenv as Dopamine.</p>"},{"location":"Python-Gym-API/#adapting-dopamines-scripts","title":"Adapting Dopamine's Scripts","text":"<p>First, open <code>dopamine/atari/run_experiment.py</code>. Alternatively, copy the entire <code>atari</code> folder, and name it something else (e.g. <code>unity</code>). If you choose the copy approach, be sure to change the package names in the import statements in <code>train.py</code> to your new directory.</p> <p>Within <code>run_experiment.py</code>, we will need to make changes to which environment is instantiated, just as in the Baselines example. At the top of the file, insert</p> <pre><code>from mlagents_envs.environment import UnityEnvironment\nfrom mlagents_envs.envs import UnityToGymWrapper\n</code></pre> <p>to import the Gym Wrapper. Navigate to the <code>create_atari_environment</code> method in the same file, and switch to instantiating a Unity environment by replacing the method with the following code.</p> <pre><code>    game_version = 'v0' if sticky_actions else 'v4'\n    full_game_name = '{}NoFrameskip-{}'.format(game_name, game_version)\n    unity_env = UnityEnvironment(&lt;path-to-environment&gt;)\n    env = UnityToGymWrapper(unity_env, uint8_visual=True)\n    return env\n</code></pre> <p><code>&lt;path-to-environment&gt;</code> is the path to your built Unity executable. For more information on building Unity environments, see here, and note the Limitations section below.</p> <p>Note that we are not using the preprocessor from Dopamine, as it uses many Atari-specific calls. Furthermore, frame-skipping can be done from within Unity, rather than on the Python side.</p>"},{"location":"Python-Gym-API/#limitations_1","title":"Limitations","text":"<p>Since Dopamine is designed around variants of DQN, it is only compatible with discrete action spaces, and specifically the Discrete Gym space. For environments that use branched discrete action spaces, you can enable the <code>flatten_branched</code> parameter in <code>UnityToGymWrapper</code>, which treats each combination of branched actions as separate actions.</p> <p>Furthermore, when building your environments, ensure that your Agent is using visual observations with greyscale enabled, and that the dimensions of the visual observations is 84 by 84 (matches the parameter found in <code>dqn_agent.py</code> and <code>rainbow_agent.py</code>). Dopamine's agents currently do not automatically adapt to the observation dimensions or number of channels.</p>"},{"location":"Python-Gym-API/#hyperparameters","title":"Hyperparameters","text":"<p>The hyperparameters provided by Dopamine are tailored to the Atari games, and you will likely need to adjust them for ML-Agents environments. Here is a sample <code>dopamine/agents/rainbow/configs/rainbow.gin</code> file that is known to work with a simple GridWorld.</p> <pre><code>import dopamine.agents.rainbow.rainbow_agent\nimport dopamine.unity.run_experiment\nimport dopamine.replay_memory.prioritized_replay_buffer\nimport gin.tf.external_configurables\n\nRainbowAgent.num_atoms = 51\nRainbowAgent.stack_size = 1\nRainbowAgent.vmax = 10.\nRainbowAgent.gamma = 0.99\nRainbowAgent.update_horizon = 3\nRainbowAgent.min_replay_history = 20000  # agent steps\nRainbowAgent.update_period = 5\nRainbowAgent.target_update_period = 50  # agent steps\nRainbowAgent.epsilon_train = 0.1\nRainbowAgent.epsilon_eval = 0.01\nRainbowAgent.epsilon_decay_period = 50000  # agent steps\nRainbowAgent.replay_scheme = 'prioritized'\nRainbowAgent.tf_device = '/cpu:0'  # use '/cpu:*' for non-GPU version\nRainbowAgent.optimizer = @tf.train.AdamOptimizer()\n\ntf.train.AdamOptimizer.learning_rate = 0.00025\ntf.train.AdamOptimizer.epsilon = 0.0003125\n\nRunner.game_name = \"Unity\" # any name can be used here\nRunner.sticky_actions = False\nRunner.num_iterations = 200\nRunner.training_steps = 10000  # agent steps\nRunner.evaluation_steps = 500  # agent steps\nRunner.max_steps_per_episode = 27000  # agent steps\n\nWrappedPrioritizedReplayBuffer.replay_capacity = 1000000\nWrappedPrioritizedReplayBuffer.batch_size = 32\n</code></pre> <p>This example assumed you copied <code>atari</code> to a separate folder named <code>unity</code>. Replace <code>unity</code> in <code>import dopamine.unity.run_experiment</code> with the folder you copied your <code>run_experiment.py</code> and <code>trainer.py</code> files to. If you directly modified the existing files, then use <code>atari</code> here.</p>"},{"location":"Python-Gym-API/#starting-a-run","title":"Starting a Run","text":"<p>You can now run Dopamine as you would normally:</p> <pre><code>python -um dopamine.unity.train \\\n  --agent_name=rainbow \\\n  --base_dir=/tmp/dopamine \\\n  --gin_files='dopamine/agents/rainbow/configs/rainbow.gin'\n</code></pre> <p>Again, we assume that you've copied <code>atari</code> into a separate folder. Remember to replace <code>unity</code> with the directory you copied your files into. If you edited the Atari files directly, this should be <code>atari</code>.</p>"},{"location":"Python-Gym-API/#example-gridworld","title":"Example: GridWorld","text":"<p>As a baseline, here are rewards over time for the three algorithms provided with Dopamine as run on the GridWorld example environment. All Dopamine (DQN, Rainbow, C51) runs were done with the same epsilon, epsilon decay, replay history, training steps, and buffer settings as specified above. Note that the first 20000 steps are used to pre-fill the training buffer, and no learning happens.</p> <p>We provide results from our PPO implementation and the DQN from Baselines as reference. Note that all runs used the same greyscale GridWorld as Dopamine. For PPO, <code>num_layers</code> was set to 2, and all other hyperparameters are the default for GridWorld in <code>config/ppo/GridWorld.yaml</code>. For Baselines DQN, the provided hyperparameters in the previous section are used. Note that Baselines implements certain features (e.g. dueling-Q) that are not enabled in Dopamine DQN.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/","title":"Table of Contents","text":"<ul> <li>mlagents_envs.base_env</li> <li>DecisionStep</li> <li>DecisionSteps<ul> <li>agent_id_to_index</li> <li>__getitem__</li> <li>empty</li> </ul> </li> <li>TerminalStep</li> <li>TerminalSteps<ul> <li>agent_id_to_index</li> <li>__getitem__</li> <li>empty</li> </ul> </li> <li>ActionTuple<ul> <li>discrete_dtype</li> </ul> </li> <li>ActionSpec<ul> <li>is_discrete</li> <li>is_continuous</li> <li>discrete_size</li> <li>empty_action</li> <li>random_action</li> <li>create_continuous</li> <li>create_discrete</li> <li>create_hybrid</li> </ul> </li> <li>DimensionProperty<ul> <li>UNSPECIFIED</li> <li>NONE</li> <li>TRANSLATIONAL_EQUIVARIANCE</li> <li>VARIABLE_SIZE</li> </ul> </li> <li>ObservationType<ul> <li>DEFAULT</li> <li>GOAL_SIGNAL</li> </ul> </li> <li>ObservationSpec</li> <li>BehaviorSpec</li> <li>BaseEnv<ul> <li>step</li> <li>reset</li> <li>close</li> <li>behavior_specs</li> <li>set_actions</li> <li>set_action_for_agent</li> <li>get_steps</li> </ul> </li> <li>mlagents_envs.environment</li> <li>UnityEnvironment<ul> <li>__init__</li> <li>close</li> </ul> </li> <li>mlagents_envs.registry</li> <li>mlagents_envs.registry.unity_env_registry</li> <li>UnityEnvRegistry<ul> <li>register</li> <li>register_from_yaml</li> <li>clear</li> <li>__getitem__</li> </ul> </li> <li>mlagents_envs.side_channel</li> <li>mlagents_envs.side_channel.raw_bytes_channel</li> <li>RawBytesChannel<ul> <li>on_message_received</li> <li>get_and_clear_received_messages</li> <li>send_raw_data</li> </ul> </li> <li>mlagents_envs.side_channel.outgoing_message</li> <li>OutgoingMessage<ul> <li>__init__</li> <li>write_bool</li> <li>write_int32</li> <li>write_float32</li> <li>write_float32_list</li> <li>write_string</li> <li>set_raw_bytes</li> </ul> </li> <li>mlagents_envs.side_channel.engine_configuration_channel</li> <li>EngineConfigurationChannel<ul> <li>on_message_received</li> <li>set_configuration_parameters</li> <li>set_configuration</li> </ul> </li> <li>mlagents_envs.side_channel.side_channel_manager</li> <li>SideChannelManager<ul> <li>process_side_channel_message</li> <li>generate_side_channel_messages</li> </ul> </li> <li>mlagents_envs.side_channel.stats_side_channel</li> <li>StatsSideChannel<ul> <li>on_message_received</li> <li>get_and_reset_stats</li> </ul> </li> <li>mlagents_envs.side_channel.incoming_message</li> <li>IncomingMessage<ul> <li>__init__</li> <li>read_bool</li> <li>read_int32</li> <li>read_float32</li> <li>read_float32_list</li> <li>read_string</li> <li>get_raw_bytes</li> </ul> </li> <li>mlagents_envs.side_channel.float_properties_channel</li> <li>FloatPropertiesChannel<ul> <li>on_message_received</li> <li>set_property</li> <li>get_property</li> <li>list_properties</li> <li>get_property_dict_copy</li> </ul> </li> <li>mlagents_envs.side_channel.environment_parameters_channel</li> <li>EnvironmentParametersChannel<ul> <li>set_float_parameter</li> <li>set_uniform_sampler_parameters</li> <li>set_gaussian_sampler_parameters</li> <li>set_multirangeuniform_sampler_parameters</li> </ul> </li> <li>mlagents_envs.side_channel.side_channel</li> <li>SideChannel<ul> <li>queue_message_to_send</li> <li>on_message_received</li> <li>channel_id</li> </ul> </li> </ul>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsbase_env","title":"mlagents_envs.base_env","text":"<p>Python Environment API for the ML-Agents Toolkit The aim of this API is to expose Agents evolving in a simulation to perform reinforcement learning on. This API supports multi-agent scenarios and groups similar Agents (same observations, actions spaces and behavior) together. These groups of Agents are identified by their BehaviorName. For performance reasons, the data of each group of agents is processed in a batched manner. Agents are identified by a unique AgentId identifier that allows tracking of Agents across simulation steps. Note that there is no guarantee that the number or order of the Agents in the state will be consistent across simulation steps. A simulation steps corresponds to moving the simulation forward until at least one agent in the simulation sends its observations to Python again. Since Agents can request decisions at different frequencies, a simulation step does not necessarily correspond to a fixed simulation time increment.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#decisionstep-objects","title":"DecisionStep Objects","text":"<pre><code>class DecisionStep(NamedTuple)\n</code></pre> <p>Contains the data a single Agent collected since the last simulation step.  - obs is a list of numpy arrays observations collected by the agent.  - reward is a float. Corresponds to the rewards collected by the agent  since the last simulation step.  - agent_id is an int and an unique identifier for the corresponding Agent.  - action_mask is an optional list of one dimensional array of booleans.  Only available when using multi-discrete actions.  Each array corresponds to an action branch. Each array contains a mask  for each action of the branch. If true, the action is not available for  the agent during this simulation step.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#decisionsteps-objects","title":"DecisionSteps Objects","text":"<pre><code>class DecisionSteps(Mapping)\n</code></pre> <p>Contains the data a batch of similar Agents collected since the last simulation step. Note that all Agents do not necessarily have new information to send at each simulation step. Therefore, the ordering of agents and the batch size of the DecisionSteps are not fixed across simulation steps.  - obs is a list of numpy arrays observations collected by the batch of  agent. Each obs has one extra dimension compared to DecisionStep: the  first dimension of the array corresponds to the batch size of the batch.  - reward is a float vector of length batch size. Corresponds to the  rewards collected by each agent since the last simulation step.  - agent_id is an int vector of length batch size containing unique  identifier for the corresponding Agent. This is used to track Agents  across simulation steps.  - action_mask is an optional list of two dimensional array of booleans.  Only available when using multi-discrete actions.  Each array corresponds to an action branch. The first dimension of each  array is the batch size and the second contains a mask for each action of  the branch. If true, the action is not available for the agent during  this simulation step.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#agent_id_to_index","title":"agent_id_to_index","text":"<pre><code> | @property\n | agent_id_to_index() -&gt; Dict[AgentId, int]\n</code></pre> <p>Returns:</p> <p>A Dict that maps agent_id to the index of those agents in this DecisionSteps.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#__getitem__","title":"__getitem__","text":"<pre><code> | __getitem__(agent_id: AgentId) -&gt; DecisionStep\n</code></pre> <p>returns the DecisionStep for a specific agent.</p> <p>Arguments:</p> <ul> <li><code>agent_id</code>: The id of the agent</li> </ul> <p>Returns:</p> <p>The DecisionStep</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#empty","title":"empty","text":"<pre><code> | @staticmethod\n | empty(spec: \"BehaviorSpec\") -&gt; \"DecisionSteps\"\n</code></pre> <p>Returns an empty DecisionSteps.</p> <p>Arguments:</p> <ul> <li><code>spec</code>: The BehaviorSpec for the DecisionSteps</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#terminalstep-objects","title":"TerminalStep Objects","text":"<pre><code>class TerminalStep(NamedTuple)\n</code></pre> <p>Contains the data a single Agent collected when its episode ended.  - obs is a list of numpy arrays observations collected by the agent.  - reward is a float. Corresponds to the rewards collected by the agent  since the last simulation step.  - interrupted is a bool. Is true if the Agent was interrupted since the last  decision step. For example, if the Agent reached the maximum number of steps for  the episode.  - agent_id is an int and an unique identifier for the corresponding Agent.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#terminalsteps-objects","title":"TerminalSteps Objects","text":"<pre><code>class TerminalSteps(Mapping)\n</code></pre> <p>Contains the data a batch of Agents collected when their episode terminated. All Agents present in the TerminalSteps have ended their episode.  - obs is a list of numpy arrays observations collected by the batch of  agent. Each obs has one extra dimension compared to DecisionStep: the  first dimension of the array corresponds to the batch size of the batch.  - reward is a float vector of length batch size. Corresponds to the  rewards collected by each agent since the last simulation step.  - interrupted is an array of booleans of length batch size. Is true if the  associated Agent was interrupted since the last decision step. For example, if the  Agent reached the maximum number of steps for the episode.  - agent_id is an int vector of length batch size containing unique  identifier for the corresponding Agent. This is used to track Agents  across simulation steps.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#agent_id_to_index_1","title":"agent_id_to_index","text":"<pre><code> | @property\n | agent_id_to_index() -&gt; Dict[AgentId, int]\n</code></pre> <p>Returns:</p> <p>A Dict that maps agent_id to the index of those agents in this TerminalSteps.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#__getitem___1","title":"__getitem__","text":"<pre><code> | __getitem__(agent_id: AgentId) -&gt; TerminalStep\n</code></pre> <p>returns the TerminalStep for a specific agent.</p> <p>Arguments:</p> <ul> <li><code>agent_id</code>: The id of the agent</li> </ul> <p>Returns:</p> <p>obs, reward, done, agent_id and optional action mask for a specific agent</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#empty_1","title":"empty","text":"<pre><code> | @staticmethod\n | empty(spec: \"BehaviorSpec\") -&gt; \"TerminalSteps\"\n</code></pre> <p>Returns an empty TerminalSteps.</p> <p>Arguments:</p> <ul> <li><code>spec</code>: The BehaviorSpec for the TerminalSteps</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#actiontuple-objects","title":"ActionTuple Objects","text":"<pre><code>class ActionTuple(_ActionTupleBase)\n</code></pre> <p>An object whose fields correspond to actions of different types. Continuous and discrete actions are numpy arrays of type float32 and int32, respectively and are type checked on construction. Dimensions are of (n_agents, continuous_size) and (n_agents, discrete_size), respectively. Note, this also holds when continuous or discrete size is zero.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#discrete_dtype","title":"discrete_dtype","text":"<pre><code> | @property\n | discrete_dtype() -&gt; np.dtype\n</code></pre> <p>The dtype of a discrete action.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#actionspec-objects","title":"ActionSpec Objects","text":"<pre><code>class ActionSpec(NamedTuple)\n</code></pre> <p>A NamedTuple containing utility functions and information about the action spaces for a group of Agents under the same behavior. - num_continuous_actions is an int corresponding to the number of floats which constitute the action. - discrete_branch_sizes is a Tuple of int where each int corresponds to the number of discrete actions available to the agent on an independent action branch.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#is_discrete","title":"is_discrete","text":"<pre><code> | is_discrete() -&gt; bool\n</code></pre> <p>Returns true if this Behavior uses discrete actions</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#is_continuous","title":"is_continuous","text":"<pre><code> | is_continuous() -&gt; bool\n</code></pre> <p>Returns true if this Behavior uses continuous actions</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#discrete_size","title":"discrete_size","text":"<pre><code> | @property\n | discrete_size() -&gt; int\n</code></pre> <p>Returns a an int corresponding to the number of discrete branches.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#empty_action","title":"empty_action","text":"<pre><code> | empty_action(n_agents: int) -&gt; ActionTuple\n</code></pre> <p>Generates ActionTuple corresponding to an empty action (all zeros) for a number of agents.</p> <p>Arguments:</p> <ul> <li><code>n_agents</code>: The number of agents that will have actions generated</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#random_action","title":"random_action","text":"<pre><code> | random_action(n_agents: int) -&gt; ActionTuple\n</code></pre> <p>Generates ActionTuple corresponding to a random action (either discrete or continuous) for a number of agents.</p> <p>Arguments:</p> <ul> <li><code>n_agents</code>: The number of agents that will have actions generated</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#create_continuous","title":"create_continuous","text":"<pre><code> | @staticmethod\n | create_continuous(continuous_size: int) -&gt; \"ActionSpec\"\n</code></pre> <p>Creates an ActionSpec that is homogenously continuous</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#create_discrete","title":"create_discrete","text":"<pre><code> | @staticmethod\n | create_discrete(discrete_branches: Tuple[int]) -&gt; \"ActionSpec\"\n</code></pre> <p>Creates an ActionSpec that is homogenously discrete</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#create_hybrid","title":"create_hybrid","text":"<pre><code> | @staticmethod\n | create_hybrid(continuous_size: int, discrete_branches: Tuple[int]) -&gt; \"ActionSpec\"\n</code></pre> <p>Creates a hybrid ActionSpace</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#dimensionproperty-objects","title":"DimensionProperty Objects","text":"<pre><code>class DimensionProperty(IntFlag)\n</code></pre> <p>The dimension property of a dimension of an observation.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#unspecified","title":"UNSPECIFIED","text":"<p>No properties specified.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#none","title":"NONE","text":"<p>No Property of the observation in that dimension. Observation can be processed with Fully connected networks.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#translational_equivariance","title":"TRANSLATIONAL_EQUIVARIANCE","text":"<p>Means it is suitable to do a convolution in this dimension.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#variable_size","title":"VARIABLE_SIZE","text":"<p>Means that there can be a variable number of observations in this dimension. The observations are unordered.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#observationtype-objects","title":"ObservationType Objects","text":"<pre><code>class ObservationType(Enum)\n</code></pre> <p>An Enum which defines the type of information carried in the observation of the agent.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#default","title":"DEFAULT","text":"<p>Observation information is generic.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#goal_signal","title":"GOAL_SIGNAL","text":"<p>Observation contains goal information for current task.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#observationspec-objects","title":"ObservationSpec Objects","text":"<pre><code>class ObservationSpec(NamedTuple)\n</code></pre> <p>A NamedTuple containing information about the observation of Agents. - shape is a Tuple of int : It corresponds to the shape of an observation's dimensions. - dimension_property is a Tuple of DimensionProperties flag, one flag for each dimension. - observation_type is an enum of ObservationType.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#behaviorspec-objects","title":"BehaviorSpec Objects","text":"<pre><code>class BehaviorSpec(NamedTuple)\n</code></pre> <p>A NamedTuple containing information about the observation and action spaces for a group of Agents under the same behavior. - observation_specs is a List of ObservationSpec NamedTuple containing information about the information of the Agent's observations such as their shapes. The order of the ObservationSpec is the same as the order of the observations of an agent. - action_spec is an ActionSpec NamedTuple.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#baseenv-objects","title":"BaseEnv Objects","text":"<pre><code>class BaseEnv(ABC)\n</code></pre>"},{"location":"Python-LLAPI-Documentation/#step","title":"step","text":"<pre><code> | @abstractmethod\n | step() -&gt; None\n</code></pre> <p>Signals the environment that it must move the simulation forward by one step.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#reset","title":"reset","text":"<pre><code> | @abstractmethod\n | reset() -&gt; None\n</code></pre> <p>Signals the environment that it must reset the simulation.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#close","title":"close","text":"<pre><code> | @abstractmethod\n | close() -&gt; None\n</code></pre> <p>Signals the environment that it must close.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#behavior_specs","title":"behavior_specs","text":"<pre><code> | @property\n | @abstractmethod\n | behavior_specs() -&gt; MappingType[str, BehaviorSpec]\n</code></pre> <p>Returns a Mapping from behavior names to behavior specs. Agents grouped under the same behavior name have the same action and observation specs, and are expected to behave similarly in the environment. Note that new keys can be added to this mapping as new policies are instantiated.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_actions","title":"set_actions","text":"<pre><code> | @abstractmethod\n | set_actions(behavior_name: BehaviorName, action: ActionTuple) -&gt; None\n</code></pre> <p>Sets the action for all of the agents in the simulation for the next step. The Actions must be in the same order as the order received in the DecisionSteps.</p> <p>Arguments:</p> <ul> <li><code>behavior_name</code>: The name of the behavior the agents are part of</li> <li><code>action</code>: ActionTuple tuple of continuous and/or discrete action. Actions are np.arrays with dimensions  (n_agents, continuous_size) and (n_agents, discrete_size), respectively.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_action_for_agent","title":"set_action_for_agent","text":"<pre><code> | @abstractmethod\n | set_action_for_agent(behavior_name: BehaviorName, agent_id: AgentId, action: ActionTuple) -&gt; None\n</code></pre> <p>Sets the action for one of the agents in the simulation for the next step.</p> <p>Arguments:</p> <ul> <li><code>behavior_name</code>: The name of the behavior the agent is part of</li> <li><code>agent_id</code>: The id of the agent the action is set for</li> <li><code>action</code>: ActionTuple tuple of continuous and/or discrete action Actions are np.arrays with dimensions  (1, continuous_size) and (1, discrete_size), respectively. Note, this initial dimensions of 1 is because this action is meant for a single agent.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#get_steps","title":"get_steps","text":"<pre><code> | @abstractmethod\n | get_steps(behavior_name: BehaviorName) -&gt; Tuple[DecisionSteps, TerminalSteps]\n</code></pre> <p>Retrieves the steps of the agents that requested a step in the simulation.</p> <p>Arguments:</p> <ul> <li><code>behavior_name</code>: The name of the behavior the agents are part of</li> </ul> <p>Returns:</p> <p>A tuple containing : - A DecisionSteps NamedTuple containing the observations, the rewards, the agent ids and the action masks for the Agents of the specified behavior. These Agents need an action this step. - A TerminalSteps NamedTuple containing the observations, rewards, agent ids and interrupted flags of the agents that had their episode terminated last step.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsenvironment","title":"mlagents_envs.environment","text":""},{"location":"Python-LLAPI-Documentation/#unityenvironment-objects","title":"UnityEnvironment Objects","text":"<pre><code>class UnityEnvironment(BaseEnv)\n</code></pre>"},{"location":"Python-LLAPI-Documentation/#__init__","title":"__init__","text":"<pre><code> | __init__(file_name: Optional[str] = None, worker_id: int = 0, base_port: Optional[int] = None, seed: int = 0, no_graphics: bool = False, no_graphics_monitor: bool = False, timeout_wait: int = 60, additional_args: Optional[List[str]] = None, side_channels: Optional[List[SideChannel]] = None, log_folder: Optional[str] = None, num_areas: int = 1)\n</code></pre> <p>Starts a new unity environment and establishes a connection with the environment. Notice: Currently communication between Unity and Python takes place over an open socket without authentication. Ensure that the network where training takes place is secure.</p> <p>:string file_name: Name of Unity environment binary. :int base_port: Baseline port number to connect to Unity environment over. worker_id increments over this. If no environment is specified (i.e. file_name is None), the DEFAULT_EDITOR_PORT will be used. :int worker_id: Offset from base_port. Used for training multiple environments simultaneously. :bool no_graphics: Whether to run the Unity simulator in no-graphics mode :bool no_graphics_monitor: Whether to run the main worker in graphics mode, with the remaining in no-graphics mode :int timeout_wait: Time (in seconds) to wait for connection from environment. :list args: Addition Unity command line arguments :list side_channels: Additional side channel for no-rl communication with Unity :str log_folder: Optional folder to write the Unity Player log file into.  Requires absolute path.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#close_1","title":"close","text":"<pre><code> | close()\n</code></pre> <p>Sends a shutdown signal to the unity environment, and closes the socket connection.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsregistry","title":"mlagents_envs.registry","text":""},{"location":"Python-LLAPI-Documentation/#mlagents_envsregistryunity_env_registry","title":"mlagents_envs.registry.unity_env_registry","text":""},{"location":"Python-LLAPI-Documentation/#unityenvregistry-objects","title":"UnityEnvRegistry Objects","text":"<pre><code>class UnityEnvRegistry(Mapping)\n</code></pre>"},{"location":"Python-LLAPI-Documentation/#unityenvregistry","title":"UnityEnvRegistry","text":"<p>Provides a library of Unity environments that can be launched without the need of downloading the Unity Editor. The UnityEnvRegistry implements a Map, to access an entry of the Registry, use:</p> <pre><code>registry = UnityEnvRegistry()\nentry = registry[&lt;environment_identifyier&gt;]\n</code></pre> <p>An entry has the following properties :  * <code>identifier</code> : Uniquely identifies this environment  * <code>expected_reward</code> : Corresponds to the reward an agent must obtained for the task  to be considered completed.  * <code>description</code> : A human readable description of the environment.</p> <p>To launch a Unity environment from a registry entry, use the <code>make</code> method:</p> <pre><code>registry = UnityEnvRegistry()\nenv = registry[&lt;environment_identifyier&gt;].make()\n</code></pre> <p></p>"},{"location":"Python-LLAPI-Documentation/#register","title":"register","text":"<pre><code> | register(new_entry: BaseRegistryEntry) -&gt; None\n</code></pre> <p>Registers a new BaseRegistryEntry to the registry. The BaseRegistryEntry.identifier value will be used as indexing key. If two are more environments are registered under the same key, the most recentry added will replace the others.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#register_from_yaml","title":"register_from_yaml","text":"<pre><code> | register_from_yaml(path_to_yaml: str) -&gt; None\n</code></pre> <p>Registers the environments listed in a yaml file (either local or remote). Note that the entries are registered lazily: the registration will only happen when an environment is accessed. The yaml file must have the following format :</p> <pre><code>environments:\n- &lt;identifier of the first environment&gt;:\n    expected_reward: &lt;expected reward of the environment&gt;\n    description: | &lt;a multi line description of the environment&gt;\n      &lt;continued multi line description&gt;\n    linux_url: &lt;The url for the Linux executable zip file&gt;\n    darwin_url: &lt;The url for the OSX executable zip file&gt;\n    win_url: &lt;The url for the Windows executable zip file&gt;\n\n- &lt;identifier of the second environment&gt;:\n    expected_reward: &lt;expected reward of the environment&gt;\n    description: | &lt;a multi line description of the environment&gt;\n      &lt;continued multi line description&gt;\n    linux_url: &lt;The url for the Linux executable zip file&gt;\n    darwin_url: &lt;The url for the OSX executable zip file&gt;\n    win_url: &lt;The url for the Windows executable zip file&gt;\n\n- ...\n</code></pre> <p>Arguments:</p> <ul> <li><code>path_to_yaml</code>: A local path or url to the yaml file</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#clear","title":"clear","text":"<pre><code> | clear() -&gt; None\n</code></pre> <p>Deletes all entries in the registry.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#__getitem___2","title":"__getitem__","text":"<pre><code> | __getitem__(identifier: str) -&gt; BaseRegistryEntry\n</code></pre> <p>Returns the BaseRegistryEntry with the provided identifier. BaseRegistryEntry can then be used to make a Unity Environment.</p> <p>Arguments:</p> <ul> <li><code>identifier</code>: The identifier of the BaseRegistryEntry</li> </ul> <p>Returns:</p> <p>The associated BaseRegistryEntry</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channel","title":"mlagents_envs.side_channel","text":""},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelraw_bytes_channel","title":"mlagents_envs.side_channel.raw_bytes_channel","text":""},{"location":"Python-LLAPI-Documentation/#rawbyteschannel-objects","title":"RawBytesChannel Objects","text":"<pre><code>class RawBytesChannel(SideChannel)\n</code></pre> <p>This is an example of what the SideChannel for raw bytes exchange would look like. Is meant to be used for general research purpose.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#on_message_received","title":"on_message_received","text":"<pre><code> | on_message_received(msg: IncomingMessage) -&gt; None\n</code></pre> <p>Is called by the environment to the side channel. Can be called multiple times per step if multiple messages are meant for that SideChannel.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#get_and_clear_received_messages","title":"get_and_clear_received_messages","text":"<pre><code> | get_and_clear_received_messages() -&gt; List[bytes]\n</code></pre> <p>returns a list of bytearray received from the environment.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#send_raw_data","title":"send_raw_data","text":"<pre><code> | send_raw_data(data: bytearray) -&gt; None\n</code></pre> <p>Queues a message to be sent by the environment at the next call to step.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channeloutgoing_message","title":"mlagents_envs.side_channel.outgoing_message","text":""},{"location":"Python-LLAPI-Documentation/#outgoingmessage-objects","title":"OutgoingMessage Objects","text":"<pre><code>class OutgoingMessage()\n</code></pre> <p>Utility class for forming the message that is written to a SideChannel. All data is written in little-endian format using the struct module.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#__init___1","title":"__init__","text":"<pre><code> | __init__()\n</code></pre> <p>Create an OutgoingMessage with an empty buffer.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#write_bool","title":"write_bool","text":"<pre><code> | write_bool(b: bool) -&gt; None\n</code></pre> <p>Append a boolean value.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#write_int32","title":"write_int32","text":"<pre><code> | write_int32(i: int) -&gt; None\n</code></pre> <p>Append an integer value.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#write_float32","title":"write_float32","text":"<pre><code> | write_float32(f: float) -&gt; None\n</code></pre> <p>Append a float value. It will be truncated to 32-bit precision.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#write_float32_list","title":"write_float32_list","text":"<pre><code> | write_float32_list(float_list: List[float]) -&gt; None\n</code></pre> <p>Append a list of float values. They will be truncated to 32-bit precision.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#write_string","title":"write_string","text":"<pre><code> | write_string(s: str) -&gt; None\n</code></pre> <p>Append a string value. Internally, it will be encoded to ascii, and the encoded length will also be written to the message.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_raw_bytes","title":"set_raw_bytes","text":"<pre><code> | set_raw_bytes(buffer: bytearray) -&gt; None\n</code></pre> <p>Set the internal buffer to a new bytearray. This will overwrite any existing data.</p> <p>Arguments:</p> <ul> <li><code>buffer</code>:</li> </ul> <p>Returns:</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelengine_configuration_channel","title":"mlagents_envs.side_channel.engine_configuration_channel","text":""},{"location":"Python-LLAPI-Documentation/#engineconfigurationchannel-objects","title":"EngineConfigurationChannel Objects","text":"<pre><code>class EngineConfigurationChannel(SideChannel)\n</code></pre> <p>This is the SideChannel for engine configuration exchange. The data in the engine configuration is as follows :  - int width;  - int height;  - int qualityLevel;  - float timeScale;  - int targetFrameRate;  - int captureFrameRate;</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#on_message_received_1","title":"on_message_received","text":"<pre><code> | on_message_received(msg: IncomingMessage) -&gt; None\n</code></pre> <p>Is called by the environment to the side channel. Can be called multiple times per step if multiple messages are meant for that SideChannel. Note that Python should never receive an engine configuration from Unity</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_configuration_parameters","title":"set_configuration_parameters","text":"<pre><code> | set_configuration_parameters(width: Optional[int] = None, height: Optional[int] = None, quality_level: Optional[int] = None, time_scale: Optional[float] = None, target_frame_rate: Optional[int] = None, capture_frame_rate: Optional[int] = None) -&gt; None\n</code></pre> <p>Sets the engine configuration. Takes as input the configurations of the engine.</p> <p>Arguments:</p> <ul> <li><code>width</code>: Defines the width of the display. (Must be set alongside height)</li> <li><code>height</code>: Defines the height of the display. (Must be set alongside width)</li> <li><code>quality_level</code>: Defines the quality level of the simulation.</li> <li><code>time_scale</code>: Defines the multiplier for the deltatime in the simulation. If set to a higher value, time will pass faster in the simulation but the physics might break.</li> <li><code>target_frame_rate</code>: Instructs simulation to try to render at a specified frame rate.</li> <li><code>capture_frame_rate</code>: Instructs the simulation to consider time between updates to always be constant, regardless of the actual frame rate.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_configuration","title":"set_configuration","text":"<pre><code> | set_configuration(config: EngineConfig) -&gt; None\n</code></pre> <p>Sets the engine configuration. Takes as input an EngineConfig.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelside_channel_manager","title":"mlagents_envs.side_channel.side_channel_manager","text":""},{"location":"Python-LLAPI-Documentation/#sidechannelmanager-objects","title":"SideChannelManager Objects","text":"<pre><code>class SideChannelManager()\n</code></pre>"},{"location":"Python-LLAPI-Documentation/#process_side_channel_message","title":"process_side_channel_message","text":"<pre><code> | process_side_channel_message(data: bytes) -&gt; None\n</code></pre> <p>Separates the data received from Python into individual messages for each registered side channel and calls on_message_received on them.</p> <p>Arguments:</p> <ul> <li><code>data</code>: The packed message sent by Unity</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#generate_side_channel_messages","title":"generate_side_channel_messages","text":"<pre><code> | generate_side_channel_messages() -&gt; bytearray\n</code></pre> <p>Gathers the messages that the registered side channels will send to Unity and combines them into a single message ready to be sent.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelstats_side_channel","title":"mlagents_envs.side_channel.stats_side_channel","text":""},{"location":"Python-LLAPI-Documentation/#statssidechannel-objects","title":"StatsSideChannel Objects","text":"<pre><code>class StatsSideChannel(SideChannel)\n</code></pre> <p>Side channel that receives (string, float) pairs from the environment, so that they can eventually be passed to a StatsReporter.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#on_message_received_2","title":"on_message_received","text":"<pre><code> | on_message_received(msg: IncomingMessage) -&gt; None\n</code></pre> <p>Receive the message from the environment, and save it for later retrieval.</p> <p>Arguments:</p> <ul> <li><code>msg</code>:</li> </ul> <p>Returns:</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#get_and_reset_stats","title":"get_and_reset_stats","text":"<pre><code> | get_and_reset_stats() -&gt; EnvironmentStats\n</code></pre> <p>Returns the current stats, and resets the internal storage of the stats.</p> <p>Returns:</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelincoming_message","title":"mlagents_envs.side_channel.incoming_message","text":""},{"location":"Python-LLAPI-Documentation/#incomingmessage-objects","title":"IncomingMessage Objects","text":"<pre><code>class IncomingMessage()\n</code></pre> <p>Utility class for reading the message written to a SideChannel. Values must be read in the order they were written.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#__init___2","title":"__init__","text":"<pre><code> | __init__(buffer: bytes, offset: int = 0)\n</code></pre> <p>Create a new IncomingMessage from the bytes.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#read_bool","title":"read_bool","text":"<pre><code> | read_bool(default_value: bool = False) -&gt; bool\n</code></pre> <p>Read a boolean value from the message buffer.</p> <p>Arguments:</p> <ul> <li><code>default_value</code>: Default value to use if the end of the message is reached.</li> </ul> <p>Returns:</p> <p>The value read from the message, or the default value if the end was reached.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#read_int32","title":"read_int32","text":"<pre><code> | read_int32(default_value: int = 0) -&gt; int\n</code></pre> <p>Read an integer value from the message buffer.</p> <p>Arguments:</p> <ul> <li><code>default_value</code>: Default value to use if the end of the message is reached.</li> </ul> <p>Returns:</p> <p>The value read from the message, or the default value if the end was reached.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#read_float32","title":"read_float32","text":"<pre><code> | read_float32(default_value: float = 0.0) -&gt; float\n</code></pre> <p>Read a float value from the message buffer.</p> <p>Arguments:</p> <ul> <li><code>default_value</code>: Default value to use if the end of the message is reached.</li> </ul> <p>Returns:</p> <p>The value read from the message, or the default value if the end was reached.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#read_float32_list","title":"read_float32_list","text":"<pre><code> | read_float32_list(default_value: List[float] = None) -&gt; List[float]\n</code></pre> <p>Read a list of float values from the message buffer.</p> <p>Arguments:</p> <ul> <li><code>default_value</code>: Default value to use if the end of the message is reached.</li> </ul> <p>Returns:</p> <p>The value read from the message, or the default value if the end was reached.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#read_string","title":"read_string","text":"<pre><code> | read_string(default_value: str = \"\") -&gt; str\n</code></pre> <p>Read a string value from the message buffer.</p> <p>Arguments:</p> <ul> <li><code>default_value</code>: Default value to use if the end of the message is reached.</li> </ul> <p>Returns:</p> <p>The value read from the message, or the default value if the end was reached.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#get_raw_bytes","title":"get_raw_bytes","text":"<pre><code> | get_raw_bytes() -&gt; bytes\n</code></pre> <p>Get a copy of the internal bytes used by the message.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelfloat_properties_channel","title":"mlagents_envs.side_channel.float_properties_channel","text":""},{"location":"Python-LLAPI-Documentation/#floatpropertieschannel-objects","title":"FloatPropertiesChannel Objects","text":"<pre><code>class FloatPropertiesChannel(SideChannel)\n</code></pre> <p>This is the SideChannel for float properties shared with Unity. You can modify the float properties of an environment with the commands set_property, get_property and list_properties.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#on_message_received_3","title":"on_message_received","text":"<pre><code> | on_message_received(msg: IncomingMessage) -&gt; None\n</code></pre> <p>Is called by the environment to the side channel. Can be called multiple times per step if multiple messages are meant for that SideChannel.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_property","title":"set_property","text":"<pre><code> | set_property(key: str, value: float) -&gt; None\n</code></pre> <p>Sets a property in the Unity Environment.</p> <p>Arguments:</p> <ul> <li><code>key</code>: The string identifier of the property.</li> <li><code>value</code>: The float value of the property.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#get_property","title":"get_property","text":"<pre><code> | get_property(key: str) -&gt; Optional[float]\n</code></pre> <p>Gets a property in the Unity Environment. If the property was not found, will return None.</p> <p>Arguments:</p> <ul> <li><code>key</code>: The string identifier of the property.</li> </ul> <p>Returns:</p> <p>The float value of the property or None.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#list_properties","title":"list_properties","text":"<pre><code> | list_properties() -&gt; List[str]\n</code></pre> <p>Returns a list of all the string identifiers of the properties currently present in the Unity Environment.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#get_property_dict_copy","title":"get_property_dict_copy","text":"<pre><code> | get_property_dict_copy() -&gt; Dict[str, float]\n</code></pre> <p>Returns a copy of the float properties.</p> <p>Returns:</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelenvironment_parameters_channel","title":"mlagents_envs.side_channel.environment_parameters_channel","text":""},{"location":"Python-LLAPI-Documentation/#environmentparameterschannel-objects","title":"EnvironmentParametersChannel Objects","text":"<pre><code>class EnvironmentParametersChannel(SideChannel)\n</code></pre> <p>This is the SideChannel for sending environment parameters to Unity. You can send parameters to an environment with the command set_float_parameter.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_float_parameter","title":"set_float_parameter","text":"<pre><code> | set_float_parameter(key: str, value: float) -&gt; None\n</code></pre> <p>Sets a float environment parameter in the Unity Environment.</p> <p>Arguments:</p> <ul> <li><code>key</code>: The string identifier of the parameter.</li> <li><code>value</code>: The float value of the parameter.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_uniform_sampler_parameters","title":"set_uniform_sampler_parameters","text":"<pre><code> | set_uniform_sampler_parameters(key: str, min_value: float, max_value: float, seed: int) -&gt; None\n</code></pre> <p>Sets a uniform environment parameter sampler.</p> <p>Arguments:</p> <ul> <li><code>key</code>: The string identifier of the parameter.</li> <li><code>min_value</code>: The minimum of the sampling distribution.</li> <li><code>max_value</code>: The maximum of the sampling distribution.</li> <li><code>seed</code>: The random seed to initialize the sampler.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_gaussian_sampler_parameters","title":"set_gaussian_sampler_parameters","text":"<pre><code> | set_gaussian_sampler_parameters(key: str, mean: float, st_dev: float, seed: int) -&gt; None\n</code></pre> <p>Sets a gaussian environment parameter sampler.</p> <p>Arguments:</p> <ul> <li><code>key</code>: The string identifier of the parameter.</li> <li><code>mean</code>: The mean of the sampling distribution.</li> <li><code>st_dev</code>: The standard deviation of the sampling distribution.</li> <li><code>seed</code>: The random seed to initialize the sampler.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_multirangeuniform_sampler_parameters","title":"set_multirangeuniform_sampler_parameters","text":"<pre><code> | set_multirangeuniform_sampler_parameters(key: str, intervals: List[Tuple[float, float]], seed: int) -&gt; None\n</code></pre> <p>Sets a multirangeuniform environment parameter sampler.</p> <p>Arguments:</p> <ul> <li><code>key</code>: The string identifier of the parameter.</li> <li><code>intervals</code>: The lists of min and max that define each uniform distribution.</li> <li><code>seed</code>: The random seed to initialize the sampler.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelside_channel","title":"mlagents_envs.side_channel.side_channel","text":""},{"location":"Python-LLAPI-Documentation/#sidechannel-objects","title":"SideChannel Objects","text":"<pre><code>class SideChannel(ABC)\n</code></pre> <p>The side channel just get access to a bytes buffer that will be shared between C# and Python. For example, We will create a specific side channel for properties that will be a list of string (fixed size) to float number, that can be modified by both C# and Python. All side channels are passed to the Env object at construction.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#queue_message_to_send","title":"queue_message_to_send","text":"<pre><code> | queue_message_to_send(msg: OutgoingMessage) -&gt; None\n</code></pre> <p>Queues a message to be sent by the environment at the next call to step.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#on_message_received_4","title":"on_message_received","text":"<pre><code> | @abstractmethod\n | on_message_received(msg: IncomingMessage) -&gt; None\n</code></pre> <p>Is called by the environment to the side channel. Can be called multiple times per step if multiple messages are meant for that SideChannel.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#channel_id","title":"channel_id","text":"<pre><code> | @property\n | channel_id() -&gt; uuid.UUID\n</code></pre> <p>Returns:</p> <p>The type of side channel used. Will influence how the data is processed in the environment.</p>"},{"location":"Python-LLAPI/","title":"Unity ML-Agents Python Low Level API","text":"<p>The <code>mlagents</code> Python package contains two components: a low level API which allows you to interact directly with a Unity Environment (<code>mlagents_envs</code>) and an entry point to train (<code>mlagents-learn</code>) which allows you to train agents in Unity Environments using our implementations of reinforcement learning or imitation learning. This document describes how to use the <code>mlagents_envs</code> API. For information on using <code>mlagents-learn</code>, see here. For Python Low Level API documentation, see here.</p> <p>The Python Low Level API can be used to interact directly with your Unity learning environment. As such, it can serve as the basis for developing and evaluating new learning algorithms.</p>"},{"location":"Python-LLAPI/#mlagents_envs","title":"mlagents_envs","text":"<p>The ML-Agents Toolkit Low Level API is a Python API for controlling the simulation loop of an environment or game built with Unity. This API is used by the training algorithms inside the ML-Agent Toolkit, but you can also write your own Python programs using this API.</p> <p>The key objects in the Python API include:</p> <ul> <li>UnityEnvironment \u2014 the main interface between the Unity application and   your code. Use UnityEnvironment to start and control a simulation or training   session.</li> <li>BehaviorName - is a string that identifies a behavior in the simulation.</li> <li>AgentId - is an <code>int</code> that serves as unique identifier for Agents in the   simulation.</li> <li>DecisionSteps \u2014 contains the data from Agents belonging to the same   \"Behavior\" in the simulation, such as observations and rewards. Only Agents   that requested a decision since the last call to <code>env.step()</code> are in the   DecisionSteps object.</li> <li>TerminalSteps \u2014 contains the data from Agents belonging to the same   \"Behavior\" in the simulation, such as observations and rewards. Only Agents   whose episode ended since the last call to <code>env.step()</code> are in the   TerminalSteps object.</li> <li>BehaviorSpec \u2014 describes the shape of the observation data inside   DecisionSteps and TerminalSteps as well as the expected action shapes.</li> </ul> <p>These classes are all defined in the base_env script.</p> <p>An Agent \"Behavior\" is a group of Agents identified by a <code>BehaviorName</code> that share the same observations and action types (described in their <code>BehaviorSpec</code>). You can think about Agent Behavior as a group of agents that will share the same policy. All Agents with the same behavior have the same goal and reward signals.</p> <p>To communicate with an Agent in a Unity environment from a Python program, the Agent in the simulation must have <code>Behavior Parameters</code> set to communicate. You must set the <code>Behavior Type</code> to <code>Default</code> and give it a <code>Behavior Name</code>.</p> <p>Notice: Currently communication between Unity and Python takes place over an open socket without authentication. As such, please make sure that the network where training takes place is secure. This will be addressed in a future release.</p>"},{"location":"Python-LLAPI/#loading-a-unity-environment","title":"Loading a Unity Environment","text":"<p>Python-side communication happens through <code>UnityEnvironment</code> which is located in <code>environment.py</code>. To load a Unity environment from a built binary file, put the file in the same directory as <code>envs</code>. For example, if the filename of your Unity environment is <code>3DBall</code>, in python, run:</p> <pre><code>from mlagents_envs.environment import UnityEnvironment\n# This is a non-blocking call that only loads the environment.\nenv = UnityEnvironment(file_name=\"3DBall\", seed=1, side_channels=[])\n# Start interacting with the environment.\nenv.reset()\nbehavior_names = env.behavior_specs.keys()\n...\n</code></pre> <p>NOTE: Please read Interacting with a Unity Environment to read more about how you can interact with the Unity environment from Python.</p> <ul> <li><code>file_name</code> is the name of the environment binary (located in the root   directory of the python project).</li> <li><code>worker_id</code> indicates which port to use for communication with the   environment. For use in parallel training regimes such as A3C.</li> <li><code>seed</code> indicates the seed to use when generating random numbers during the   training process. In environments which are stochastic, setting the seed   enables reproducible experimentation by ensuring that the environment and   trainers utilize the same random seed.</li> <li><code>side_channels</code> provides a way to exchange data with the Unity simulation that   is not related to the reinforcement learning loop. For example: configurations   or properties. More on them in the Side Channels doc.</li> </ul> <p>If you want to directly interact with the Editor, you need to use <code>file_name=None</code>, then press the Play button in the Editor when the message \"Start training by pressing the Play button in the Unity Editor\" is displayed on the screen</p>"},{"location":"Python-LLAPI/#interacting-with-a-unity-environment","title":"Interacting with a Unity Environment","text":""},{"location":"Python-LLAPI/#the-baseenv-interface","title":"The BaseEnv interface","text":"<p>A <code>BaseEnv</code> has the following methods:</p> <ul> <li>Reset : <code>env.reset()</code> Sends a signal to reset the environment. Returns   None.</li> <li>Step : <code>env.step()</code> Sends a signal to step the environment. Returns None.   Note that a \"step\" for Python does not correspond to either Unity <code>Update</code> nor   <code>FixedUpdate</code>. When <code>step()</code> or <code>reset()</code> is called, the Unity simulation will   move forward until an Agent in the simulation needs a input from Python to   act.</li> <li>Close : <code>env.close()</code> Sends a shutdown signal to the environment and   terminates the communication.</li> <li>Behavior Specs : <code>env.behavior_specs</code> Returns a Mapping of   <code>BehaviorName</code> to <code>BehaviorSpec</code> objects (read only).   A <code>BehaviorSpec</code> contains the observation shapes and the   <code>ActionSpec</code> (which defines the action shape). Note that   the <code>BehaviorSpec</code> for a specific group is fixed throughout the simulation.   The number of entries in the Mapping can change over time in the simulation   if new Agent behaviors are created in the simulation.</li> <li>Get Steps : <code>env.get_steps(behavior_name: str)</code> Returns a tuple   <code>DecisionSteps, TerminalSteps</code> corresponding to the behavior_name given as   input. The <code>DecisionSteps</code> contains information about the state of the agents   that need an action this step and have the behavior behavior_name. The   <code>TerminalSteps</code> contains information about the state of the agents whose   episode ended and have the behavior behavior_name. Both <code>DecisionSteps</code> and   <code>TerminalSteps</code> contain information such as the observations, the rewards and   the agent identifiers. <code>DecisionSteps</code> also contains action masks for the next   action while <code>TerminalSteps</code> contains the reason for termination (did the   Agent reach its maximum step and was interrupted). The data is in <code>np.array</code>   of which the first dimension is always the number of agents note that the   number of agents is not guaranteed to remain constant during the simulation   and it is not unusual to have either <code>DecisionSteps</code> or <code>TerminalSteps</code>   contain no Agents at all.</li> <li>Set Actions :<code>env.set_actions(behavior_name: str, action: ActionTuple)</code> Sets   the actions for a whole agent group. <code>action</code> is an <code>ActionTuple</code>, which   is made up of a 2D <code>np.array</code> of <code>dtype=np.int32</code> for discrete actions, and   <code>dtype=np.float32</code> for continuous actions. The first dimension of <code>np.array</code>   in the tuple is the number of agents that requested a decision since the   last call to <code>env.step()</code>. The second dimension is the number of discrete or   continuous actions for the corresponding array.</li> <li>Set Action for Agent :   <code>env.set_action_for_agent(agent_group: str, agent_id: int, action: ActionTuple)</code>   Sets the action for a specific Agent in an agent group. <code>agent_group</code> is the   name of the group the Agent belongs to and <code>agent_id</code> is the integer   identifier of the Agent. <code>action</code> is an <code>ActionTuple</code> as described above. Note: If no action is provided for an agent group between two calls to <code>env.step()</code> then the default action will be all zeros.</li> </ul>"},{"location":"Python-LLAPI/#decisionsteps-and-decisionstep","title":"DecisionSteps and DecisionStep","text":"<p><code>DecisionSteps</code> (with <code>s</code>) contains information about a whole batch of Agents while <code>DecisionStep</code> (no <code>s</code>) only contains information about a single Agent.</p> <p>A <code>DecisionSteps</code> has the following fields :</p> <ul> <li><code>obs</code> is a list of numpy arrays observations collected by the group of agent.   The first dimension of the array corresponds to the batch size of the group   (number of agents requesting a decision since the last call to <code>env.step()</code>).</li> <li><code>reward</code> is a float vector of length batch size. Corresponds to the rewards   collected by each agent since the last simulation step.</li> <li><code>agent_id</code> is an int vector of length batch size containing unique identifier   for the corresponding Agent. This is used to track Agents across simulation   steps.</li> <li><code>action_mask</code> is an optional list of two dimensional arrays of booleans which is only   available when using multi-discrete actions. Each array corresponds to an   action branch. The first dimension of each array is the batch size and the   second contains a mask for each action of the branch. If true, the action is   not available for the agent during this simulation step.</li> </ul> <p>It also has the two following methods:</p> <ul> <li><code>len(DecisionSteps)</code> Returns the number of agents requesting a decision since   the last call to <code>env.step()</code>.</li> <li><code>DecisionSteps[agent_id]</code> Returns a <code>DecisionStep</code> for the Agent with the   <code>agent_id</code> unique identifier.</li> </ul> <p>A <code>DecisionStep</code> has the following fields:</p> <ul> <li><code>obs</code> is a list of numpy arrays observations collected by the agent. (Each   array has one less dimension than the arrays in <code>DecisionSteps</code>)</li> <li><code>reward</code> is a float. Corresponds to the rewards collected by the agent since   the last simulation step.</li> <li><code>agent_id</code> is an int and an unique identifier for the corresponding Agent.</li> <li><code>action_mask</code> is an optional list of one dimensional arrays of booleans which is only   available when using multi-discrete actions. Each array corresponds to an   action branch. Each array contains a mask for each action of the branch. If   true, the action is not available for the agent during this simulation step.</li> </ul>"},{"location":"Python-LLAPI/#terminalsteps-and-terminalstep","title":"TerminalSteps and TerminalStep","text":"<p>Similarly to <code>DecisionSteps</code> and <code>DecisionStep</code>, <code>TerminalSteps</code> (with <code>s</code>) contains information about a whole batch of Agents while <code>TerminalStep</code> (no <code>s</code>) only contains information about a single Agent.</p> <p>A <code>TerminalSteps</code> has the following fields :</p> <ul> <li><code>obs</code> is a list of numpy arrays observations collected by the group of agent.   The first dimension of the array corresponds to the batch size of the group   (number of agents requesting a decision since the last call to <code>env.step()</code>).</li> <li><code>reward</code> is a float vector of length batch size. Corresponds to the rewards   collected by each agent since the last simulation step.</li> <li><code>agent_id</code> is an int vector of length batch size containing unique identifier   for the corresponding Agent. This is used to track Agents across simulation   steps.</li> <li><code>interrupted</code> is an array of booleans of length batch size. Is true if the  associated Agent was interrupted since the last decision step. For example,  if the Agent reached the maximum number of steps for the episode.</li> </ul> <p>It also has the two following methods:</p> <ul> <li><code>len(TerminalSteps)</code> Returns the number of agents requesting a decision since   the last call to <code>env.step()</code>.</li> <li><code>TerminalSteps[agent_id]</code> Returns a <code>TerminalStep</code> for the Agent with the   <code>agent_id</code> unique identifier.</li> </ul> <p>A <code>TerminalStep</code> has the following fields:</p> <ul> <li><code>obs</code> is a list of numpy arrays observations collected by the agent. (Each   array has one less dimension than the arrays in <code>TerminalSteps</code>)</li> <li><code>reward</code> is a float. Corresponds to the rewards collected by the agent since   the last simulation step.</li> <li><code>agent_id</code> is an int and an unique identifier for the corresponding Agent.</li> <li><code>interrupted</code> is a bool. Is true if the Agent was interrupted since the last  decision step. For example, if the Agent reached the maximum number of steps for  the episode.</li> </ul>"},{"location":"Python-LLAPI/#behaviorspec","title":"BehaviorSpec","text":"<p>A <code>BehaviorSpec</code> has the following fields :</p> <ul> <li><code>observation_specs</code> is a List of <code>ObservationSpec</code> objects : Each <code>ObservationSpec</code>   corresponds to an observation's properties: <code>shape</code> is a tuple of ints that   corresponds to the shape of the observation (without the number of agents dimension).   <code>dimension_property</code> is a tuple of flags containing extra information about how the   data should be processed in the corresponding dimension. <code>observation_type</code> is an enum   corresponding to what type of observation is generating the data (i.e., default, goal,   etc). Note that the <code>ObservationSpec</code> have the same ordering as the ordering of observations   in the DecisionSteps, DecisionStep, TerminalSteps and TerminalStep.</li> <li><code>action_spec</code> is an <code>ActionSpec</code> namedtuple that defines the number and types   of actions for the Agent.</li> </ul> <p>An <code>ActionSpec</code> has the following fields and properties: - <code>continuous_size</code> is the number of floats that constitute the continuous actions. - <code>discrete_size</code> is the number of branches (the number of independent actions) that   constitute the multi-discrete actions. - <code>discrete_branches</code> is a Tuple of ints. Each int corresponds to the number of   different options for each branch of the action. For example:   In a game direction input (no movement, left, right) and   jump input (no jump, jump) there will be two branches (direction and jump),   the first one with 3 options and the second with 2 options. (<code>discrete_size = 2</code>   and <code>discrete_action_branches = (3,2,)</code>)</p>"},{"location":"Python-LLAPI/#communicating-additional-information-with-the-environment","title":"Communicating additional information with the Environment","text":"<p>In addition to the means of communicating between Unity and python described above, we also provide methods for sharing agent-agnostic information. These additional methods are referred to as side channels. ML-Agents includes two ready-made side channels, described below. It is also possible to create custom side channels to communicate any additional data between a Unity environment and Python. Instructions for creating custom side channels can be found here.</p> <p>Side channels exist as separate classes which are instantiated, and then passed as list to the <code>side_channels</code> argument of the constructor of the <code>UnityEnvironment</code> class.</p> <pre><code>channel = MyChannel()\n\nenv = UnityEnvironment(side_channels = [channel])\n</code></pre> <p>Note : A side channel will only send/receive messages when <code>env.step</code> or <code>env.reset()</code> is called.</p>"},{"location":"Python-LLAPI/#engineconfigurationchannel","title":"EngineConfigurationChannel","text":"<p>The <code>EngineConfiguration</code> side channel allows you to modify the time-scale, resolution, and graphics quality of the environment. This can be useful for adjusting the environment to perform better during training, or be more interpretable during inference.</p> <p><code>EngineConfigurationChannel</code> has two methods :</p> <ul> <li><code>set_configuration_parameters</code> which takes the following arguments:</li> <li><code>width</code>: Defines the width of the display. (Must be set alongside height)</li> <li><code>height</code>: Defines the height of the display. (Must be set alongside width)</li> <li><code>quality_level</code>: Defines the quality level of the simulation.</li> <li><code>time_scale</code>: Defines the multiplier for the deltatime in the simulation. If     set to a higher value, time will pass faster in the simulation but the     physics may perform unpredictably.</li> <li><code>target_frame_rate</code>: Instructs simulation to try to render at a specified     frame rate.</li> <li><code>capture_frame_rate</code> Instructs the simulation to consider time between     updates to always be constant, regardless of the actual frame rate.</li> <li><code>set_configuration</code> with argument config which is an <code>EngineConfig</code> NamedTuple   object.</li> </ul> <p>For example, the following code would adjust the time-scale of the simulation to be 2x realtime.</p> <pre><code>from mlagents_envs.environment import UnityEnvironment\nfrom mlagents_envs.side_channel.engine_configuration_channel import EngineConfigurationChannel\n\nchannel = EngineConfigurationChannel()\n\nenv = UnityEnvironment(side_channels=[channel])\n\nchannel.set_configuration_parameters(time_scale = 2.0)\n\ni = env.reset()\n...\n</code></pre>"},{"location":"Python-LLAPI/#environmentparameters","title":"EnvironmentParameters","text":"<p>The <code>EnvironmentParameters</code> will allow you to get and set pre-defined numerical values in the environment. This can be useful for adjusting environment-specific settings, or for reading non-agent related information from the environment. You can call <code>get_property</code> and <code>set_property</code> on the side channel to read and write properties.</p> <p><code>EnvironmentParametersChannel</code> has one methods:</p> <ul> <li><code>set_float_parameter</code> Sets a float parameter in the Unity Environment.</li> <li>key: The string identifier of the property.</li> <li>value: The float value of the property.</li> </ul> <pre><code>from mlagents_envs.environment import UnityEnvironment\nfrom mlagents_envs.side_channel.environment_parameters_channel import EnvironmentParametersChannel\n\nchannel = EnvironmentParametersChannel()\n\nenv = UnityEnvironment(side_channels=[channel])\n\nchannel.set_float_parameter(\"parameter_1\", 2.0)\n\ni = env.reset()\n...\n</code></pre> <p>Once a property has been modified in Python, you can access it in C# after the next call to <code>step</code> as follows:</p> <pre><code>var envParameters = Academy.Instance.EnvironmentParameters;\nfloat property1 = envParameters.GetWithDefault(\"parameter_1\", 0.0f);\n</code></pre>"},{"location":"Python-LLAPI/#custom-side-channels","title":"Custom side channels","text":"<p>For information on how to make custom side channels for sending additional data types, see the documentation here.</p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/","title":"Table of Contents","text":"<ul> <li>mlagents.trainers.trainer.on_policy_trainer</li> <li>OnPolicyTrainer<ul> <li>__init__</li> <li>add_policy</li> </ul> </li> <li>mlagents.trainers.trainer.off_policy_trainer</li> <li>OffPolicyTrainer<ul> <li>__init__</li> <li>save_model</li> <li>save_replay_buffer</li> <li>load_replay_buffer</li> <li>add_policy</li> </ul> </li> <li>mlagents.trainers.trainer.rl_trainer</li> <li>RLTrainer<ul> <li>end_episode</li> <li>create_optimizer</li> <li>save_model</li> <li>advance</li> </ul> </li> <li>mlagents.trainers.trainer.trainer</li> <li>Trainer<ul> <li>__init__</li> <li>stats_reporter</li> <li>parameters</li> <li>get_max_steps</li> <li>get_step</li> <li>threaded</li> <li>should_still_train</li> <li>reward_buffer</li> <li>save_model</li> <li>end_episode</li> <li>create_policy</li> <li>add_policy</li> <li>get_policy</li> <li>advance</li> <li>publish_policy_queue</li> <li>subscribe_trajectory_queue</li> </ul> </li> <li>mlagents.trainers.settings</li> <li>deep_update_dict</li> <li>RewardSignalSettings<ul> <li>structure</li> </ul> </li> <li>ParameterRandomizationSettings<ul> <li>__str__</li> <li>structure</li> <li>unstructure</li> <li>apply</li> </ul> </li> <li>ConstantSettings<ul> <li>__str__</li> <li>apply</li> </ul> </li> <li>UniformSettings<ul> <li>__str__</li> <li>apply</li> </ul> </li> <li>GaussianSettings<ul> <li>__str__</li> <li>apply</li> </ul> </li> <li>MultiRangeUniformSettings<ul> <li>__str__</li> <li>apply</li> </ul> </li> <li>CompletionCriteriaSettings<ul> <li>need_increment</li> </ul> </li> <li>Lesson</li> <li>EnvironmentParameterSettings<ul> <li>structure</li> </ul> </li> <li>TrainerSettings<ul> <li>structure</li> </ul> </li> <li>CheckpointSettings<ul> <li>prioritize_resume_init</li> </ul> </li> <li>RunOptions<ul> <li>from_argparse</li> </ul> </li> </ul>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#mlagentstrainerstraineron_policy_trainer","title":"mlagents.trainers.trainer.on_policy_trainer","text":""},{"location":"Python-On-Off-Policy-Trainer-Documentation/#onpolicytrainer-objects","title":"OnPolicyTrainer Objects","text":"<pre><code>class OnPolicyTrainer(RLTrainer)\n</code></pre> <p>The PPOTrainer is an implementation of the PPO algorithm.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__init__","title":"__init__","text":"<pre><code> | __init__(behavior_name: str, reward_buff_cap: int, trainer_settings: TrainerSettings, training: bool, load: bool, seed: int, artifact_path: str)\n</code></pre> <p>Responsible for collecting experiences and training an on-policy model.</p> <p>Arguments:</p> <ul> <li><code>behavior_name</code>: The name of the behavior associated with trainer config</li> <li><code>reward_buff_cap</code>: Max reward history to track in the reward buffer</li> <li><code>trainer_settings</code>: The parameters for the trainer.</li> <li><code>training</code>: Whether the trainer is set for training.</li> <li><code>load</code>: Whether the model should be loaded.</li> <li><code>seed</code>: The seed the model will be initialized with</li> <li><code>artifact_path</code>: The directory within which to store artifacts from this trainer.</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#add_policy","title":"add_policy","text":"<pre><code> | add_policy(parsed_behavior_id: BehaviorIdentifiers, policy: Policy) -&gt; None\n</code></pre> <p>Adds policy to trainer.</p> <p>Arguments:</p> <ul> <li><code>parsed_behavior_id</code>: Behavior identifiers that the policy should belong to.</li> <li><code>policy</code>: Policy to associate with name_behavior_id.</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#mlagentstrainerstraineroff_policy_trainer","title":"mlagents.trainers.trainer.off_policy_trainer","text":""},{"location":"Python-On-Off-Policy-Trainer-Documentation/#offpolicytrainer-objects","title":"OffPolicyTrainer Objects","text":"<pre><code>class OffPolicyTrainer(RLTrainer)\n</code></pre> <p>The SACTrainer is an implementation of the SAC algorithm, with support for discrete actions and recurrent networks.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__init___1","title":"__init__","text":"<pre><code> | __init__(behavior_name: str, reward_buff_cap: int, trainer_settings: TrainerSettings, training: bool, load: bool, seed: int, artifact_path: str)\n</code></pre> <p>Responsible for collecting experiences and training an off-policy model.</p> <p>Arguments:</p> <ul> <li><code>behavior_name</code>: The name of the behavior associated with trainer config</li> <li><code>reward_buff_cap</code>: Max reward history to track in the reward buffer</li> <li><code>trainer_settings</code>: The parameters for the trainer.</li> <li><code>training</code>: Whether the trainer is set for training.</li> <li><code>load</code>: Whether the model should be loaded.</li> <li><code>seed</code>: The seed the model will be initialized with</li> <li><code>artifact_path</code>: The directory within which to store artifacts from this trainer.</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#save_model","title":"save_model","text":"<pre><code> | save_model() -&gt; None\n</code></pre> <p>Saves the final training model to memory Overrides the default to save the replay buffer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#save_replay_buffer","title":"save_replay_buffer","text":"<pre><code> | save_replay_buffer() -&gt; None\n</code></pre> <p>Save the training buffer's update buffer to a pickle file.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#load_replay_buffer","title":"load_replay_buffer","text":"<pre><code> | load_replay_buffer() -&gt; None\n</code></pre> <p>Loads the last saved replay buffer from a file.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#add_policy_1","title":"add_policy","text":"<pre><code> | add_policy(parsed_behavior_id: BehaviorIdentifiers, policy: Policy) -&gt; None\n</code></pre> <p>Adds policy to trainer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#mlagentstrainerstrainerrl_trainer","title":"mlagents.trainers.trainer.rl_trainer","text":""},{"location":"Python-On-Off-Policy-Trainer-Documentation/#rltrainer-objects","title":"RLTrainer Objects","text":"<pre><code>class RLTrainer(Trainer)\n</code></pre> <p>This class is the base class for trainers that use Reward Signals.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#end_episode","title":"end_episode","text":"<pre><code> | end_episode() -&gt; None\n</code></pre> <p>A signal that the Episode has ended. The buffer must be reset. Get only called when the academy resets.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#create_optimizer","title":"create_optimizer","text":"<pre><code> | @abc.abstractmethod\n | create_optimizer() -&gt; TorchOptimizer\n</code></pre> <p>Creates an Optimizer object</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#save_model_1","title":"save_model","text":"<pre><code> | save_model() -&gt; None\n</code></pre> <p>Saves the policy associated with this trainer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#advance","title":"advance","text":"<pre><code> | advance() -&gt; None\n</code></pre> <p>Steps the trainer, taking in trajectories and updates if ready. Will block and wait briefly if there are no trajectories.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#mlagentstrainerstrainertrainer","title":"mlagents.trainers.trainer.trainer","text":""},{"location":"Python-On-Off-Policy-Trainer-Documentation/#trainer-objects","title":"Trainer Objects","text":"<pre><code>class Trainer(abc.ABC)\n</code></pre> <p>This class is the base class for the mlagents_envs.trainers</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__init___2","title":"__init__","text":"<pre><code> | __init__(brain_name: str, trainer_settings: TrainerSettings, training: bool, load: bool, artifact_path: str, reward_buff_cap: int = 1)\n</code></pre> <p>Responsible for collecting experiences and training a neural network model.</p> <p>Arguments:</p> <ul> <li><code>brain_name</code>: Brain name of brain to be trained.</li> <li><code>trainer_settings</code>: The parameters for the trainer (dictionary).</li> <li><code>training</code>: Whether the trainer is set for training.</li> <li><code>artifact_path</code>: The directory within which to store artifacts from this trainer</li> <li><code>reward_buff_cap</code>:</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#stats_reporter","title":"stats_reporter","text":"<pre><code> | @property\n | stats_reporter()\n</code></pre> <p>Returns the stats reporter associated with this Trainer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#parameters","title":"parameters","text":"<pre><code> | @property\n | parameters() -&gt; TrainerSettings\n</code></pre> <p>Returns the trainer parameters of the trainer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#get_max_steps","title":"get_max_steps","text":"<pre><code> | @property\n | get_max_steps() -&gt; int\n</code></pre> <p>Returns the maximum number of steps. Is used to know when the trainer should be stopped.</p> <p>Returns:</p> <p>The maximum number of steps of the trainer</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#get_step","title":"get_step","text":"<pre><code> | @property\n | get_step() -&gt; int\n</code></pre> <p>Returns the number of steps the trainer has performed</p> <p>Returns:</p> <p>the step count of the trainer</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#threaded","title":"threaded","text":"<pre><code> | @property\n | threaded() -&gt; bool\n</code></pre> <p>Whether or not to run the trainer in a thread. True allows the trainer to update the policy while the environment is taking steps. Set to False to enforce strict on-policy updates (i.e. don't update the policy when taking steps.)</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#should_still_train","title":"should_still_train","text":"<pre><code> | @property\n | should_still_train() -&gt; bool\n</code></pre> <p>Returns whether or not the trainer should train. A Trainer could stop training if it wasn't training to begin with, or if max_steps is reached.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#reward_buffer","title":"reward_buffer","text":"<pre><code> | @property\n | reward_buffer() -&gt; Deque[float]\n</code></pre> <p>Returns the reward buffer. The reward buffer contains the cumulative rewards of the most recent episodes completed by agents using this trainer.</p> <p>Returns:</p> <p>the reward buffer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#save_model_2","title":"save_model","text":"<pre><code> | @abc.abstractmethod\n | save_model() -&gt; None\n</code></pre> <p>Saves model file(s) for the policy or policies associated with this trainer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#end_episode_1","title":"end_episode","text":"<pre><code> | @abc.abstractmethod\n | end_episode()\n</code></pre> <p>A signal that the Episode has ended. The buffer must be reset. Get only called when the academy resets.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#create_policy","title":"create_policy","text":"<pre><code> | @abc.abstractmethod\n | create_policy(parsed_behavior_id: BehaviorIdentifiers, behavior_spec: BehaviorSpec) -&gt; Policy\n</code></pre> <p>Creates a Policy object</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#add_policy_2","title":"add_policy","text":"<pre><code> | @abc.abstractmethod\n | add_policy(parsed_behavior_id: BehaviorIdentifiers, policy: Policy) -&gt; None\n</code></pre> <p>Adds policy to trainer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#get_policy","title":"get_policy","text":"<pre><code> | get_policy(name_behavior_id: str) -&gt; Policy\n</code></pre> <p>Gets policy associated with name_behavior_id</p> <p>Arguments:</p> <ul> <li><code>name_behavior_id</code>: Fully qualified behavior name</li> </ul> <p>Returns:</p> <p>Policy associated with name_behavior_id</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#advance_1","title":"advance","text":"<pre><code> | @abc.abstractmethod\n | advance() -&gt; None\n</code></pre> <p>Advances the trainer. Typically, this means grabbing trajectories from all subscribed trajectory queues (self.trajectory_queues), and updating a policy using the steps in them, and if needed pushing a new policy onto the right policy queues (self.policy_queues).</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#publish_policy_queue","title":"publish_policy_queue","text":"<pre><code> | publish_policy_queue(policy_queue: AgentManagerQueue[Policy]) -&gt; None\n</code></pre> <p>Adds a policy queue to the list of queues to publish to when this Trainer makes a policy update</p> <p>Arguments:</p> <ul> <li><code>policy_queue</code>: Policy queue to publish to.</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#subscribe_trajectory_queue","title":"subscribe_trajectory_queue","text":"<pre><code> | subscribe_trajectory_queue(trajectory_queue: AgentManagerQueue[Trajectory]) -&gt; None\n</code></pre> <p>Adds a trajectory queue to the list of queues for the trainer to ingest Trajectories from.</p> <p>Arguments:</p> <ul> <li><code>trajectory_queue</code>: Trajectory queue to read from.</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#mlagentstrainerssettings","title":"mlagents.trainers.settings","text":""},{"location":"Python-On-Off-Policy-Trainer-Documentation/#deep_update_dict","title":"deep_update_dict","text":"<pre><code>deep_update_dict(d: Dict, update_d: Mapping) -&gt; None\n</code></pre> <p>Similar to dict.update(), but works for nested dicts of dicts as well.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#rewardsignalsettings-objects","title":"RewardSignalSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass RewardSignalSettings()\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#structure","title":"structure","text":"<pre><code> | @staticmethod\n | structure(d: Mapping, t: type) -&gt; Any\n</code></pre> <p>Helper method to structure a Dict of RewardSignalSettings class. Meant to be registered with cattr.register_structure_hook() and called with cattr.structure(). This is needed to handle the special Enum selection of RewardSignalSettings classes.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#parameterrandomizationsettings-objects","title":"ParameterRandomizationSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass ParameterRandomizationSettings(abc.ABC)\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__str__","title":"__str__","text":"<pre><code> | __str__() -&gt; str\n</code></pre> <p>Helper method to output sampler stats to console.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#structure_1","title":"structure","text":"<pre><code> | @staticmethod\n | structure(d: Union[Mapping, float], t: type) -&gt; \"ParameterRandomizationSettings\"\n</code></pre> <p>Helper method to a ParameterRandomizationSettings class. Meant to be registered with cattr.register_structure_hook() and called with cattr.structure(). This is needed to handle the special Enum selection of ParameterRandomizationSettings classes.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#unstructure","title":"unstructure","text":"<pre><code> | @staticmethod\n | unstructure(d: \"ParameterRandomizationSettings\") -&gt; Mapping\n</code></pre> <p>Helper method to a ParameterRandomizationSettings class. Meant to be registered with cattr.register_unstructure_hook() and called with cattr.unstructure().</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#apply","title":"apply","text":"<pre><code> | @abc.abstractmethod\n | apply(key: str, env_channel: EnvironmentParametersChannel) -&gt; None\n</code></pre> <p>Helper method to send sampler settings over EnvironmentParametersChannel Calls the appropriate sampler type set method.</p> <p>Arguments:</p> <ul> <li><code>key</code>: environment parameter to be sampled</li> <li><code>env_channel</code>: The EnvironmentParametersChannel to communicate sampler settings to environment</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#constantsettings-objects","title":"ConstantSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass ConstantSettings(ParameterRandomizationSettings)\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__str___1","title":"__str__","text":"<pre><code> | __str__() -&gt; str\n</code></pre> <p>Helper method to output sampler stats to console.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#apply_1","title":"apply","text":"<pre><code> | apply(key: str, env_channel: EnvironmentParametersChannel) -&gt; None\n</code></pre> <p>Helper method to send sampler settings over EnvironmentParametersChannel Calls the constant sampler type set method.</p> <p>Arguments:</p> <ul> <li><code>key</code>: environment parameter to be sampled</li> <li><code>env_channel</code>: The EnvironmentParametersChannel to communicate sampler settings to environment</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#uniformsettings-objects","title":"UniformSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass UniformSettings(ParameterRandomizationSettings)\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__str___2","title":"__str__","text":"<pre><code> | __str__() -&gt; str\n</code></pre> <p>Helper method to output sampler stats to console.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#apply_2","title":"apply","text":"<pre><code> | apply(key: str, env_channel: EnvironmentParametersChannel) -&gt; None\n</code></pre> <p>Helper method to send sampler settings over EnvironmentParametersChannel Calls the uniform sampler type set method.</p> <p>Arguments:</p> <ul> <li><code>key</code>: environment parameter to be sampled</li> <li><code>env_channel</code>: The EnvironmentParametersChannel to communicate sampler settings to environment</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#gaussiansettings-objects","title":"GaussianSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass GaussianSettings(ParameterRandomizationSettings)\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__str___3","title":"__str__","text":"<pre><code> | __str__() -&gt; str\n</code></pre> <p>Helper method to output sampler stats to console.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#apply_3","title":"apply","text":"<pre><code> | apply(key: str, env_channel: EnvironmentParametersChannel) -&gt; None\n</code></pre> <p>Helper method to send sampler settings over EnvironmentParametersChannel Calls the gaussian sampler type set method.</p> <p>Arguments:</p> <ul> <li><code>key</code>: environment parameter to be sampled</li> <li><code>env_channel</code>: The EnvironmentParametersChannel to communicate sampler settings to environment</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#multirangeuniformsettings-objects","title":"MultiRangeUniformSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass MultiRangeUniformSettings(ParameterRandomizationSettings)\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__str___4","title":"__str__","text":"<pre><code> | __str__() -&gt; str\n</code></pre> <p>Helper method to output sampler stats to console.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#apply_4","title":"apply","text":"<pre><code> | apply(key: str, env_channel: EnvironmentParametersChannel) -&gt; None\n</code></pre> <p>Helper method to send sampler settings over EnvironmentParametersChannel Calls the multirangeuniform sampler type set method.</p> <p>Arguments:</p> <ul> <li><code>key</code>: environment parameter to be sampled</li> <li><code>env_channel</code>: The EnvironmentParametersChannel to communicate sampler settings to environment</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#completioncriteriasettings-objects","title":"CompletionCriteriaSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass CompletionCriteriaSettings()\n</code></pre> <p>CompletionCriteriaSettings contains the information needed to figure out if the next lesson must start.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#need_increment","title":"need_increment","text":"<pre><code> | need_increment(progress: float, reward_buffer: List[float], smoothing: float) -&gt; Tuple[bool, float]\n</code></pre> <p>Given measures, this method returns a boolean indicating if the lesson needs to change now, and a float corresponding to the new smoothed value.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#lesson-objects","title":"Lesson Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass Lesson()\n</code></pre> <p>Gathers the data of one lesson for one environment parameter including its name, the condition that must be fullfiled for the lesson to be completed and a sampler for the environment parameter. If the completion_criteria is None, then this is the last lesson in the curriculum.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#environmentparametersettings-objects","title":"EnvironmentParameterSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass EnvironmentParameterSettings()\n</code></pre> <p>EnvironmentParameterSettings is an ordered list of lessons for one environment parameter.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#structure_2","title":"structure","text":"<pre><code> | @staticmethod\n | structure(d: Mapping, t: type) -&gt; Dict[str, \"EnvironmentParameterSettings\"]\n</code></pre> <p>Helper method to structure a Dict of EnvironmentParameterSettings class. Meant to be registered with cattr.register_structure_hook() and called with cattr.structure().</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#trainersettings-objects","title":"TrainerSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass TrainerSettings(ExportableSettings)\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#structure_3","title":"structure","text":"<pre><code> | @staticmethod\n | structure(d: Mapping, t: type) -&gt; Any\n</code></pre> <p>Helper method to structure a TrainerSettings class. Meant to be registered with cattr.register_structure_hook() and called with cattr.structure().</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#checkpointsettings-objects","title":"CheckpointSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass CheckpointSettings()\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#prioritize_resume_init","title":"prioritize_resume_init","text":"<pre><code> | prioritize_resume_init() -&gt; None\n</code></pre> <p>Prioritize explicit command line resume/init over conflicting yaml options. if both resume/init are set at one place use resume</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#runoptions-objects","title":"RunOptions Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass RunOptions(ExportableSettings)\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#from_argparse","title":"from_argparse","text":"<pre><code> | @staticmethod\n | from_argparse(args: argparse.Namespace) -&gt; \"RunOptions\"\n</code></pre> <p>Takes an argparse.Namespace as specified in <code>parse_command_line</code>, loads input configuration files from file paths, and converts to a RunOptions instance.</p> <p>Arguments:</p> <ul> <li><code>args</code>: collection of command-line parameters passed to mlagents-learn</li> </ul> <p>Returns:</p> <p>RunOptions representing the passed in arguments, with trainer config, curriculum and sampler configs loaded from files.</p>"},{"location":"Python-Optimizer-Documentation/","title":"Table of Contents","text":"<ul> <li>mlagents.trainers.optimizer.torch_optimizer</li> <li>TorchOptimizer<ul> <li>create_reward_signals</li> <li>get_trajectory_value_estimates</li> </ul> </li> <li>mlagents.trainers.optimizer.optimizer</li> <li>Optimizer<ul> <li>update</li> </ul> </li> </ul>"},{"location":"Python-Optimizer-Documentation/#mlagentstrainersoptimizertorch_optimizer","title":"mlagents.trainers.optimizer.torch_optimizer","text":""},{"location":"Python-Optimizer-Documentation/#torchoptimizer-objects","title":"TorchOptimizer Objects","text":"<pre><code>class TorchOptimizer(Optimizer)\n</code></pre>"},{"location":"Python-Optimizer-Documentation/#create_reward_signals","title":"create_reward_signals","text":"<pre><code> | create_reward_signals(reward_signal_configs: Dict[RewardSignalType, RewardSignalSettings]) -&gt; None\n</code></pre> <p>Create reward signals</p> <p>Arguments:</p> <ul> <li><code>reward_signal_configs</code>: Reward signal config.</li> </ul> <p></p>"},{"location":"Python-Optimizer-Documentation/#get_trajectory_value_estimates","title":"get_trajectory_value_estimates","text":"<pre><code> | get_trajectory_value_estimates(batch: AgentBuffer, next_obs: List[np.ndarray], done: bool, agent_id: str = \"\") -&gt; Tuple[Dict[str, np.ndarray], Dict[str, float], Optional[AgentBufferField]]\n</code></pre> <p>Get value estimates and memories for a trajectory, in batch form.</p> <p>Arguments:</p> <ul> <li><code>batch</code>: An AgentBuffer that consists of a trajectory.</li> <li><code>next_obs</code>: the next observation (after the trajectory). Used for boostrapping     if this is not a termiinal trajectory.</li> <li><code>done</code>: Set true if this is a terminal trajectory.</li> <li><code>agent_id</code>: Agent ID of the agent that this trajectory belongs to.</li> </ul> <p>Returns:</p> <p>A Tuple of the Value Estimates as a Dict of [name, np.ndarray(trajectory_len)],     the final value estimate as a Dict of [name, float], and optionally (if using memories)     an AgentBufferField of initial critic memories to be used during update.</p> <p></p>"},{"location":"Python-Optimizer-Documentation/#mlagentstrainersoptimizeroptimizer","title":"mlagents.trainers.optimizer.optimizer","text":""},{"location":"Python-Optimizer-Documentation/#optimizer-objects","title":"Optimizer Objects","text":"<pre><code>class Optimizer(abc.ABC)\n</code></pre> <p>Creates loss functions and auxillary networks (e.g. Q or Value) needed for training. Provides methods to update the Policy.</p> <p></p>"},{"location":"Python-Optimizer-Documentation/#update","title":"update","text":"<pre><code> | @abc.abstractmethod\n | update(batch: AgentBuffer, num_sequences: int) -&gt; Dict[str, float]\n</code></pre> <p>Update the Policy based on the batch that was passed in.</p> <p>Arguments:</p> <ul> <li><code>batch</code>: AgentBuffer that contains the minibatch of data used for this update.</li> <li><code>num_sequences</code>: Number of recurrent sequences found in the minibatch.</li> </ul> <p>Returns:</p> <p>A Dict containing statistics (name, value) from the update (e.g. loss)</p>"},{"location":"Python-PettingZoo-API-Documentation/","title":"Table of Contents","text":"<ul> <li>mlagents_envs.envs.pettingzoo_env_factory</li> <li>PettingZooEnvFactory<ul> <li>env</li> </ul> </li> <li>mlagents_envs.envs.unity_aec_env</li> <li>UnityAECEnv<ul> <li>__init__</li> <li>step</li> <li>observe</li> <li>last</li> </ul> </li> <li>mlagents_envs.envs.unity_parallel_env</li> <li>UnityParallelEnv<ul> <li>__init__</li> <li>reset</li> </ul> </li> <li>mlagents_envs.envs.unity_pettingzoo_base_env</li> <li>UnityPettingzooBaseEnv<ul> <li>observation_spaces</li> <li>observation_space</li> <li>action_spaces</li> <li>action_space</li> <li>side_channel</li> <li>reset</li> <li>seed</li> <li>render</li> <li>close</li> </ul> </li> </ul>"},{"location":"Python-PettingZoo-API-Documentation/#mlagents_envsenvspettingzoo_env_factory","title":"mlagents_envs.envs.pettingzoo_env_factory","text":""},{"location":"Python-PettingZoo-API-Documentation/#pettingzooenvfactory-objects","title":"PettingZooEnvFactory Objects","text":"<pre><code>class PettingZooEnvFactory()\n</code></pre>"},{"location":"Python-PettingZoo-API-Documentation/#env","title":"env","text":"<pre><code> | env(seed: Optional[int] = None, **kwargs: Union[List, int, bool, None]) -&gt; UnityAECEnv\n</code></pre> <p>Creates the environment with env_id from unity's default_registry and wraps it in a UnityToPettingZooWrapper</p> <p>Arguments:</p> <ul> <li><code>seed</code>: The seed for the action spaces of the agents.</li> <li><code>kwargs</code>: Any argument accepted by <code>UnityEnvironment</code>class except file_name</li> </ul> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#mlagents_envsenvsunity_aec_env","title":"mlagents_envs.envs.unity_aec_env","text":""},{"location":"Python-PettingZoo-API-Documentation/#unityaecenv-objects","title":"UnityAECEnv Objects","text":"<pre><code>class UnityAECEnv(UnityPettingzooBaseEnv,  AECEnv)\n</code></pre> <p>Unity AEC (PettingZoo) environment wrapper.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#__init__","title":"__init__","text":"<pre><code> | __init__(env: BaseEnv, seed: Optional[int] = None)\n</code></pre> <p>Initializes a Unity AEC environment wrapper.</p> <p>Arguments:</p> <ul> <li><code>env</code>: The UnityEnvironment that is being wrapped.</li> <li><code>seed</code>: The seed for the action spaces of the agents.</li> </ul> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#step","title":"step","text":"<pre><code> | step(action: Any) -&gt; None\n</code></pre> <p>Sets the action of the active agent and get the observation, reward, done and info of the next agent.</p> <p>Arguments:</p> <ul> <li><code>action</code>: The action for the active agent</li> </ul> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#observe","title":"observe","text":"<pre><code> | observe(agent_id)\n</code></pre> <p>Returns the observation an agent currently can make. <code>last()</code> calls this function.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#last","title":"last","text":"<pre><code> | last(observe=True)\n</code></pre> <p>returns observation, cumulative reward, done, info for the current agent (specified by self.agent_selection)</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#mlagents_envsenvsunity_parallel_env","title":"mlagents_envs.envs.unity_parallel_env","text":""},{"location":"Python-PettingZoo-API-Documentation/#unityparallelenv-objects","title":"UnityParallelEnv Objects","text":"<pre><code>class UnityParallelEnv(UnityPettingzooBaseEnv,  ParallelEnv)\n</code></pre> <p>Unity Parallel (PettingZoo) environment wrapper.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#__init___1","title":"__init__","text":"<pre><code> | __init__(env: BaseEnv, seed: Optional[int] = None)\n</code></pre> <p>Initializes a Unity Parallel environment wrapper.</p> <p>Arguments:</p> <ul> <li><code>env</code>: The UnityEnvironment that is being wrapped.</li> <li><code>seed</code>: The seed for the action spaces of the agents.</li> </ul> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#reset","title":"reset","text":"<pre><code> | reset() -&gt; Dict[str, Any]\n</code></pre> <p>Resets the environment.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#mlagents_envsenvsunity_pettingzoo_base_env","title":"mlagents_envs.envs.unity_pettingzoo_base_env","text":""},{"location":"Python-PettingZoo-API-Documentation/#unitypettingzoobaseenv-objects","title":"UnityPettingzooBaseEnv Objects","text":"<pre><code>class UnityPettingzooBaseEnv()\n</code></pre> <p>Unity Petting Zoo base environment.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#observation_spaces","title":"observation_spaces","text":"<pre><code> | @property\n | observation_spaces() -&gt; Dict[str, spaces.Space]\n</code></pre> <p>Return the observation spaces of all the agents.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#observation_space","title":"observation_space","text":"<pre><code> | observation_space(agent: str) -&gt; Optional[spaces.Space]\n</code></pre> <p>The observation space of the current agent.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#action_spaces","title":"action_spaces","text":"<pre><code> | @property\n | action_spaces() -&gt; Dict[str, spaces.Space]\n</code></pre> <p>Return the action spaces of all the agents.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#action_space","title":"action_space","text":"<pre><code> | action_space(agent: str) -&gt; Optional[spaces.Space]\n</code></pre> <p>The action space of the current agent.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#side_channel","title":"side_channel","text":"<pre><code> | @property\n | side_channel() -&gt; Dict[str, Any]\n</code></pre> <p>The side channels of the environment. You can access the side channels of an environment with <code>env.side_channel[&lt;name-of-channel&gt;]</code>.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#reset_1","title":"reset","text":"<pre><code> | reset()\n</code></pre> <p>Resets the environment.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#seed","title":"seed","text":"<pre><code> | seed(seed=None)\n</code></pre> <p>Reseeds the environment (making the resulting environment deterministic). <code>reset()</code> must be called after <code>seed()</code>, and before <code>step()</code>.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#render","title":"render","text":"<pre><code> | render(mode=\"human\")\n</code></pre> <p>NOT SUPPORTED.</p> <p>Displays a rendered frame from the environment, if supported. Alternate render modes in the default environments are <code>'rgb_array'</code> which returns a numpy array and is supported by all environments outside of classic, and <code>'ansi'</code> which returns the strings printed (specific to classic environments).</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#close","title":"close","text":"<pre><code> | close() -&gt; None\n</code></pre> <p>Close the environment.</p>"},{"location":"Python-PettingZoo-API/","title":"Unity ML-Agents PettingZoo Wrapper","text":"<p>With the increasing interest in multi-agent training with a gym-like API, we provide a PettingZoo Wrapper around the Petting Zoo API. Our wrapper provides interfaces on top of our <code>UnityEnvironment</code> class, which is the default way of interfacing with a Unity environment via Python.</p>"},{"location":"Python-PettingZoo-API/#installation-and-examples","title":"Installation and Examples","text":"<p>The PettingZoo wrapper is part of the <code>mlgents_envs</code> package. Please refer to the mlagents_envs installation instructions.</p> <p>[Colab] PettingZoo Wrapper Example</p> <p>This colab notebook demonstrates the example usage of the wrapper, including installation, basic usages, and an example with our Striker vs Goalie environment which is a multi-agents environment with multiple different behavior names.</p>"},{"location":"Python-PettingZoo-API/#api-interface","title":"API interface","text":"<p>This wrapper is compatible with PettingZoo API. Please check out PettingZoo API page for more details. Here's an example of interacting with wrapped environment:</p> <pre><code>from mlagents_envs.environment import UnityEnvironment\nfrom mlagents_envs.envs import UnityToPettingZooWrapper\n\nunity_env = UnityEnvironment(\"StrikersVsGoalie\")\nenv = UnityToPettingZooWrapper(unity_env)\nenv.reset()\nfor agent in env.agent_iter():\n    observation, reward, done, info = env.last()\n    action = policy(observation, agent)\n    env.step(action)\n</code></pre>"},{"location":"Python-PettingZoo-API/#notes","title":"Notes","text":"<ul> <li>There is support for both AEC   and Parallel PettingZoo APIs.</li> <li>The AEC wrapper is compatible with PettingZoo (PZ) API interface but works in a slightly   different way under the hood. For the AEC API, Instead of stepping the environment in every <code>env.step(action)</code>,   the PZ wrapper will store the action, and will only perform environment stepping when all the   agents requesting for actions in the current step have been assigned an action. This is for   performance, considering that the communication between Unity and python is more efficient   when data are sent in batches.</li> <li>Since the actions for the AEC wrapper are stored without applying them to the environment until   all the actions are queued, some components of the API might behave in unexpected way. For example, a call   to <code>env.reward</code> should return the instantaneous reward for that particular step, but the true   reward would only be available when an actual environment step is performed. It's recommended that   you follow the API definition for training (access rewards from <code>env.last()</code> instead of   <code>env.reward</code>) and the underlying mechanism shouldn't affect training results.</li> <li>The environments will automatically reset when it's done, so <code>env.agent_iter(max_step)</code> will   keep going on until the specified max step is reached (default: <code>2**63</code>). There is no need to   call <code>env.reset()</code> except for the very beginning of instantiating an environment.</li> </ul>"},{"location":"Readme/","title":"Unity ML-Agents Toolkit","text":"<p>(latest release) (all releases)</p> <p>The Unity Machine Learning Agents Toolkit (ML-Agents) is an open-source project that enables games and simulations to serve as environments for training intelligent agents. We provide implementations (based on PyTorch) of state-of-the-art algorithms to enable game developers and hobbyists to easily train intelligent agents for 2D, 3D and VR/AR games. Researchers can also use the provided simple-to-use Python API to train Agents using reinforcement learning, imitation learning, neuroevolution, or any other methods. These trained agents can be used for multiple purposes, including controlling NPC behavior (in a variety of settings such as multi-agent and adversarial), automated testing of game builds and evaluating different game design decisions pre-release. The ML-Agents Toolkit is mutually beneficial for both game developers and AI researchers as it provides a central platform where advances in AI can be evaluated on Unity\u2019s rich environments and then made accessible to the wider research and game developer communities.</p>"},{"location":"Readme/#features","title":"Features","text":"<ul> <li>17+ example Unity environments</li> <li>Support for multiple environment configurations and training scenarios</li> <li>Flexible Unity SDK that can be integrated into your game or custom Unity scene</li> <li>Support for training single-agent, multi-agent cooperative, and multi-agent   competitive scenarios via several Deep Reinforcement Learning algorithms (PPO, SAC, MA-POCA, self-play).</li> <li>Support for learning from demonstrations through two Imitation Learning algorithms (BC and GAIL).</li> <li>Quickly and easily add your own custom training algorithm and/or components.</li> <li>Easily definable Curriculum Learning scenarios for complex tasks</li> <li>Train robust agents using environment randomization</li> <li>Flexible agent control with On Demand Decision Making</li> <li>Train using multiple concurrent Unity environment instances</li> <li>Utilizes the Sentis to   provide native cross-platform support</li> <li>Unity environment control from Python</li> <li>Wrap Unity learning environments as a gym environment</li> <li>Wrap Unity learning environments as a PettingZoo environment</li> </ul> <p>See our ML-Agents Overview page for detailed descriptions of all these features. Or go straight to our web docs.</p>"},{"location":"Readme/#releases-documentation","title":"Releases &amp; Documentation","text":"<p>Our latest, stable release is <code>Release 21</code>. Click here to get started with the latest release of ML-Agents.</p> <p>You can also check out our new web docs!</p> <p>The table below lists all our releases, including our <code>main</code> branch which is under active development and may be unstable. A few helpful guidelines: - The Versioning page overviews how we manage our GitHub   releases and the versioning process for each of the ML-Agents components. - The Releases page   contains details of the changes between releases. - The Migration page contains details on how to upgrade   from earlier releases of the ML-Agents Toolkit. - The Documentation links in the table below include installation and usage   instructions specific to each release. Remember to always use the   documentation that corresponds to the release version you're using. - The <code>com.unity.ml-agents</code> package is verified   for Unity 2020.1 and later. Verified packages releases are numbered 1.0.x.</p> Version Release Date Source Documentation Download Python Package Unity Package develop (unstable) -- source docs download -- -- Release 21 October 9, 2023 source docs download 1.0.0 3.0.0 <p>If you are a researcher interested in a discussion of Unity as an AI platform, see a pre-print of our reference paper on Unity and the ML-Agents Toolkit.</p> <p>If you use Unity or the ML-Agents Toolkit to conduct research, we ask that you cite the following paper as a reference:</p> <pre><code>@article{juliani2020,\n  title={Unity: A general platform for intelligent agents},\n  author={Juliani, Arthur and Berges, Vincent-Pierre and Teng, Ervin and Cohen, Andrew and Harper, Jonathan and Elion, Chris and Goy, Chris and Gao, Yuan and Henry, Hunter and Mattar, Marwan and Lange, Danny},\n  journal={arXiv preprint arXiv:1809.02627},\n  url={https://arxiv.org/pdf/1809.02627.pdf},\n  year={2020}\n}\n</code></pre> <p>Additionally, if you use the MA-POCA trainer in your research, we ask that you cite the following paper as a reference:</p> <pre><code>@article{cohen2022,\n  title={On the Use and Misuse of Absorbing States in Multi-agent Reinforcement Learning},\n  author={Cohen, Andrew and Teng, Ervin and Berges, Vincent-Pierre and Dong, Ruo-Ping and Henry, Hunter and Mattar, Marwan and Zook, Alexander and Ganguly, Sujoy},\n  journal={RL in Games Workshop AAAI 2022},\n  url={http://aaai-rlg.mlanctot.info/papers/AAAI22-RLG_paper_32.pdf},\n  year={2022}\n}\n</code></pre>"},{"location":"Readme/#additional-resources","title":"Additional Resources","text":"<p>We have a Unity Learn course, ML-Agents: Hummingbirds, that provides a gentle introduction to Unity and the ML-Agents Toolkit.</p> <p>We've also partnered with CodeMonkeyUnity to create a series of tutorial videos on how to implement and use the ML-Agents Toolkit.</p> <p>We have also published a series of blog posts that are relevant for ML-Agents:</p> <ul> <li>(July 12, 2021)   ML-Agents plays Dodgeball</li> <li>(May 5, 2021)   ML-Agents v2.0 release: Now supports training complex cooperative behaviors</li> <li>(December 28, 2020)   Happy holidays from the Unity ML-Agents team!</li> <li>(November 20, 2020)   How Eidos-Montr\u00e9al created Grid Sensors to improve observations for training agents</li> <li>(November 11, 2020)   2020 AI@Unity interns shoutout</li> <li>(May 12, 2020)   Announcing ML-Agents Unity Package v1.0!</li> <li>(February 28, 2020)   Training intelligent adversaries using self-play with ML-Agents</li> <li>(November 11, 2019)   Training your agents 7 times faster with ML-Agents</li> <li>(October 21, 2019)   The AI@Unity interns help shape the world</li> <li>(April 15, 2019)   Unity ML-Agents Toolkit v0.8: Faster training on real games</li> <li>(March 1, 2019)   Unity ML-Agents Toolkit v0.7: A leap towards cross-platform inference</li> <li>(December 17, 2018)   ML-Agents Toolkit v0.6: Improved usability of Brains and Imitation Learning</li> <li>(October 2, 2018)   Puppo, The Corgi: Cuteness Overload with the Unity ML-Agents Toolkit</li> <li>(September 11, 2018)   ML-Agents Toolkit v0.5, new resources for AI researchers available now</li> <li>(June 26, 2018)   Solving sparse-reward tasks with Curiosity</li> <li>(June 19, 2018)   Unity ML-Agents Toolkit v0.4 and Udacity Deep Reinforcement Learning Nanodegree</li> <li>(May 24, 2018)   Imitation Learning in Unity: The Workflow</li> <li>(March 15, 2018)   ML-Agents Toolkit v0.3 Beta released: Imitation Learning, feedback-driven features, and more</li> <li>(December 11, 2017)   Using Machine Learning Agents in a real game: a beginner\u2019s guide</li> <li>(December 8, 2017)   Introducing ML-Agents Toolkit v0.2: Curriculum Learning, new environments, and more</li> <li>(September 19, 2017)   Introducing: Unity Machine Learning Agents Toolkit</li> <li>Overviewing reinforcement learning concepts   (multi-armed bandit   and   Q-learning)</li> </ul>"},{"location":"Readme/#more-from-unity","title":"More from Unity","text":"<ul> <li>Unity Sentis</li> <li>Introductin Unity Muse and Sentis</li> </ul>"},{"location":"Readme/#community-and-feedback","title":"Community and Feedback","text":"<p>The ML-Agents Toolkit is an open-source project and we encourage and welcome contributions. If you wish to contribute, be sure to review our contribution guidelines and code of conduct.</p> <p>For problems with the installation and setup of the ML-Agents Toolkit, or discussions about how to best setup or train your agents, please create a new thread on the Unity ML-Agents forum and make sure to include as much detail as possible. If you run into any other problems using the ML-Agents Toolkit or have a specific feature request, please submit a GitHub issue.</p> <p>Please tell us which samples you would like to see shipped with the ML-Agents Unity package by replying to this forum thread.</p> <p>Your opinion matters a great deal to us. Only by hearing your thoughts on the Unity ML-Agents Toolkit can we continue to improve and grow. Please take a few minutes to let us know about it.</p> <p>For any other questions or feedback, connect directly with the ML-Agents team at ml-agents@unity3d.com.</p>"},{"location":"Readme/#privacy","title":"Privacy","text":"<p>In order to improve the developer experience for Unity ML-Agents Toolkit, we have added in-editor analytics. Please refer to \"Information that is passively collected by Unity\" in the Unity Privacy Policy.</p>"},{"location":"Sentis/","title":"Sentis","text":"<p>The ML-Agents Toolkit allows you to use pre-trained neural network models inside your Unity games. This support is possible thanks to the Sentis (codenamed Sentis). Sentis uses compute shaders to run the neural network within Unity.</p>"},{"location":"Sentis/#supported-devices","title":"Supported devices","text":"<p>See the Sentis documentation for a list of the supported platforms.</p> <p>Scripting Backends : Sentis is generally faster with IL2CPP than with Mono for Standalone builds. In the Editor, It is not possible to use Sentis with GPU device selected when Editor Graphics Emulation is set to OpenGL(ES) 3.0 or 2.0 emulation. Also there might be non-fatal build time errors when target platform includes Graphics API that does not support Unity Compute Shaders.</p>"},{"location":"Sentis/#using-sentis","title":"Using Sentis","text":"<p>When using a model, drag the model file into the Model field in the Inspector of the Agent. Select the Inference Device : CPU or GPU you want to use for Inference.</p> <p>Note: For most of the models generated with the ML-Agents Toolkit, CPU will be faster than GPU. You should use the GPU only if you use the ResNet visual encoder or have a large number of agents with visual observations.</p>"},{"location":"Sentis/#unsupported-use-cases","title":"Unsupported use cases","text":""},{"location":"Sentis/#externally-trained-models","title":"Externally trained models","text":"<p>The ML-Agents Toolkit only supports the models created with our trainers. Model loading expects certain conventions for constants and tensor names. While it is possible to construct a model that follows these conventions, we don't provide any additional help for this. More details can be found in TensorNames.cs and SentisModelParamLoader.cs.</p> <p>If you wish to run inference on an externally trained model, you should use Sentis directly, instead of trying to run it through ML-Agents.</p>"},{"location":"Sentis/#model-inference-outside-of-unity","title":"Model inference outside of Unity","text":"<p>We do not provide support for inference anywhere outside of Unity. The <code>.onnx</code> files produced by training use the open format ONNX; if you wish to convert a <code>.onnx</code> file to another format or run inference with them, refer to their documentation.</p>"},{"location":"Training-Configuration-File/","title":"Training Configuration File","text":"<p>Table of Contents</p> <ul> <li>Common Trainer Configurations</li> <li>Trainer-specific Configurations</li> <li>PPO-specific Configurations</li> <li>SAC-specific Configurations</li> <li>Reward Signals</li> <li>Extrinsic Rewards</li> <li>Curiosity Intrinsic Reward</li> <li>GAIL Intrinsic Reward</li> <li>RND Intrinsic Reward</li> <li>Reward Signal Settings for SAC</li> <li>Behavioral Cloning</li> <li>Memory-enhanced Agents using Recurrent Neural Networks</li> <li>Self-Play</li> <li>Note on Reward Signals</li> <li>Note on Swap Steps</li> </ul>"},{"location":"Training-Configuration-File/#common-trainer-configurations","title":"Common Trainer Configurations","text":"<p>One of the first decisions you need to make regarding your training run is which trainer to use: PPO, SAC, or POCA. There are some training configurations that are common to both trainers (which we review now) and others that depend on the choice of the trainer (which we review on subsequent sections).</p> Setting Description <code>trainer_type</code> (default = <code>ppo</code>) The type of trainer to use: <code>ppo</code>,  <code>sac</code>, or <code>poca</code>. <code>summary_freq</code> (default = <code>50000</code>) Number of experiences that needs to be collected before generating and displaying training statistics. This determines the granularity of the graphs in Tensorboard. <code>time_horizon</code> (default = <code>64</code>) How many steps of experience to collect per-agent before adding it to the experience buffer. When this limit is reached before the end of an episode, a value estimate is used to predict the overall expected reward from the agent's current state. As such, this parameter trades off between a less biased, but higher variance estimate (long time horizon) and more biased, but less varied estimate (short time horizon). In cases where there are frequent rewards within an episode, or episodes are prohibitively large, a smaller number can be more ideal. This number should be large enough to capture all the important behavior within a sequence of an agent's actions.  Typical range: <code>32</code> - <code>2048</code> <code>max_steps</code> (default = <code>500000</code>) Total number of steps (i.e., observation collected and action taken) that must be taken in the environment (or across all environments if using multiple in parallel) before ending the training process. If you have multiple agents with the same behavior name within your environment, all steps taken by those agents will contribute to the same <code>max_steps</code> count. Typical range: <code>5e5</code> - <code>1e7</code> <code>keep_checkpoints</code> (default = <code>5</code>) The maximum number of model checkpoints to keep. Checkpoints are saved after the number of steps specified by the checkpoint_interval option. Once the maximum number of checkpoints has been reached, the oldest checkpoint is deleted when saving a new checkpoint. <code>even_checkpoints</code> (default = <code>false</code>) If set to true, ignores <code>checkpoint_interval</code> and evenly distributes checkpoints throughout training based on <code>keep_checkpoints</code>and <code>max_steps</code>, i.e. <code>checkpoint_interval = max_steps / keep_checkpoints</code>. Useful for cataloging agent behavior throughout training. <code>checkpoint_interval</code> (default = <code>500000</code>) The number of experiences collected between each checkpoint by the trainer. A maximum of <code>keep_checkpoints</code> checkpoints are saved before old ones are deleted. Each checkpoint saves the <code>.onnx</code> files in <code>results/</code> folder. <code>init_path</code> (default = None) Initialize trainer from a previously saved model. Note that the prior run should have used the same trainer configurations as the current run, and have been saved with the same version of ML-Agents. You can provide either the file name or the full path to the checkpoint, e.g. <code>{checkpoint_name.pt}</code> or <code>./models/{run-id}/{behavior_name}/{checkpoint_name.pt}</code>. This option is provided in case you want to initialize different behaviors from different runs or initialize from an older checkpoint; in most cases, it is sufficient to use the <code>--initialize-from</code> CLI parameter to initialize all models from the same run. <code>threaded</code> (default = <code>false</code>) Allow environments to step while updating the model. This might result in a training speedup, especially when using SAC. For best performance, leave setting to <code>false</code> when using self-play. <code>hyperparameters -&gt; learning_rate</code> (default = <code>3e-4</code>) Initial learning rate for gradient descent. Corresponds to the strength of each gradient descent update step. This should typically be decreased if training is unstable, and the reward does not consistently increase. Typical range: <code>1e-5</code> - <code>1e-3</code> <code>hyperparameters -&gt; batch_size</code> Number of experiences in each iteration of gradient descent. This should always be multiple times smaller than <code>buffer_size</code>. If you are using continuous actions, this value should be large (on the order of 1000s). If you are using only discrete actions, this value should be smaller (on the order of 10s).  Typical range: (Continuous - PPO): <code>512</code> - <code>5120</code>; (Continuous - SAC): <code>128</code> - <code>1024</code>; (Discrete, PPO &amp; SAC): <code>32</code> - <code>512</code>. <code>hyperparameters -&gt; buffer_size</code> (default = <code>10240</code> for PPO and <code>50000</code> for SAC) PPO: Number of experiences to collect before updating the policy model. Corresponds to how many experiences should be collected before we do any learning or updating of the model. This should be multiple times larger than <code>batch_size</code>. Typically a larger <code>buffer_size</code> corresponds to more stable training updates.  SAC: The max size of the experience buffer - on the order of thousands of times longer than your episodes, so that SAC can learn from old as well as new experiences. Typical range: PPO: <code>2048</code> - <code>409600</code>; SAC: <code>50000</code> - <code>1000000</code> <code>hyperparameters -&gt; learning_rate_schedule</code> (default = <code>linear</code> for PPO and <code>constant</code> for SAC) Determines how learning rate changes over time. For PPO, we recommend decaying learning rate until max_steps so learning converges more stably. However, for some cases (e.g. training for an unknown amount of time) this feature can be disabled. For SAC, we recommend holding learning rate constant so that the agent can continue to learn until its Q function converges naturally. <code>linear</code> decays the learning_rate linearly, reaching 0 at max_steps, while <code>constant</code> keeps the learning rate constant for the entire training run. <code>network_settings -&gt; hidden_units</code> (default = <code>128</code>) Number of units in the hidden layers of the neural network. Correspond to how many units are in each fully connected layer of the neural network. For simple problems where the correct action is a straightforward combination of the observation inputs, this should be small. For problems where the action is a very complex interaction between the observation variables, this should be larger.  Typical range: <code>32</code> - <code>512</code> <code>network_settings -&gt; num_layers</code> (default = <code>2</code>) The number of hidden layers in the neural network. Corresponds to how many hidden layers are present after the observation input, or after the CNN encoding of the visual observation. For simple problems, fewer layers are likely to train faster and more efficiently. More layers may be necessary for more complex control problems.  Typical range: <code>1</code> - <code>3</code> <code>network_settings -&gt; normalize</code> (default = <code>false</code>) Whether normalization is applied to the vector observation inputs. This normalization is based on the running average and variance of the vector observation. Normalization can be helpful in cases with complex continuous control problems, but may be harmful with simpler discrete control problems. <code>network_settings -&gt; vis_encode_type</code> (default = <code>simple</code>) Encoder type for encoding visual observations.  <code>simple</code> (default) uses a simple encoder which consists of two convolutional layers, <code>nature_cnn</code> uses the CNN implementation proposed by Mnih et al., consisting of three convolutional layers, and <code>resnet</code> uses the IMPALA Resnet consisting of three stacked layers, each with two residual blocks, making a much larger network than the other two. <code>match3</code> is a smaller CNN (Gudmundsoon et al.) that can capture more granular spatial relationships and is optimized for board games. <code>fully_connected</code> uses a single fully connected dense layer as encoder without any convolutional layers.  Due to the size of convolution kernel, there is a minimum observation size limitation that each encoder type can handle - <code>simple</code>: 20x20, <code>nature_cnn</code>: 36x36, <code>resnet</code>: 15 x 15, <code>match3</code>: 5x5.  <code>fully_connected</code> doesn't have convolutional layers and thus no size limits, but since it has less representation power it should be reserved for very small inputs. Note that using the <code>match3</code> CNN with very large visual input might result in a huge observation encoding and thus potentially slow down training or cause memory issues. <code>network_settings -&gt; conditioning_type</code> (default = <code>hyper</code>) Conditioning type for the policy using goal observations.  <code>none</code> treats the goal observations as regular observations, <code>hyper</code> (default) uses a HyperNetwork with goal observations as input to generate some of the weights of the policy. Note that when using <code>hyper</code> the number of parameters of the network increases greatly. Therefore, it is recommended to reduce the number of <code>hidden_units</code> when using this <code>conditioning_type</code>"},{"location":"Training-Configuration-File/#trainer-specific-configurations","title":"Trainer-specific Configurations","text":"<p>Depending on your choice of a trainer, there are additional trainer-specific configurations. We present them below in two separate tables, but keep in mind that you only need to include the configurations for the trainer selected (i.e. the <code>trainer</code> setting above).</p>"},{"location":"Training-Configuration-File/#ppo-specific-configurations","title":"PPO-specific Configurations","text":"Setting Description <code>hyperparameters -&gt; beta</code> (default = <code>5.0e-3</code>) Strength of the entropy regularization, which makes the policy \"more random.\" This ensures that agents properly explore the action space during training. Increasing this will ensure more random actions are taken. This should be adjusted such that the entropy (measurable from TensorBoard) slowly decreases alongside increases in reward. If entropy drops too quickly, increase beta. If entropy drops too slowly, decrease <code>beta</code>. Typical range: <code>1e-4</code> - <code>1e-2</code> <code>hyperparameters -&gt; epsilon</code> (default = <code>0.2</code>) Influences how rapidly the policy can evolve during training. Corresponds to the acceptable threshold of divergence between the old and new policies during gradient descent updating. Setting this value small will result in more stable updates, but will also slow the training process. Typical range: <code>0.1</code> - <code>0.3</code> <code>hyperparameters -&gt; beta_schedule</code> (default = <code>learning_rate_schedule</code>) Determines how beta changes over time. <code>linear</code> decays beta linearly, reaching 0 at max_steps, while <code>constant</code> keeps beta constant for the entire training run. If not explicitly set, the default beta schedule will be set to <code>hyperparameters -&gt; learning_rate_schedule</code>. <code>hyperparameters -&gt; epsilon_schedule</code> (default = <code>learning_rate_schedule</code>) Determines how epsilon changes over time (PPO only). <code>linear</code> decays epsilon linearly, reaching 0 at max_steps, while <code>constant</code> keeps the epsilon constant for the entire training run. If not explicitly set, the default epsilon schedule will be set to <code>hyperparameters -&gt; learning_rate_schedule</code>. <code>hyperparameters -&gt; lambd</code> (default = <code>0.95</code>) Regularization parameter (lambda) used when calculating the Generalized Advantage Estimate (GAE). This can be thought of as how much the agent relies on its current value estimate when calculating an updated value estimate. Low values correspond to relying more on the current value estimate (which can be high bias), and high values correspond to relying more on the actual rewards received in the environment (which can be high variance). The parameter provides a trade-off between the two, and the right value can lead to a more stable training process. Typical range: <code>0.9</code> - <code>0.95</code> <code>hyperparameters -&gt; num_epoch</code> (default = <code>3</code>) Number of passes to make through the experience buffer when performing gradient descent optimization.The larger the batch_size, the larger it is acceptable to make this. Decreasing this will ensure more stable updates, at the cost of slower learning. Typical range: <code>3</code> - <code>10</code> <code>hyperparameters -&gt; shared_critic</code> (default = <code>False</code>) Whether or not the policy and value function networks share a backbone. It may be useful to use a shared backbone when learning from image observations."},{"location":"Training-Configuration-File/#sac-specific-configurations","title":"SAC-specific Configurations","text":"Setting Description <code>hyperparameters -&gt; buffer_init_steps</code> (default = <code>0</code>) Number of experiences to collect into the buffer before updating the policy model. As the untrained policy is fairly random, pre-filling the buffer with random actions is useful for exploration. Typically, at least several episodes of experiences should be pre-filled. Typical range: <code>1000</code> - <code>10000</code> <code>hyperparameters -&gt; init_entcoef</code> (default = <code>1.0</code>) How much the agent should explore in the beginning of training. Corresponds to the initial entropy coefficient set at the beginning of training. In SAC, the agent is incentivized to make its actions entropic to facilitate better exploration. The entropy coefficient weighs the true reward with a bonus entropy reward. The entropy coefficient is automatically adjusted to a preset target entropy, so the <code>init_entcoef</code> only corresponds to the starting value of the entropy bonus. Increase init_entcoef to explore more in the beginning, decrease to converge to a solution faster. Typical range: (Continuous): <code>0.5</code> - <code>1.0</code>; (Discrete): <code>0.05</code> - <code>0.5</code> <code>hyperparameters -&gt; save_replay_buffer</code> (default = <code>false</code>) Whether to save and load the experience replay buffer as well as the model when quitting and re-starting training. This may help resumes go more smoothly, as the experiences collected won't be wiped. Note that replay buffers can be very large, and will take up a considerable amount of disk space. For that reason, we disable this feature by default. <code>hyperparameters -&gt; tau</code> (default = <code>0.005</code>) How aggressively to update the target network used for bootstrapping value estimation in SAC. Corresponds to the magnitude of the target Q update during the SAC model update. In SAC, there are two neural networks: the target and the policy. The target network is used to bootstrap the policy's estimate of the future rewards at a given state, and is fixed while the policy is being updated. This target is then slowly updated according to tau. Typically, this value should be left at 0.005. For simple problems, increasing tau to 0.01 might reduce the time it takes to learn, at the cost of stability. Typical range: <code>0.005</code> - <code>0.01</code> <code>hyperparameters -&gt; steps_per_update</code> (default = <code>1</code>) Average ratio of agent steps (actions) taken to updates made of the agent's policy. In SAC, a single \"update\" corresponds to grabbing a batch of size <code>batch_size</code> from the experience replay buffer, and using this mini batch to update the models. Note that it is not guaranteed that after exactly <code>steps_per_update</code> steps an update will be made, only that the ratio will hold true over many steps. Typically, <code>steps_per_update</code> should be greater than or equal to 1. Note that setting <code>steps_per_update</code> lower will improve sample efficiency (reduce the number of steps required to train) but increase the CPU time spent performing updates. For most environments where steps are fairly fast (e.g. our example environments) <code>steps_per_update</code> equal to the number of agents in the scene is a good balance. For slow environments (steps take 0.1 seconds or more) reducing <code>steps_per_update</code> may improve training speed. We can also change <code>steps_per_update</code> to lower than 1 to update more often than once per step, though this will usually result in a slowdown unless the environment is very slow. Typical range: <code>1</code> - <code>20</code> <code>hyperparameters -&gt; reward_signal_num_update</code> (default = <code>steps_per_update</code>) Number of steps per mini batch sampled and used for updating the reward signals. By default, we update the reward signals once every time the main policy is updated. However, to imitate the training procedure in certain imitation learning papers (e.g. Kostrikov et. al, Blond\u00e9 et. al), we may want to update the reward signal (GAIL) M times for every update of the policy. We can change <code>steps_per_update</code> of SAC to N, as well as <code>reward_signal_steps_per_update</code> under <code>reward_signals</code> to N / M to accomplish this. By default, <code>reward_signal_steps_per_update</code> is set to <code>steps_per_update</code>."},{"location":"Training-Configuration-File/#ma-poca-specific-configurations","title":"MA-POCA-specific Configurations","text":"<p>MA-POCA uses the same configurations as PPO, and there are no additional POCA-specific parameters.</p> <p>NOTE: Reward signals other than Extrinsic Rewards have not been extensively tested with MA-POCA, though they can still be added and used for training on a your-mileage-may-vary basis.</p>"},{"location":"Training-Configuration-File/#reward-signals","title":"Reward Signals","text":"<p>The <code>reward_signals</code> section enables the specification of settings for both extrinsic (i.e. environment-based) and intrinsic reward signals (e.g. curiosity and GAIL). Each reward signal should define at least two parameters, <code>strength</code> and <code>gamma</code>, in addition to any class-specific hyperparameters. Note that to remove a reward signal, you should delete its entry entirely from <code>reward_signals</code>. At least one reward signal should be left defined at all times. Provide the following configurations to design the reward signal for your training run.</p>"},{"location":"Training-Configuration-File/#extrinsic-rewards","title":"Extrinsic Rewards","text":"<p>Enable these settings to ensure that your training run incorporates your environment-based reward signal:</p> Setting Description <code>extrinsic -&gt; strength</code> (default = <code>1.0</code>) Factor by which to multiply the reward given by the environment. Typical ranges will vary depending on the reward signal. Typical range: <code>1.00</code> <code>extrinsic -&gt; gamma</code> (default = <code>0.99</code>) Discount factor for future rewards coming from the environment. This can be thought of as how far into the future the agent should care about possible rewards. In situations when the agent should be acting in the present in order to prepare for rewards in the distant future, this value should be large. In cases when rewards are more immediate, it can be smaller. Must be strictly smaller than 1. Typical range: <code>0.8</code> - <code>0.995</code>"},{"location":"Training-Configuration-File/#curiosity-intrinsic-reward","title":"Curiosity Intrinsic Reward","text":"<p>To enable curiosity, provide these settings:</p> Setting Description <code>curiosity -&gt; strength</code> (default = <code>1.0</code>) Magnitude of the curiosity reward generated by the intrinsic curiosity module. This should be scaled in order to ensure it is large enough to not be overwhelmed by extrinsic reward signals in the environment. Likewise it should not be too large to overwhelm the extrinsic reward signal. Typical range: <code>0.001</code> - <code>0.1</code> <code>curiosity -&gt; gamma</code> (default = <code>0.99</code>) Discount factor for future rewards. Typical range: <code>0.8</code> - <code>0.995</code> <code>curiosity -&gt; network_settings</code> Please see the documentation for <code>network_settings</code> under Common Trainer Configurations. The network specs used by the intrinsic curiosity model. The value should of <code>hidden_units</code> should be small enough to encourage the ICM to compress the original observation, but also not too small to prevent it from learning to differentiate between expected and actual observations. Typical range: <code>64</code> - <code>256</code> <code>curiosity -&gt; learning_rate</code> (default = <code>3e-4</code>) Learning rate used to update the intrinsic curiosity module. This should typically be decreased if training is unstable, and the curiosity loss is unstable. Typical range: <code>1e-5</code> - <code>1e-3</code>"},{"location":"Training-Configuration-File/#gail-intrinsic-reward","title":"GAIL Intrinsic Reward","text":"<p>To enable GAIL (assuming you have recorded demonstrations), provide these settings:</p> Setting Description <code>gail -&gt; strength</code> (default = <code>1.0</code>) Factor by which to multiply the raw reward. Note that when using GAIL with an Extrinsic Signal, this value should be set lower if your demonstrations are suboptimal (e.g. from a human), so that a trained agent will focus on receiving extrinsic rewards instead of exactly copying the demonstrations. Keep the strength below about 0.1 in those cases. Typical range: <code>0.01</code> - <code>1.0</code> <code>gail -&gt; gamma</code> (default = <code>0.99</code>) Discount factor for future rewards. Typical range: <code>0.8</code> - <code>0.9</code> <code>gail -&gt; demo_path</code> (Required, no default) The path to your .demo file or directory of .demo files. <code>gail -&gt; network_settings</code> Please see the documentation for <code>network_settings</code> under Common Trainer Configurations. The network specs for the GAIL discriminator. The value of <code>hidden_units</code> should be small enough to encourage the discriminator to compress the original observation, but also not too small to prevent it from learning to differentiate between demonstrated and actual behavior. Dramatically increasing this size will also negatively affect training times. Typical range: <code>64</code> - <code>256</code> <code>gail -&gt; learning_rate</code> (Optional, default = <code>3e-4</code>) Learning rate used to update the discriminator. This should typically be decreased if training is unstable, and the GAIL loss is unstable. Typical range: <code>1e-5</code> - <code>1e-3</code> <code>gail -&gt; use_actions</code> (default = <code>false</code>) Determines whether the discriminator should discriminate based on both observations and actions, or just observations. Set to True if you want the agent to mimic the actions from the demonstrations, and False if you'd rather have the agent visit the same states as in the demonstrations but with possibly different actions. Setting to False is more likely to be stable, especially with imperfect demonstrations, but may learn slower. <code>gail -&gt; use_vail</code> (default = <code>false</code>) Enables a variational bottleneck within the GAIL discriminator. This forces the discriminator to learn a more general representation and reduces its tendency to be \"too good\" at discriminating, making learning more stable. However, it does increase training time. Enable this if you notice your imitation learning is unstable, or unable to learn the task at hand."},{"location":"Training-Configuration-File/#rnd-intrinsic-reward","title":"RND Intrinsic Reward","text":"<p>Random Network Distillation (RND) is only available for the PyTorch trainers. To enable RND, provide these settings:</p> Setting Description <code>rnd -&gt; strength</code> (default = <code>1.0</code>) Magnitude of the curiosity reward generated by the intrinsic rnd module. This should be scaled in order to ensure it is large enough to not be overwhelmed by extrinsic reward signals in the environment. Likewise it should not be too large to overwhelm the extrinsic reward signal. Typical range: <code>0.001</code> - <code>0.01</code> <code>rnd -&gt; gamma</code> (default = <code>0.99</code>) Discount factor for future rewards. Typical range: <code>0.8</code> - <code>0.995</code> <code>rnd -&gt; network_settings</code> Please see the documentation for <code>network_settings</code> under Common Trainer Configurations. The network specs for the RND model. <code>curiosity -&gt; learning_rate</code> (default = <code>3e-4</code>) Learning rate used to update the RND module. This should be large enough for the RND module to quickly learn the state representation, but small enough to allow for stable learning. Typical range: <code>1e-5</code> - <code>1e-3</code>"},{"location":"Training-Configuration-File/#behavioral-cloning","title":"Behavioral Cloning","text":"<p>To enable Behavioral Cloning as a pre-training option (assuming you have recorded demonstrations), provide the following configurations under the <code>behavioral_cloning</code> section:</p> Setting Description <code>demo_path</code> (Required, no default) The path to your .demo file or directory of .demo files. <code>strength</code> (default = <code>1.0</code>) Learning rate of the imitation relative to the learning rate of PPO, and roughly corresponds to how strongly we allow BC to influence the policy. Typical range: <code>0.1</code> - <code>0.5</code> <code>steps</code> (default = <code>0</code>) During BC, it is often desirable to stop using demonstrations after the agent has \"seen\" rewards, and allow it to optimize past the available demonstrations and/or generalize outside of the provided demonstrations. steps corresponds to the training steps over which BC is active. The learning rate of BC will anneal over the steps. Set the steps to 0 for constant imitation over the entire training run. <code>batch_size</code> (default = <code>batch_size</code> of trainer) Number of demonstration experiences used for one iteration of a gradient descent update. If not specified, it will default to the <code>batch_size</code> of the trainer. Typical range: (Continuous): <code>512</code> - <code>5120</code>; (Discrete): <code>32</code> - <code>512</code> <code>num_epoch</code> (default = <code>num_epoch</code> of trainer) Number of passes through the experience buffer during gradient descent. If not specified, it will default to the number of epochs set for PPO. Typical range: <code>3</code> - <code>10</code> <code>samples_per_update</code> (default = <code>0</code>) Maximum number of samples to use during each imitation update. You may want to lower this if your demonstration dataset is very large to avoid overfitting the policy on demonstrations. Set to 0 to train over all of the demonstrations at each update step. Typical range: <code>buffer_size</code>"},{"location":"Training-Configuration-File/#memory-enhanced-agents-using-recurrent-neural-networks","title":"Memory-enhanced Agents using Recurrent Neural Networks","text":"<p>You can enable your agents to use memory by adding a <code>memory</code> section under <code>network_settings</code>, and setting <code>memory_size</code> and <code>sequence_length</code>:</p> Setting Description <code>network_settings -&gt; memory -&gt; memory_size</code> (default = <code>128</code>) Size of the memory an agent must keep. In order to use a LSTM, training requires a sequence of experiences instead of single experiences. Corresponds to the size of the array of floating point numbers used to store the hidden state of the recurrent neural network of the policy. This value must be a multiple of 2, and should scale with the amount of information you expect the agent will need to remember in order to successfully complete the task. Typical range: <code>32</code> - <code>256</code> <code>network_settings -&gt; memory -&gt; sequence_length</code> (default = <code>64</code>) Defines how long the sequences of experiences must be while training. Note that if this number is too small, the agent will not be able to remember things over longer periods of time. If this number is too large, the neural network will take longer to train. Typical range: <code>4</code> - <code>128</code> <p>A few considerations when deciding to use memory:</p> <ul> <li>LSTM does not work well with continuous actions. Please use   discrete actions for better results.</li> <li>Adding a recurrent layer increases the complexity of the neural network, it is   recommended to decrease <code>num_layers</code> when using recurrent.</li> <li>It is required that <code>memory_size</code> be divisible by 2.</li> </ul>"},{"location":"Training-Configuration-File/#self-play","title":"Self-Play","text":"<p>Training with self-play adds additional confounding factors to the usual issues faced by reinforcement learning. In general, the tradeoff is between the skill level and generality of the final policy and the stability of learning. Training against a set of slowly or unchanging adversaries with low diversity results in a more stable learning process than training against a set of quickly changing adversaries with high diversity. With this context, this guide discusses the exposed self-play hyperparameters and intuitions for tuning them.</p> <p>If your environment contains multiple agents that are divided into teams, you can leverage our self-play training option by providing these configurations for each Behavior:</p> Setting Description <code>save_steps</code> (default = <code>20000</code>) Number of trainer steps between snapshots. For example, if <code>save_steps=10000</code> then a snapshot of the current policy will be saved every <code>10000</code> trainer steps. Note, trainer steps are counted per agent. For more information, please see the migration doc after v0.13. A larger value of <code>save_steps</code> will yield a set of opponents that cover a wider range of skill levels and possibly play styles since the policy receives more training. As a result, the agent trains against a wider variety of opponents. Learning a policy to defeat more diverse opponents is a harder problem and so may require more overall training steps but also may lead to more general and robust policy at the end of training. This value is also dependent on how intrinsically difficult the environment is for the agent.  Typical range: <code>10000</code> - <code>100000</code> <code>team_change</code> (default = <code>5 * save_steps</code>) Number of trainer_steps between switching the learning team. This is the number of trainer steps the teams associated with a specific ghost trainer will train before a different team becomes the new learning team. It is possible that, in asymmetric games, opposing teams require fewer trainer steps to make similar performance gains. This enables users to train a more complicated team of agents for more trainer steps than a simpler team of agents per team switch. A larger value of <code>team-change</code> will allow the agent to train longer against it's opponents. The longer an agent trains against the same set of opponents the more able it will be to defeat them. However, training against them for too long may result in overfitting to the particular opponent strategies and so the agent may fail against the next batch of opponents.  The value of <code>team-change</code> will determine how many snapshots of the agent's policy are saved to be used as opponents for the other team. So, we recommend setting this value as a function of the <code>save_steps</code> parameter discussed previously.  Typical range: 4x-10x where x=<code>save_steps</code> <code>swap_steps</code> (default = <code>10000</code>) Number of ghost steps (not trainer steps) between swapping the opponents policy with a different snapshot. A 'ghost step' refers to a step taken by an agent that is following a fixed policy and not learning. The reason for this distinction is that in asymmetric games, we may have teams with an unequal number of agents e.g. a 2v1 scenario like our Strikers Vs Goalie example environment. The team with two agents collects twice as many agent steps per environment step as the team with one agent. Thus, these two values will need to be distinct to ensure that the same number of trainer steps corresponds to the same number of opponent swaps for each team. The formula for <code>swap_steps</code> if a user desires <code>x</code> swaps of a team with <code>num_agents</code> agents against an opponent team with <code>num_opponent_agents</code> agents during <code>team-change</code> total steps is: <code>(num_agents / num_opponent_agents) * (team_change / x)</code>  Typical range: <code>10000</code> - <code>100000</code> <code>play_against_latest_model_ratio</code> (default = <code>0.5</code>) Probability an agent will play against the latest opponent policy. With probability 1 - <code>play_against_latest_model_ratio</code>, the agent will play against a snapshot of its opponent from a past iteration.  A larger value of <code>play_against_latest_model_ratio</code> indicates that an agent will be playing against the current opponent more often. Since the agent is updating it's policy, the opponent will be different from iteration to iteration. This can lead to an unstable learning environment, but poses the agent with an auto-curricula of more increasingly challenging situations which may lead to a stronger final policy.  Typical range: <code>0.0</code> - <code>1.0</code> <code>window</code> (default = <code>10</code>) Size of the sliding window of past snapshots from which the agent's opponents are sampled. For example, a <code>window</code> size of 5 will save the last 5 snapshots taken. Each time a new snapshot is taken, the oldest is discarded. A larger value of <code>window</code> means that an agent's pool of opponents will contain a larger diversity of behaviors since it will contain policies from earlier in the training run. Like in the <code>save_steps</code> hyperparameter, the agent trains against a wider variety of opponents. Learning a policy to defeat more diverse opponents is a harder problem and so may require more overall training steps but also may lead to more general and robust policy at the end of training.  Typical range: <code>5</code> - <code>30</code>"},{"location":"Training-Configuration-File/#note-on-reward-signals","title":"Note on Reward Signals","text":"<p>We make the assumption that the final reward in a trajectory corresponds to the outcome of an episode. A final reward of +1 indicates winning, -1 indicates losing and 0 indicates a draw. The ELO calculation (discussed below) depends on this final reward being either +1, 0, -1.</p> <p>The reward signal should still be used as described in the documentation for the other trainers. However, we encourage users to be a bit more conservative when shaping reward functions due to the instability and non-stationarity of learning in adversarial games. Specifically, we encourage users to begin with the simplest possible reward function (+1 winning, -1 losing) and to allow for more iterations of training to compensate for the sparsity of reward.</p>"},{"location":"Training-Configuration-File/#note-on-swap-steps","title":"Note on Swap Steps","text":"<p>As an example, in a 2v1 scenario, if we want the swap to occur x=4 times during team-change=200000 steps, the swap_steps for the team of one agent is:</p> <p>swap_steps = (1 / 2) * (200000 / 4) = 25000 The swap_steps for the team of two agents is:</p> <p>swap_steps = (2 / 1) * (200000 / 4) = 100000 Note, with equal team sizes, the first term is equal to 1 and swap_steps can be calculated by just dividing the total steps by the desired number of swaps.</p> <p>A larger value of swap_steps means that an agent will play against the same fixed opponent for a longer number of training iterations. This results in a more stable training scenario, but leaves the agent open to the risk of overfitting it's behavior for this particular opponent. Thus, when a new opponent is swapped, the agent may lose more often than expected.</p>"},{"location":"Training-ML-Agents/","title":"Training ML-Agents","text":"<p>Table of Contents</p> <ul> <li>Training ML-Agents</li> <li>Training with mlagents-learn<ul> <li>Starting Training</li> <li>Observing Training</li> <li>Stopping and Resuming Training</li> <li>Loading an Existing Model</li> </ul> </li> <li>Training Configurations<ul> <li>Adding CLI Arguments to the Training Configuration file</li> <li>Environment settings</li> <li>Engine settings</li> <li>Checkpoint settings</li> <li>Torch settings:</li> <li>Behavior Configurations</li> <li>Default Behavior Settings</li> <li>Environment Parameters</li> <li>Environment Parameter Randomization<ul> <li>Supported Sampler Types</li> <li>Training with Environment Parameter Randomization</li> </ul> </li> <li>Curriculum<ul> <li>Training with a Curriculum</li> </ul> </li> <li>Training Using Concurrent Unity Instances</li> </ul> </li> </ul> <p>For a broad overview of reinforcement learning, imitation learning and all the training scenarios, methods and options within the ML-Agents Toolkit, see ML-Agents Toolkit Overview.</p> <p>Once your learning environment has been created and is ready for training, the next step is to initiate a training run. Training in the ML-Agents Toolkit is powered by a dedicated Python package, <code>mlagents</code>. This package exposes a command <code>mlagents-learn</code> that is the single entry point for all training workflows (e.g. reinforcement leaning, imitation learning, curriculum learning). Its implementation can be found at ml-agents/mlagents/trainers/learn.py.</p>"},{"location":"Training-ML-Agents/#training-with-mlagents-learn","title":"Training with mlagents-learn","text":""},{"location":"Training-ML-Agents/#starting-training","title":"Starting Training","text":"<p><code>mlagents-learn</code> is the main training utility provided by the ML-Agents Toolkit. It accepts a number of CLI options in addition to a YAML configuration file that contains all the configurations and hyperparameters to be used during training. The set of configurations and hyperparameters to include in this file depend on the agents in your environment and the specific training method you wish to utilize. Keep in mind that the hyperparameter values can have a big impact on the training performance (i.e. your agent's ability to learn a policy that solves the task). In this page, we will review all the hyperparameters for all training methods and provide guidelines and advice on their values.</p> <p>To view a description of all the CLI options accepted by <code>mlagents-learn</code>, use the <code>--help</code>:</p> <pre><code>mlagents-learn --help\n</code></pre> <p>The basic command for training is:</p> <pre><code>mlagents-learn &lt;trainer-config-file&gt; --env=&lt;env_name&gt; --run-id=&lt;run-identifier&gt;\n</code></pre> <p>where</p> <ul> <li><code>&lt;trainer-config-file&gt;</code> is the file path of the trainer configuration YAML.   This contains all the hyperparameter values. We offer a detailed guide on the   structure of this file and the meaning of the hyperparameters (and advice on   how to set them) in the dedicated   Training Configurations section below.</li> <li><code>&lt;env_name&gt;</code>(Optional) is the name (including path) of your   Unity executable containing the agents   to be trained. If <code>&lt;env_name&gt;</code> is not passed, the training will happen in the   Editor. Press the Play button in Unity when the message \"Start training   by pressing the Play button in the Unity Editor\" is displayed on the screen.</li> <li><code>&lt;run-identifier&gt;</code> is a unique name you can use to identify the results of   your training runs.</li> </ul> <p>See the Getting Started Guide for a sample execution of the <code>mlagents-learn</code> command.</p>"},{"location":"Training-ML-Agents/#observing-training","title":"Observing Training","text":"<p>Regardless of which training methods, configurations or hyperparameters you provide, the training process will always generate three artifacts, all found in the <code>results/&lt;run-identifier&gt;</code> folder:</p> <ol> <li>Summaries: these are training metrics that    are updated throughout the training process. They are helpful to monitor your    training performance and may help inform how to update your hyperparameter    values. See Using TensorBoard for more details on how    to visualize the training metrics.</li> <li>Models: these contain the model checkpoints that    are updated throughout training and the final model file (<code>.onnx</code>). This final    model file is generated once either when training completes or is    interrupted.</li> <li>Timers file (under <code>results/&lt;run-identifier&gt;/run_logs</code>): this contains aggregated    metrics on your training process, including time spent on specific code    blocks. See Profiling in Python for more information    on the timers generated.</li> </ol> <p>These artifacts are updated throughout the training process and finalized when training is completed or is interrupted.</p>"},{"location":"Training-ML-Agents/#stopping-and-resuming-training","title":"Stopping and Resuming Training","text":"<p>To interrupt training and save the current progress, hit <code>Ctrl+C</code> once and wait for the model(s) to be saved out.</p> <p>To resume a previously interrupted or completed training run, use the <code>--resume</code> flag and make sure to specify the previously used run ID.</p> <p>If you would like to re-run a previously interrupted or completed training run and re-use the same run ID (in this case, overwriting the previously generated artifacts), then use the <code>--force</code> flag.</p>"},{"location":"Training-ML-Agents/#loading-an-existing-model","title":"Loading an Existing Model","text":"<p>You can also use this mode to run inference of an already-trained model in Python by using both the <code>--resume</code> and <code>--inference</code> flags. Note that if you want to run inference in Unity, you should use the Sentis.</p> <p>Additionally, if the network architecture changes, you may still load an existing model, but ML-Agents will only load the parts of the model it can load and ignore all others. For instance, if you add a new reward signal, the existing model will load but the new reward signal will be initialized from scratch. If you have a model with a visual encoder (CNN) but change the <code>hidden_units</code>, the CNN will be loaded but the body of the network will be initialized from scratch.</p> <p>Alternatively, you might want to start a new training run but initialize it using an already-trained model. You may want to do this, for instance, if your environment changed and you want a new model, but the old behavior is still better than random. You can do this by specifying <code>--initialize-from=&lt;run-identifier&gt;</code>, where <code>&lt;run-identifier&gt;</code> is the old run ID.</p>"},{"location":"Training-ML-Agents/#training-configurations","title":"Training Configurations","text":"<p>The Unity ML-Agents Toolkit provides a wide range of training scenarios, methods and options. As such, specific training runs may require different training configurations and may generate different artifacts and TensorBoard statistics. This section offers a detailed guide into how to manage the different training set-ups withing the toolkit.</p> <p>More specifically, this section offers a detailed guide on the command-line flags for <code>mlagents-learn</code> that control the training configurations:</p> <ul> <li><code>&lt;trainer-config-file&gt;</code>: defines the training hyperparameters for each   Behavior in the scene, and the set-ups for the environment parameters   (Curriculum Learning and Environment Parameter Randomization)</li> </ul> <p>It is important to highlight that successfully training a Behavior in the ML-Agents Toolkit involves tuning the training hyperparameters and configuration. This guide contains some best practices for tuning the training process when the default parameters don't seem to be giving the level of performance you would like. We provide sample configuration files for our example environments in the config/ directory. The <code>config/ppo/3DBall.yaml</code> was used to train the 3D Balance Ball in the Getting Started guide. That configuration file uses the PPO trainer, but we also have configuration files for SAC and GAIL.</p> <p>Additionally, the set of configurations you provide depend on the training functionalities you use (see ML-Agents Toolkit Overview for a description of all the training functionalities). Each functionality you add typically has its own training configurations. For instance:</p> <ul> <li>Use PPO or SAC?</li> <li>Use Recurrent Neural Networks for adding memory to your agents?</li> <li>Use the intrinsic curiosity module?</li> <li>Ignore the environment reward signal?</li> <li>Pre-train using behavioral cloning? (Assuming you have recorded   demonstrations.)</li> <li>Include the GAIL intrinsic reward signals? (Assuming you have recorded   demonstrations.)</li> <li>Use self-play? (Assuming your environment includes multiple agents.)</li> </ul> <p>The trainer config file, <code>&lt;trainer-config-file&gt;</code>, determines the features you will use during training, and the answers to the above questions will dictate its contents. The rest of this guide breaks down the different sub-sections of the trainer config file and explains the possible settings for each. If you need a list of all the trainer configurations, please see Training Configuration File.</p> <p>NOTE: The configuration file format has been changed between 0.17.0 and 0.18.0 and between 0.18.0 and onwards. To convert an old set of configuration files (trainer config, curriculum, and sampler files) to the new format, a script has been provided. Run <code>python -m mlagents.trainers.upgrade_config -h</code> in your console to see the script's usage.</p>"},{"location":"Training-ML-Agents/#adding-cli-arguments-to-the-training-configuration-file","title":"Adding CLI Arguments to the Training Configuration file","text":"<p>Additionally, within the training configuration YAML file, you can also add the CLI arguments (such as <code>--num-envs</code>).</p> <p>Reminder that a detailed description of all the CLI arguments can be found by using the help utility:</p> <pre><code>mlagents-learn --help\n</code></pre> <p>These additional CLI arguments are grouped into environment, engine, checkpoint and torch. The available settings and example values are shown below.</p>"},{"location":"Training-ML-Agents/#environment-settings","title":"Environment settings","text":"<pre><code>env_settings:\n  env_path: FoodCollector\n  env_args: null\n  base_port: 5005\n  num_envs: 1\n  timeout_wait: 10\n  seed: -1\n  max_lifetime_restarts: 10\n  restarts_rate_limit_n: 1\n  restarts_rate_limit_period_s: 60\n</code></pre>"},{"location":"Training-ML-Agents/#engine-settings","title":"Engine settings","text":"<pre><code>engine_settings:\n  width: 84\n  height: 84\n  quality_level: 5\n  time_scale: 20\n  target_frame_rate: -1\n  capture_frame_rate: 60\n  no_graphics: false\n</code></pre>"},{"location":"Training-ML-Agents/#checkpoint-settings","title":"Checkpoint settings","text":"<pre><code>checkpoint_settings:\n  run_id: foodtorch\n  initialize_from: null\n  load_model: false\n  resume: false\n  force: true\n  train_model: false\n  inference: false\n</code></pre>"},{"location":"Training-ML-Agents/#torch-settings","title":"Torch settings:","text":"<pre><code>torch_settings:\n  device: cpu\n</code></pre>"},{"location":"Training-ML-Agents/#behavior-configurations","title":"Behavior Configurations","text":"<p>The primary section of the trainer config file is a set of configurations for each Behavior in your scene. These are defined under the sub-section <code>behaviors</code> in your trainer config file. Some of the configurations are required while others are optional. To help us get started, below is a sample file that includes all the possible settings if we're using a PPO trainer with all the possible training functionalities enabled (memory, behavioral cloning, curiosity, GAIL and self-play). You will notice that curriculum and environment parameter randomization settings are not part of the <code>behaviors</code> configuration, but in their own section called <code>environment_parameters</code>.</p> <pre><code>behaviors:\n  BehaviorPPO:\n    trainer_type: ppo\n\n    hyperparameters:\n      # Hyperparameters common to PPO and SAC\n      batch_size: 1024\n      buffer_size: 10240\n      learning_rate: 3.0e-4\n      learning_rate_schedule: linear\n\n      # PPO-specific hyperparameters\n      beta: 5.0e-3\n      beta_schedule: constant\n      epsilon: 0.2\n      epsilon_schedule: linear\n      lambd: 0.95\n      num_epoch: 3\n      shared_critic: False\n\n    # Configuration of the neural network (common to PPO/SAC)\n    network_settings:\n      vis_encode_type: simple\n      normalize: false\n      hidden_units: 128\n      num_layers: 2\n      # memory\n      memory:\n        sequence_length: 64\n        memory_size: 256\n\n    # Trainer configurations common to all trainers\n    max_steps: 5.0e5\n    time_horizon: 64\n    summary_freq: 10000\n    keep_checkpoints: 5\n    checkpoint_interval: 50000\n    threaded: false\n    init_path: null\n\n    # behavior cloning\n    behavioral_cloning:\n      demo_path: Project/Assets/ML-Agents/Examples/Pyramids/Demos/ExpertPyramid.demo\n      strength: 0.5\n      steps: 150000\n      batch_size: 512\n      num_epoch: 3\n      samples_per_update: 0\n\n    reward_signals:\n      # environment reward (default)\n      extrinsic:\n        strength: 1.0\n        gamma: 0.99\n\n      # curiosity module\n      curiosity:\n        strength: 0.02\n        gamma: 0.99\n        encoding_size: 256\n        learning_rate: 3.0e-4\n\n      # GAIL\n      gail:\n        strength: 0.01\n        gamma: 0.99\n        encoding_size: 128\n        demo_path: Project/Assets/ML-Agents/Examples/Pyramids/Demos/ExpertPyramid.demo\n        learning_rate: 3.0e-4\n        use_actions: false\n        use_vail: false\n\n    # self-play\n    self_play:\n      window: 10\n      play_against_latest_model_ratio: 0.5\n      save_steps: 50000\n      swap_steps: 2000\n      team_change: 100000\n</code></pre> <p>Here is an equivalent file if we use an SAC trainer instead. Notice that the configurations for the additional functionalities (memory, behavioral cloning, curiosity and self-play) remain unchanged.</p> <pre><code>behaviors:\n  BehaviorSAC:\n    trainer_type: sac\n\n    # Trainer configs common to PPO/SAC (excluding reward signals)\n    # same as PPO config\n\n    # SAC-specific configs (replaces the hyperparameters section above)\n    hyperparameters:\n      # Hyperparameters common to PPO and SAC\n      # Same as PPO config\n\n      # SAC-specific hyperparameters\n      # Replaces the \"PPO-specific hyperparameters\" section above\n      buffer_init_steps: 0\n      tau: 0.005\n      steps_per_update: 10.0\n      save_replay_buffer: false\n      init_entcoef: 0.5\n      reward_signal_steps_per_update: 10.0\n\n    # Configuration of the neural network (common to PPO/SAC)\n    network_settings:\n      # Same as PPO config\n\n    # Trainer configurations common to all trainers\n      # &lt;Same as PPO config&gt;\n\n    # pre-training using behavior cloning\n    behavioral_cloning:\n      # same as PPO config\n\n    reward_signals:\n      # environment reward\n      extrinsic:\n        # same as PPO config\n\n      # curiosity module\n      curiosity:\n        # same as PPO config\n\n      # GAIL\n      gail:\n        # same as PPO config\n\n    # self-play\n    self_play:\n      # same as PPO config\n</code></pre> <p>We now break apart the components of the configuration file and describe what each of these parameters mean and provide guidelines on how to set them. See Training Configuration File for a detailed description of all the configurations listed above, along with their defaults. Unless otherwise specified, omitting a configuration will revert it to its default.</p>"},{"location":"Training-ML-Agents/#default-behavior-settings","title":"Default Behavior Settings","text":"<p>In some cases, you may want to specify a set of default configurations for your Behaviors. This may be useful, for instance, if your Behavior names are generated procedurally by the environment and not known before runtime, or if you have many Behaviors with very similar settings. To specify a default configuraton, insert a <code>default_settings</code> section in your YAML. This section should be formatted exactly like a configuration for a Behavior.</p> <pre><code>default_settings:\n  # &lt; Same as Behavior configuration &gt;\nbehaviors:\n  # &lt; Same as above &gt;\n</code></pre> <p>Behaviors found in the environment that aren't specified in the YAML will now use the <code>default_settings</code>, and unspecified settings in behavior configurations will default to the values in <code>default_settings</code> if specified there.</p>"},{"location":"Training-ML-Agents/#environment-parameters","title":"Environment Parameters","text":"<p>In order to control the <code>EnvironmentParameters</code> in the Unity simulation during training, you need to add a section called <code>environment_parameters</code>. For example you can set the value of an <code>EnvironmentParameter</code> called <code>my_environment_parameter</code> to <code>3.0</code> with the following code :</p> <pre><code>behaviors:\n  BehaviorY:\n    # &lt; Same as above &gt;\n\n# Add this section\nenvironment_parameters:\n  my_environment_parameter: 3.0\n</code></pre> <p>Inside the Unity simulation, you can access your Environment Parameters by doing :</p> <pre><code>Academy.Instance.EnvironmentParameters.GetWithDefault(\"my_environment_parameter\", 0.0f);\n</code></pre>"},{"location":"Training-ML-Agents/#environment-parameter-randomization","title":"Environment Parameter Randomization","text":"<p>To enable environment parameter randomization, you need to edit the <code>environment_parameters</code> section of your training configuration yaml file. Instead of providing a single float value for your environment parameter, you can specify a sampler instead. Here is an example with three environment parameters called <code>mass</code>, <code>length</code> and <code>scale</code>:</p> <pre><code>behaviors:\n  BehaviorY:\n    # &lt; Same as above &gt;\n\n# Add this section\nenvironment_parameters:\n  mass:\n    sampler_type: uniform\n    sampler_parameters:\n        min_value: 0.5\n        max_value: 10\n\n  length:\n    sampler_type: multirangeuniform\n    sampler_parameters:\n        intervals: [[7, 10], [15, 20]]\n\n  scale:\n    sampler_type: gaussian\n    sampler_parameters:\n        mean: 2\n        st_dev: .3\n</code></pre> Setting Description <code>sampler_type</code> A string identifier for the type of sampler to use for this <code>Environment Parameter</code>. <code>sampler_parameters</code> The parameters for a given <code>sampler_type</code>. Samplers of different types can have different <code>sampler_parameters</code>"},{"location":"Training-ML-Agents/#supported-sampler-types","title":"Supported Sampler Types","text":"<p>Below is a list of the <code>sampler_type</code> values supported by the toolkit.</p> <ul> <li><code>uniform</code> - Uniform sampler</li> <li>Uniformly samples a single float value from a range with a given minimum     and maximum value (inclusive).</li> <li>parameters - <code>min_value</code>, <code>max_value</code></li> <li><code>gaussian</code> - Gaussian sampler</li> <li>Samples a single float value from a normal distribution with a given mean     and standard deviation.</li> <li>parameters - <code>mean</code>, <code>st_dev</code></li> <li><code>multirange_uniform</code> - Multirange uniform sampler</li> <li>First, samples an interval from a set of intervals in proportion to relative     length of the intervals. Then, uniformly samples a single float value from the     sampled interval (inclusive). This sampler can take an arbitrary number of     intervals in a list in the following format:     [[<code>interval_1_min</code>, <code>interval_1_max</code>], [<code>interval_2_min</code>,     <code>interval_2_max</code>], ...]</li> <li>parameters - <code>intervals</code></li> </ul> <p>The implementation of the samplers can be found in the Samplers.cs file.</p>"},{"location":"Training-ML-Agents/#training-with-environment-parameter-randomization","title":"Training with Environment Parameter Randomization","text":"<p>After the sampler configuration is defined, we proceed by launching <code>mlagents-learn</code> and specify trainer configuration with  parameter randomization enabled. For example, if we wanted to train the 3D ball agent with parameter randomization, we would run</p> <pre><code>mlagents-learn config/ppo/3DBall_randomize.yaml --run-id=3D-Ball-randomize\n</code></pre> <p>We can observe progress and metrics via TensorBoard.</p>"},{"location":"Training-ML-Agents/#curriculum","title":"Curriculum","text":"<p>To enable curriculum learning, you need to add a <code>curriculum</code> sub-section to your environment parameter. Here is one example with the environment parameter <code>my_environment_parameter</code> :</p> <pre><code>behaviors:\n  BehaviorY:\n    # &lt; Same as above &gt;\n\n# Add this section\nenvironment_parameters:\n  my_environment_parameter:\n    curriculum:\n      - name: MyFirstLesson # The '-' is important as this is a list\n        completion_criteria:\n          measure: progress\n          behavior: my_behavior\n          signal_smoothing: true\n          min_lesson_length: 100\n          threshold: 0.2\n        value: 0.0\n      - name: MySecondLesson # This is the start of the second lesson\n        completion_criteria:\n          measure: progress\n          behavior: my_behavior\n          signal_smoothing: true\n          min_lesson_length: 100\n          threshold: 0.6\n          require_reset: true\n        value:\n          sampler_type: uniform\n          sampler_parameters:\n            min_value: 4.0\n            max_value: 7.0\n      - name: MyLastLesson\n        value: 8.0\n</code></pre> <p>Note that this curriculum only applies to <code>my_environment_parameter</code>. The <code>curriculum</code> section contains a list of <code>Lessons</code>. In the example, the lessons are named <code>MyFirstLesson</code>, <code>MySecondLesson</code> and <code>MyLastLesson</code>. Each <code>Lesson</code> has 3 fields :</p> <ul> <li><code>name</code> which is a user defined name for the lesson (The name of the lesson will be displayed in  the console when the lesson changes)</li> <li><code>completion_criteria</code> which determines what needs to happen in the simulation before the lesson  can be considered complete. When that condition is met, the curriculum moves on to the next  <code>Lesson</code>. Note that you do not need to specify a <code>completion_criteria</code> for the last <code>Lesson</code></li> <li><code>value</code> which is the value the environment parameter will take during the lesson. Note that this  can be a float or a sampler.</li> </ul> <p>There are the different settings of the <code>completion_criteria</code> :</p> Setting Description <code>measure</code> What to measure learning progress, and advancement in lessons by. <code>reward</code> uses a measure of received reward, <code>progress</code> uses the ratio of steps/max_steps, while <code>Elo</code> is available only for self-play situations and uses Elo score as a curriculum completion measure. <code>behavior</code> Specifies which behavior is being tracked. There can be multiple behaviors with different names, each at different points of training. This setting allows the curriculum to track only one of them. <code>threshold</code> Determines at what point in value of <code>measure</code> the lesson should be increased. <code>min_lesson_length</code> The minimum number of episodes that should be completed before the lesson can change. If <code>measure</code> is set to <code>reward</code>, the average cumulative reward of the last <code>min_lesson_length</code> episodes will be used to determine if the lesson should change. Must be nonnegative.  Important: the average reward that is compared to the thresholds is different than the mean reward that is logged to the console. For example, if <code>min_lesson_length</code> is <code>100</code>, the lesson will increment after the average cumulative reward of the last <code>100</code> episodes exceeds the current threshold. The mean reward logged to the console is dictated by the <code>summary_freq</code> parameter defined above. <code>signal_smoothing</code> Whether to weight the current progress measure by previous values. <code>require_reset</code> Whether changing lesson requires the environment to reset (default: false) ##### Training with a Curriculum <p>Once we have specified our metacurriculum and curricula, we can launch <code>mlagents-learn</code> to point to the config file containing our curricula and PPO will train using Curriculum Learning. For example, to train agents in the Wall Jump environment with curriculum learning, we can run:</p> <pre><code>mlagents-learn config/ppo/WallJump_curriculum.yaml --run-id=wall-jump-curriculum\n</code></pre> <p>We can then keep track of the current lessons and progresses via TensorBoard. If you've terminated the run, you can resume it using <code>--resume</code> and lesson progress will start off where it ended.</p>"},{"location":"Training-ML-Agents/#training-using-concurrent-unity-instances","title":"Training Using Concurrent Unity Instances","text":"<p>In order to run concurrent Unity instances during training, set the number of environment instances using the command line option <code>--num-envs=&lt;n&gt;</code> when you invoke <code>mlagents-learn</code>. Optionally, you can also set the <code>--base-port</code>, which is the starting port used for the concurrent Unity instances.</p> <p>Some considerations:</p> <ul> <li>Buffer Size - If you are having trouble getting an agent to train, even   with multiple concurrent Unity instances, you could increase <code>buffer_size</code> in   the trainer config file. A common practice is to multiply   <code>buffer_size</code> by <code>num-envs</code>.</li> <li>Resource Constraints - Invoking concurrent Unity instances is constrained   by the resources on the machine. Please use discretion when setting   <code>--num-envs=&lt;n&gt;</code>.</li> <li>Result Variation Using Concurrent Unity Instances - If you keep all the   hyperparameters the same, but change <code>--num-envs=&lt;n&gt;</code>, the results and model   would likely change.</li> </ul>"},{"location":"Training-Plugins/","title":"Customizing Training via Plugins","text":"<p>ML-Agents provides support for running your own python implementations of specific interfaces during the training process. These interfaces are currently fairly limited, but will be expanded in the future.</p> <p>Note: Plugin interfaces should currently be considered \"in beta\", and they may change in future releases.</p>"},{"location":"Training-Plugins/#how-to-write-your-own-plugin","title":"How to Write Your Own Plugin","text":"<p>This video explains the basics of how to create a plugin system using setuptools, and is the same approach that ML-Agents' plugin system is based on.</p> <p>The <code>ml-agents-plugin-examples</code> directory contains a reference implementation of each plugin interface, so it's a good starting point.</p>"},{"location":"Training-Plugins/#setuppy","title":"setup.py","text":"<p>If you don't already have a <code>setup.py</code> file for your python code, you'll need to add one. <code>ml-agents-plugin-examples</code> has a minimal example of this.</p> <p>In the call to <code>setup()</code>, you'll need to add to the <code>entry_points</code> dictionary for each plugin interface that you implement. The form of this is <code>{entry point name}={plugin module}:{plugin function}</code>. For example, in  <code>ml-agents-plugin-examples</code>:</p> <pre><code>entry_points={\n    ML_AGENTS_STATS_WRITER: [\n        \"example=mlagents_plugin_examples.example_stats_writer:get_example_stats_writer\"\n    ]\n}\n</code></pre> <ul> <li><code>ML_AGENTS_STATS_WRITER</code> (which is a string constant, <code>mlagents.stats_writer</code>) is the name of the plugin interface. This must be one of the provided interfaces (see below).</li> <li><code>example</code> is the plugin implementation name. This can be anything.</li> <li><code>mlagents_plugin_examples.example_stats_writer</code> is the plugin module. This points to the module where the plugin registration function is defined.</li> <li><code>get_example_stats_writer</code> is the plugin registration function. This is called when running <code>mlagents-learn</code>. The arguments and expected return type for this are different for each plugin interface.</li> </ul>"},{"location":"Training-Plugins/#local-installation","title":"Local Installation","text":"<p>Once you've defined <code>entry_points</code> in your <code>setup.py</code>, you will need to run</p> <pre><code>pip install -e [path to your plugin code]\n</code></pre> <p>in the same python virtual environment that you have <code>mlagents</code> installed.</p>"},{"location":"Training-Plugins/#plugin-interfaces","title":"Plugin Interfaces","text":""},{"location":"Training-Plugins/#statswriter","title":"StatsWriter","text":"<p>The StatsWriter class receives various information from the training process, such as the average Agent reward in each summary period. By default, we log this information to the console and write it to TensorBoard.</p>"},{"location":"Training-Plugins/#interface","title":"Interface","text":"<p>The <code>StatsWriter.write_stats()</code> method must be implemented in any derived classes. It takes a \"category\" parameter, which typically is the behavior name of the Agents being trained, and a dictionary of <code>StatSummary</code> values with string keys. Additionally, <code>StatsWriter.on_add_stat()</code> may be extended to register a callback handler for each stat emission.</p>"},{"location":"Training-Plugins/#registration","title":"Registration","text":"<p>The <code>StatsWriter</code> registration function takes a <code>RunOptions</code> argument and returns a list of <code>StatsWriter</code>s. An example implementation is provided in <code>mlagents_plugin_examples</code></p>"},{"location":"Training-on-Amazon-Web-Service/","title":"Training on Amazon Web Service","text":"<p>:warning: Note: We no longer use this guide ourselves and so it may not work correctly. We've decided to keep it up just in case it is helpful to you.</p> <p>This page contains instructions for setting up an EC2 instance on Amazon Web Service for training ML-Agents environments.</p>"},{"location":"Training-on-Amazon-Web-Service/#pre-configured-ami","title":"Pre-configured AMI","text":"<p>We've prepared a pre-configured AMI for you with the ID: <code>ami-016ff5559334f8619</code> in the <code>us-east-1</code> region. It was created as a modification of Deep Learning AMI (Ubuntu). The AMI has been tested with p2.xlarge instance. Furthermore, if you want to train without headless mode, you need to enable X Server.</p> <p>After launching your EC2 instance using the ami and ssh into it, run the following commands to enable it:</p> <pre><code># Start the X Server, press Enter to come to the command line\n$ sudo /usr/bin/X :0 &amp;\n\n# Check if Xorg process is running\n# You will have a list of processes running on the GPU, Xorg should be in the\n# list, as shown below\n$ nvidia-smi\n\n# Thu Jun 14 20:27:26 2018\n# +-----------------------------------------------------------------------------+\n# | NVIDIA-SMI 390.67                 Driver Version: 390.67                    |\n# |-------------------------------+----------------------+----------------------+\n# | GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n# | Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n# |===============================+======================+======================|\n# |   0  Tesla K80           On   | 00000000:00:1E.0 Off |                    0 |\n# | N/A   35C    P8    31W / 149W |      9MiB / 11441MiB |      0%      Default |\n# +-------------------------------+----------------------+----------------------+\n#\n# +-----------------------------------------------------------------------------+\n# | Processes:                                                       GPU Memory |\n# |  GPU       PID   Type   Process name                             Usage      |\n# |=============================================================================|\n# |    0      2331      G   /usr/lib/xorg/Xorg                             8MiB |\n# +-----------------------------------------------------------------------------+\n\n# Make the ubuntu use X Server for display\n$ export DISPLAY=:0\n</code></pre>"},{"location":"Training-on-Amazon-Web-Service/#configuring-your-own-instance","title":"Configuring your own instance","text":"<p>You could also choose to configure your own instance. To begin with, you will need an EC2 instance which contains the latest Nvidia drivers, CUDA9, and cuDNN. In this tutorial we used the Deep Learning AMI (Ubuntu) listed under AWS Marketplace with a p2.xlarge instance.</p>"},{"location":"Training-on-Amazon-Web-Service/#installing-the-ml-agents-toolkit-on-the-instance","title":"Installing the ML-Agents Toolkit on the instance","text":"<p>After launching your EC2 instance using the ami and ssh into it:</p> <ol> <li>Activate the python3 environment</li> </ol> <p><code>sh    source activate python3</code></p> <ol> <li>Clone the ML-Agents repo and install the required Python packages</li> </ol> <p><code>sh    git clone --branch release_21 https://github.com/Unity-Technologies/ml-agents.git    cd ml-agents/ml-agents/    pip3 install -e .</code></p>"},{"location":"Training-on-Amazon-Web-Service/#setting-up-x-server-optional","title":"Setting up X Server (optional)","text":"<p>X Server setup is only necessary if you want to do training that requires visual observation input. Instructions here are adapted from this Medium post on running general Unity applications in the cloud.</p> <p>Current limitations of the Unity Engine require that a screen be available to render to when using visual observations. In order to make this possible when training on a remote server, a virtual screen is required. We can do this by installing Xorg and creating a virtual screen. Once installed and created, we can display the Unity environment in the virtual environment, and train as we would on a local machine. Ensure that <code>headless</code> mode is disabled when building linux executables which use visual observations.</p>"},{"location":"Training-on-Amazon-Web-Service/#install-and-setup-xorg","title":"Install and setup Xorg:","text":"<pre><code>```sh\n# Install Xorg\n$ sudo apt-get update\n$ sudo apt-get install -y xserver-xorg mesa-utils\n$ sudo nvidia-xconfig -a --use-display-device=None --virtual=1280x1024\n\n# Get the BusID information\n$ nvidia-xconfig --query-gpu-info\n\n# Add the BusID information to your /etc/X11/xorg.conf file\n$ sudo sed -i 's/    BoardName      \"Tesla K80\"/    BoardName      \"Tesla K80\"\\n    BusID          \"0:30:0\"/g' /etc/X11/xorg.conf\n\n# Remove the Section \"Files\" from the /etc/X11/xorg.conf file\n# And remove two lines that contain Section \"Files\" and EndSection\n$ sudo vim /etc/X11/xorg.conf\n```\n</code></pre>"},{"location":"Training-on-Amazon-Web-Service/#update-and-setup-nvidia-driver","title":"Update and setup Nvidia driver:","text":"<pre><code>```sh\n# Download and install the latest Nvidia driver for ubuntu\n# Please refer to http://download.nvidia.com/XFree86/Linux-#x86_64/latest.txt\n$ wget http://download.nvidia.com/XFree86/Linux-x86_64/390.87/NVIDIA-Linux-x86_64-390.87.run\n$ sudo /bin/bash ./NVIDIA-Linux-x86_64-390.87.run --accept-license --no-questions --ui=none\n\n# Disable Nouveau as it will clash with the Nvidia driver\n$ sudo echo 'blacklist nouveau'  | sudo tee -a /etc/modprobe.d/blacklist.conf\n$ sudo echo 'options nouveau modeset=0'  | sudo tee -a /etc/modprobe.d/blacklist.conf\n$ sudo echo options nouveau modeset=0 | sudo tee -a /etc/modprobe.d/nouveau-kms.conf\n$ sudo update-initramfs -u\n```\n</code></pre>"},{"location":"Training-on-Amazon-Web-Service/#restart-the-ec2-instance","title":"Restart the EC2 instance:","text":"<pre><code>```sh\nsudo reboot now\n```\n</code></pre>"},{"location":"Training-on-Amazon-Web-Service/#make-sure-there-are-no-xorg-processes-running","title":"Make sure there are no Xorg processes running:","text":"<pre><code># Kill any possible running Xorg processes\n# Note that you might have to run this command multiple times depending on\n# how Xorg is configured.\n$ sudo killall Xorg\n\n# Check if there is any Xorg process left\n# You will have a list of processes running on the GPU, Xorg should not be in\n# the list, as shown below.\n$ nvidia-smi\n\n# Thu Jun 14 20:21:11 2018\n# +-----------------------------------------------------------------------------+\n# | NVIDIA-SMI 390.67                 Driver Version: 390.67                    |\n# |-------------------------------+----------------------+----------------------+\n# | GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n# | Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n# |===============================+======================+======================|\n# |   0  Tesla K80           On   | 00000000:00:1E.0 Off |                    0 |\n# | N/A   37C    P8    31W / 149W |      0MiB / 11441MiB |      0%      Default |\n# +-------------------------------+----------------------+----------------------+\n#\n# +-----------------------------------------------------------------------------+\n# | Processes:                                                       GPU Memory |\n# |  GPU       PID   Type   Process name                             Usage      |\n# |=============================================================================|\n# |  No running processes found                                                 |\n# +-----------------------------------------------------------------------------+\n\n</code></pre>"},{"location":"Training-on-Amazon-Web-Service/#start-x-server-and-make-the-ubuntu-use-x-server-for-display","title":"Start X Server and make the ubuntu use X Server for display:","text":"<pre><code>```console\n# Start the X Server, press Enter to come back to the command line\n$ sudo /usr/bin/X :0 &amp;\n\n# Check if Xorg process is running\n# You will have a list of processes running on the GPU, Xorg should be in the list.\n$ nvidia-smi\n\n# Make the ubuntu use X Server for display\n$ export DISPLAY=:0\n```\n</code></pre>"},{"location":"Training-on-Amazon-Web-Service/#ensure-the-xorg-is-correctly-configured","title":"Ensure the Xorg is correctly configured:","text":"<pre><code>```sh\n# For more information on glxgears, see ftp://www.x.org/pub/X11R6.8.1/doc/glxgears.1.html.\n$ glxgears\n# If Xorg is configured correctly, you should see the following message\n\n# Running synchronized to the vertical refresh.  The framerate should be\n# approximately the same as the monitor refresh rate.\n# 137296 frames in 5.0 seconds = 27459.053 FPS\n# 141674 frames in 5.0 seconds = 28334.779 FPS\n# 141490 frames in 5.0 seconds = 28297.875 FPS\n\n```\n</code></pre>"},{"location":"Training-on-Amazon-Web-Service/#training-on-ec2-instance","title":"Training on EC2 instance","text":"<ol> <li>In the Unity Editor, load a project containing an ML-Agents environment (you    can use one of the example environments if you have not created your own).</li> <li>Open the Build Settings window (menu: File &gt; Build Settings).</li> <li>Select Linux as the Target Platform, and x86_64 as the target architecture    (the default x86 currently does not work).</li> <li>Check Headless Mode if you have not setup the X Server. (If you do not use    Headless Mode, you have to setup the X Server to enable training.)</li> <li>Click Build to build the Unity environment executable.</li> <li>Upload the executable to your EC2 instance within <code>ml-agents</code> folder.</li> <li>Change the permissions of the executable.</li> </ol> <p><code>sh    chmod +x &lt;your_env&gt;.x86_64</code></p> <ol> <li>(Without Headless Mode) Start X Server and use it for display:</li> </ol> <p>```sh    # Start the X Server, press Enter to come back to the command line    $ sudo /usr/bin/X :0 &amp;</p> <p># Check if Xorg process is running    # You will have a list of processes running on the GPU, Xorg should be in the list.    $ nvidia-smi</p> <p># Make the ubuntu use X Server for display    $ export DISPLAY=:0    ```</p> <ol> <li>Test the instance setup from Python using:</li> </ol> <p>```python    from mlagents_envs.environment import UnityEnvironment</p> <p>env = UnityEnvironment()    ``` <p>Where <code>&lt;your_env&gt;</code> corresponds to the path to your environment executable.</p> <p>You should receive a message confirming that the environment was loaded    successfully.</p> <ol> <li> <p>Train your models</p> <p><code>console mlagents-learn &lt;trainer-config-file&gt; --env=&lt;your_env&gt; --train</code></p> </li> </ol>"},{"location":"Training-on-Amazon-Web-Service/#faq","title":"FAQ","text":""},{"location":"Training-on-Amazon-Web-Service/#the-_data-folder-hasnt-been-copied-cover","title":"The _Data folder hasn't been copied cover <p>If you've built your Linux executable, but forget to copy over the corresponding _Data folder, you will see error message like the following: <pre><code>Set current directory to /home/ubuntu/ml-agents/ml-agents\nFound path: /home/ubuntu/ml-agents/ml-agents/3dball_linux.x86_64\nno boot config - using default values\n\n(Filename:  Line: 403)\n\nThere is no data folder\n</code></pre>","text":""},{"location":"Training-on-Amazon-Web-Service/#unity-environment-not-responding","title":"Unity Environment not responding <p>If you didn't setup X Server or hasn't launched it properly, or your environment somehow crashes, or you haven't <code>chmod +x</code> your Unity Environment, all of these will cause connection between Unity and Python to fail. Then you will see something like this:</p> <pre><code>Logging to /home/ubuntu/.config/unity3d/&lt;Some_Path&gt;/Player.log\nTraceback (most recent call last):\n  File \"&lt;stdin&gt;\", line 1, in &lt;module&gt;\n  File \"/home/ubuntu/ml-agents/ml-agents/mlagents_envs/environment.py\", line 63, in __init__\n    aca_params = self.send_academy_parameters(rl_init_parameters_in)\n  File \"/home/ubuntu/ml-agents/ml-agents/mlagents_envs/environment.py\", line 489, in send_academy_parameters\n    return self.communicator.initialize(inputs).rl_initialization_output\n  File \"/home/ubuntu/ml-agents/ml-agents/mlagents_envs/rpc_communicator.py\", line 60, in initialize\nmlagents_envs.exception.UnityTimeOutException: The Unity environment took too long to respond. Make sure that :\n         The environment does not need user interaction to launch\n         The environment and the Python interface have compatible versions.\n</code></pre> <p>It would be also really helpful to check your /home/ubuntu/.config/unity3d//Player.log to see what happens with your Unity environment.","text":""},{"location":"Training-on-Amazon-Web-Service/#could-not-launch-x-server","title":"Could not launch X Server <p>When you execute:</p> <pre><code>sudo /usr/bin/X :0 &amp;\n</code></pre> <p>You might see something like:</p> <pre><code>X.Org X Server 1.18.4\n...\n(==) Log file: \"/var/log/Xorg.0.log\", Time: Thu Oct 11 21:10:38 2018\n(==) Using config file: \"/etc/X11/xorg.conf\"\n(==) Using system config directory \"/usr/share/X11/xorg.conf.d\"\n(EE)\nFatal server error:\n(EE) no screens found(EE)\n(EE)\nPlease consult the X.Org Foundation support\n         at http://wiki.x.org\n for help.\n(EE) Please also check the log file at \"/var/log/Xorg.0.log\" for additional information.\n(EE)\n(EE) Server terminated with error (1). Closing log file.\n</code></pre> <p>And when you execute:</p> <pre><code>nvidia-smi\n</code></pre> <p>You might see something like:</p> <pre><code>NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.\n</code></pre> <p>This means the NVIDIA's driver needs to be updated. Refer to this section for more information.</p>","text":""},{"location":"Training-on-Microsoft-Azure/","title":"Training on Microsoft Azure (works with ML-Agents Toolkit v0.3)","text":"<p>:warning: Note: We no longer use this guide ourselves and so it may not work correctly. We've decided to keep it up just in case it is helpful to you.</p> <p>This page contains instructions for setting up training on Microsoft Azure through either Azure Container Instances or Virtual Machines. Non \"headless\" training has not yet been tested to verify support.</p>"},{"location":"Training-on-Microsoft-Azure/#pre-configured-azure-virtual-machine","title":"Pre-Configured Azure Virtual Machine","text":"<p>A pre-configured virtual machine image is available in the Azure Marketplace and is nearly completely ready for training. You can start by deploying the Data Science Virtual Machine for Linux (Ubuntu) into your Azure subscription.</p> <p>Note that, if you choose to deploy the image to an N-Series GPU optimized VM, training will, by default, run on the GPU. If you choose any other type of VM, training will run on the CPU.</p>"},{"location":"Training-on-Microsoft-Azure/#configuring-your-own-instance","title":"Configuring your own Instance","text":"<p>Setting up your own instance requires a number of package installations. Please view the documentation for doing so here.</p>"},{"location":"Training-on-Microsoft-Azure/#installing-ml-agents","title":"Installing ML-Agents","text":"<ol> <li>Move    the <code>ml-agents</code> sub-folder of this ml-agents repo to the remote Azure    instance, and set it as the working directory.</li> <li>Install the required packages:    Torch: <code>pip3 install torch==1.7.0 -f https://download.pytorch.org/whl/torch_stable.html</code> and    MLAgents: <code>python -m pip install mlagents==1.0.0</code></li> </ol>"},{"location":"Training-on-Microsoft-Azure/#testing","title":"Testing","text":"<p>To verify that all steps worked correctly:</p> <ol> <li>In the Unity Editor, load a project containing an ML-Agents environment (you    can use one of the example environments if you have not created your own).</li> <li>Open the Build Settings window (menu: File &gt; Build Settings).</li> <li>Select Linux as the Target Platform, and x86_64 as the target architecture.</li> <li>Check Headless Mode.</li> <li>Click Build to build the Unity environment executable.</li> <li>Upload the resulting files to your Azure instance.</li> <li>Test the instance setup from Python using:</li> </ol> <pre><code>from mlagents_envs.environment import UnityEnvironment\n\nenv = UnityEnvironment(file_name=\"&lt;your_env&gt;\", seed=1, side_channels=[])\n</code></pre> <p>Where <code>&lt;your_env&gt;</code> corresponds to the path to your environment executable (i.e. <code>/home/UserName/Build/yourFile</code>).</p> <p>You should receive a message confirming that the environment was loaded successfully.</p> <p>Note: When running your environment in headless mode, you must append <code>--no-graphics</code> to your mlagents-learn command, as it won't train otherwise. You can test this simply by aborting a training and check if it says \"Model Saved\" or \"Aborted\", or see if it generated the .onnx in the result folder.</p>"},{"location":"Training-on-Microsoft-Azure/#running-training-on-your-virtual-machine","title":"Running Training on your Virtual Machine","text":"<p>To run your training on the VM:</p> <ol> <li>Move    your built Unity application to your Virtual Machine.</li> <li>Set the directory where the ML-Agents Toolkit was installed to your working    directory.</li> <li>Run the following command:</li> </ol> <pre><code>mlagents-learn &lt;trainer_config&gt; --env=&lt;your_app&gt; --run-id=&lt;run_id&gt; --train\n</code></pre> <p>Where <code>&lt;your_app&gt;</code> is the path to your app (i.e. <code>~/unity-volume/3DBallHeadless</code>) and <code>&lt;run_id&gt;</code> is an identifier you would like to identify your training run with.</p> <p>If you've selected to run on a N-Series VM with GPU support, you can verify that the GPU is being used by running <code>nvidia-smi</code> from the command line.</p>"},{"location":"Training-on-Microsoft-Azure/#monitoring-your-training-run-with-tensorboard","title":"Monitoring your Training Run with TensorBoard","text":"<p>Once you have started training, you can use TensorBoard to observe the training.</p> <ol> <li> <p>Start by    opening the appropriate port for web traffic to connect to your VM.</p> </li> <li> <p>Note that you don't need to generate a new <code>Network Security Group</code> but      instead, go to the Networking tab under Settings for your VM.</p> </li> <li> <p>As an example, you could use the following settings to open the Port with      the following Inbound Rule settings:</p> <ul> <li>Source: Any</li> <li>Source Port Ranges: *</li> <li>Destination: Any</li> <li>Destination Port Ranges: 6006</li> <li>Protocol: Any</li> <li>Action: Allow</li> <li>Priority: (Leave as default)</li> </ul> </li> <li> <p>Unless you started the training as a background process, connect to your VM    from another terminal instance.</p> </li> <li>Run the following command from your terminal    <code>tensorboard --logdir results --host 0.0.0.0</code></li> <li>You should now be able to open a browser and navigate to    <code>&lt;Your_VM_IP_Address&gt;:6060</code> to view the TensorBoard report.</li> </ol>"},{"location":"Training-on-Microsoft-Azure/#running-on-azure-container-instances","title":"Running on Azure Container Instances","text":"<p>Azure Container Instances allow you to spin up a container, on demand, that will run your training and then be shut down. This ensures you aren't leaving a billable VM running when it isn't needed. Using ACI enables you to offload training of your models without needing to install Python and TensorFlow on your own computer.</p>"},{"location":"Training-on-Microsoft-Azure/#custom-instances","title":"Custom Instances","text":"<p>This page contains instructions for setting up a custom Virtual Machine on Microsoft Azure so you can running ML-Agents training in the cloud.</p> <ol> <li>Start by    deploying an Azure VM    with Ubuntu Linux (tests were done with 16.04 LTS). To use GPU support, use a    N-Series VM.</li> <li>SSH into your VM.</li> <li>Start with the following commands to install the Nvidia driver:</li> </ol> <p>```sh    wget http://us.download.nvidia.com/tesla/375.66/nvidia-diag-driver-local-repo-ubuntu1604_375.66-1_amd64.deb</p> <p>sudo dpkg -i nvidia-diag-driver-local-repo-ubuntu1604_375.66-1_amd64.deb</p> <p>sudo apt-get update</p> <p>sudo apt-get install cuda-drivers</p> <p>sudo reboot    ```</p> <ol> <li>After a minute you should be able to reconnect to your VM and install the    CUDA toolkit:</li> </ol> <p>```sh    wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/cuda-repo-ubuntu1604_8.0.61-1_amd64.deb</p> <p>sudo dpkg -i cuda-repo-ubuntu1604_8.0.61-1_amd64.deb</p> <p>sudo apt-get update</p> <p>sudo apt-get install cuda-8-0    ```</p> <ol> <li> <p>You'll next need to download cuDNN from the Nvidia developer site. This    requires a registered account.</p> </li> <li> <p>Navigate to http://developer.nvidia.com and    create an account and verify it.</p> </li> <li> <p>Download (to your own computer) cuDNN from    this url.</p> </li> <li> <p>Copy the deb package to your VM:</p> </li> </ol> <p><code>sh    scp libcudnn6_6.0.21-1+cuda8.0_amd64.deb &lt;VMUserName&gt;@&lt;VMIPAddress&gt;:libcudnn6_6.0.21-1+cuda8.0_amd64.deb</code></p> <ol> <li>SSH back to your VM and execute the following:</li> </ol> <p>```console    sudo dpkg -i libcudnn6_6.0.21-1+cuda8.0_amd64.deb</p> <p>export LD_LIBRARY_PATH=/usr/local/cuda/lib64/:/usr/lib/x86_64-linux-gnu/:$LD_LIBRARY_PATH    . ~/.profile</p> <p>sudo reboot    ```</p> <ol> <li> <p>After a minute, you should be able to SSH back into your VM. After doing so,     run the following:</p> <p><code>sh sudo apt install python-pip sudo apt install python3-pip</code></p> </li> <li> <p>At this point, you need to install TensorFlow. The version you install     should be tied to if you are using GPU to train:</p> <p><code>sh pip3 install tensorflow-gpu==1.4.0 keras==2.0.6</code></p> <p>Or CPU to train:</p> <p><code>sh pip3 install tensorflow==1.4.0 keras==2.0.6</code></p> </li> <li> <p>You'll then need to install additional dependencies:</p> <p><code>sh pip3 install pillow pip3 install numpy</code></p> </li> </ol>"},{"location":"Tutorial-Custom-Trainer-Plugin/","title":"Custom Trainer Plugin","text":""},{"location":"Tutorial-Custom-Trainer-Plugin/#how-to-write-a-custom-trainer-plugin","title":"How to write a custom trainer plugin","text":""},{"location":"Tutorial-Custom-Trainer-Plugin/#step-1-write-your-custom-trainer-class","title":"Step 1: Write your custom trainer class","text":"<p>Before you start writing your code, make sure to use your favorite environment management tool(e.g. <code>venv</code> or <code>conda</code>) to create and activate a Python virtual environment. The following command uses <code>conda</code>, but other tools work similarly:</p> <pre><code>conda create -n trainer-env python=3.10.12\nconda activate trainer-env\n</code></pre> <p>Users of the plug-in system are responsible for implementing the trainer class subject to the API standard. Let us follow an example by implementing a custom trainer named \"YourCustomTrainer\". You can either extend <code>OnPolicyTrainer</code> or <code>OffPolicyTrainer</code> classes depending on the training strategies you choose.</p> <p>Please refer to the internal PPO implementation for a complete code example. We will not provide a workable code in the document. The purpose of the tutorial is to introduce you to the core components and interfaces of our plugin framework. We use code snippets and patterns to demonstrate the control and data flow.</p> <p>Your custom trainers are responsible for collecting experiences and training the models. Your custom trainer class acts like a co-ordinator to the policy and optimizer. To start implementing methods in the class, create a policy class objects from method <code>create_policy</code>:</p> <pre><code>def create_policy(\n    self, parsed_behavior_id: BehaviorIdentifiers, behavior_spec: BehaviorSpec\n) -&gt; TorchPolicy:\n\n    actor_cls: Union[Type[SimpleActor], Type[SharedActorCritic]] = SimpleActor\n    actor_kwargs: Dict[str, Any] = {\n        \"conditional_sigma\": False,\n        \"tanh_squash\": False,\n    }\n    if self.shared_critic:\n        reward_signal_configs = self.trainer_settings.reward_signals\n        reward_signal_names = [\n            key.value for key, _ in reward_signal_configs.items()\n        ]\n        actor_cls = SharedActorCritic\n        actor_kwargs.update({\"stream_names\": reward_signal_names})\n\n    policy = TorchPolicy(\n        self.seed,\n        behavior_spec,\n        self.trainer_settings.network_settings,\n        actor_cls,\n        actor_kwargs,\n    )\n    return policy\n\n</code></pre> <p>Depending on whether you use shared or separate network architecture for your policy, we provide <code>SimpleActor</code> and <code>SharedActorCritic</code> from <code>mlagents.trainers.torch_entities.networks</code> that you can choose from. In our example above, we use a <code>SimpleActor</code>.</p> <p>Next, create an optimizer class object from <code>create_optimizer</code> method and connect it to the policy object you created above:</p> <pre><code>def create_optimizer(self) -&gt; TorchOptimizer:\n    return TorchPPOOptimizer(  # type: ignore\n        cast(TorchPolicy, self.policy), self.trainer_settings  # type: ignore\n    )  # type: ignore\n\n</code></pre> <p>There are a couple of abstract methods(<code>_process_trajectory</code> and <code>_update_policy</code>) inherited from <code>RLTrainer</code> that you need to implement in your custom trainer class. <code>_process_trajectory</code> takes a trajectory and processes it, putting it into the update buffer. Processing involves calculating value and advantage targets for the model updating step. Given input <code>trajectory: Trajectory</code>, users are responsible for processing the data in the trajectory and append <code>agent_buffer_trajectory</code> to the back of the update buffer by calling <code>self._append_to_update_buffer(agent_buffer_trajectory)</code>, whose output will be used in updating the model in <code>optimizer</code> class.</p> <p>A typical <code>_process_trajectory</code> function(incomplete) will convert a trajectory object to an agent buffer then get all value estimates from the trajectory by calling <code>self.optimizer.get_trajectory_value_estimates</code>. From the returned dictionary of value estimates we extract reward signals keyed by their names:</p> <pre><code>def _process_trajectory(self, trajectory: Trajectory) -&gt; None:\n    super()._process_trajectory(trajectory)\n    agent_id = trajectory.agent_id  # All the agents should have the same ID\n\n    agent_buffer_trajectory = trajectory.to_agentbuffer()\n\n    # Get all value estimates\n    (\n        value_estimates,\n        value_next,\n        value_memories,\n    ) =  self.optimizer.get_trajectory_value_estimates(\n        agent_buffer_trajectory,\n        trajectory.next_obs,\n        trajectory.done_reached and not trajectory.interrupted,\n    )\n\n    for name, v in value_estimates.items():\n        agent_buffer_trajectory[RewardSignalUtil.value_estimates_key(name)].extend(\n            v\n        )\n        self._stats_reporter.add_stat(\n            f\"Policy/{self.optimizer.reward_signals[name].name.capitalize()} Value Estimate\",\n            np.mean(v),\n        )\n\n    # Evaluate all reward functions\n    self.collected_rewards[\"environment\"][agent_id] += np.sum(\n        agent_buffer_trajectory[BufferKey.ENVIRONMENT_REWARDS]\n    )\n    for name, reward_signal in self.optimizer.reward_signals.items():\n        evaluate_result = (\n            reward_signal.evaluate(agent_buffer_trajectory) * reward_signal.strength\n        )\n        agent_buffer_trajectory[RewardSignalUtil.rewards_key(name)].extend(\n            evaluate_result\n        )\n        # Report the reward signals\n        self.collected_rewards[name][agent_id] += np.sum(evaluate_result)\n\n    self._append_to_update_buffer(agent_buffer_trajectory)\n\n</code></pre> <p>A trajectory will be a list of dictionaries of strings mapped to <code>Anything</code>. When calling <code>forward</code> on a policy, the argument will include an \u201cexperience\u201d dictionary from the last step. The <code>forward</code> method will generate an action and the next \u201cexperience\u201d dictionary. Examples of fields in the \u201cexperience\u201d dictionary include observation, action, reward, done status, group_reward, LSTM memory state, etc.</p>"},{"location":"Tutorial-Custom-Trainer-Plugin/#step-2-implement-your-custom-optimizer-for-the-trainer","title":"Step 2: implement your custom optimizer for the trainer.","text":"<p>We will show you an example we implemented - <code>class TorchPPOOptimizer(TorchOptimizer)</code>, which takes a Policy and a Dict of trainer parameters and creates an Optimizer that connects to the policy. Your optimizer should include a value estimator and a loss function in the <code>update</code> method.</p> <p>Before writing your optimizer class, first define setting class <code>class PPOSettings(OnPolicyHyperparamSettings)</code> for your custom optimizer:</p> <pre><code>class PPOSettings(OnPolicyHyperparamSettings):\n    beta: float = 5.0e-3\n    epsilon: float = 0.2\n    lambd: float = 0.95\n    num_epoch: int = 3\n    shared_critic: bool = False\n    learning_rate_schedule: ScheduleType = ScheduleType.LINEAR\n    beta_schedule: ScheduleType = ScheduleType.LINEAR\n    epsilon_schedule: ScheduleType = ScheduleType.LINEAR\n\n</code></pre> <p>You should implement <code>update</code> function following interface:</p> <pre><code>def update(self, batch: AgentBuffer, num_sequences: int) -&gt; Dict[str, float]:\n\n</code></pre> <p>In which losses and other metrics are calculated from an <code>AgentBuffer</code> that is generated from your trainer class, depending on which model you choose to implement the loss functions will be different. In our case we calculate value loss from critic and trust region policy loss. A typical pattern(incomplete) of the calculations will look like the following:</p> <pre><code>run_out = self.policy.actor.get_stats(\n    current_obs,\n    actions,\n    masks=act_masks,\n    memories=memories,\n    sequence_length=self.policy.sequence_length,\n)\n\nlog_probs = run_out[\"log_probs\"]\nentropy = run_out[\"entropy\"]\n\nvalues, _ = self.critic.critic_pass(\n    current_obs,\n    memories=value_memories,\n    sequence_length=self.policy.sequence_length,\n)\npolicy_loss = ModelUtils.trust_region_policy_loss(\n    ModelUtils.list_to_tensor(batch[BufferKey.ADVANTAGES]),\n    log_probs,\n    old_log_probs,\n    loss_masks,\n    decay_eps,\n)\nloss = (\n    policy_loss\n    + 0.5 * value_loss\n    - decay_bet * ModelUtils.masked_mean(entropy, loss_masks)\n)\n\n</code></pre> <p>Finally update the model and return the a dictionary including calculated losses and updated decay learning rate:</p> <pre><code>ModelUtils.update_learning_rate(self.optimizer, decay_lr)\nself.optimizer.zero_grad()\nloss.backward()\n\nself.optimizer.step()\nupdate_stats = {\n    \"Losses/Policy Loss\": torch.abs(policy_loss).item(),\n    \"Losses/Value Loss\": value_loss.item(),\n    \"Policy/Learning Rate\": decay_lr,\n    \"Policy/Epsilon\": decay_eps,\n    \"Policy/Beta\": decay_bet,\n}\n\n</code></pre>"},{"location":"Tutorial-Custom-Trainer-Plugin/#step-3-integrate-your-custom-trainer-into-the-plugin-system","title":"Step 3: Integrate your custom trainer into the plugin system","text":"<p>By integrating a custom trainer into the plugin system, a user can use their published packages which have their implementations. To do that, you need to add a setup.py file. In the call to setup(), you'll need to add to the entry_points dictionary for each plugin interface that you implement. The form of this is {entry point name}={plugin module}:{plugin function}. For example:</p> <pre><code>entry_points={\n        ML_AGENTS_TRAINER_TYPE: [\n            \"your_trainer_type=your_package.your_custom_trainer:get_type_and_setting\"\n        ]\n    },\n</code></pre> <p>Some key elements in the code:</p> <pre><code>ML_AGENTS_TRAINER_TYPE: a string constant for trainer type\nyour_trainer_type: name your trainer type, used in configuration file\nyour_package: your pip installable package containing custom trainer implementation\n</code></pre> <p>Also define <code>get_type_and_setting</code> method in <code>YourCustomTrainer</code> class:</p> <pre><code>def get_type_and_setting():\n    return {YourCustomTrainer.get_trainer_name(): YourCustomTrainer}, {\n        YourCustomTrainer.get_trainer_name(): YourCustomSetting\n    }\n\n</code></pre> <p>Finally, specify trainer type in the config file:</p> <pre><code>behaviors:\n  3DBall:\n    trainer_type: your_trainer_type\n...\n</code></pre>"},{"location":"Tutorial-Custom-Trainer-Plugin/#step-4-install-your-custom-trainer-and-run-training","title":"Step 4: Install your custom trainer and run training:","text":"<p>Before installing your custom trainer package, make sure you have <code>ml-agents-env</code> and <code>ml-agents</code> installed</p> <pre><code>pip3 install -e ./ml-agents-envs &amp;&amp; pip3 install -e ./ml-agents\n</code></pre> <p>Install your cutom trainer package(if your package is pip installable):</p> <pre><code>pip3 install your_custom_package\n</code></pre> <p>Or follow our internal implementations:</p> <pre><code>pip3 install -e ./ml-agents-trainer-plugin\n</code></pre> <p>Following the previous installations your package is added as an entrypoint and you can use a config file with new trainers:</p> <pre><code>mlagents-learn ml-agents-trainer-plugin/mlagents_trainer_plugin/a2c/a2c_3DBall.yaml --run-id &lt;run-id-name&gt;\n--env &lt;env-executable&gt;\n</code></pre>"},{"location":"Tutorial-Custom-Trainer-Plugin/#validate-your-implementations","title":"Validate your implementations:","text":"<p>Create a clean Python environment with Python 3.10.12 and activate it before you start, if you haven't done so already:</p> <pre><code>conda create -n trainer-env python=3.10.12\nconda activate trainer-env\n</code></pre> <p>Make sure you follow previous steps and install all required packages. We are testing internal implementations in this tutorial, but ML-Agents users can run similar validations once they have their own implementations installed:</p> <pre><code>pip3 install -e ./ml-agents-envs &amp;&amp; pip3 install -e ./ml-agents\npip3 install -e ./ml-agents-trainer-plugin\n</code></pre> <p>Once your package is added as an <code>entrypoint</code>, you can add to the config file the new trainer type. Check if trainer type is specified in the config file <code>a2c_3DBall.yaml</code>:</p> <pre><code>trainer_type: a2c\n</code></pre> <p>Test if custom trainer package is installed by running:</p> <pre><code>mlagents-learn ml-agents-trainer-plugin/mlagents_trainer_plugin/a2c/a2c_3DBall.yaml --run-id test-trainer\n</code></pre> <p>You can also list all trainers installed in the registry. Type <code>python</code> in your shell to open a REPL session. Run the python code below, you should be able to see all trainer types currently installed:</p> <pre><code>&gt;&gt;&gt; import pkg_resources\n&gt;&gt;&gt; for entry in pkg_resources.iter_entry_points('mlagents.trainer_type'):\n...     print(entry)\n...\ndefault = mlagents.plugins.trainer_type:get_default_trainer_types\na2c = mlagents_trainer_plugin.a2c.a2c_trainer:get_type_and_setting\ndqn = mlagents_trainer_plugin.dqn.dqn_trainer:get_type_and_setting\n</code></pre> <p>If it is properly installed, you will see Unity logo and message indicating training will start:</p> <pre><code>[INFO] Listening on port 5004. Start training by pressing the Play button in the Unity Editor.\n</code></pre> <p>If you see the following error message, it could be due to trainer type is wrong or the trainer type specified is not installed:</p> <pre><code>mlagents.trainers.exception.TrainerConfigError: Invalid trainer type a2c was found\n</code></pre>"},{"location":"Unity-Environment-Registry/","title":"Unity Environment Registry [Experimental]","text":"<p>The Unity Environment Registry is a database of pre-built Unity environments that can be easily used without having to install the Unity Editor. It is a great way to get started with our UnityEnvironment API.</p>"},{"location":"Unity-Environment-Registry/#loading-an-environment-from-the-registry","title":"Loading an Environment from the Registry","text":"<p>To get started, you can access the default registry we provide with our Example Environments. The Unity Environment Registry implements a Mapping, therefore, you can access an entry with its identifier with the square brackets <code>[ ]</code>. Use the following code to list all of the environment identifiers present in the default registry:</p> <pre><code>from mlagents_envs.registry import default_registry\n\nenvironment_names = list(default_registry.keys())\nfor name in environment_names:\n   print(name)\n</code></pre> <p>The <code>make()</code> method on a registry value will return a <code>UnityEnvironment</code> ready to be used. All arguments passed to the make method will be passed to the constructor of the <code>UnityEnvironment</code> as well. Refer to the documentation on the Python-API for more information about the arguments of the <code>UnityEnvironment</code> constructor. For example, the following code will create the environment under the identifier <code>\"my-env\"</code>, reset it, perform a few steps and finally close it:</p> <pre><code>from mlagents_envs.registry import default_registry\n\nenv = default_registry[\"my-env\"].make()\nenv.reset()\nfor _ in range(10):\n  env.step()\nenv.close()\n</code></pre>"},{"location":"Unity-Environment-Registry/#create-and-share-your-own-registry","title":"Create and share your own registry","text":"<p>In order to share the <code>UnityEnvironemnt</code> you created, you must :  - Create a Unity executable of your environment for each platform (Linux, OSX and/or Windows)  - Place each executable in a <code>zip</code> compressed folder  - Upload each zip file online to your preferred hosting platform  - Create a <code>yaml</code> file that will contain the description and path to your environment  - Upload the <code>yaml</code> file online The <code>yaml</code> file must have the following format :</p> <pre><code>environments:\n  - &lt;environment-identifier&gt;:\n     expected_reward: &lt;expected-reward-float&gt;\n     description: &lt;description-of-the-environment&gt;\n     linux_url: &lt;url-to-the-linux-zip-folder&gt;\n     darwin_url: &lt;url-to-the-osx-zip-folder&gt;\n     win_url: &lt;url-to-the-windows-zip-folder&gt;\n     additional_args:\n      - &lt;an-optional-list-of-command-line-arguments-for-the-executable&gt;\n      - ...\n</code></pre> <p>Your users can now use your environment with the following code :</p> <pre><code>from mlagents_envs.registry import UnityEnvRegistry\n\nregistry = UnityEnvRegistry()\nregistry.register_from_yaml(\"url-or-path-to-your-yaml-file\")\n</code></pre> <p>Note: The <code>\"url-or-path-to-your-yaml-file\"</code> can be either a url or a local path.</p>"},{"location":"Using-Docker/","title":"Using Docker For ML-Agents (Deprecated)","text":"<p>:warning: Note: We no longer use this guide ourselves and so it may not work correctly. We've decided to keep it up just in case it is helpful to you.</p> <p>We currently offer a solution for Windows and Mac users who would like to do training or inference using Docker. This option may be appealing to those who would like to avoid installing Python and TensorFlow themselves. The current setup forces both TensorFlow and Unity to only rely on the CPU for computations. Consequently, our Docker simulation does not use a GPU and uses <code>Xvfb</code> to do visual rendering. <code>Xvfb</code> is a utility that enables <code>ML-Agents</code> (or any other application) to do rendering virtually i.e. it does not assume that the machine running <code>ML-Agents</code> has a GPU or a display attached to it. This means that rich environments which involve agents using camera-based visual observations might be slower.</p>"},{"location":"Using-Docker/#requirements","title":"Requirements","text":"<ul> <li>Docker</li> <li>Unity Linux Build Support Component. Make sure to select the Linux Build   Support component when installing Unity.</li> </ul>"},{"location":"Using-Docker/#setup","title":"Setup","text":"<ul> <li> <p>Download the Unity Installer and add   the Linux Build Support Component</p> </li> <li> <p>Download and install   Docker if you don't have it setup on your machine.</p> </li> <li> <p>Since Docker runs a container in an environment that is isolated from the host   machine, a mounted directory in your host machine is used to share data, e.g.   the trainer configuration file, Unity executable and   TensorFlow graph. For convenience, we created an empty <code>unity-volume</code>   directory at the root of the repository for this purpose, but feel free to use   any other directory. The remainder of this guide assumes that the   <code>unity-volume</code> directory is the one used.</p> </li> </ul>"},{"location":"Using-Docker/#usage","title":"Usage","text":"<p>Using Docker for ML-Agents involves three steps: building the Unity environment with specific flags, building a Docker container and, finally, running the container. If you are not familiar with building a Unity environment for ML-Agents, please read through our Getting Started with the 3D Balance Ball Example guide first.</p>"},{"location":"Using-Docker/#build-the-environment-optional","title":"Build the Environment (Optional)","text":"<p>If you want to used the Editor to perform training, you can skip this step.</p> <p>Since Docker typically runs a container sharing a (linux) kernel with the host machine, the Unity environment has to be built for the linux platform. When building a Unity environment, please select the following options from the the Build Settings window:</p> <ul> <li>Set the Target Platform to <code>Linux</code></li> <li>Set the Architecture to <code>x86_64</code></li> </ul> <p>Then click <code>Build</code>, pick an environment name (e.g. <code>3DBall</code>) and set the output directory to <code>unity-volume</code>. After building, ensure that the file <code>&lt;environment-name&gt;.x86_64</code> and subdirectory <code>&lt;environment-name&gt;_Data/</code> are created under <code>unity-volume</code>.</p> <p></p>"},{"location":"Using-Docker/#build-the-docker-container","title":"Build the Docker Container","text":"<p>First, make sure the Docker engine is running on your machine. Then build the Docker container by calling the following command at the top-level of the repository:</p> <pre><code>docker build -t &lt;image-name&gt; .\n</code></pre> <p>Replace <code>&lt;image-name&gt;</code> with a name for the Docker image, e.g. <code>balance.ball.v0.1</code>.</p>"},{"location":"Using-Docker/#run-the-docker-container","title":"Run the Docker Container","text":"<p>Run the Docker container by calling the following command at the top-level of the repository:</p> <pre><code>docker run -it --name &lt;container-name&gt; \\\n           --mount type=bind,source=\"$(pwd)\"/unity-volume,target=/unity-volume \\\n           -p 5005:5005 \\\n           -p 6006:6006 \\\n           &lt;image-name&gt;:latest \\\n           &lt;trainer-config-file&gt; \\\n           --env=&lt;environment-name&gt; \\\n           --train \\\n           --run-id=&lt;run-id&gt;\n</code></pre> <p>Notes on argument values:</p> <ul> <li><code>&lt;container-name&gt;</code> is used to identify the container (in case you want to   interrupt and terminate it). This is optional and Docker will generate a   random name if this is not set. Note that this must be unique for every run   of a Docker image.</li> <li><code>&lt;image-name&gt;</code> references the image name used when building the container.</li> <li><code>&lt;environment-name&gt;</code> (Optional): If you are training with a linux   executable, this is the name of the executable. If you are training in the   Editor, do not pass a <code>&lt;environment-name&gt;</code> argument and press the Play   button in Unity when the message \"Start training by pressing the Play button   in the Unity Editor\" is displayed on the screen.</li> <li><code>source</code>: Reference to the path in your host OS where you will store the Unity   executable.</li> <li><code>target</code>: Tells Docker to mount the <code>source</code> path as a disk with this name.</li> <li><code>trainer-config-file</code>, <code>train</code>, <code>run-id</code>: ML-Agents arguments passed to   <code>mlagents-learn</code>. <code>trainer-config-file</code> is the filename of the trainer config   file, <code>train</code> trains the algorithm, and <code>run-id</code> is used to tag each   experiment with a unique identifier. We recommend placing the trainer-config   file inside <code>unity-volume</code> so that the container has access to the file.</li> </ul> <p>To train with a <code>3DBall</code> environment executable, the command would be:</p> <pre><code>docker run -it --name 3DBallContainer.first.trial \\\n           --mount type=bind,source=\"$(pwd)\"/unity-volume,target=/unity-volume \\\n           -p 5005:5005 \\\n           -p 6006:6006 \\\n           balance.ball.v0.1:latest 3DBall \\\n           /unity-volume/trainer_config.yaml \\\n           --env=/unity-volume/3DBall \\\n           --train \\\n           --run-id=3dball_first_trial\n</code></pre> <p>For more detail on Docker mounts, check out these docs from Docker.</p> <p>NOTE If you are training using docker for environments that use visual observations, you may need to increase the default memory that Docker allocates for the container. For example, see here for instructions for Docker for Mac.</p>"},{"location":"Using-Docker/#running-tensorboard","title":"Running Tensorboard","text":"<p>You can run Tensorboard to monitor your training instance on http://localhost:6006:</p> <pre><code>docker exec -it &lt;container-name&gt; tensorboard --logdir /unity-volume/results --host 0.0.0.0\n</code></pre> <p>With our previous 3DBall example, this command would look like this:</p> <pre><code>docker exec -it 3DBallContainer.first.trial tensorboard --logdir /unity-volume/results --host 0.0.0.0\n</code></pre> <p>For more details on Tensorboard, check out the documentation about Using Tensorboard.</p>"},{"location":"Using-Docker/#stopping-container-and-saving-state","title":"Stopping Container and Saving State","text":"<p>If you are satisfied with the training progress, you can stop the Docker container while saving state by either using <code>Ctrl+C</code> or <code>\u2318+C</code> (Mac) or by using the following command:</p> <pre><code>docker kill --signal=SIGINT &lt;container-name&gt;\n</code></pre> <p><code>&lt;container-name&gt;</code> is the name of the container specified in the earlier <code>docker run</code> command. If you didn't specify one, you can find the randomly generated identifier by running <code>docker container ls</code>.</p>"},{"location":"Using-Tensorboard/","title":"Using TensorBoard to Observe Training","text":"<p>The ML-Agents Toolkit saves statistics during learning session that you can view with a TensorFlow utility named, TensorBoard.</p> <p>The <code>mlagents-learn</code> command saves training statistics to a folder named <code>results</code>, organized by the <code>run-id</code> value you assign to a training session.</p> <p>In order to observe the training process, either during training or afterward, start TensorBoard:</p> <ol> <li>Open a terminal or console window:</li> <li>Navigate to the directory where the ML-Agents Toolkit is installed.</li> <li>From the command line run: <code>tensorboard --logdir results --port 6006</code></li> <li>Open a browser window and navigate to    localhost:6006.</li> </ol> <p>Note: The default port TensorBoard uses is 6006. If there is an existing session running on port 6006 a new session can be launched on an open port using the --port option.</p> <p>Note: If you don't assign a <code>run-id</code> identifier, <code>mlagents-learn</code> uses the default string, \"ppo\". You can delete the folders under the <code>results</code> directory to clear out old statistics.</p> <p>On the left side of the TensorBoard window, you can select which of the training runs you want to display. You can select multiple run-ids to compare statistics. The TensorBoard window also provides options for how to display and smooth graphs.</p>"},{"location":"Using-Tensorboard/#the-ml-agents-toolkit-training-statistics","title":"The ML-Agents Toolkit training statistics","text":"<p>The ML-Agents training program saves the following statistics:</p> <p></p>"},{"location":"Using-Tensorboard/#environment-statistics","title":"Environment Statistics","text":"<ul> <li> <p><code>Environment/Lesson</code> - Plots the progress from lesson to lesson. Only   interesting when performing curriculum training.</p> </li> <li> <p><code>Environment/Cumulative Reward</code> - The mean cumulative episode reward over all   agents. Should increase during a successful training session.</p> </li> <li> <p><code>Environment/Episode Length</code> - The mean length of each episode in the   environment for all agents.</p> </li> </ul>"},{"location":"Using-Tensorboard/#is-training","title":"Is Training","text":"<ul> <li><code>Is Training</code> - A boolean indicating if the agent is updating its model.</li> </ul>"},{"location":"Using-Tensorboard/#policy-statistics","title":"Policy Statistics","text":"<ul> <li> <p><code>Policy/Entropy</code> (PPO; SAC) - How random the decisions of the model are.   Should slowly decrease during a successful training process. If it decreases   too quickly, the <code>beta</code> hyperparameter should be increased.</p> </li> <li> <p><code>Policy/Learning Rate</code> (PPO; SAC) - How large a step the training algorithm   takes as it searches for the optimal policy. Should decrease over time.</p> </li> <li> <p><code>Policy/Entropy Coefficient</code> (SAC) - Determines the relative importance of the   entropy term. This value is adjusted automatically so that the agent retains   some amount of randomness during training.</p> </li> <li> <p><code>Policy/Extrinsic Reward</code> (PPO; SAC) - This corresponds to the mean cumulative   reward received from the environment per-episode.</p> </li> <li> <p><code>Policy/Value Estimate</code> (PPO; SAC) - The mean value estimate for all states   visited by the agent. Should increase during a successful training session.</p> </li> <li> <p><code>Policy/Curiosity Reward</code> (PPO/SAC+Curiosity) - This corresponds to the mean   cumulative intrinsic reward generated per-episode.</p> </li> <li> <p><code>Policy/Curiosity Value Estimate</code> (PPO/SAC+Curiosity) - The agent's value   estimate for the curiosity reward.</p> </li> <li> <p><code>Policy/GAIL Reward</code> (PPO/SAC+GAIL) - This corresponds to the mean cumulative   discriminator-based reward generated per-episode.</p> </li> <li> <p><code>Policy/GAIL Value Estimate</code> (PPO/SAC+GAIL) - The agent's value estimate for   the GAIL reward.</p> </li> <li> <p><code>Policy/GAIL Policy Estimate</code> (PPO/SAC+GAIL) - The discriminator's estimate   for states and actions generated by the policy.</p> </li> <li> <p><code>Policy/GAIL Expert Estimate</code> (PPO/SAC+GAIL) - The discriminator's estimate   for states and actions drawn from expert demonstrations.</p> </li> </ul>"},{"location":"Using-Tensorboard/#learning-loss-functions","title":"Learning Loss Functions","text":"<ul> <li> <p><code>Losses/Policy Loss</code> (PPO; SAC) - The mean magnitude of policy loss function.   Correlates to how much the policy (process for deciding actions) is changing.   The magnitude of this should decrease during a successful training session.</p> </li> <li> <p><code>Losses/Value Loss</code> (PPO; SAC) - The mean loss of the value function update.   Correlates to how well the model is able to predict the value of each state.   This should increase while the agent is learning, and then decrease once the   reward stabilizes.</p> </li> <li> <p><code>Losses/Forward Loss</code> (PPO/SAC+Curiosity) - The mean magnitude of the forward   model loss function. Corresponds to how well the model is able to predict the   new observation encoding.</p> </li> <li> <p><code>Losses/Inverse Loss</code> (PPO/SAC+Curiosity) - The mean magnitude of the inverse   model loss function. Corresponds to how well the model is able to predict the   action taken between two observations.</p> </li> <li> <p><code>Losses/Pretraining Loss</code> (BC) - The mean magnitude of the behavioral cloning   loss. Corresponds to how well the model imitates the demonstration data.</p> </li> <li> <p><code>Losses/GAIL Loss</code> (GAIL) - The mean magnitude of the GAIL discriminator loss.   Corresponds to how well the model imitates the demonstration data.</p> </li> </ul>"},{"location":"Using-Tensorboard/#self-play","title":"Self-Play","text":"<ul> <li><code>Self-Play/ELO</code> (Self-Play) -   ELO measures the relative   skill level between two players. In a proper training run, the ELO of the   agent should steadily increase.</li> </ul>"},{"location":"Using-Tensorboard/#exporting-data-from-tensorboard","title":"Exporting Data from TensorBoard","text":"<p>To export timeseries data in CSV or JSON format, check the \"Show data download links\" in the upper left. This will enable download links below each chart.</p> <p></p>"},{"location":"Using-Tensorboard/#custom-metrics-from-unity","title":"Custom Metrics from Unity","text":"<p>To get custom metrics from a C# environment into TensorBoard, you can use the <code>StatsRecorder</code>:</p> <pre><code>var statsRecorder = Academy.Instance.StatsRecorder;\nstatsRecorder.Add(\"MyMetric\", 1.0);\n</code></pre>"},{"location":"Using-Virtual-Environment/","title":"Using Virtual Environment","text":""},{"location":"Using-Virtual-Environment/#what-is-a-virtual-environment","title":"What is a Virtual Environment?","text":"<p>A Virtual Environment is a self contained directory tree that contains a Python installation for a particular version of Python, plus a number of additional packages. To learn more about Virtual Environments see here.</p>"},{"location":"Using-Virtual-Environment/#why-should-i-use-a-virtual-environment","title":"Why should I use a Virtual Environment?","text":"<p>A Virtual Environment keeps all dependencies for the Python project separate from dependencies of other projects. This has a few advantages:</p> <ol> <li>It makes dependency management for the project easy.</li> <li>It enables using and testing of different library versions by quickly    spinning up a new environment and verifying the compatibility of the code    with the different version.</li> </ol>"},{"location":"Using-Virtual-Environment/#python-version-requirement-required","title":"Python Version Requirement (Required)","text":"<p>This guide has been tested with Python 3.10.12. Newer versions might not have support for the dependent libraries, so are not recommended.</p>"},{"location":"Using-Virtual-Environment/#use-conda-or-mamba","title":"Use Conda (or Mamba)","text":"<p>While there are many options for setting up virtual environments for python, by far the most</p>"},{"location":"Using-Virtual-Environment/#installing-pip-required","title":"Installing Pip (Required)","text":"<ol> <li>Download the <code>get-pip.py</code> file using the command    <code>curl https://bootstrap.pypa.io/get-pip.py -o get-pip.py</code></li> <li>Run the following <code>python3 get-pip.py</code></li> <li>Check pip version using <code>pip3 -V</code></li> </ol> <p>Note (for Ubuntu users): If the <code>ModuleNotFoundError: No module named 'distutils.util'</code> error is encountered, then python3-distutils needs to be installed. Install python3-distutils using <code>sudo apt-get install python3-distutils</code></p>"},{"location":"Using-Virtual-Environment/#mac-os-x-setup","title":"Mac OS X Setup","text":"<ol> <li>Create a folder where the virtual environments will reside    <code>$ mkdir ~/python-envs</code></li> <li>To create a new environment named <code>sample-env</code> execute    <code>$ python3 -m venv ~/python-envs/sample-env</code></li> <li>To activate the environment execute    <code>$ source ~/python-envs/sample-env/bin/activate</code></li> <li>Upgrade to the latest pip version using <code>$ pip3 install --upgrade pip</code></li> <li>Upgrade to the latest setuptools version using    <code>$ pip3 install --upgrade setuptools</code></li> <li>To deactivate the environment execute <code>$ deactivate</code> (you can reactivate the    environment using the same <code>activate</code> command listed above)</li> </ol>"},{"location":"Using-Virtual-Environment/#ubuntu-setup","title":"Ubuntu Setup","text":"<ol> <li>Install the python3-venv package using <code>$ sudo apt-get install python3-venv</code></li> <li>Follow the steps in the Mac OS X installation.</li> </ol>"},{"location":"Using-Virtual-Environment/#windows-setup","title":"Windows Setup","text":"<ol> <li>Create a folder where the virtual environments will reside <code>md python-envs</code></li> <li>To create a new environment named <code>sample-env</code> execute    <code>python -m venv python-envs\\sample-env</code></li> <li>To activate the environment execute <code>python-envs\\sample-env\\Scripts\\activate</code></li> <li>Upgrade to the latest pip version using <code>pip install --upgrade pip</code></li> <li>To deactivate the environment execute <code>deactivate</code> (you can reactivate the    environment using the same <code>activate</code> command listed above)</li> </ol> <p>Note: - Verify that you are using Python version 3.10.12. Launch a   command prompt using <code>cmd</code> and execute <code>python --version</code> to verify the version. - Python3 installation may require admin privileges on Windows. - This guide is for Windows 10 using a 64-bit architecture only.</p>"},{"location":"Versioning/","title":"ML-Agents Versioning","text":""},{"location":"Versioning/#context","title":"Context","text":"<p>As the ML-Agents project evolves into a more mature product, we want to communicate the process we use to version our packages and the data that flows into, through, and out of them clearly. Our project now has four packages (1 Unity, 3 Python) along with artifacts that are produced as well as consumed.  This document covers the versioning for these packages and artifacts.</p>"},{"location":"Versioning/#github-releases","title":"GitHub Releases","text":"<p>Up until now, all packages were in lockstep in-terms of versioning. As a result, the GitHub releases were tagged with the version of all those packages (e.g. v0.15.0, v0.15.1) and labeled accordingly. With the decoupling of package versions, we now need to revisit our GitHub release tagging. The proposal is that we move towards an integer release numbering for our repo and each such release will call out specific version upgrades of each package. For instance, with the April 30th release, we will have: - GitHub Release 1 (branch name: release_1_branch)   - com.unity.ml-agents release 1.0.0   - ml-agents release 0.16.0   - ml-agents-envs release 0.16.0   - gym-unity release 0.16.0</p> <p>Our release cadence will not be affected by these versioning changes.  We will keep having monthly releases to fix bugs and release new features.</p>"},{"location":"Versioning/#packages","title":"Packages","text":"<p>All of the software packages, and their generated artifacts will be versioned.  Any automation tools will not be versioned.</p>"},{"location":"Versioning/#unity-package","title":"Unity package","text":"<p>Package name: com.unity.ml-agents - Versioned following Semantic Versioning Guidelines - This package consumes an artifact of the training process: the <code>.nn</code> file.  These files     are integer versioned and currently at version 2. The com.unity.ml-agents package     will need to support the version of <code>.nn</code> files which existed at its 1.0.0 release.     For example, consider that com.unity.ml-agents is at version 1.0.0 and the NN files     are at version 2.  If the NN files change to version 3, the next release of     com.unity.ml-agents at version 1.1.0 guarantees it will be able to read both of these     formats.  If the NN files were to change to version 4 and com.unity.ml-agents to     version 2.0.0, support for NN versions 2 and 3 could be dropped for com.unity.ml-agents     version 2.0.0. - This package produces one artifact, the <code>.demo</code> files.  These files will have integer     versioning. This means their version will increment by 1 at each change.  The     com.unity.ml-agents package must be backward compatible with version changes     that occur between minor versions. - To summarize, the artifacts produced and consumed by com.unity.ml-agents are guaranteed     to be supported for 1.x.x versions of com.unity.ml-agents.  We intend to provide stability     for our users by moving to a 1.0.0 release of com.unity.ml-agents.</p>"},{"location":"Versioning/#python-packages","title":"Python Packages","text":"<p>Package names: ml-agents / ml-agents-envs / gym-unity - The python packages remain in \"Beta.\"  This means that breaking changes to the public     API of the python packages can change without having to have a major version bump.     Historically, the python and C# packages were in version lockstep.  This is no longer     the case.  The python packages will remain in lockstep with each other for now, while the     C# package will follow its own versioning as is appropriate.  However, the python package     versions may diverge in the future. - While the python packages will remain in Beta for now, we acknowledge that the most     heavily used portion of our python interface is the <code>mlagents-learn</code> CLI and strive     to make this part of our API backward compatible. We are actively working on this and     expect to have a stable CLI in the next few weeks.</p>"},{"location":"Versioning/#communicator","title":"Communicator","text":"<p>Packages which communicate: com.unity.ml-agents / ml-agents-envs</p> <p>Another entity of the ML-Agents Toolkit that requires versioning is the communication layer between C# and Python, which will follow also semantic versioning.  This guarantees a level of backward compatibility between different versions of C# and Python packages which communicate. Any Communicator version 1.x.x of the Unity package should be compatible with any 1.x.x Communicator Version in Python.</p> <p>An RLCapabilities struct keeps track of which features exist. This struct is passed from C# to Python, and another from Python to C#.  With this feature level granularity, we can notify users more specifically about feature limitations based on what's available in both C# and Python. These notifications will be logged to the python terminal, or to the Unity Editor Console.</p>"},{"location":"Versioning/#side-channels","title":"Side Channels","text":"<p>The communicator is what manages data transfer between Unity and Python for the core training loop. Side Channels are another means of data transfer between Unity and Python. Side Channels are not versioned, but have been designed to support backward compatibility for what they are. As of today, we provide 4 side channels: - FloatProperties: shared float data between Unity - Python (bidirectional) - RawBytes: raw data that can be sent Unity - Python (bidirectional) - EngineConfig: a set of numeric fields in a pre-defined order sent from Python to Unity - Stats: (name, value, agg) messages sent from Unity to Python</p> <p>Aside from the specific implementations of side channels we provide (and use ourselves), the Side Channel interface is made available for users to create their own custom side channels. As such, we guarantee that the built in SideChannel interface between Unity and Python is backward compatible in packages that share the same major version.</p>"},{"location":"com.unity.ml-agents.extensions/","title":"About ML-Agents Extensions package (<code>com.unity.ml-agents.extensions</code>)","text":"<p>The Unity ML-Agents Extensions package contains optional add-ons to the C# SDK for the Unity ML-Agents Toolkit.</p> <p>These extensions are all considered experimental, and their API or behavior may change between versions.</p>"},{"location":"com.unity.ml-agents.extensions/#package-contents","title":"Package contents","text":"<p>The following table describes the package folder structure:</p> Location Description Documentation~ Contains the documentation for the Unity package. Editor Contains utilities for Editor windows and drawers. Runtime Contains core C# APIs for integrating ML-Agents into your Unity scene. Tests Contains the unit tests for the package. <p>The Runtime directory currently contains these features:  * Physics-based sensors  * Input System Package Integration  * Custom Grid-based Sensors</p>"},{"location":"com.unity.ml-agents.extensions/#installation","title":"Installation","text":"<p>The ML-Agents Extensions package is not currently available in the Package Manager. There are two recommended ways to install the package:</p>"},{"location":"com.unity.ml-agents.extensions/#local-installation","title":"Local Installation","text":"<p>Clone the repository and follow the Local Installation for Development directions (substituting <code>com.unity.ml-agents.extensions</code> for the package name).</p>"},{"location":"com.unity.ml-agents.extensions/#github-via-package-manager","title":"Github via Package Manager","text":"<p>In Unity 2019.4 or later, open the Package Manager, hit the \"+\" button, and select \"Add package from git URL\".</p> <p></p> <p>In the dialog that appears, enter  ``` git+https://github.com/Unity-Technologies/ml-agents.git?path=com.unity.ml-agents.extensions#release_21</p> <pre><code>\nYou can also edit your project's `manifest.json` directly and add the following line to the `dependencies`\nsection:\n</code></pre> <p>\"com.unity.ml-agents.extensions\": \"git+https://github.com/Unity-Technologies/ml-agents.git?path=com.unity.ml-agents.extensions#release_21\", ``` See Git dependencies for more information. Note that this may take several minutes to resolve the packages the first time that you add it.</p>"},{"location":"com.unity.ml-agents.extensions/#requirements","title":"Requirements","text":"<p>This version of the Unity ML-Agents package is compatible with the following versions of the Unity Editor:</p> <ul> <li>2019.4 and later</li> </ul> <p>If using the <code>InputActuatorComponent</code> - install the <code>com.unity.inputsystem</code> package version <code>1.1.0-preview.3</code> or later.</p>"},{"location":"com.unity.ml-agents.extensions/#known-limitations","title":"Known Limitations","text":"<ul> <li>For the <code>InputActuatorComponent</code><ul> <li>Limited implementation of <code>InputControls</code></li> <li>No way to customize the action space of the <code>InputActuatorComponent</code></li> </ul> </li> </ul>"},{"location":"com.unity.ml-agents.extensions/#need-help","title":"Need Help?","text":"<p>The main README contains links for contacting the team or getting support.</p>"},{"location":"com.unity.ml-agents/","title":"About ML-Agents package (<code>com.unity.ml-agents</code>)","text":"<p>The ML-Agents package contains the primary C# SDK for the Unity ML-Agents Toolkit.</p> <p>The package allows you to convert any Unity scene into a learning environment and train character behaviors using a variety of machine learning algorithms. Additionally, it allows you to embed these trained behaviors back into Unity scenes to control your characters. More specifically, the package provides the following core functionalities:</p> <ul> <li>Define Agents: entities, or characters, whose behavior will be learned. Agents   are entities that generate observations (through sensors), take actions, and   receive rewards from the environment.</li> <li>Define Behaviors: entities that specify how an agent should act. Multiple   agents can share the same Behavior and a scene may have multiple Behaviors.</li> <li>Record demonstrations of an agent within the Editor. You can use   demonstrations to help train a behavior for that agent.</li> <li>Embedding a trained behavior into the scene via the Unity Inference Engine.   Embedded behaviors allow you to switch an Agent between learning and   inference.</li> </ul> <p>Note that the ML-Agents package does not contain the machine learning algorithms for training behaviors. The ML-Agents package only supports instrumenting a Unity scene, setting it up for training, and then embedding the trained model back into your Unity scene. The machine learning algorithms that orchestrate training are part of the companion Python package.</p> <p>Note that we also provide an ML-Agents Extensions package (<code>com.unity.ml-agents.extensions</code>) that contains early/experimental features that you may find useful. This package is only available from the ML-Agents GitHub repo.</p>"},{"location":"com.unity.ml-agents/#package-contents","title":"Package contents","text":"<p>The following table describes the package folder structure:</p> Location Description Documentation~ Contains the documentation for the Unity package. Editor Contains utilities for Editor windows and drawers. Plugins Contains third-party DLLs. Runtime Contains core C# APIs for integrating ML-Agents into your Unity scene. Runtime/Integrations Contains utilities for integrating ML-Agents into specific game genres. Tests Contains the unit tests for the package. <p></p>"},{"location":"com.unity.ml-agents/#installation","title":"Installation","text":"<p>To install this ML-Agents package, follow the instructions in the Package Manager documentation.</p> <p>To install the companion Python package to enable training behaviors, follow the installation instructions on our GitHub repository.</p>"},{"location":"com.unity.ml-agents/#advanced-installation","title":"Advanced Installation","text":"<p>With the changes to Unity Package Manager in 2021, experimental packages will not show up in the package list and have to be installed manually. There are two recommended ways to install the package manually:</p>"},{"location":"com.unity.ml-agents/#github-via-package-manager","title":"Github via Package Manager","text":"<p>In Unity 2019.4 or later, open the Package Manager, hit the \"+\" button, and select \"Add package from git URL\".</p> <p></p> <p>In the dialog that appears, enter</p> <pre><code>git+https://github.com/Unity-Technologies/ml-agents.git?path=com.unity.ml-agents#release_21\n</code></pre> <p>You can also edit your project's <code>manifest.json</code> directly and add the following line to the <code>dependencies</code> section:</p> <pre><code>\"com.unity.ml-agents\": \"git+https://github.com/Unity-Technologies/ml-agents.git?path=com.unity.ml-agents#release_21\",\n</code></pre> <p>See Git dependencies for more information. Note that this may take several minutes to resolve the packages the first time that you add it.</p>"},{"location":"com.unity.ml-agents/#local-installation-for-development","title":"Local Installation for Development","text":"<p>Clone the repository and follow the Local Installation for Development directions.</p>"},{"location":"com.unity.ml-agents/#requirements","title":"Requirements","text":"<p>This version of the Unity ML-Agents package is compatible with the following versions of the Unity Editor:</p> <ul> <li>2019.4 and later</li> </ul>"},{"location":"com.unity.ml-agents/#known-limitations","title":"Known Limitations","text":""},{"location":"com.unity.ml-agents/#training","title":"Training","text":"<p>Training is limited to the Unity Editor and Standalone builds on Windows, MacOS, and Linux with the Mono scripting backend. Currently, training does not work with the IL2CPP scripting backend. Your environment will default to inference mode if training is not supported or is not currently running.</p>"},{"location":"com.unity.ml-agents/#inference","title":"Inference","text":"<p>Inference is executed via the Unity Inference Engine.</p> <p>CPU</p> <p>All platforms supported.</p> <p>GPU</p> <p>All platforms supported except:</p> <ul> <li>WebGL and GLES 3/2 on Android / iPhone</li> </ul> <p>NOTE: Mobile platform support includes:</p> <ul> <li>Vulkan for Android</li> <li>Metal for iOS.</li> </ul>"},{"location":"com.unity.ml-agents/#headless-mode","title":"Headless Mode","text":"<p>If you enable Headless mode, you will not be able to collect visual observations from your agents.</p>"},{"location":"com.unity.ml-agents/#rendering-speed-and-synchronization","title":"Rendering Speed and Synchronization","text":"<p>Currently the speed of the game physics can only be increased to 100x real-time. The Academy also moves in time with FixedUpdate() rather than Update(), so game behavior implemented in Update() may be out of sync with the agent decision making. See Execution Order of Event Functions for more information.</p> <p>You can control the frequency of Academy stepping by calling <code>Academy.Instance.DisableAutomaticStepping()</code>, and then calling <code>Academy.Instance.EnvironmentStep()</code></p>"},{"location":"com.unity.ml-agents/#unity-inference-engine-models","title":"Unity Inference Engine Models","text":"<p>Currently, only models created with our trainers are supported for running ML-Agents with a neural network behavior.</p>"},{"location":"com.unity.ml-agents/#helpful-links","title":"Helpful links","text":"<p>If you are new to the Unity ML-Agents package, or have a question after reading the documentation, you can checkout our GitHub Repository, which also includes a number of ways to connect with us including our ML-Agents Forum.</p> <p>In order to improve the developer experience for Unity ML-Agents Toolkit, we have added in-editor analytics. Please refer to \"Information that is passively collected by Unity\" in the Unity Privacy Policy.</p>"},{"location":"ml-agents-envs/","title":"Unity ML-Agents Python Interface","text":"<p>The <code>mlagents_envs</code> Python package is part of the ML-Agents Toolkit. <code>mlagents_envs</code> provides three Python APIs that allows direct interaction with the Unity game engine: - A single agent API (Gym API) - A gym-like multi-agent API (PettingZoo API) - A low-level API (LLAPI)</p> <p>The LLAPI is used by the trainer implementation in <code>mlagents</code>. <code>mlagents_envs</code> can be used independently of <code>mlagents</code> for Python communication.</p>"},{"location":"ml-agents-envs/#installation","title":"Installation","text":"<p>Install the <code>mlagents_envs</code> package with:</p> <pre><code>python -m pip install mlagents_envs==1.0.0\n</code></pre>"},{"location":"ml-agents-envs/#usage-more-information","title":"Usage &amp; More Information","text":"<p>See - Gym API Guide - PettingZoo API Guide - Python API Guide</p> <p>for more information on how to use the API to interact with a Unity environment.</p> <p>For more information on the ML-Agents Toolkit and how to instrument a Unity scene with the ML-Agents SDK, check out the main ML-Agents Toolkit documentation.</p>"},{"location":"ml-agents-envs/#limitations","title":"Limitations","text":"<ul> <li><code>mlagents_envs</code> uses localhost ports to exchange data between Unity and   Python. As such, multiple instances can have their ports collide, leading to   errors. Make sure to use a different port if you are using multiple instances   of <code>UnityEnvironment</code>.</li> <li>Communication between Unity and the Python <code>UnityEnvironment</code> is not secure.</li> <li>On Linux, ports are not released immediately after the communication closes.   As such, you cannot reuse ports right after closing a <code>UnityEnvironment</code>.</li> </ul>"},{"location":"ml-agents/","title":"Unity ML-Agents Trainers","text":"<p>The <code>mlagents</code> Python package is part of the ML-Agents Toolkit. <code>mlagents</code> provides a set of reinforcement and imitation learning algorithms designed to be used with Unity environments. The algorithms interface with the Python API provided by the <code>mlagents_envs</code> package. See here for more information on <code>mlagents_envs</code>.</p> <p>The algorithms can be accessed using the: <code>mlagents-learn</code> access point. See here for more information on using this package.</p>"},{"location":"ml-agents/#installation","title":"Installation","text":"<p>Install the <code>mlagents</code> package with:</p> <pre><code>python -m pip install mlagents==1.0.0\n</code></pre>"},{"location":"ml-agents/#usage-more-information","title":"Usage &amp; More Information","text":"<p>For more information on the ML-Agents Toolkit and how to instrument a Unity scene with the ML-Agents SDK, check out the main ML-Agents Toolkit documentation.</p>"},{"location":"ml-agents/#limitations","title":"Limitations","text":"<ul> <li>Resuming self-play from a checkpoint resets the reported ELO to the default   value.</li> </ul>"},{"location":"doxygen/Readme/","title":"Doxygen files","text":"<p>To generate the API reference as HTML files, run:</p> <pre><code>doxygen dox-ml-agents.conf\n</code></pre>"}]}
\ No newline at end of file
+{"config":{"lang":["en"],"separator":"[\\s\\-]+","pipeline":["stopWordFilter"]},"docs":[{"location":"","title":"Home","text":""},{"location":"#unity-ml-agents-toolkit","title":"Unity ML-Agents Toolkit","text":"<p>(latest release) (all releases)</p> <p>The Unity Machine Learning Agents Toolkit (ML-Agents) is an open-source project that enables games and simulations to serve as environments for training intelligent agents. We provide implementations (based on PyTorch) of state-of-the-art algorithms to enable game developers and hobbyists to easily train intelligent agents for 2D, 3D and VR/AR games. Researchers can also use the provided simple-to-use Python API to train Agents using reinforcement learning, imitation learning, neuroevolution, or any other methods. These trained agents can be used for multiple purposes, including controlling NPC behavior (in a variety of settings such as multi-agent and adversarial), automated testing of game builds and evaluating different game design decisions pre-release. The ML-Agents Toolkit is mutually beneficial for both game developers and AI researchers as it provides a central platform where advances in AI can be evaluated on Unity\u2019s rich environments and then made accessible to the wider research and game developer communities.</p>"},{"location":"#features","title":"Features","text":"<ul> <li>17+ example Unity environments</li> <li>Support for multiple environment configurations and training scenarios</li> <li>Flexible Unity SDK that can be integrated into your game or custom Unity scene</li> <li>Support for training single-agent, multi-agent cooperative, and multi-agent   competitive scenarios via several Deep Reinforcement Learning algorithms (PPO, SAC, MA-POCA, self-play).</li> <li>Support for learning from demonstrations through two Imitation Learning algorithms (BC and GAIL).</li> <li>Quickly and easily add your own custom training algorithm and/or components.</li> <li>Easily definable Curriculum Learning scenarios for complex tasks</li> <li>Train robust agents using environment randomization</li> <li>Flexible agent control with On Demand Decision Making</li> <li>Train using multiple concurrent Unity environment instances</li> <li>Utilizes the Sentis to   provide native cross-platform support</li> <li>Unity environment control from Python</li> <li>Wrap Unity learning environments as a gym environment</li> <li>Wrap Unity learning environments as a PettingZoo environment</li> </ul> <p>See our ML-Agents Overview page for detailed descriptions of all these features. Or go straight to our web docs.</p>"},{"location":"#releases-documentation","title":"Releases &amp; Documentation","text":"<p>Our latest, stable release is <code>Release 21</code>. Click here to get started with the latest release of ML-Agents.</p> <p>You can also check out our new web docs!</p> <p>The table below lists all our releases, including our <code>main</code> branch which is under active development and may be unstable. A few helpful guidelines:</p> <ul> <li>The Versioning page overviews how we manage our GitHub   releases and the versioning process for each of the ML-Agents components.</li> <li>The Releases page   contains details of the changes between releases.</li> <li>The Migration page contains details on how to upgrade   from earlier releases of the ML-Agents Toolkit.</li> <li>The Documentation links in the table below include installation and usage   instructions specific to each release. Remember to always use the   documentation that corresponds to the release version you're using.</li> <li>The <code>com.unity.ml-agents</code> package is verified   for Unity 2020.1 and later. Verified packages releases are numbered 1.0.x.</li> </ul> Version Release Date Source Documentation Download Python Package Unity Package develop (unstable) -- source docs download -- -- Release 21 October 9, 2023 source docs download 1.0.0 3.0.0 <p>If you are a researcher interested in a discussion of Unity as an AI platform, see a pre-print of our reference paper on Unity and the ML-Agents Toolkit.</p> <p>If you use Unity or the ML-Agents Toolkit to conduct research, we ask that you cite the following paper as a reference:</p> <pre><code>@article{juliani2020,\n  title={Unity: A general platform for intelligent agents},\n  author={Juliani, Arthur and Berges, Vincent-Pierre and Teng, Ervin and Cohen, Andrew and Harper, Jonathan and Elion, Chris and Goy, Chris and Gao, Yuan and Henry, Hunter and Mattar, Marwan and Lange, Danny},\n  journal={arXiv preprint arXiv:1809.02627},\n  url={https://arxiv.org/pdf/1809.02627.pdf},\n  year={2020}\n}\n</code></pre> <p>Additionally, if you use the MA-POCA trainer in your research, we ask that you cite the following paper as a reference:</p> <pre><code>@article{cohen2022,\n  title={On the Use and Misuse of Absorbing States in Multi-agent Reinforcement Learning},\n  author={Cohen, Andrew and Teng, Ervin and Berges, Vincent-Pierre and Dong, Ruo-Ping and Henry, Hunter and Mattar, Marwan and Zook, Alexander and Ganguly, Sujoy},\n  journal={RL in Games Workshop AAAI 2022},\n  url={http://aaai-rlg.mlanctot.info/papers/AAAI22-RLG_paper_32.pdf},\n  year={2022}\n}\n</code></pre>"},{"location":"#additional-resources","title":"Additional Resources","text":"<p>We have a Unity Learn course, ML-Agents: Hummingbirds, that provides a gentle introduction to Unity and the ML-Agents Toolkit.</p> <p>We've also partnered with CodeMonkeyUnity to create a series of tutorial videos on how to implement and use the ML-Agents Toolkit.</p> <p>We have also published a series of blog posts that are relevant for ML-Agents:</p> <ul> <li>(July 12, 2021)   ML-Agents plays Dodgeball</li> <li>(May 5, 2021)   ML-Agents v2.0 release: Now supports training complex cooperative behaviors</li> <li>(December 28, 2020)   Happy holidays from the Unity ML-Agents team!</li> <li>(November 20, 2020)   How Eidos-Montr\u00e9al created Grid Sensors to improve observations for training agents</li> <li>(November 11, 2020)   2020 AI@Unity interns shoutout</li> <li>(May 12, 2020)   Announcing ML-Agents Unity Package v1.0!</li> <li>(February 28, 2020)   Training intelligent adversaries using self-play with ML-Agents</li> <li>(November 11, 2019)   Training your agents 7 times faster with ML-Agents</li> <li>(October 21, 2019)   The AI@Unity interns help shape the world</li> <li>(April 15, 2019)   Unity ML-Agents Toolkit v0.8: Faster training on real games</li> <li>(March 1, 2019)   Unity ML-Agents Toolkit v0.7: A leap towards cross-platform inference</li> <li>(December 17, 2018)   ML-Agents Toolkit v0.6: Improved usability of Brains and Imitation Learning</li> <li>(October 2, 2018)   Puppo, The Corgi: Cuteness Overload with the Unity ML-Agents Toolkit</li> <li>(September 11, 2018)   ML-Agents Toolkit v0.5, new resources for AI researchers available now</li> <li>(June 26, 2018)   Solving sparse-reward tasks with Curiosity</li> <li>(June 19, 2018)   Unity ML-Agents Toolkit v0.4 and Udacity Deep Reinforcement Learning Nanodegree</li> <li>(May 24, 2018)   Imitation Learning in Unity: The Workflow</li> <li>(March 15, 2018)   ML-Agents Toolkit v0.3 Beta released: Imitation Learning, feedback-driven features, and more</li> <li>(December 11, 2017)   Using Machine Learning Agents in a real game: a beginner\u2019s guide</li> <li>(December 8, 2017)   Introducing ML-Agents Toolkit v0.2: Curriculum Learning, new environments, and more</li> <li>(September 19, 2017)   Introducing: Unity Machine Learning Agents Toolkit</li> <li>Overviewing reinforcement learning concepts   (multi-armed bandit   and   Q-learning)</li> </ul>"},{"location":"#more-from-unity","title":"More from Unity","text":"<ul> <li>Unity Sentis</li> <li>Introducing Unity Muse and Sentis</li> </ul>"},{"location":"#community-and-feedback","title":"Community and Feedback","text":"<p>The ML-Agents Toolkit is an open-source project and we encourage and welcome contributions. If you wish to contribute, be sure to review our contribution guidelines and code of conduct.</p> <p>For problems with the installation and setup of the ML-Agents Toolkit, or discussions about how to best setup or train your agents, please create a new thread on the Unity ML-Agents forum and make sure to include as much detail as possible. If you run into any other problems using the ML-Agents Toolkit or have a specific feature request, please submit a GitHub issue.</p> <p>Please tell us which samples you would like to see shipped with the ML-Agents Unity package by replying to this forum thread.</p> <p>Your opinion matters a great deal to us. Only by hearing your thoughts on the Unity ML-Agents Toolkit can we continue to improve and grow. Please take a few minutes to let us know about it.</p> <p>For any other questions or feedback, connect directly with the ML-Agents team at ml-agents@unity3d.com.</p>"},{"location":"#privacy","title":"Privacy","text":"<p>In order to improve the developer experience for Unity ML-Agents Toolkit, we have added in-editor analytics. Please refer to \"Information that is passively collected by Unity\" in the Unity Privacy Policy.</p>"},{"location":"API-Reference/","title":"API Reference","text":"<p>Our developer-facing C# classes have been documented to be compatible with Doxygen for auto-generating HTML documentation.</p> <p>To generate the API reference, download Doxygen and run the following command within the <code>docs/</code> directory:</p> <pre><code>doxygen dox-ml-agents.conf\n</code></pre> <p><code>dox-ml-agents.conf</code> is a Doxygen configuration file for the ML-Agents Toolkit that includes the classes that have been properly formatted. The generated HTML files will be placed in the <code>html/</code> subdirectory. Open <code>index.html</code> within that subdirectory to navigate to the API reference home. Note that <code>html/</code> is already included in the repository's <code>.gitignore</code> file.</p> <p>In the near future, we aim to expand our documentation to include the Python classes.</p>"},{"location":"Background-Machine-Learning/","title":"Background: Machine Learning","text":"<p>Given that a number of users of the ML-Agents Toolkit might not have a formal machine learning background, this page provides an overview to facilitate the understanding of the ML-Agents Toolkit. However, we will not attempt to provide a thorough treatment of machine learning as there are fantastic resources online.</p> <p>Machine learning, a branch of artificial intelligence, focuses on learning patterns from data. The three main classes of machine learning algorithms include: unsupervised learning, supervised learning and reinforcement learning. Each class of algorithm learns from a different type of data. The following paragraphs provide an overview for each of these classes of machine learning, as well as introductory examples.</p>"},{"location":"Background-Machine-Learning/#unsupervised-learning","title":"Unsupervised Learning","text":"<p>The goal of unsupervised learning is to group or cluster similar items in a data set. For example, consider the players of a game. We may want to group the players depending on how engaged they are with the game. This would enable us to target different groups (e.g. for highly-engaged players we might invite them to be beta testers for new features, while for unengaged players we might email them helpful tutorials). Say that we wish to split our players into two groups. We would first define basic attributes of the players, such as the number of hours played, total money spent on in-app purchases and number of levels completed. We can then feed this data set (three attributes for every player) to an unsupervised learning algorithm where we specify the number of groups to be two. The algorithm would then split the data set of players into two groups where the players within each group would be similar to each other. Given the attributes we used to describe each player, in this case, the output would be a split of all the players into two groups, where one group would semantically represent the engaged players and the second group would semantically represent the unengaged players.</p> <p>With unsupervised learning, we did not provide specific examples of which players are considered engaged and which are considered unengaged. We just defined the appropriate attributes and relied on the algorithm to uncover the two groups on its own. This type of data set is typically called an unlabeled data set as it is lacking these direct labels. Consequently, unsupervised learning can be helpful in situations where these labels can be expensive or hard to produce. In the next paragraph, we overview supervised learning algorithms which accept input labels in addition to attributes.</p>"},{"location":"Background-Machine-Learning/#supervised-learning","title":"Supervised Learning","text":"<p>In supervised learning, we do not want to just group similar items but directly learn a mapping from each item to the group (or class) that it belongs to. Returning to our earlier example of clustering players, let's say we now wish to predict which of our players are about to churn (that is stop playing the game for the next 30 days). We can look into our historical records and create a data set that contains attributes of our players in addition to a label indicating whether they have churned or not. Note that the player attributes we use for this churn prediction task may be different from the ones we used for our earlier clustering task. We can then feed this data set (attributes and label for each player) into a supervised learning algorithm which would learn a mapping from the player attributes to a label indicating whether that player will churn or not. The intuition is that the supervised learning algorithm will learn which values of these attributes typically correspond to players who have churned and not churned (for example, it may learn that players who spend very little and play for very short periods will most likely churn). Now given this learned model, we can provide it the attributes of a new player (one that recently started playing the game) and it would output a predicted label for that player. This prediction is the algorithms expectation of whether the player will churn or not. We can now use these predictions to target the players who are expected to churn and entice them to continue playing the game.</p> <p>As you may have noticed, for both supervised and unsupervised learning, there are two tasks that need to be performed: attribute selection and model selection. Attribute selection (also called feature selection) pertains to selecting how we wish to represent the entity of interest, in this case, the player. Model selection, on the other hand, pertains to selecting the algorithm (and its parameters) that perform the task well. Both of these tasks are active areas of machine learning research and, in practice, require several iterations to achieve good performance.</p> <p>We now switch to reinforcement learning, the third class of machine learning algorithms, and arguably the one most relevant for the ML-Agents Toolkit.</p>"},{"location":"Background-Machine-Learning/#reinforcement-learning","title":"Reinforcement Learning","text":"<p>Reinforcement learning can be viewed as a form of learning for sequential decision making that is commonly associated with controlling robots (but is, in fact, much more general). Consider an autonomous firefighting robot that is tasked with navigating into an area, finding the fire and neutralizing it. At any given moment, the robot perceives the environment through its sensors (e.g. camera, heat, touch), processes this information and produces an action (e.g. move to the left, rotate the water hose, turn on the water). In other words, it is continuously making decisions about how to interact in this environment given its view of the world (i.e. sensors input) and objective (i.e. neutralizing the fire). Teaching a robot to be a successful firefighting machine is precisely what reinforcement learning is designed to do.</p> <p>More specifically, the goal of reinforcement learning is to learn a policy, which is essentially a mapping from observations to actions. An observation is what the robot can measure from its environment (in this case, all its sensory inputs) and an action, in its most raw form, is a change to the configuration of the robot (e.g. position of its base, position of its water hose and whether the hose is on or off).</p> <p>The last remaining piece of the reinforcement learning task is the reward signal. The robot is trained to learn a policy that maximizes its overall rewards. When training a robot to be a mean firefighting machine, we provide it with rewards (positive and negative) indicating how well it is doing on completing the task. Note that the robot does not know how to put out fires before it is trained. It learns the objective because it receives a large positive reward when it puts out the fire and a small negative reward for every passing second. The fact that rewards are sparse (i.e. may not be provided at every step, but only when a robot arrives at a success or failure situation), is a defining characteristic of reinforcement learning and precisely why learning good policies can be difficult (and/or time-consuming) for complex environments.</p> <p>Learning a policy usually requires many trials and iterative policy updates. More specifically, the robot is placed in several fire situations and over time learns an optimal policy which allows it to put out fires more effectively. Obviously, we cannot expect to train a robot repeatedly in the real world, particularly when fires are involved. This is precisely why the use of Unity as a simulator serves as the perfect training grounds for learning such behaviors. While our discussion of reinforcement learning has centered around robots, there are strong parallels between robots and characters in a game. In fact, in many ways, one can view a non-playable character (NPC) as a virtual robot, with its own observations about the environment, its own set of actions and a specific objective. Thus it is natural to explore how we can train behaviors within Unity using reinforcement learning. This is precisely what the ML-Agents Toolkit offers. The video linked below includes a reinforcement learning demo showcasing training character behaviors using the ML-Agents Toolkit.</p> <p> </p> <p>Similar to both unsupervised and supervised learning, reinforcement learning also involves two tasks: attribute selection and model selection. Attribute selection is defining the set of observations for the robot that best help it complete its objective, while model selection is defining the form of the policy (mapping from observations to actions) and its parameters. In practice, training behaviors is an iterative process that may require changing the attribute and model choices.</p>"},{"location":"Background-Machine-Learning/#training-and-inference","title":"Training and Inference","text":"<p>One common aspect of all three branches of machine learning is that they all involve a training phase and an inference phase. While the details of the training and inference phases are different for each of the three, at a high-level, the training phase involves building a model using the provided data, while the inference phase involves applying this model to new, previously unseen, data. More specifically:</p> <ul> <li>For our unsupervised learning example, the training phase learns the optimal   two clusters based on the data describing existing players, while the   inference phase assigns a new player to one of these two clusters.</li> <li>For our supervised learning example, the training phase learns the mapping   from player attributes to player label (whether they churned or not), and the   inference phase predicts whether a new player will churn or not based on that   learned mapping.</li> <li>For our reinforcement learning example, the training phase learns the optimal   policy through guided trials, and in the inference phase, the agent observes   and takes actions in the wild using its learned policy.</li> </ul> <p>To briefly summarize: all three classes of algorithms involve training and inference phases in addition to attribute and model selections. What ultimately separates them is the type of data available to learn from. In unsupervised learning our data set was a collection of attributes, in supervised learning our data set was a collection of attribute-label pairs, and, lastly, in reinforcement learning our data set was a collection of observation-action-reward tuples.</p>"},{"location":"Background-Machine-Learning/#deep-learning","title":"Deep Learning","text":"<p>Deep learning is a family of algorithms that can be used to address any of the problems introduced above. More specifically, they can be used to solve both attribute and model selection tasks. Deep learning has gained popularity in recent years due to its outstanding performance on several challenging machine learning tasks. One example is AlphaGo, a computer Go program, that leverages deep learning, that was able to beat Lee Sedol (a Go world champion).</p> <p>A key characteristic of deep learning algorithms is their ability to learn very complex functions from large amounts of training data. This makes them a natural choice for reinforcement learning tasks when a large amount of data can be generated, say through the use of a simulator or engine such as Unity. By generating hundreds of thousands of simulations of the environment within Unity, we can learn policies for very complex environments (a complex environment is one where the number of observations an agent perceives and the number of actions they can take are large). Many of the algorithms we provide in ML-Agents use some form of deep learning, built on top of the open-source library, PyTorch.</p>"},{"location":"Background-PyTorch/","title":"Background: PyTorch","text":"<p>As discussed in our machine learning background page, many of the algorithms we provide in the ML-Agents Toolkit leverage some form of deep learning. More specifically, our implementations are built on top of the open-source library PyTorch. In this page we provide a brief overview of PyTorch and TensorBoard that we leverage within the ML-Agents Toolkit.</p>"},{"location":"Background-PyTorch/#pytorch","title":"PyTorch","text":"<p>PyTorch is an open source library for performing computations using data flow graphs, the underlying representation of deep learning models. It facilitates training and inference on CPUs and GPUs in a desktop, server, or mobile device. Within the ML-Agents Toolkit, when you train the behavior of an agent, the output is a model (.onnx) file that you can then associate with an Agent. Unless you implement a new algorithm, the use of PyTorch is mostly abstracted away and behind the scenes.</p>"},{"location":"Background-PyTorch/#tensorboard","title":"TensorBoard","text":"<p>One component of training models with PyTorch is setting the values of certain model attributes (called hyperparameters). Finding the right values of these hyperparameters can require a few iterations. Consequently, we leverage a visualization tool called TensorBoard. It allows the visualization of certain agent attributes (e.g. reward) throughout training which can be helpful in both building intuitions for the different hyperparameters and setting the optimal values for your Unity environment. We provide more details on setting the hyperparameters in the Training ML-Agents page. If you are unfamiliar with TensorBoard we recommend our guide on using TensorBoard with ML-Agents or this tutorial.</p>"},{"location":"Background-Unity/","title":"Background: Unity","text":"<p>If you are not familiar with the Unity Engine, we highly recommend the Unity Manual and Tutorials page. The Roll-a-ball tutorial is a fantastic resource to learn all the basic concepts of Unity to get started with the ML-Agents Toolkit:</p> <ul> <li>Editor</li> <li>Scene</li> <li>GameObject</li> <li>Rigidbody</li> <li>Camera</li> <li>Scripting</li> <li>Physics</li> <li>Ordering of event functions   (e.g. FixedUpdate, Update)</li> <li>Prefabs</li> </ul>"},{"location":"CODE_OF_CONDUCT/","title":"Contributor Covenant Code of Conduct","text":""},{"location":"CODE_OF_CONDUCT/#our-pledge","title":"Our Pledge","text":"<p>In the interest of fostering an open and welcoming environment, we as contributors and maintainers pledge to making participation in our project and our community a harassment-free experience for everyone, regardless of age, body size, disability, ethnicity, gender identity and expression, level of experience, nationality, personal appearance, race, religion, or sexual identity and orientation.</p>"},{"location":"CODE_OF_CONDUCT/#our-standards","title":"Our Standards","text":"<p>Examples of behavior that contributes to creating a positive environment include:</p> <ul> <li>Using welcoming and inclusive language</li> <li>Being respectful of differing viewpoints and experiences</li> <li>Gracefully accepting constructive criticism</li> <li>Focusing on what is best for the community</li> <li>Showing empathy towards other community members</li> </ul> <p>Examples of unacceptable behavior by participants include:</p> <ul> <li>The use of sexualized language or imagery and unwelcome sexual attention or   advances</li> <li>Trolling, insulting/derogatory comments, and personal or political attacks</li> <li>Public or private harassment</li> <li>Publishing others' private information, such as a physical or electronic   address, without explicit permission</li> <li>Other conduct which could reasonably be considered inappropriate in a   professional setting</li> </ul>"},{"location":"CODE_OF_CONDUCT/#our-responsibilities","title":"Our Responsibilities","text":"<p>Project maintainers are responsible for clarifying the standards of acceptable behavior and are expected to take appropriate and fair corrective action in response to any instances of unacceptable behavior.</p> <p>Project maintainers have the right and responsibility to remove, edit, or reject comments, commits, code, wiki edits, issues, and other contributions that are not aligned to this Code of Conduct, or to ban temporarily or permanently any contributor for other behaviors that they deem inappropriate, threatening, offensive, or harmful.</p>"},{"location":"CODE_OF_CONDUCT/#scope","title":"Scope","text":"<p>This Code of Conduct applies both within project spaces and in public spaces when an individual is representing the project or its community. Examples of representing a project or community include using an official project e-mail address, posting via an official social media account, or acting as an appointed representative at an online or offline event. Representation of a project may be further defined and clarified by project maintainers.</p>"},{"location":"CODE_OF_CONDUCT/#enforcement","title":"Enforcement","text":"<p>Instances of abusive, harassing, or otherwise unacceptable behavior may be reported by contacting the project team at ml-agents@unity3d.com. All complaints will be reviewed and investigated and will result in a response that is deemed necessary and appropriate to the circumstances. The project team is obligated to maintain confidentiality with regard to the reporter of an incident. Further details of specific enforcement policies may be posted separately.</p> <p>Project maintainers who do not follow or enforce the Code of Conduct in good faith may face temporary or permanent repercussions as determined by other members of the project's leadership.</p>"},{"location":"CODE_OF_CONDUCT/#attribution","title":"Attribution","text":"<p>This Code of Conduct is adapted from the Contributor Covenant, version 1.4, available at https://www.contributor-covenant.org/version/1/4/code-of-conduct/</p>"},{"location":"CONTRIBUTING/","title":"How to Contribute to ML-Agents","text":""},{"location":"CONTRIBUTING/#1fork-the-repository","title":"1.Fork the repository","text":"<p>Fork the ML-Agents repository by clicking on the \"Fork\" button in the top right corner of the GitHub page. This creates a copy of the repository under your GitHub account.</p>"},{"location":"CONTRIBUTING/#2-set-up-your-development-environment","title":"2. Set up your development environment","text":"<p>Clone the forked repository to your local machine using Git. Install the necessary dependencies and follow the instructions provided in the project's documentation to set up your development environment properly.</p>"},{"location":"CONTRIBUTING/#3-choose-an-issue-or-feature","title":"3. Choose an issue or feature","text":"<p>Browse the project's issue tracker or discussions to find an open issue or feature that you would like to contribute to. Read the guidelines and comments associated with the issue to understand the requirements and constraints.</p>"},{"location":"CONTRIBUTING/#4-make-your-changes","title":"4. Make your changes","text":"<p>Create a new branch for your changes based on the main branch of the ML-Agents repository. Implement your code changes or add new features as necessary. Ensure that your code follows the project's coding style and conventions.</p> <ul> <li>Example: Let's say you want to add support for a new type of reward function in the ML-Agents framework. You can create a new branch named feature/reward-function to implement this feature.</li> </ul>"},{"location":"CONTRIBUTING/#5-test-your-changes","title":"5. Test your changes","text":"<p>Run the appropriate tests to ensure your changes work as intended. If necessary, add new tests to cover your code and verify that it doesn't introduce regressions.</p> <ul> <li>Example: For the reward function feature, you would write tests to check different scenarios and expected outcomes of the new reward function.</li> </ul>"},{"location":"CONTRIBUTING/#6-submit-a-pull-request","title":"6. Submit a pull request","text":"<p>Push your branch to your forked repository and submit a pull request (PR) to the ML-Agents main repository. Provide a clear and concise description of your changes, explaining the problem you solved or the feature you added.</p> <ul> <li>Example: In the pull request description, you would explain how the new reward function works, its benefits, and any relevant implementation details.</li> </ul>"},{"location":"CONTRIBUTING/#7-respond-to-feedback","title":"7. Respond to feedback","text":"<p>Be responsive to any feedback or comments provided by the project maintainers. Address the feedback by making necessary revisions to your code and continue the discussion if required.</p>"},{"location":"CONTRIBUTING/#8-continuous-integration-and-code-review","title":"8. Continuous integration and code review","text":"<p>The ML-Agents project utilizes automated continuous integration (CI) systems to run tests on pull requests. Address any issues flagged by the CI system and actively participate in the code review process by addressing comments from reviewers.</p>"},{"location":"CONTRIBUTING/#9-merge-your-changes","title":"9. Merge your changes","text":"<p>Once your pull request has been approved and meets all the project's requirements, a project maintainer will merge your changes into the main repository. Congratulations, your contribution has been successfully integrated!</p> <p>Remember to always adhere to the project's code of conduct, be respectful, and follow any specific contribution guidelines provided by the ML-Agents project. Happy contributing!</p>"},{"location":"Custom-SideChannels/","title":"Custom Side Channels","text":"<p>You can create your own side channel in C# and Python and use it to communicate custom data structures between the two. This can be useful for situations in which the data to be sent is too complex or structured for the built-in <code>EnvironmentParameters</code>, or is not related to any specific agent, and therefore inappropriate as an agent observation.</p>"},{"location":"Custom-SideChannels/#overview","title":"Overview","text":"<p>In order to use a side channel, it must be implemented as both Unity and Python classes.</p>"},{"location":"Custom-SideChannels/#unity-side","title":"Unity side","text":"<p>The side channel will have to implement the <code>SideChannel</code> abstract class and the following method.</p> <ul> <li><code>OnMessageReceived(IncomingMessage msg)</code> : You must implement this method and   read the data from IncomingMessage. The data must be read in the order that it   was written.</li> </ul> <p>The side channel must also assign a <code>ChannelId</code> property in the constructor. The <code>ChannelId</code> is a Guid (or UUID in Python) used to uniquely identify a side channel. This Guid must be the same on C# and Python. There can only be one side channel of a certain id during communication.</p> <p>To send data from C# to Python, create an <code>OutgoingMessage</code> instance, add data to it, call the <code>base.QueueMessageToSend(msg)</code> method inside the side channel, and call the <code>OutgoingMessage.Dispose()</code> method.</p> <p>To register a side channel on the Unity side, call <code>SideChannelManager.RegisterSideChannel</code> with the side channel as only argument.</p>"},{"location":"Custom-SideChannels/#python-side","title":"Python side","text":"<p>The side channel will have to implement the <code>SideChannel</code> abstract class. You must implement :</p> <ul> <li><code>on_message_received(self, msg: \"IncomingMessage\") -&gt; None</code> : You must   implement this method and read the data from IncomingMessage. The data must be   read in the order that it was written.</li> </ul> <p>The side channel must also assign a <code>channel_id</code> property in the constructor. The <code>channel_id</code> is a UUID (referred in C# as Guid) used to uniquely identify a side channel. This number must be the same on C# and Python. There can only be one side channel of a certain id during communication.</p> <p>To assign the <code>channel_id</code> call the abstract class constructor with the appropriate <code>channel_id</code> as follows:</p> <pre><code>super().__init__(my_channel_id)\n</code></pre> <p>To send a byte array from Python to C#, create an <code>OutgoingMessage</code> instance, add data to it, and call the <code>super().queue_message_to_send(msg)</code> method inside the side channel.</p> <p>To register a side channel on the Python side, pass the side channel as argument when creating the <code>UnityEnvironment</code> object. One of the arguments of the constructor (<code>side_channels</code>) is a list of side channels.</p>"},{"location":"Custom-SideChannels/#example-implementation","title":"Example implementation","text":"<p>Below is a simple implementation of a side channel that will exchange ASCII encoded strings between a Unity environment and Python.</p>"},{"location":"Custom-SideChannels/#example-unity-c-code","title":"Example Unity C# code","text":"<p>The first step is to create the <code>StringLogSideChannel</code> class within the Unity project. Here is an implementation of a <code>StringLogSideChannel</code> that will listen for messages from python and print them to the Unity debug log, as well as send error messages from Unity to python.</p> <pre><code>using UnityEngine;\nusing Unity.MLAgents;\nusing Unity.MLAgents.SideChannels;\nusing System.Text;\nusing System;\n\npublic class StringLogSideChannel : SideChannel\n{\n    public StringLogSideChannel()\n    {\n        ChannelId = new Guid(\"621f0a70-4f87-11ea-a6bf-784f4387d1f7\");\n    }\n\n    protected override void OnMessageReceived(IncomingMessage msg)\n    {\n        var receivedString = msg.ReadString();\n        Debug.Log(\"From Python : \" + receivedString);\n    }\n\n    public void SendDebugStatementToPython(string logString, string stackTrace, LogType type)\n    {\n        if (type == LogType.Error)\n        {\n            var stringToSend = type.ToString() + \": \" + logString + \"\\n\" + stackTrace;\n            using (var msgOut = new OutgoingMessage())\n            {\n                msgOut.WriteString(stringToSend);\n                QueueMessageToSend(msgOut);\n            }\n        }\n    }\n}\n</code></pre> <p>Once we have defined our custom side channel class, we need to ensure that it is instantiated and registered. This can typically be done wherever the logic of the side channel makes sense to be associated, for example on a MonoBehaviour object that might need to access data from the side channel. Here we show a simple MonoBehaviour object which instantiates and registers the new side channel. If you have not done it already, make sure that the MonoBehaviour which registers the side channel is attached to a GameObject which will be live in your Unity scene.</p> <pre><code>using UnityEngine;\nusing Unity.MLAgents;\n\n\npublic class RegisterStringLogSideChannel : MonoBehaviour\n{\n\n    StringLogSideChannel stringChannel;\n    public void Awake()\n    {\n        // We create the Side Channel\n        stringChannel = new StringLogSideChannel();\n\n        // When a Debug.Log message is created, we send it to the stringChannel\n        Application.logMessageReceived += stringChannel.SendDebugStatementToPython;\n\n        // The channel must be registered with the SideChannelManager class\n        SideChannelManager.RegisterSideChannel(stringChannel);\n    }\n\n    public void OnDestroy()\n    {\n        // De-register the Debug.Log callback\n        Application.logMessageReceived -= stringChannel.SendDebugStatementToPython;\n        if (Academy.IsInitialized){\n            SideChannelManager.UnregisterSideChannel(stringChannel);\n        }\n    }\n\n    public void Update()\n    {\n        // Optional : If the space bar is pressed, raise an error !\n        if (Input.GetKeyDown(KeyCode.Space))\n        {\n            Debug.LogError(\"This is a fake error. Space bar was pressed in Unity.\");\n        }\n    }\n}\n</code></pre>"},{"location":"Custom-SideChannels/#example-python-code","title":"Example Python code","text":"<p>Now that we have created the necessary Unity C# classes, we can create their Python counterparts.</p> <pre><code>from mlagents_envs.environment import UnityEnvironment\nfrom mlagents_envs.side_channel.side_channel import (\n    SideChannel,\n    IncomingMessage,\n    OutgoingMessage,\n)\nimport numpy as np\nimport uuid\n\n\n# Create the StringLogChannel class\nclass StringLogChannel(SideChannel):\n\n    def __init__(self) -&gt; None:\n        super().__init__(uuid.UUID(\"621f0a70-4f87-11ea-a6bf-784f4387d1f7\"))\n\n    def on_message_received(self, msg: IncomingMessage) -&gt; None:\n        \"\"\"\n        Note: We must implement this method of the SideChannel interface to\n        receive messages from Unity\n        \"\"\"\n        # We simply read a string from the message and print it.\n        print(msg.read_string())\n\n    def send_string(self, data: str) -&gt; None:\n        # Add the string to an OutgoingMessage\n        msg = OutgoingMessage()\n        msg.write_string(data)\n        # We call this method to queue the data we want to send\n        super().queue_message_to_send(msg)\n</code></pre> <p>We can then instantiate the new side channel, launch a <code>UnityEnvironment</code> with that side channel active, and send a series of messages to the Unity environment from Python using it.</p> <pre><code># Create the channel\nstring_log = StringLogChannel()\n\n# We start the communication with the Unity Editor and pass the string_log side channel as input\nenv = UnityEnvironment(side_channels=[string_log])\nenv.reset()\nstring_log.send_string(\"The environment was reset\")\n\ngroup_name = list(env.behavior_specs.keys())[0]  # Get the first group_name\ngroup_spec = env.behavior_specs[group_name]\nfor i in range(1000):\n    decision_steps, terminal_steps = env.get_steps(group_name)\n    # We send data to Unity : A string with the number of Agent at each\n    string_log.send_string(\n        f\"Step {i} occurred with {len(decision_steps)} deciding agents and \"\n        f\"{len(terminal_steps)} terminal agents\"\n    )\n    env.step()  # Move the simulation forward\n\nenv.close()\n</code></pre> <p>Now, if you run this script and press <code>Play</code> the Unity Editor when prompted, the console in the Unity Editor will display a message at every Python step. Additionally, if you press the Space Bar in the Unity Engine, a message will appear in the terminal.</p>"},{"location":"ELO-Rating-System/","title":"ELO Rating System","text":"<p>In adversarial games, the cumulative environment reward may not be a meaningful metric by which to track learning progress.</p> <p>This is because the cumulative reward is entirely dependent on the skill of the opponent.</p> <p>An agent at a particular skill level will get more or less reward against a worse or better agent, respectively.</p> <p>Instead, it's better to use ELO rating system, a method to calculate the relative skill level between two players in a zero-sum game.</p> <p>If the training performs correctly, this value should steadily increase.</p>"},{"location":"ELO-Rating-System/#what-is-a-zero-sum-game","title":"What is a zero-sum game?","text":"<p>A zero-sum game is a game where each player's gain or loss of utility is exactly balanced by the gain or loss of the utility of the opponent.</p> <p>Simply explained, we face a zero-sum game when one agent gets +1.0, its opponent gets -1.0 reward.</p> <p>For instance, Tennis is a zero-sum game: if you win the point you get +1.0 and your opponent gets -1.0 reward.</p>"},{"location":"ELO-Rating-System/#how-works-the-elo-rating-system","title":"How works the ELO Rating System","text":"<ul> <li> <p>Each player has an initial ELO score. It's defined in the <code>initial_elo</code> trainer config hyperparameter.</p> </li> <li> <p>The difference in rating between the two players serves as the predictor of the outcomes of a match.</p> </li> </ul> <p> For instance, if player A has an Elo score of 2100 and player B has an ELO score of 1800 the chance that player A wins is 85% against 15% for player b.</p> <ul> <li>We calculate the expected score of each player using this formula:</li> </ul> <p></p> <ul> <li>At the end of the game, based on the outcome we update the player\u2019s actual Elo score, we use a linear adjustment proportional to the amount by which the player over-performed or under-performed. The winning player takes points from the losing one:</li> <li>If the higher-rated player wins \u2192 a few points will be taken from the lower-rated player.</li> <li>If the lower-rated player wins \u2192 a lot of points will be taken from the high-rated player.</li> <li> <p>If it\u2019s a draw \u2192 the lower-rated player gains a few points from higher.</p> </li> <li> <p>We update players rating using this formula:</p> </li> </ul> <p></p>"},{"location":"ELO-Rating-System/#the-tennis-example","title":"The Tennis example","text":"<ul> <li>We start to train our agents.</li> <li>Both of them have the same skills. So ELO score for each of them that we defined using parameter <code>initial_elo = 1200.0</code>.</li> </ul> <p>We calculate the expected score E: Ea = 0.5 Eb = 0.5</p> <p>So it means that each player has 50% chances of winning the point.</p> <p>If A wins, the new rating R would be:</p> <p>Ra = 1200 + 16 * (1 - 0.5) \u2192 1208</p> <p>Rb = 1200 + 16 * (0 - 0.5) \u2192 1192</p> <p>Player A has now an ELO score of 1208 and Player B an ELO score of 1192. Therefore, Player A is now a little bit better than Player B.</p>"},{"location":"FAQ/","title":"Frequently Asked Questions","text":""},{"location":"FAQ/#installation-problems","title":"Installation problems","text":""},{"location":"FAQ/#environment-permission-error","title":"Environment Permission Error","text":"<p>If you directly import your Unity environment without building it in the editor, you might need to give it additional permissions to execute it.</p> <p>If you receive such a permission error on macOS, run:</p> <pre><code>chmod -R 755 *.app\n</code></pre> <p>or on Linux:</p> <pre><code>chmod -R 755 *.x86_64\n</code></pre> <p>On Windows, you can find instructions.</p>"},{"location":"FAQ/#environment-connection-timeout","title":"Environment Connection Timeout","text":"<p>If you are able to launch the environment from <code>UnityEnvironment</code> but then receive a timeout error like this:</p> <pre><code>UnityAgentsException: The Communicator was unable to connect. Please make sure the External process is ready to accept communication with Unity.\n</code></pre> <p>There may be a number of possible causes:</p> <ul> <li>Cause: There may be no agent in the scene</li> <li>Cause: On OSX, the firewall may be preventing communication with the   environment. Solution: Add the built environment binary to the list of   exceptions on the firewall by following   instructions.</li> <li>Cause: An error happened in the Unity Environment preventing communication.   Solution: Look into the   log files generated by the   Unity Environment to figure what error happened.</li> <li>Cause: You have assigned <code>HTTP_PROXY</code> and <code>HTTPS_PROXY</code> values in your   environment variables. Solution: Remove these values and try again.</li> <li>Cause: You are running in a headless environment (e.g. remotely connected   to a server). Solution: Pass <code>--no-graphics</code> to <code>mlagents-learn</code>, or   <code>no_graphics=True</code> to <code>RemoteRegistryEntry.make()</code> or the <code>UnityEnvironment</code>   initializer. If you need graphics for visual observations, you will need to   set up <code>xvfb</code> (or equivalent).</li> </ul>"},{"location":"FAQ/#communication-port-still-in-use","title":"Communication port {} still in use","text":"<p>If you receive an exception <code>\"Couldn't launch new environment because communication port {} is still in use. \"</code>, you can change the worker number in the Python script when calling</p> <pre><code>UnityEnvironment(file_name=filename, worker_id=X)\n</code></pre>"},{"location":"FAQ/#mean-reward-nan","title":"Mean reward : nan","text":"<p>If you receive a message <code>Mean reward : nan</code> when attempting to train a model using PPO, this is due to the episodes of the Learning Environment not terminating. In order to address this, set <code>Max Steps</code> for the Agents within the Scene Inspector to a value greater than 0. Alternatively, it is possible to manually set <code>done</code> conditions for episodes from within scripts for custom episode-terminating events.</p>"},{"location":"FAQ/#file-name-cannot-be-opened-because-the-developer-cannot-be-verified","title":"\"File name\" cannot be opened because the developer cannot be verified.","text":"<p>If you have downloaded the repository using the github website on macOS 10.15 (Catalina) or later, you may see this error when attempting to play scenes in the Unity project. Workarounds include installing the package using the Unity Package Manager (this is the officially supported approach - see here), or following the instructions here to verify the relevant files on your machine on a file-by-file basis.</p>"},{"location":"Getting-Started/","title":"Getting Started Guide","text":"<p>This guide walks through the end-to-end process of opening one of our example environments in Unity, training an Agent in it, and embedding the trained model into the Unity environment. After reading this tutorial, you should be able to train any of the example environments. If you are not familiar with the Unity Engine, view our Background: Unity page for helpful pointers. Additionally, if you're not familiar with machine learning, view our Background: Machine Learning page for a brief overview and helpful pointers.</p> <p></p> <p>For this guide, we'll use the 3D Balance Ball environment which contains a number of agent cubes and balls (which are all copies of each other). Each agent cube tries to keep its ball from falling by rotating either horizontally or vertically. In this environment, an agent cube is an Agent that receives a reward for every step that it balances the ball. An agent is also penalized with a negative reward for dropping the ball. The goal of the training process is to have the agents learn to balance the ball on their head.</p> <p>Let's get started!</p>"},{"location":"Getting-Started/#installation","title":"Installation","text":"<p>If you haven't already, follow the installation instructions. Afterwards, open the Unity Project that contains all the example environments:</p> <ol> <li>Open the Package Manager Window by navigating to <code>Window -&gt; Package Manager</code>    in the menu.</li> <li>Navigate to the ML-Agents Package and click on it.</li> <li>Find the <code>3D Ball</code> sample and click <code>Import</code>.</li> <li>In the Project window, go to the    <code>Assets/ML-Agents/Examples/3DBall/Scenes</code> folder and open the <code>3DBall</code> scene    file.</li> </ol>"},{"location":"Getting-Started/#understanding-a-unity-environment","title":"Understanding a Unity Environment","text":"<p>An agent is an autonomous actor that observes and interacts with an environment. In the context of Unity, an environment is a scene containing one or more Agent objects, and, of course, the other entities that an agent interacts with.</p> <p></p> <p>Note: In Unity, the base object of everything in a scene is the GameObject. The GameObject is essentially a container for everything else, including behaviors, graphics, physics, etc. To see the components that make up a GameObject, select the GameObject in the Scene window, and open the Inspector window. The Inspector shows every component on a GameObject.</p> <p>The first thing you may notice after opening the 3D Balance Ball scene is that it contains not one, but several agent cubes. Each agent cube in the scene is an independent agent, but they all share the same Behavior. 3D Balance Ball does this to speed up training since all twelve agents contribute to training in parallel.</p>"},{"location":"Getting-Started/#agent","title":"Agent","text":"<p>The Agent is the actor that observes and takes actions in the environment. In the 3D Balance Ball environment, the Agent components are placed on the twelve \"Agent\" GameObjects. The base Agent object has a few properties that affect its behavior:</p> <ul> <li>Behavior Parameters \u2014 Every Agent must have a Behavior. The Behavior   determines how an Agent makes decisions.</li> <li>Max Step \u2014 Defines how many simulation steps can occur before the Agent's   episode ends. In 3D Balance Ball, an Agent restarts after 5000 steps.</li> </ul>"},{"location":"Getting-Started/#behavior-parameters-vector-observation-space","title":"Behavior Parameters : Vector Observation Space","text":"<p>Before making a decision, an agent collects its observation about its state in the world. The vector observation is a vector of floating point numbers which contain relevant information for the agent to make decisions.</p> <p>The Behavior Parameters of the 3D Balance Ball example uses a <code>Space Size</code> of 8. This means that the feature vector containing the Agent's observations contains eight elements: the <code>x</code> and <code>z</code> components of the agent cube's rotation and the <code>x</code>, <code>y</code>, and <code>z</code> components of the ball's relative position and velocity.</p>"},{"location":"Getting-Started/#behavior-parameters-actions","title":"Behavior Parameters : Actions","text":"<p>An Agent is given instructions in the form of actions. ML-Agents Toolkit classifies actions into two types: continuous and discrete. The 3D Balance Ball example is programmed to use continuous actions, which are a vector of floating-point numbers that can vary continuously. More specifically, it uses a <code>Space Size</code> of 2 to control the amount of <code>x</code> and <code>z</code> rotations to apply to itself to keep the ball balanced on its head.</p>"},{"location":"Getting-Started/#running-a-pre-trained-model","title":"Running a pre-trained model","text":"<p>We include pre-trained models for our agents (<code>.onnx</code> files) and we use the Sentis to run these models inside Unity. In this section, we will use the pre-trained model for the 3D Ball example.</p> <ol> <li>In the Project window, go to the    <code>Assets/ML-Agents/Examples/3DBall/Prefabs</code> folder. Expand <code>3DBall</code> and click    on the <code>Agent</code> prefab. You should see the <code>Agent</code> prefab in the Inspector    window.</li> </ol> <p>Note: The platforms in the <code>3DBall</code> scene were created using the <code>3DBall</code>    prefab. Instead of updating all 12 platforms individually, you can update the    <code>3DBall</code> prefab instead.</p> <p></p> <ol> <li>In the Project window, drag the 3DBall Model located in    <code>Assets/ML-Agents/Examples/3DBall/TFModels</code> into the <code>Model</code> property under    <code>Behavior Parameters (Script)</code> component in the Agent GameObject    Inspector window.</li> </ol> <p></p> <ol> <li>You should notice that each <code>Agent</code> under each <code>3DBall</code> in the Hierarchy    windows now contains 3DBall as <code>Model</code> on the <code>Behavior Parameters</code>.    Note : You can modify multiple game objects in a scene by selecting them    all at once using the search bar in the Scene Hierarchy.</li> <li>Set the Inference Device to use for this model as <code>CPU</code>.</li> <li>Click the Play button in the Unity Editor and you will see the platforms    balance the balls using the pre-trained model.</li> </ol>"},{"location":"Getting-Started/#training-a-new-model-with-reinforcement-learning","title":"Training a new model with Reinforcement Learning","text":"<p>While we provide pre-trained models for the agents in this environment, any environment you make yourself will require training agents from scratch to generate a new model file. In this section we will demonstrate how to use the reinforcement learning algorithms that are part of the ML-Agents Python package to accomplish this. We have provided a convenient command <code>mlagents-learn</code> which accepts arguments used to configure both training and inference phases.</p>"},{"location":"Getting-Started/#training-the-environment","title":"Training the environment","text":"<ol> <li>Open a command or terminal window.</li> <li>Navigate to the folder where you cloned the <code>ml-agents</code> repository. Note:    If you followed the default installation, then you should    be able to run <code>mlagents-learn</code> from any directory.</li> <li>Run <code>mlagents-learn config/ppo/3DBall.yaml --run-id=first3DBallRun</code>.</li> <li><code>config/ppo/3DBall.yaml</code> is the path to a default training      configuration file that we provide. The <code>config/ppo</code> folder includes training configuration      files for all our example environments, including 3DBall.</li> <li><code>run-id</code> is a unique name for this training session.</li> <li>When the message \"Start training by pressing the Play button in the Unity    Editor\" is displayed on the screen, you can press the Play button in    Unity to start training in the Editor.</li> </ol> <p>If <code>mlagents-learn</code> runs correctly and starts training, you should see something like this:</p> <pre><code>INFO:mlagents_envs:\n'Ball3DAcademy' started successfully!\nUnity Academy name: Ball3DAcademy\n\nINFO:mlagents_envs:Connected new brain:\nUnity brain name: 3DBallLearning\n        Number of Visual Observations (per agent): 0\n        Vector Observation space size (per agent): 8\n        Number of stacked Vector Observation: 1\nINFO:mlagents_envs:Hyperparameters for the PPO Trainer of brain 3DBallLearning:\n        batch_size:          64\n        beta:                0.001\n        buffer_size:         12000\n        epsilon:             0.2\n        gamma:               0.995\n        hidden_units:        128\n        lambd:               0.99\n        learning_rate:       0.0003\n        max_steps:           5.0e4\n        normalize:           True\n        num_epoch:           3\n        num_layers:          2\n        time_horizon:        1000\n        sequence_length:     64\n        summary_freq:        1000\n        use_recurrent:       False\n        memory_size:         256\n        use_curiosity:       False\n        curiosity_strength:  0.01\n        curiosity_enc_size:  128\n        output_path: ./results/first3DBallRun/3DBallLearning\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 1000. Mean Reward: 1.242. Std of Reward: 0.746. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 2000. Mean Reward: 1.319. Std of Reward: 0.693. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 3000. Mean Reward: 1.804. Std of Reward: 1.056. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 4000. Mean Reward: 2.151. Std of Reward: 1.432. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 5000. Mean Reward: 3.175. Std of Reward: 2.250. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 6000. Mean Reward: 4.898. Std of Reward: 4.019. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 7000. Mean Reward: 6.716. Std of Reward: 5.125. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 8000. Mean Reward: 12.124. Std of Reward: 11.929. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 9000. Mean Reward: 18.151. Std of Reward: 16.871. Training.\nINFO:mlagents.trainers: first3DBallRun: 3DBallLearning: Step: 10000. Mean Reward: 27.284. Std of Reward: 28.667. Training.\n</code></pre> <p>Note how the <code>Mean Reward</code> value printed to the screen increases as training progresses. This is a positive sign that training is succeeding.</p> <p>Note: You can train using an executable rather than the Editor. To do so, follow the instructions in Using an Executable.</p>"},{"location":"Getting-Started/#observing-training-progress","title":"Observing Training Progress","text":"<p>Once you start training using <code>mlagents-learn</code> in the way described in the previous section, the <code>ml-agents</code> directory will contain a <code>results</code> directory. In order to observe the training process in more detail, you can use TensorBoard. From the command line run:</p> <pre><code>tensorboard --logdir results\n</code></pre> <p>Then navigate to <code>localhost:6006</code> in your browser to view the TensorBoard summary statistics as shown below. For the purposes of this section, the most important statistic is <code>Environment/Cumulative Reward</code> which should increase throughout training, eventually converging close to <code>100</code> which is the maximum reward the agent can accumulate.</p> <p></p>"},{"location":"Getting-Started/#embedding-the-model-into-the-unity-environment","title":"Embedding the model into the Unity Environment","text":"<p>Once the training process completes, and the training process saves the model (denoted by the <code>Saved Model</code> message) you can add it to the Unity project and use it with compatible Agents (the Agents that generated the model). Note: Do not just close the Unity Window once the <code>Saved Model</code> message appears. Either wait for the training process to close the window or press <code>Ctrl+C</code> at the command-line prompt. If you close the window manually, the <code>.onnx</code> file containing the trained model is not exported into the ml-agents folder.</p> <p>If you've quit the training early using <code>Ctrl+C</code> and want to resume training, run the same command again, appending the <code>--resume</code> flag:</p> <pre><code>mlagents-learn config/ppo/3DBall.yaml --run-id=first3DBallRun --resume\n</code></pre> <p>Your trained model will be at <code>results/&lt;run-identifier&gt;/&lt;behavior_name&gt;.onnx</code> where <code>&lt;behavior_name&gt;</code> is the name of the <code>Behavior Name</code> of the agents corresponding to the model. This file corresponds to your model's latest checkpoint. You can now embed this trained model into your Agents by following the steps below, which is similar to the steps described above.</p> <ol> <li>Move your model file into    <code>Project/Assets/ML-Agents/Examples/3DBall/TFModels/</code>.</li> <li>Open the Unity Editor, and select the 3DBall scene as described above.</li> <li>Select the 3DBall prefab Agent object.</li> <li>Drag the <code>&lt;behavior_name&gt;.onnx</code> file from the Project window of the Editor to    the Model placeholder in the Ball3DAgent inspector window.</li> <li>Press the Play button at the top of the Editor.</li> </ol>"},{"location":"Getting-Started/#next-steps","title":"Next Steps","text":"<ul> <li>For more information on the ML-Agents Toolkit, in addition to helpful   background, check out the ML-Agents Toolkit Overview   page.</li> <li>For a \"Hello World\" introduction to creating your own Learning Environment,   check out the   Making a New Learning Environment page.</li> <li>For an overview on the more complex example environments that are provided in   this toolkit, check out the   Example Environments page.</li> <li>For more information on the various training options available, check out the   Training ML-Agents page.</li> </ul>"},{"location":"Glossary/","title":"ML-Agents Toolkit Glossary","text":"<ul> <li>Academy - Singleton object which controls timing, reset, and   training/inference settings of the environment.</li> <li>Action - The carrying-out of a decision on the part of an agent within the   environment.</li> <li>Agent - Unity Component which produces observations and takes actions in   the environment. Agents actions are determined by decisions produced by a   Policy.</li> <li>Decision - The specification produced by a Policy for an action to be   carried out given an observation.</li> <li>Editor - The Unity Editor, which may include any pane (e.g. Hierarchy,   Scene, Inspector).</li> <li>Environment - The Unity scene which contains Agents.</li> <li>Experience - Corresponds to a tuple of [Agent observations, actions,   rewards] of a single Agent obtained after a Step.</li> <li>External Coordinator - ML-Agents class responsible for communication with   outside processes (in this case, the Python API).</li> <li>FixedUpdate - Unity method called each time the game engine is stepped.   ML-Agents logic should be placed here.</li> <li>Frame - An instance of rendering the main camera for the display.   Corresponds to each <code>Update</code> call of the game engine.</li> <li>Observation - Partial information describing the state of the environment   available to a given agent. (e.g. Vector, Visual)</li> <li>Policy - The decision making mechanism for producing decisions from   observations, typically a neural network model.</li> <li>Reward - Signal provided at every step used to indicate desirability of an   agent\u2019s action within the current state of the environment.</li> <li>State - The underlying properties of the environment (including all agents   within it) at a given time.</li> <li>Step - Corresponds to an atomic change of the engine that happens between   Agent decisions.</li> <li>Trainer - Python class which is responsible for training a given group of   Agents.</li> <li>Update - Unity function called each time a frame is rendered. ML-Agents   logic should not be placed here.</li> </ul>"},{"location":"Hugging-Face-Integration/","title":"The Hugging Face Integration","text":"<p>The Hugging Face Hub \ud83e\udd17 is a central place where anyone can share and download models.</p> <p>It allows you to: - Host your trained models. - Download trained models from the community. - Visualize your agents playing directly on your browser.</p> <p>You can see the list of ml-agents models here.</p> <p>We wrote a complete tutorial to learn to train your first agent using ML-Agents and publish it to the Hub:</p> <ul> <li>A short tutorial where you teach Huggy the Dog to fetch the stick and then play with him directly in your browser</li> <li>A more in-depth tutorial</li> </ul>"},{"location":"Hugging-Face-Integration/#download-a-model-from-the-hub","title":"Download a model from the Hub","text":"<p>You can simply download a model from the Hub using <code>mlagents-load-from-hf</code>.</p> <p>You need to define two parameters:</p> <ul> <li><code>--repo-id</code>: the name of the Hugging Face repo you want to download.</li> <li><code>--local-dir</code>: the path to download the model.</li> </ul> <p>For instance, I want to load the model with model-id \"ThomasSimonini/MLAgents-Pyramids\" and put it in the downloads directory:</p> <pre><code>mlagents-load-from-hf --repo-id=\"ThomasSimonini/MLAgents-Pyramids\" --local-dir=\"./downloads\"\n</code></pre>"},{"location":"Hugging-Face-Integration/#upload-a-model-to-the-hub","title":"Upload a model to the Hub","text":"<p>You can simply upload a model to the Hub using <code>mlagents-push-to-hf</code></p> <p>You need to define four parameters:</p> <ul> <li><code>--run-id</code>: the name of the training run id.</li> <li><code>--local-dir</code>: where the model was saved</li> <li><code>--repo-id</code>: the name of the Hugging Face repo you want to create or update. It\u2019s always / If the repo does not exist it will be created automatically <li><code>--commit-message</code>: since HF repos are git repositories you need to give a commit message.</li> <p>For instance, I want to upload my model trained with run-id \"SnowballTarget1\" to the repo-id: ThomasSimonini/ppo-SnowballTarget:</p> <pre><code>  mlagents-push-to-hf --run-id=\"SnowballTarget1\" --local-dir=\"./results/SnowballTarget1\" --repo-id=\"ThomasSimonini/ppo-SnowballTarget\" --commit-message=\"First Push\"\n</code></pre>"},{"location":"Hugging-Face-Integration/#visualize-an-agent-playing","title":"Visualize an agent playing","text":"<p>You can watch your agent playing directly in your browser (if the environment is from the ML-Agents official environments)</p> <ul> <li>Step 1: Go to https://huggingface.co/unity and select the environment demo.</li> <li>Step 2: Find your model_id in the list.</li> <li>Step 3: Select your .nn /.onnx file.</li> <li>Step 4: Click on Watch the agent play</li> </ul>"},{"location":"Installation-Anaconda-Windows/","title":"Installing ML-Agents Toolkit for Windows (Deprecated)","text":"<p>:warning: Note: We no longer use this guide ourselves and so it may not work correctly. We've decided to keep it up just in case it is helpful to you.</p> <p>The ML-Agents Toolkit supports Windows 10. While it might be possible to run the ML-Agents Toolkit using other versions of Windows, it has not been tested on other versions. Furthermore, the ML-Agents Toolkit has not been tested on a Windows VM such as Bootcamp or Parallels.</p> <p>To use the ML-Agents Toolkit, you install Python and the required Python packages as outlined below. This guide also covers how set up GPU-based training (for advanced users). GPU-based training is not currently required for the ML-Agents Toolkit. However, training on a GPU might be required by future versions and features.</p>"},{"location":"Installation-Anaconda-Windows/#step-1-install-python-via-anaconda","title":"Step 1: Install Python via Anaconda","text":"<p>Download and install Anaconda for Windows. By using Anaconda, you can manage separate environments for different distributions of Python. Python 3.7.2 or higher is required as we no longer support Python 2. In this guide, we are using Python version 3.7 and Anaconda version 5.1 (64-bit or 32-bit direct links).</p> <p> </p> <p>We recommend the default advanced installation options. However, select the options appropriate for your specific situation.</p> <p> </p> <p>After installation, you must open Anaconda Navigator to finish the setup. From the Windows search bar, type anaconda navigator. You can close Anaconda Navigator after it opens.</p> <p>If environment variables were not created, you will see error \"conda is not recognized as internal or external command\" when you type <code>conda</code> into the command line. To solve this you will need to set the environment variable correctly.</p> <p>Type <code>environment variables</code> in the search bar (this can be reached by hitting the Windows key or the bottom left Windows button). You should see an option called Edit the system environment variables.</p> <p> </p> <p>From here, click the Environment Variables button. Double click \"Path\" under System variable to edit the \"Path\" variable, click New to add the following new paths.</p> <pre><code>%UserProfile%\\Anaconda3\\Scripts\n%UserProfile%\\Anaconda3\\Scripts\\conda.exe\n%UserProfile%\\Anaconda3\n%UserProfile%\\Anaconda3\\python.exe\n</code></pre>"},{"location":"Installation-Anaconda-Windows/#step-2-setup-and-activate-a-new-conda-environment","title":"Step 2: Setup and Activate a New Conda Environment","text":"<p>You will create a new Conda environment to be used with the ML-Agents Toolkit. This means that all the packages that you install are localized to just this environment. It will not affect any other installation of Python or other environments. Whenever you want to run ML-Agents, you will need activate this Conda environment.</p> <p>To create a new Conda environment, open a new Anaconda Prompt (Anaconda Prompt in the search bar) and type in the following command:</p> <pre><code>conda create -n ml-agents python=3.7\n</code></pre> <p>You may be asked to install new packages. Type <code>y</code> and press enter (make sure you are connected to the Internet). You must install these required packages. The new Conda environment is called ml-agents and uses Python version 3.7.</p> <p> </p> <p>To use this environment, you must activate it. (To use this environment In the future, you can run the same command). In the same Anaconda Prompt, type in the following command:</p> <pre><code>activate ml-agents\n</code></pre> <p>You should see <code>(ml-agents)</code> prepended on the last line.</p> <p>Next, install <code>tensorflow</code>. Install this package using <code>pip</code> - which is a package management system used to install Python packages. Latest versions of TensorFlow won't work, so you will need to make sure that you install version 1.7.1. In the same Anaconda Prompt, type in the following command (make sure you are connected to the Internet):</p> <pre><code>pip install tensorflow==1.7.1\n</code></pre>"},{"location":"Installation-Anaconda-Windows/#step-3-install-required-python-packages","title":"Step 3: Install Required Python Packages","text":"<p>The ML-Agents Toolkit depends on a number of Python packages. Use <code>pip</code> to install these Python dependencies.</p> <p>If you haven't already, clone the ML-Agents Toolkit Github repository to your local computer. You can do this using Git (download here) and running the following commands in an Anaconda Prompt (if you open a new prompt, be sure to activate the ml-agents Conda environment by typing <code>activate ml-agents</code>):</p> <pre><code>git clone --branch release_21 https://github.com/Unity-Technologies/ml-agents.git\n</code></pre> <p>The <code>--branch release_21</code> option will switch to the tag of the latest stable release. Omitting that will get the <code>main</code> branch which is potentially unstable.</p> <p>If you don't want to use Git, you can find download links on the releases page.</p> <p>The <code>com.unity.ml-agents</code> subdirectory contains the core code to add to your projects. The <code>Project</code> subdirectory contains many example environments to help you get started.</p> <p>The <code>ml-agents</code> subdirectory contains a Python package which provides deep reinforcement learning trainers to use with Unity environments.</p> <p>The <code>ml-agents-envs</code> subdirectory contains a Python API to interface with Unity, which the <code>ml-agents</code> package depends on.</p> <p>The <code>gym-unity</code> subdirectory contains a package to interface with OpenAI Gym.</p> <p>Keep in mind where the files were downloaded, as you will need the trainer config files in this directory when running <code>mlagents-learn</code>. Make sure you are connected to the Internet and then type in the Anaconda Prompt:</p> <pre><code>python -m pip install mlagents==1.0.0\n</code></pre> <p>This will complete the installation of all the required Python packages to run the ML-Agents Toolkit.</p> <p>Sometimes on Windows, when you use pip to install certain Python packages, the pip will get stuck when trying to read the cache of the package. If you see this, you can try:</p> <pre><code>python -m pip install mlagents==1.0.0 --no-cache-dir\n</code></pre> <p>This <code>--no-cache-dir</code> tells the pip to disable the cache.</p>"},{"location":"Installation-Anaconda-Windows/#installing-for-development","title":"Installing for Development","text":"<p>If you intend to make modifications to <code>ml-agents</code> or <code>ml-agents-envs</code>, you should install the packages from the cloned repo rather than from PyPi. To do this, you will need to install <code>ml-agents</code> and <code>ml-agents-envs</code> separately.</p> <p>In our example, the files are located in <code>C:\\Downloads</code>. After you have either cloned or downloaded the files, from the Anaconda Prompt, change to the ml-agents subdirectory inside the ml-agents directory:</p> <pre><code>cd C:\\Downloads\\ml-agents\n</code></pre> <p>From the repo's main directory, now run:</p> <pre><code>cd ml-agents-envs\npip install -e .\ncd ..\ncd ml-agents\npip install -e .\n</code></pre> <p>Running pip with the <code>-e</code> flag will let you make changes to the Python files directly and have those reflected when you run <code>mlagents-learn</code>. It is important to install these packages in this order as the <code>mlagents</code> package depends on <code>mlagents_envs</code>, and installing it in the other order will download <code>mlagents_envs</code> from PyPi.</p>"},{"location":"Installation-Anaconda-Windows/#optional-step-4-gpu-training-using-the-ml-agents-toolkit","title":"(Optional) Step 4: GPU Training using The ML-Agents Toolkit","text":"<p>GPU is not required for the ML-Agents Toolkit and won't speed up the PPO algorithm a lot during training(but something in the future will benefit from GPU). This is a guide for advanced users who want to train using GPUs. Additionally, you will need to check if your GPU is CUDA compatible. Please check Nvidia's page here.</p> <p>Currently for the ML-Agents Toolkit, only CUDA v9.0 and cuDNN v7.0.5 is supported.</p>"},{"location":"Installation-Anaconda-Windows/#install-nvidia-cuda-toolkit","title":"Install Nvidia CUDA toolkit","text":"<p>Download and install the CUDA toolkit 9.0 from Nvidia's archive. The toolkit includes GPU-accelerated libraries, debugging and optimization tools, a C/C++ (Step Visual Studio 2017) compiler and a runtime library and is needed to run the ML-Agents Toolkit. In this guide, we are using version 9.0.176).</p> <p>Before installing, please make sure you close any running instances of Unity or Visual Studio.</p> <p>Run the installer and select the Express option. Note the directory where you installed the CUDA toolkit. In this guide, we installed in the directory <code>C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v9.0</code></p>"},{"location":"Installation-Anaconda-Windows/#install-nvidia-cudnn-library","title":"Install Nvidia cuDNN library","text":"<p>Download and install the cuDNN library from Nvidia. cuDNN is a GPU-accelerated library of primitives for deep neural networks. Before you can download, you will need to sign up for free to the Nvidia Developer Program.</p> <p> </p> <p>Once you've signed up, go back to the cuDNN downloads page. You may or may not be asked to fill out a short survey. When you get to the list cuDNN releases, make sure you are downloading the right version for the CUDA toolkit you installed in Step 1. In this guide, we are using version 7.0.5 for CUDA toolkit version 9.0 (direct link).</p> <p>After you have downloaded the cuDNN files, you will need to extract the files into the CUDA toolkit directory. In the cuDNN zip file, there are three folders called <code>bin</code>, <code>include</code>, and <code>lib</code>.</p> <p> </p> <p>Copy these three folders into the CUDA toolkit directory. The CUDA toolkit directory is located at <code>C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v9.0</code></p> <p> </p>"},{"location":"Installation-Anaconda-Windows/#set-environment-variables","title":"Set Environment Variables","text":"<p>You will need to add one environment variable and two path variables.</p> <p>To set the environment variable, type <code>environment variables</code> in the search bar (this can be reached by hitting the Windows key or the bottom left Windows button). You should see an option called Edit the system environment variables.</p> <p> </p> <p>From here, click the Environment Variables button. Click New to add a new system variable (make sure you do this under System variables and not User variables.</p> <p> </p> <p>For Variable Name, enter <code>CUDA_HOME</code>. For the variable value, put the directory location for the CUDA toolkit. In this guide, the directory location is <code>C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v9.0</code>. Press OK once.</p> <p> </p> <p>To set the two path variables, inside the same Environment Variables window and under the second box called System Variables, find a variable called <code>Path</code> and click Edit. You will add two directories to the list. For this guide, the two entries would look like:</p> <pre><code>C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v9.0\\lib\\x64\nC:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v9.0\\extras\\CUPTI\\libx64\n</code></pre> <p>Make sure to replace the relevant directory location with the one you have installed. Please note that case sensitivity matters.</p> <p> </p>"},{"location":"Installation-Anaconda-Windows/#install-tensorflow-gpu","title":"Install TensorFlow GPU","text":"<p>Next, install <code>tensorflow-gpu</code> using <code>pip</code>. You'll need version 1.7.1. In an Anaconda Prompt with the Conda environment ml-agents activated, type in the following command to uninstall TensorFlow for cpu and install TensorFlow for gpu (make sure you are connected to the Internet):</p> <pre><code>pip uninstall tensorflow\npip install tensorflow-gpu==1.7.1\n</code></pre> <p>Lastly, you should test to see if everything installed properly and that TensorFlow can identify your GPU. In the same Anaconda Prompt, open Python in the Prompt by calling:</p> <pre><code>python\n</code></pre> <p>And then type the following commands:</p> <pre><code>import tensorflow as tf\n\nsess = tf.Session(config=tf.ConfigProto(log_device_placement=True))\n</code></pre> <p>You should see something similar to:</p> <pre><code>Found device 0 with properties ...\n</code></pre>"},{"location":"Installation-Anaconda-Windows/#acknowledgments","title":"Acknowledgments","text":"<p>We would like to thank Jason Weimann and Nitish S. Mutha for writing the original articles which were used to create this guide.</p>"},{"location":"Installation/","title":"Installation","text":"<p>The ML-Agents Toolkit contains several components:</p> <ul> <li>Unity package (<code>com.unity.ml-agents</code>) contains the   Unity C# SDK that will be integrated into your Unity project.  This package contains   a sample to help you get started with ML-Agents.</li> <li>Unity package   (<code>com.unity.ml-agents.extensions</code>)   contains experimental C#/Unity components that are not yet ready to be part   of the base <code>com.unity.ml-agents</code> package. <code>com.unity.ml-agents.extensions</code>   has a direct dependency on <code>com.unity.ml-agents</code>.</li> <li>Two Python packages:</li> <li><code>mlagents</code> contains the machine learning algorithms that     enables you to train behaviors in your Unity scene. Most users of ML-Agents     will only need to directly install <code>mlagents</code>.</li> <li><code>mlagents_envs</code> contains a set of Python APIs to interact with     a Unity scene. It is a foundational layer that facilitates data messaging     between Unity scene and the Python machine learning algorithms.     Consequently, <code>mlagents</code> depends on <code>mlagents_envs</code>.</li> <li>Unity Project that contains several   example environments that highlight the   various features of the toolkit to help you get started.</li> </ul> <p>Consequently, to install and use the ML-Agents Toolkit you will need to:</p> <ul> <li>Install Unity (2023.2 or later)</li> <li>Install Python (3.10.12 or higher)</li> <li>Clone this repository (Recommended for the latest version and bug fixes)</li> <li>Note: If you do not clone the repository, then you will not be   able to access the example environments and training configurations or the   <code>com.unity.ml-agents.extensions</code> package. Additionally, the   Getting Started Guide assumes that you have cloned the   repository.</li> <li>Install the <code>com.unity.ml-agents</code> Unity package</li> <li>Install the <code>com.unity.ml-agents.extensions</code> Unity package (Optional)</li> <li>Install the <code>mlagents-envs</code></li> <li>Install the <code>mlagents</code> Python package</li> </ul>"},{"location":"Installation/#install-unity-20232-or-later","title":"Install Unity 2023.2 or Later","text":"<p>Download and install Unity. We strongly recommend that you install Unity through the Unity Hub as it will enable you to manage multiple Unity versions.</p>"},{"location":"Installation/#install-python-31012","title":"Install Python 3.10.12","text":"<p>We recommend installing Python 3.10.12. If you are using Windows, please install the x86-64 version and not x86. If your Python environment doesn't include <code>pip3</code>, see these instructions on installing it. We also recommend using conda or mamba to manage your python virtual environments.</p>"},{"location":"Installation/#conda-python-setup","title":"Conda python setup","text":"<p>Once conda has been installed in your system, open a terminal and execute the following commands to setup a python 3.10.12 virtual environment and activate it.</p> <pre><code>conda create -n mlagents python=3.10.12 &amp;&amp; conda activate mlagents\n</code></pre>"},{"location":"Installation/#clone-the-ml-agents-toolkit-repository-recommended","title":"Clone the ML-Agents Toolkit Repository (Recommended)","text":"<p>Now that you have installed Unity and Python, you can now install the Unity and Python packages. You do not need to clone the repository to install those packages, but you may choose to clone the repository if you'd like download our example environments and training configurations to experiment with them (some of our tutorials / guides assume you have access to our example environments).</p> <p>NOTE: There are samples shipped with the Unity Package.  You only need to clone the repository if you would like to explore more examples.</p> <pre><code>git clone --branch release_21 https://github.com/Unity-Technologies/ml-agents.git\n</code></pre> <p>The <code>--branch release_21</code> option will switch to the tag of the latest stable release. Omitting that will get the <code>main</code> branch which is potentially unstable.</p>"},{"location":"Installation/#advanced-local-installation-for-development","title":"Advanced: Local Installation for Development","text":"<p>You will need to clone the repository if you plan to modify or extend the ML-Agents Toolkit for your purposes. If you plan to contribute those changes back, make sure to clone the <code>develop</code> branch (by omitting <code>--branch release_21</code> from the command above). See our Contributions Guidelines for more information on contributing to the ML-Agents Toolkit.</p>"},{"location":"Installation/#install-the-comunityml-agents-unity-package","title":"Install the <code>com.unity.ml-agents</code> Unity package","text":"<p>The Unity ML-Agents C# SDK is a Unity Package. You can install the <code>com.unity.ml-agents</code> package directly from the Package Manager registry. Please make sure you enable 'Preview Packages' in the 'Advanced' dropdown in order to find the latest Preview release of the package.</p> <p>NOTE: If you do not see the ML-Agents package listed in the Package Manager please follow the advanced installation instructions below.</p>"},{"location":"Installation/#advanced-local-installation-for-development_1","title":"Advanced: Local Installation for Development","text":"<p>You can add the local <code>com.unity.ml-agents</code> package (from the repository that you just cloned) to your project by:</p> <ol> <li>navigating to the menu <code>Window</code> -&gt; <code>Package Manager</code>.</li> <li>In the package manager window click on the <code>+</code> button on the top left of the packages list).</li> <li>Select <code>Add package from disk...</code></li> <li>Navigate into the <code>com.unity.ml-agents</code> folder.</li> <li>Select the <code>package.json</code> file.</li> </ol> <p> </p> <p>If you are going to follow the examples from our documentation, you can open the <code>Project</code> folder in Unity and start tinkering immediately.</p>"},{"location":"Installation/#install-the-comunityml-agentsextensions-unity-package-optional","title":"Install the <code>com.unity.ml-agents.extensions</code> Unity package (Optional)","text":"<p>To install the <code>com.unity.ml-agents.extensions</code> package, you need to first clone the repo and then complete a local installation similar to what was outlined in the previous Advanced: Local Installation for Development section. Complete installation steps can be found in the package documentation.</p>"},{"location":"Installation/#install-the-mlagents-python-package","title":"Install the <code>mlagents</code> Python package","text":"<p>Installing the <code>mlagents</code> Python package involves installing other Python packages that <code>mlagents</code> depends on. So you may run into installation issues if your machine has older versions of any of those dependencies already installed. Consequently, our supported path for installing <code>mlagents</code> is to leverage Python Virtual Environments. Virtual Environments provide a mechanism for isolating the dependencies for each project and are supported on Mac / Windows / Linux. We offer a dedicated guide on Virtual Environments.</p>"},{"location":"Installation/#windows-installing-pytorch","title":"(Windows) Installing PyTorch","text":"<p>On Windows, you'll have to install the PyTorch package separately prior to installing ML-Agents in order to make sure the cuda-enabled version is used, rather than the CPU-only version. Activate your virtual environment and run from the command line:</p> <pre><code>pip3 install torch~=2.2.1 --index-url https://download.pytorch.org/whl/cu121\n</code></pre> <p>Note that on Windows, you may also need Microsoft's Visual C++ Redistributable if you don't have it already. See the PyTorch installation guide for more installation options and versions.</p>"},{"location":"Installation/#os-x-installing-grpc-libraries","title":"(OS X) Installing GRPC libraries","text":"<p>On OS X, you may need to explicitly install the GRPC runtime libraries to avoid hitting errors when training like <code>dlopen(/Users/alex.mccarthy/miniconda3/envs/mlagents/lib/python3.10/site-packages/grpc/_cython/cygrpc.cpython-310-darwin.so, 0x0002): symbol not found in flat namespace '_CFRelease'</code>.</p> <pre><code>pip3 install grpcio\n</code></pre>"},{"location":"Installation/#installing-mlagents","title":"Installing <code>mlagents</code>","text":"<p>To install the <code>mlagents</code> Python package, activate your virtual environment and run from the command line:</p> <pre><code>cd /path/to/ml-agents\npython -m pip install ./ml-agents-envs\npython -m pip install ./ml-agents\n</code></pre> <p>Note that this will install <code>mlagents</code> from the cloned repository, not from the PyPi repository. If you installed this correctly, you should be able to run <code>mlagents-learn --help</code>, after which you will see the command line parameters you can use with <code>mlagents-learn</code>.</p> <p>NOTE: Since ML-Agents development has slowed, PyPi releases will be less frequent. However, you can install from PyPi by executing the following command:</p> <pre><code>python -m pip install mlagents==1.0.0\n</code></pre> <p>which will install the latest version of ML-Agents and associated dependencies available on PyPi. Note, you need to have the matching version of the Unity packages with the particular release of the python packages. You can find the release history here</p> <p>By installing the <code>mlagents</code> package, the dependencies listed in the setup.py file are also installed. These include PyTorch.</p>"},{"location":"Installation/#advanced-local-installation-for-development_2","title":"Advanced: Local Installation for Development","text":"<p>If you intend to make modifications to <code>mlagents</code> or <code>mlagents_envs</code>, you should install the packages from the cloned repository rather than from PyPi. To do this, you will need to install <code>mlagents</code> and <code>mlagents_envs</code> separately. From the repository's root directory, run:</p> <pre><code>pip3 install torch -f https://download.pytorch.org/whl/torch_stable.html\npip3 install -e ./ml-agents-envs\npip3 install -e ./ml-agents\n</code></pre> <p>Running pip with the <code>-e</code> flag will let you make changes to the Python files directly and have those reflected when you run <code>mlagents-learn</code>. It is important to install these packages in this order as the <code>mlagents</code> package depends on <code>mlagents_envs</code>, and installing it in the other order will download <code>mlagents_envs</code> from PyPi.</p>"},{"location":"Installation/#next-steps","title":"Next Steps","text":"<p>The Getting Started guide contains several short tutorials on setting up the ML-Agents Toolkit within Unity, running a pre-trained model, in addition to building and training environments.</p>"},{"location":"Installation/#help","title":"Help","text":"<p>If you run into any problems regarding ML-Agents, refer to our FAQ and our Limitations pages. If you can't find anything please submit an issue and make sure to cite relevant information on OS, Python version, and exact error message (whenever possible).</p>"},{"location":"Integrations-Match3/","title":"Match-3 with ML-Agents","text":""},{"location":"Integrations-Match3/#getting-started","title":"Getting started","text":"<p>The C# code for Match-3 exists inside of the Unity package (<code>com.unity.ml-agents</code>). The good first step would be to take a look at how we have implemented the C# code in the example Match-3 scene (located under /Project/Assets/ML-Agents/Examples/match3). Once you have some familiarity, then the next step would be to implement the C# code for Match-3 from the extensions package.</p> <p>Additionally, see below for additional technical specifications on the C# code for Match-3. Please note the Match-3 game isn't human playable as implemented and can be only played via training.</p>"},{"location":"Integrations-Match3/#technical-specifications-for-match-3-with-ml-agents","title":"Technical specifications for Match-3 with ML-Agents","text":""},{"location":"Integrations-Match3/#abstractboard-class","title":"AbstractBoard class","text":"<p>The <code>AbstractBoard</code> is the bridge between ML-Agents and your game. It allows ML-Agents to * ask your game what the current and maximum sizes (rows, columns, and potential piece types) of the board are * ask your game what the \"color\" of a cell is * ask whether the cell is a \"special\" piece type or not * ask your game whether a move is allowed * request that your game make a move</p> <p>These are handled by implementing the abstract methods of <code>AbstractBoard</code>.</p>"},{"location":"Integrations-Match3/#public-abstract-boardsize-getmaxboardsize","title":"<code>public abstract BoardSize GetMaxBoardSize()</code>","text":"<p>Returns the largest <code>BoardSize</code> that the game can use. This is used to determine the sizes of observations and sensors, so don't make it larger than necessary.</p>"},{"location":"Integrations-Match3/#public-virtual-boardsize-getcurrentboardsize","title":"<code>public virtual BoardSize GetCurrentBoardSize()</code>","text":"<p>Returns the current size of the board. Each field on this BoardSize must be less than or equal to the corresponding field returned by <code>GetMaxBoardSize()</code>. This method is optional; if your always use the same size board, you don't need to override it.</p> <p>If the current board size is smaller than the maximum board size, <code>GetCellType()</code> and <code>GetSpecialType()</code> will not be called for cells outside the current board size, and <code>IsValidMove</code> won't be called for moves that would go outside of the current board size.</p>"},{"location":"Integrations-Match3/#public-abstract-int-getcelltypeint-row-int-col","title":"<code>public abstract int GetCellType(int row, int col)</code>","text":"<p>Returns the \"color\" of piece at the given row and column. This should be between 0 and BoardSize.NumCellTypes-1 (inclusive). The actual order of the values doesn't matter.</p>"},{"location":"Integrations-Match3/#public-abstract-int-getspecialtypeint-row-int-col","title":"<code>public abstract int GetSpecialType(int row, int col)</code>","text":"<p>Returns the special type of the piece at the given row and column. This should be between 0 and BoardSize.NumSpecialTypes (inclusive). The actual order of the values doesn't matter.</p>"},{"location":"Integrations-Match3/#public-abstract-bool-ismovevalidmove-m","title":"<code>public abstract bool IsMoveValid(Move m)</code>","text":"<p>Check whether the particular <code>Move</code> is valid for the game. The actual results will depend on the rules of the game, but we provide the <code>SimpleIsMoveValid()</code> method that handles basic match3 rules with no special or immovable pieces.</p>"},{"location":"Integrations-Match3/#public-abstract-bool-makemovemove-m","title":"<code>public abstract bool MakeMove(Move m)</code>","text":"<p>Instruct the game to make the given move. Returns true if the move was made. Note that during training, a move that was marked as invalid may occasionally still be requested. If this happens, it is safe to do nothing and request another move.</p>"},{"location":"Integrations-Match3/#move-struct","title":"<code>Move</code> struct","text":"<p>The Move struct encapsulates a swap of two adjacent cells. You can get the number of potential moves for a board of a given size with. <code>Move.NumPotentialMoves(maxBoardSize)</code>. There are two helper functions to create a new <code>Move</code>: * <code>public static Move FromMoveIndex(int moveIndex, BoardSize maxBoardSize)</code> can be used to iterate over all potential moves for the board by looping from 0 to <code>Move.NumPotentialMoves()</code> * <code>public static Move FromPositionAndDirection(int row, int col, Direction dir, BoardSize maxBoardSize)</code> creates a <code>Move</code> from a row, column, and direction (and board size).</p>"},{"location":"Integrations-Match3/#boardsize-struct","title":"<code>BoardSize</code> struct","text":"<p>Describes the \"size\" of the board, including the number of potential piece types that the board can have. This is returned by the AbstractBoard.GetMaxBoardSize() and GetCurrentBoardSize() methods.</p>"},{"location":"Integrations-Match3/#match3sensor-and-match3sensorcomponent-classes","title":"<code>Match3Sensor</code> and <code>Match3SensorComponent</code> classes","text":"<p>The <code>Match3Sensor</code> generates observations about the state using the <code>AbstractBoard</code> interface. You can choose whether to use vector or \"visual\" observations; in theory, visual observations should perform better because they are 2-dimensional like the board, but we need to experiment more on this.</p> <p>A <code>Match3SensorComponent</code> generates <code>Match3Sensor</code>s (the exact number of sensors depends on your configuration) at runtime, and should be added to the same GameObject as your <code>Agent</code> implementation. You do not need to write any additional code to use them.</p>"},{"location":"Integrations-Match3/#match3actuator-and-match3actuatorcomponent-classes","title":"<code>Match3Actuator</code> and <code>Match3ActuatorComponent</code> classes","text":"<p>The <code>Match3Actuator</code> converts actions from training or inference into a <code>Move</code> that is sent to<code>AbstractBoard.MakeMove()</code> It also checks <code>AbstractBoard.IsMoveValid</code> for each potential move and uses this to set the action mask for Agent.</p> <p>A <code>Match3ActuatorComponent</code> generates a <code>Match3Actuator</code> at runtime, and should be added to the same GameObject as your <code>Agent</code> implementation.  You do not need to write any additional code to use them.</p>"},{"location":"Integrations-Match3/#setting-up-match-3-simulation","title":"Setting up Match-3 simulation","text":"<ul> <li>Implement the <code>AbstractBoard</code> methods to integrate with your game.</li> <li>Give the <code>Agent</code> rewards when it does what you want it to (match multiple pieces in a row, clears pieces of a certain type, etc).</li> <li>Add the <code>Agent</code>, <code>AbstractBoard</code> implementation, <code>Match3SensorComponent</code>, and <code>Match3ActuatorComponent</code> to the same <code>GameObject</code>.</li> <li>Call <code>Agent.RequestDecision()</code> when you're ready for the <code>Agent</code> to make a move on the next <code>Academy</code> step. During the next <code>Academy</code> step, the <code>MakeMove()</code> method on the board will be called.</li> </ul>"},{"location":"Integrations-Match3/#implementation-details","title":"Implementation Details","text":""},{"location":"Integrations-Match3/#action-space","title":"Action Space","text":"<p>The indexing for actions is the same as described in Human Like Playtesting with Deep Learning (for example, Figure 2b). The horizontal moves are enumerated first, then the vertical ones. </p>"},{"location":"Integrations/","title":"Game Integrations","text":"<p>ML-Agents provides some utilities to make it easier to integrate with some common genres of games.</p>"},{"location":"Integrations/#match-3","title":"Match-3","text":"<p>The Match-3 integration provides an abstraction of a match-3 game board and moves, along with a sensor to observe the game state, and an actuator to translate the ML-Agent actions into game moves.</p>"},{"location":"Integrations/#interested-in-more-game-templates","title":"Interested in more game templates?","text":"<p>Do you have a type of game you are interested for ML-Agents? If so, please post a forum issue with <code>[GAME TEMPLATE]</code> in the title.</p>"},{"location":"LICENSE/","title":"LICENSE","text":"<p>Copyright 2017-2021 Unity Technologies</p> <p>Licensed under the Apache License, Version 2.0 (the \"License\"); you may not use this file except in compliance with the License. You may obtain a copy of the License at</p> <pre><code>http://www.apache.org/licenses/LICENSE-2.0\n</code></pre> <p>Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.</p> <pre><code>                             Apache License\n                       Version 2.0, January 2004\n                    http://www.apache.org/licenses/\n</code></pre> <p>TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION</p> <ol> <li> <p>Definitions.</p> <p>\"License\" shall mean the terms and conditions for use, reproduction,   and distribution as defined by Sections 1 through 9 of this document.</p> <p>\"Licensor\" shall mean the copyright owner or entity authorized by   the copyright owner that is granting the License.</p> <p>\"Legal Entity\" shall mean the union of the acting entity and all   other entities that control, are controlled by, or are under common   control with that entity. For the purposes of this definition,   \"control\" means (i) the power, direct or indirect, to cause the   direction or management of such entity, whether by contract or   otherwise, or (ii) ownership of fifty percent (50%) or more of the   outstanding shares, or (iii) beneficial ownership of such entity.</p> <p>\"You\" (or \"Your\") shall mean an individual or Legal Entity   exercising permissions granted by this License.</p> <p>\"Source\" form shall mean the preferred form for making modifications,   including but not limited to software source code, documentation   source, and configuration files.</p> <p>\"Object\" form shall mean any form resulting from mechanical   transformation or translation of a Source form, including but   not limited to compiled object code, generated documentation,   and conversions to other media types.</p> <p>\"Work\" shall mean the work of authorship, whether in Source or   Object form, made available under the License, as indicated by a   copyright notice that is included in or attached to the work   (an example is provided in the Appendix below).</p> <p>\"Derivative Works\" shall mean any work, whether in Source or Object   form, that is based on (or derived from) the Work and for which the   editorial revisions, annotations, elaborations, or other modifications   represent, as a whole, an original work of authorship. For the purposes   of this License, Derivative Works shall not include works that remain   separable from, or merely link (or bind by name) to the interfaces of,   the Work and Derivative Works thereof.</p> <p>\"Contribution\" shall mean any work of authorship, including   the original version of the Work and any modifications or additions   to that Work or Derivative Works thereof, that is intentionally   submitted to Licensor for inclusion in the Work by the copyright owner   or by an individual or Legal Entity authorized to submit on behalf of   the copyright owner. For the purposes of this definition, \"submitted\"   means any form of electronic, verbal, or written communication sent   to the Licensor or its representatives, including but not limited to   communication on electronic mailing lists, source code control systems,   and issue tracking systems that are managed by, or on behalf of, the   Licensor for the purpose of discussing and improving the Work, but   excluding communication that is conspicuously marked or otherwise   designated in writing by the copyright owner as \"Not a Contribution.\"</p> <p>\"Contributor\" shall mean Licensor and any individual or Legal Entity   on behalf of whom a Contribution has been received by Licensor and   subsequently incorporated within the Work.</p> </li> <li> <p>Grant of Copyright License. Subject to the terms and conditions of       this License, each Contributor hereby grants to You a perpetual,       worldwide, non-exclusive, no-charge, royalty-free, irrevocable       copyright license to reproduce, prepare Derivative Works of,       publicly display, publicly perform, sublicense, and distribute the       Work and such Derivative Works in Source or Object form.</p> </li> <li> <p>Grant of Patent License. Subject to the terms and conditions of       this License, each Contributor hereby grants to You a perpetual,       worldwide, non-exclusive, no-charge, royalty-free, irrevocable       (except as stated in this section) patent license to make, have made,       use, offer to sell, sell, import, and otherwise transfer the Work,       where such license applies only to those patent claims licensable       by such Contributor that are necessarily infringed by their       Contribution(s) alone or by combination of their Contribution(s)       with the Work to which such Contribution(s) was submitted. If You       institute patent litigation against any entity (including a       cross-claim or counterclaim in a lawsuit) alleging that the Work       or a Contribution incorporated within the Work constitutes direct       or contributory patent infringement, then any patent licenses       granted to You under this License for that Work shall terminate       as of the date such litigation is filed.</p> </li> <li> <p>Redistribution. You may reproduce and distribute copies of the       Work or Derivative Works thereof in any medium, with or without       modifications, and in Source or Object form, provided that You       meet the following conditions:</p> <p>(a) You must give any other recipients of the Work or       Derivative Works a copy of this License; and</p> <p>(b) You must cause any modified files to carry prominent notices       stating that You changed the files; and</p> <p>(c) You must retain, in the Source form of any Derivative Works       that You distribute, all copyright, patent, trademark, and       attribution notices from the Source form of the Work,       excluding those notices that do not pertain to any part of       the Derivative Works; and</p> <p>(d) If the Work includes a \"NOTICE\" text file as part of its       distribution, then any Derivative Works that You distribute must       include a readable copy of the attribution notices contained       within such NOTICE file, excluding those notices that do not       pertain to any part of the Derivative Works, in at least one       of the following places: within a NOTICE text file distributed       as part of the Derivative Works; within the Source form or       documentation, if provided along with the Derivative Works; or,       within a display generated by the Derivative Works, if and       wherever such third-party notices normally appear. The contents       of the NOTICE file are for informational purposes only and       do not modify the License. You may add Your own attribution       notices within Derivative Works that You distribute, alongside       or as an addendum to the NOTICE text from the Work, provided       that such additional attribution notices cannot be construed       as modifying the License.</p> <p>You may add Your own copyright statement to Your modifications and   may provide additional or different license terms and conditions   for use, reproduction, or distribution of Your modifications, or   for any such Derivative Works as a whole, provided Your use,   reproduction, and distribution of the Work otherwise complies with   the conditions stated in this License.</p> </li> <li> <p>Submission of Contributions. Unless You explicitly state otherwise,       any Contribution intentionally submitted for inclusion in the Work       by You to the Licensor shall be under the terms and conditions of       this License, without any additional terms or conditions.       Notwithstanding the above, nothing herein shall supersede or modify       the terms of any separate license agreement you may have executed       with Licensor regarding such Contributions.</p> </li> <li> <p>Trademarks. This License does not grant permission to use the trade       names, trademarks, service marks, or product names of the Licensor,       except as required for reasonable and customary use in describing the       origin of the Work and reproducing the content of the NOTICE file.</p> </li> <li> <p>Disclaimer of Warranty. Unless required by applicable law or       agreed to in writing, Licensor provides the Work (and each       Contributor provides its Contributions) on an \"AS IS\" BASIS,       WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or       implied, including, without limitation, any warranties or conditions       of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A       PARTICULAR PURPOSE. You are solely responsible for determining the       appropriateness of using or redistributing the Work and assume any       risks associated with Your exercise of permissions under this License.</p> </li> <li> <p>Limitation of Liability. In no event and under no legal theory,       whether in tort (including negligence), contract, or otherwise,       unless required by applicable law (such as deliberate and grossly       negligent acts) or agreed to in writing, shall any Contributor be       liable to You for damages, including any direct, indirect, special,       incidental, or consequential damages of any character arising as a       result of this License or out of the use or inability to use the       Work (including but not limited to damages for loss of goodwill,       work stoppage, computer failure or malfunction, or any and all       other commercial damages or losses), even if such Contributor       has been advised of the possibility of such damages.</p> </li> <li> <p>Accepting Warranty or Additional Liability. While redistributing       the Work or Derivative Works thereof, You may choose to offer,       and charge a fee for, acceptance of support, warranty, indemnity,       or other liability obligations and/or rights consistent with this       License. However, in accepting such obligations, You may act only       on Your own behalf and on Your sole responsibility, not on behalf       of any other Contributor, and only if You agree to indemnify,       defend, and hold each Contributor harmless for any liability       incurred by, or claims asserted against, such Contributor by reason       of your accepting any such warranty or additional liability.</p> </li> </ol> <p>END OF TERMS AND CONDITIONS</p> <p>APPENDIX: How to apply the Apache License to your work.</p> <pre><code>  To apply the Apache License to your work, attach the following\n  boilerplate notice, with the fields enclosed by brackets \"{}\"\n  replaced with your own identifying information. (Don't include\n  the brackets!)  The text should be enclosed in the appropriate\n  comment syntax for the file format. We also recommend that a\n  file or class name and description of purpose be included on the\n  same \"printed page\" as the copyright notice for easier\n  identification within third-party archives.\n</code></pre>"},{"location":"Learning-Environment-Create-New/","title":"Making a New Learning Environment","text":"<p>This tutorial walks through the process of creating a Unity Environment from scratch. We recommend first reading the Getting Started guide to understand the concepts presented here first in an already-built environment.</p> <p></p> <p>In this example, we will create an agent capable of controlling a ball on a platform. We will then train the agent to roll the ball toward the cube while avoiding falling off the platform.</p>"},{"location":"Learning-Environment-Create-New/#overview","title":"Overview","text":"<p>Using the ML-Agents Toolkit in a Unity project involves the following basic steps:</p> <ol> <li>Create an environment for your agents to live in. An environment can range    from a simple physical simulation containing a few objects to an entire game    or ecosystem.</li> <li>Implement your Agent subclasses. An Agent subclass defines the code an Agent    uses to observe its environment, to carry out assigned actions, and to    calculate the rewards used for reinforcement training. You can also implement    optional methods to reset the Agent when it has finished or failed its task.</li> <li>Add your Agent subclasses to appropriate GameObjects, typically, the object    in the scene that represents the Agent in the simulation.</li> </ol> <p>Note: If you are unfamiliar with Unity, refer to the Unity manual if an Editor task isn't explained sufficiently in this tutorial.</p> <p>If you haven't already, follow the installation instructions.</p>"},{"location":"Learning-Environment-Create-New/#set-up-the-unity-project","title":"Set Up the Unity Project","text":"<p>The first task to accomplish is simply creating a new Unity project and importing the ML-Agents assets into it:</p> <ol> <li>Launch Unity Hub and create a new 3D project named \"RollerBall\".</li> <li>Add the ML-Agents Unity package    to your project.</li> </ol> <p>Your Unity Project window should contain the following assets:</p> <p></p>"},{"location":"Learning-Environment-Create-New/#create-the-environment","title":"Create the Environment","text":"<p>Next, we will create a very simple scene to act as our learning environment. The \"physical\" components of the environment include a Plane to act as the floor for the Agent to move around on, a Cube to act as the goal or target for the agent to seek, and a Sphere to represent the Agent itself.</p>"},{"location":"Learning-Environment-Create-New/#create-the-floor-plane","title":"Create the Floor Plane","text":"<ol> <li>Right click in Hierarchy window, select 3D Object &gt; Plane.</li> <li>Name the GameObject \"Floor\".</li> <li>Select the Floor Plane to view its properties in the Inspector window.</li> <li>Set Transform to Position = <code>(0, 0, 0)</code>, Rotation = <code>(0, 0, 0)</code>, Scale =    <code>(1, 1, 1)</code>.</li> </ol>"},{"location":"Learning-Environment-Create-New/#add-the-target-cube","title":"Add the Target Cube","text":"<ol> <li>Right click in Hierarchy window, select 3D Object &gt; Cube.</li> <li>Name the GameObject \"Target\".</li> <li>Select the Target Cube to view its properties in the Inspector window.</li> <li>Set Transform to Position = <code>(3, 0.5, 3)</code>, Rotation = <code>(0, 0, 0)</code>, Scale =    <code>(1, 1, 1)</code>.</li> </ol>"},{"location":"Learning-Environment-Create-New/#add-the-agent-sphere","title":"Add the Agent Sphere","text":"<ol> <li>Right click in Hierarchy window, select 3D Object &gt; Sphere.</li> <li>Name the GameObject \"RollerAgent\".</li> <li>Select the RollerAgent Sphere to view its properties in the Inspector window.</li> <li>Set Transform to Position = <code>(0, 0.5, 0)</code>, Rotation = <code>(0, 0, 0)</code>, Scale =    <code>(1, 1, 1)</code>.</li> <li>Click Add Component.</li> <li>Add the <code>Rigidbody</code> component to the Sphere.</li> </ol>"},{"location":"Learning-Environment-Create-New/#group-into-training-area","title":"Group into Training Area","text":"<p>Group the floor, target and agent under a single, empty, GameObject. This will simplify some of our subsequent steps.</p> <p>To do so:</p> <ol> <li>Right-click on your Project Hierarchy and create a new empty GameObject. Name    it TrainingArea.</li> <li>Reset the TrainingArea\u2019s Transform so that it is at <code>(0,0,0)</code> with Rotation    <code>(0,0,0)</code> and Scale <code>(1,1,1)</code>.</li> <li>Drag the Floor, Target, and RollerAgent GameObjects in the Hierarchy into the    TrainingArea GameObject.</li> </ol> <p></p>"},{"location":"Learning-Environment-Create-New/#implement-an-agent","title":"Implement an Agent","text":"<p>To create the Agent Script:</p> <ol> <li>Select the RollerAgent GameObject to view it in the Inspector window.</li> <li>Click Add Component.</li> <li>Click New Script in the list of components (at the bottom).</li> <li>Name the script \"RollerAgent\".</li> <li>Click Create and Add.</li> </ol> <p>Then, edit the new <code>RollerAgent</code> script:</p> <ol> <li>In the Unity Project window, double-click the <code>RollerAgent</code> script to open it    in your code editor.</li> <li>Import ML-Agent package by adding</li> </ol> <p><code>csharp    using Unity.MLAgents;    using Unity.MLAgents.Sensors;    using Unity.MLAgents.Actuators;</code>    then change the base class from <code>MonoBehaviour</code> to <code>Agent</code>. 1. Delete <code>Update()</code> since we are not using it, but keep <code>Start()</code>.</p> <p>So far, these are the basic steps that you would use to add ML-Agents to any Unity project. Next, we will add the logic that will let our Agent learn to roll to the cube using reinforcement learning. More specifically, we will need to extend three methods from the <code>Agent</code> base class:</p> <ul> <li><code>OnEpisodeBegin()</code></li> <li><code>CollectObservations(VectorSensor sensor)</code></li> <li><code>OnActionReceived(ActionBuffers actionBuffers)</code></li> </ul> <p>We overview each of these in more detail in the dedicated subsections below.</p>"},{"location":"Learning-Environment-Create-New/#initialization-and-resetting-the-agent","title":"Initialization and Resetting the Agent","text":"<p>The process of training in the ML-Agents Toolkit involves running episodes where the Agent (Sphere) attempts to solve the task. Each episode lasts until the Agents solves the task (i.e. reaches the cube), fails (rolls off the platform) or times out (takes too long to solve or fail at the task). At the start of each episode, <code>OnEpisodeBegin()</code> is called to set-up the environment for a new episode. Typically the scene is initialized in a random manner to enable the agent to learn to solve the task under a variety of conditions.</p> <p>In this example, each time the Agent (Sphere) reaches its target (Cube), the episode ends and the target (Cube) is moved to a new random location; and if the Agent rolls off the platform, it will be put back onto the floor. These are all handled in <code>OnEpisodeBegin()</code>.</p> <p>To move the target (Cube), we need a reference to its Transform (which stores a GameObject's position, orientation and scale in the 3D world). To get this reference, add a public field of type <code>Transform</code> to the RollerAgent class. Public fields of a component in Unity get displayed in the Inspector window, allowing you to choose which GameObject to use as the target in the Unity Editor.</p> <p>To reset the Agent's velocity (and later to apply force to move the agent) we need a reference to the Rigidbody component. A Rigidbody is Unity's primary element for physics simulation. (See Physics for full documentation of Unity physics.) Since the Rigidbody component is on the same GameObject as our Agent script, the best way to get this reference is using <code>GameObject.GetComponent&lt;T&gt;()</code>, which we can call in our script's <code>Start()</code> method.</p> <p>So far, our RollerAgent script looks like:</p> <pre><code>using System.Collections.Generic;\nusing UnityEngine;\nusing Unity.MLAgents;\nusing Unity.MLAgents.Sensors;\n\npublic class RollerAgent : Agent\n{\n    Rigidbody rBody;\n    void Start () {\n        rBody = GetComponent&lt;Rigidbody&gt;();\n    }\n\n    public Transform Target;\n    public override void OnEpisodeBegin()\n    {\n       // If the Agent fell, zero its momentum\n        if (this.transform.localPosition.y &lt; 0)\n        {\n            this.rBody.angularVelocity = Vector3.zero;\n            this.rBody.velocity = Vector3.zero;\n            this.transform.localPosition = new Vector3( 0, 0.5f, 0);\n        }\n\n        // Move the target to a new spot\n        Target.localPosition = new Vector3(Random.value * 8 - 4,\n                                           0.5f,\n                                           Random.value * 8 - 4);\n    }\n}\n</code></pre> <p>Next, let's implement the <code>Agent.CollectObservations(VectorSensor sensor)</code> method.</p>"},{"location":"Learning-Environment-Create-New/#observing-the-environment","title":"Observing the Environment","text":"<p>The Agent sends the information we collect to the Brain, which uses it to make a decision. When you train the Agent (or use a trained model), the data is fed into a neural network as a feature vector. For an Agent to successfully learn a task, we need to provide the correct information. A good rule of thumb for deciding what information to collect is to consider what you would need to calculate an analytical solution to the problem.</p> <p>In our case, the information our Agent collects includes the position of the target, the position of the agent itself, and the velocity of the agent. This helps the Agent learn to control its speed so it doesn't overshoot the target and roll off the platform. In total, the agent observation contains 8 values as implemented below:</p> <pre><code>public override void CollectObservations(VectorSensor sensor)\n{\n    // Target and Agent positions\n    sensor.AddObservation(Target.localPosition);\n    sensor.AddObservation(this.transform.localPosition);\n\n    // Agent velocity\n    sensor.AddObservation(rBody.velocity.x);\n    sensor.AddObservation(rBody.velocity.z);\n}\n</code></pre>"},{"location":"Learning-Environment-Create-New/#taking-actions-and-assigning-rewards","title":"Taking Actions and Assigning Rewards","text":"<p>The final part of the Agent code is the <code>Agent.OnActionReceived()</code> method, which receives actions and assigns the reward.</p>"},{"location":"Learning-Environment-Create-New/#actions","title":"Actions","text":"<p>To solve the task of moving towards the target, the Agent (Sphere) needs to be able to move in the <code>x</code> and <code>z</code> directions. As such, the agent needs 2 actions: the first determines the force applied along the x-axis; and the second determines the force applied along the z-axis. (If we allowed the Agent to move in three dimensions, then we would need a third action.)</p> <p>The RollerAgent applies the values from the <code>action[]</code> array to its Rigidbody component <code>rBody</code>, using <code>Rigidbody.AddForce()</code>:</p> <pre><code>Vector3 controlSignal = Vector3.zero;\ncontrolSignal.x = action[0];\ncontrolSignal.z = action[1];\nrBody.AddForce(controlSignal * forceMultiplier);\n</code></pre>"},{"location":"Learning-Environment-Create-New/#rewards","title":"Rewards","text":"<p>Reinforcement learning requires rewards to signal which decisions are good and which are bad. The learning algorithm uses the rewards to determine whether it is giving the Agent the optimal actions. You want to reward an Agent for completing the assigned task. In this case, the Agent is given a reward of 1.0 for reaching the Target cube.</p> <p>Rewards are assigned in <code>OnActionReceived()</code>. The RollerAgent calculates the distance to detect when it reaches the target. When it does, the code calls <code>Agent.SetReward()</code> to assign a reward of 1.0 and marks the agent as finished by calling <code>EndEpisode()</code> on the Agent.</p> <pre><code>float distanceToTarget = Vector3.Distance(this.transform.localPosition, Target.localPosition);\n// Reached target\nif (distanceToTarget &lt; 1.42f)\n{\n    SetReward(1.0f);\n    EndEpisode();\n}\n</code></pre> <p>Finally, if the Agent falls off the platform, end the episode so that it can reset itself:</p> <pre><code>// Fell off platform\nif (this.transform.localPosition.y &lt; 0)\n{\n    EndEpisode();\n}\n</code></pre>"},{"location":"Learning-Environment-Create-New/#onactionreceived","title":"OnActionReceived()","text":"<p>With the action and reward logic outlined above, the final version of <code>OnActionReceived()</code> looks like:</p> <pre><code>public float forceMultiplier = 10;\npublic override void OnActionReceived(ActionBuffers actionBuffers)\n{\n    // Actions, size = 2\n    Vector3 controlSignal = Vector3.zero;\n    controlSignal.x = actionBuffers.ContinuousActions[0];\n    controlSignal.z = actionBuffers.ContinuousActions[1];\n    rBody.AddForce(controlSignal * forceMultiplier);\n\n    // Rewards\n    float distanceToTarget = Vector3.Distance(this.transform.localPosition, Target.localPosition);\n\n    // Reached target\n    if (distanceToTarget &lt; 1.42f)\n    {\n        SetReward(1.0f);\n        EndEpisode();\n    }\n\n    // Fell off platform\n    else if (this.transform.localPosition.y &lt; 0)\n    {\n        EndEpisode();\n    }\n}\n</code></pre> <p>Note the <code>forceMultiplier</code> class variable is defined before the method definition. Since <code>forceMultiplier</code> is public, you can set the value from the Inspector window.</p>"},{"location":"Learning-Environment-Create-New/#final-agent-setup-in-editor","title":"Final Agent Setup in Editor","text":"<p>Now that all the GameObjects and ML-Agent components are in place, it is time to connect everything together in the Unity Editor. This involves adding and setting some of the Agent Component's properties so that they are compatible with our Agent script.</p> <ol> <li>Select the RollerAgent GameObject to show its properties in the Inspector    window.</li> <li>Drag the Target GameObject in the Hierarchy into the <code>Target</code> field in RollerAgent Script.</li> <li>Add a <code>Decision Requester</code> script with the Add Component button.    Set the Decision Period to <code>10</code>. For more information on decisions,    see the Agent documentation</li> <li>Add a <code>Behavior Parameters</code> script with the Add Component button.    Set the Behavior Parameters of the Agent to the following:</li> <li><code>Behavior Name</code>: RollerBall</li> <li><code>Vector Observation</code> &gt; <code>Space Size</code> = 8</li> <li><code>Actions</code> &gt; <code>Continuous Actions</code> = 2</li> </ol> <p>In the inspector, the <code>RollerAgent</code> should look like this now:</p> <p></p> <p>Now you are ready to test the environment before training.</p>"},{"location":"Learning-Environment-Create-New/#testing-the-environment","title":"Testing the Environment","text":"<p>It is always a good idea to first test your environment by controlling the Agent using the keyboard. To do so, you will need to extend the <code>Heuristic()</code> method in the <code>RollerAgent</code> class. For our example, the heuristic will generate an action corresponding to the values of the \"Horizontal\" and \"Vertical\" input axis (which correspond to the keyboard arrow keys):</p> <pre><code>public override void Heuristic(in ActionBuffers actionsOut)\n{\n    var continuousActionsOut = actionsOut.ContinuousActions;\n    continuousActionsOut[0] = Input.GetAxis(\"Horizontal\");\n    continuousActionsOut[1] = Input.GetAxis(\"Vertical\");\n}\n</code></pre> <p>In order for the Agent to use the Heuristic, You will need to set the <code>Behavior Type</code> to <code>Heuristic Only</code> in the <code>Behavior Parameters</code> of the RollerAgent.</p> <p>Press Play to run the scene and use the arrows keys to move the Agent around the platform. Make sure that there are no errors displayed in the Unity Editor Console window and that the Agent resets when it reaches its target or falls from the platform.</p>"},{"location":"Learning-Environment-Create-New/#training-the-environment","title":"Training the Environment","text":"<p>The process is the same as described in the Getting Started Guide.</p> <p>The hyperparameters for training are specified in a configuration file that you pass to the <code>mlagents-learn</code> program. Create a new <code>rollerball_config.yaml</code> file under <code>config/</code> and include the following hyperparameter values:</p> <pre><code>behaviors:\n  RollerBall:\n    trainer_type: ppo\n    hyperparameters:\n      batch_size: 10\n      buffer_size: 100\n      learning_rate: 3.0e-4\n      beta: 5.0e-4\n      epsilon: 0.2\n      lambd: 0.99\n      num_epoch: 3\n      learning_rate_schedule: linear\n      beta_schedule: constant\n      epsilon_schedule: linear\n    network_settings:\n      normalize: false\n      hidden_units: 128\n      num_layers: 2\n    reward_signals:\n      extrinsic:\n        gamma: 0.99\n        strength: 1.0\n    max_steps: 500000\n    time_horizon: 64\n    summary_freq: 10000\n</code></pre> <p>Hyperparameters are explained in the training configuration file documentation</p> <p>Since this example creates a very simple training environment with only a few inputs and outputs, using small batch and buffer sizes speeds up the training considerably. However, if you add more complexity to the environment or change the reward or observation functions, you might also find that training performs better with different hyperparameter values. In addition to setting these hyperparameter values, the Agent DecisionFrequency parameter has a large effect on training time and success. A larger value reduces the number of decisions the training algorithm has to consider and, in this simple environment, speeds up training.</p> <p>To train your agent, run the following command before pressing Play in the Editor:</p> <pre><code>mlagents-learn config/rollerball_config.yaml --run-id=RollerBall\n</code></pre> <p>To monitor the statistics of Agent performance during training, use TensorBoard.</p> <p></p> <p>In particular, the cumulative_reward and value_estimate statistics show how well the Agent is achieving the task. In this example, the maximum reward an Agent can earn is 1.0, so these statistics approach that value when the Agent has successfully solved the problem.</p>"},{"location":"Learning-Environment-Create-New/#optional-multiple-training-areas-within-the-same-scene","title":"Optional: Multiple Training Areas within the Same Scene","text":"<p>In many of the example environments, many copies of the training area are instantiated in the scene. This generally speeds up training, allowing the environment to gather many experiences in parallel. This can be achieved simply by instantiating many Agents with the same <code>Behavior Name</code>. Note that we've already simplified our transition to using multiple areas by creating the <code>TrainingArea</code> GameObject and relying on local positions in <code>RollerAgent.cs</code>. Use the following steps to parallelize your RollerBall environment:</p> <ol> <li>Drag the TrainingArea GameObject, along with its attached GameObjects, into    your Assets browser, turning it into a prefab.</li> <li>You can now instantiate copies of the TrainingArea prefab. Drag them into    your scene, positioning them so that they do not overlap.</li> </ol> <p>Alternatively, you can use the <code>TrainingAreaReplicator</code> to replicate training areas. Use the following steps:</p> <ol> <li>Create a new empty Game Object in the scene.</li> <li>Click on the new object and add a TrainingAreaReplicator component to the empty Game Object through the inspector.</li> <li>Drag the training area to <code>Base Area</code> in the Training Area Replicator.</li> <li>Specify the number of areas to replicate and the separation between areas.</li> <li>Hit play and the areas will be replicated automatically!</li> </ol>"},{"location":"Learning-Environment-Create-New/#optional-training-using-concurrent-unity-instances","title":"Optional: Training Using Concurrent Unity Instances","text":"<p>Another level of parallelization comes by training using concurrent Unity instances. For example,</p> <pre><code>mlagents-learn config/rollerball_config.yaml --run-id=RollerBall --num-envs=2\n</code></pre> <p>will start ML Agents training with two environment instances. Combining multiple training areas within the same scene, with concurrent Unity instances, effectively gives you two levels of parallelism to speed up training. The command line option <code>--num-envs=&lt;n&gt;</code> controls the number of concurrent Unity instances that are executed in parallel during training.</p>"},{"location":"Learning-Environment-Design-Agents/","title":"Agents","text":"<p>Table of Contents:</p> <ul> <li>Decisions</li> <li>Observations and Sensors</li> <li>Generating Observations<ul> <li>Agent.CollectObservations()</li> <li>Observable Fields and Properties</li> <li>ISensor interface and SensorComponents</li> </ul> </li> <li>Vector Observations<ul> <li>One-hot encoding categorical information</li> <li>Normalization</li> <li>Stacking</li> <li>Vector Observation Summary &amp; Best Practices</li> </ul> </li> <li>Visual Observations<ul> <li>Visual Observation Summary &amp; Best Practices</li> </ul> </li> <li>Raycast Observations<ul> <li>RayCast Observation Summary &amp; Best Practices</li> </ul> </li> <li>Variable Length Observations<ul> <li>Variable Length Observation Summary &amp; Best Practices</li> </ul> </li> <li>Goal Signal<ul> <li>Goal Signal Summary &amp; Best Practices</li> </ul> </li> <li>Actions and Actuators</li> <li>Continuous Actions</li> <li>Discrete Actions<ul> <li>Masking Discrete Actions</li> </ul> </li> <li>Actions Summary &amp; Best Practices</li> <li>Rewards</li> <li>Examples</li> <li>Rewards Summary &amp; Best Practices</li> <li>Agent Properties</li> <li>Destroying an Agent</li> <li>Defining Multi-agent Scenarios</li> <li>Teams for Adversarial Scenarios</li> <li>Groups for Cooperative Scenarios</li> <li>Recording Demonstrations</li> </ul> <p>An agent is an entity that can observe its environment, decide on the best course of action using those observations, and execute those actions within its environment. Agents can be created in Unity by extending the <code>Agent</code> class. The most important aspects of creating agents that can successfully learn are the observations the agent collects, and the reward you assign to estimate the value of the agent's current state toward accomplishing its tasks.</p> <p>An Agent passes its observations to its Policy. The Policy then makes a decision and passes the chosen action back to the agent. Your agent code must execute the action, for example, move the agent in one direction or another. In order to train an agent using reinforcement learning, your agent must calculate a reward value at each action. The reward is used to discover the optimal decision-making policy.</p> <p>The <code>Policy</code> class abstracts out the decision making logic from the Agent itself so that you can use the same Policy in multiple Agents. How a Policy makes its decisions depends on the <code>Behavior Parameters</code> associated with the agent. If you set <code>Behavior Type</code> to <code>Heuristic Only</code>, the Agent will use its <code>Heuristic()</code> method to make decisions which can allow you to control the Agent manually or write your own Policy. If the Agent has a <code>Model</code> file, its Policy will use the neural network <code>Model</code> to take decisions.</p> <p>When you create an Agent, you should usually extend the base Agent class. This includes implementing the following methods:</p> <ul> <li><code>Agent.OnEpisodeBegin()</code> \u2014 Called at the beginning of an Agent's episode,   including at the beginning of the simulation.</li> <li><code>Agent.CollectObservations(VectorSensor sensor)</code> \u2014 Called every step that the Agent   requests a decision. This is one possible way for collecting the Agent's   observations of the environment; see Generating Observations   below for more options.</li> <li><code>Agent.OnActionReceived()</code> \u2014 Called every time the Agent receives an action to   take. Receives the action chosen by the Agent. It is also common to assign a   reward in this method.</li> <li><code>Agent.Heuristic()</code> - When the <code>Behavior Type</code> is set to <code>Heuristic Only</code> in   the Behavior Parameters of the Agent, the Agent will use the <code>Heuristic()</code>   method to generate the actions of the Agent. As such, the <code>Heuristic()</code> method   writes to the array of floats provided to the Heuristic method as argument.   Note: Do not create a new float array of action in the <code>Heuristic()</code> method,   as this will prevent writing floats to the original action array.</li> </ul> <p>As a concrete example, here is how the Ball3DAgent class implements these methods:</p> <ul> <li><code>Agent.OnEpisodeBegin()</code> \u2014 Resets the agent cube and ball to their starting   positions. The function randomizes the reset values so that the training   generalizes to more than a specific starting position and agent cube   orientation.</li> <li><code>Agent.CollectObservations(VectorSensor sensor)</code> \u2014 Adds information about the   orientation of the agent cube, the ball velocity, and the relative position   between the ball and the cube. Since the  <code>CollectObservations()</code>   method calls <code>VectorSensor.AddObservation()</code> such that vector size adds up to 8,   the Behavior Parameters of the Agent are set with vector observation space   with a state size of 8.</li> <li><code>Agent.OnActionReceived()</code> \u2014 The action results   in a small change in the agent cube's rotation at each step. In this example,   an Agent receives a small positive reward for each step it keeps the ball on the   agent cube's head and a larger, negative reward for dropping the ball. An   Agent's episode is also ended when it drops the ball so that it will reset   with a new ball for the next simulation step.</li> <li><code>Agent.Heuristic()</code> - Converts the keyboard inputs into actions.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#decisions","title":"Decisions","text":"<p>The observation-decision-action-reward cycle repeats each time the Agent request a decision. Agents will request a decision when <code>Agent.RequestDecision()</code> is called. If you need the Agent to request decisions on its own at regular intervals, add a <code>Decision Requester</code> component to the Agent's GameObject. Making decisions at regular step intervals is generally most appropriate for physics-based simulations. For example, an agent in a robotic simulator that must provide fine-control of joint torques should make its decisions every step of the simulation. In games such as real-time strategy, where many agents make their decisions at regular intervals, the decision timing for each agent can be staggered by setting the <code>DecisionStep</code> parameter in the <code>Decision Requester</code> component for each agent. On the other hand, an agent that only needs to make decisions when certain game or simulation events occur, such as in a turn-based game, should call <code>Agent.RequestDecision()</code> manually.</p>"},{"location":"Learning-Environment-Design-Agents/#observations-and-sensors","title":"Observations and Sensors","text":"<p>In order for an agent to learn, the observations should include all the information an agent needs to accomplish its task. Without sufficient and relevant information, an agent may learn poorly or may not learn at all. A reasonable approach for determining what information should be included is to consider what you would need to calculate an analytical solution to the problem, or what you would expect a human to be able to use to solve the problem.</p>"},{"location":"Learning-Environment-Design-Agents/#generating-observations","title":"Generating Observations","text":"<p>ML-Agents provides multiple ways for an Agent to make observations:   1. Overriding the <code>Agent.CollectObservations()</code> method and passing the     observations to the provided <code>VectorSensor</code>.   1. Adding the <code>[Observable]</code> attribute to fields and properties on the Agent.   1. Implementing the <code>ISensor</code> interface, using a <code>SensorComponent</code> attached to   the Agent to create the <code>ISensor</code>.</p>"},{"location":"Learning-Environment-Design-Agents/#agentcollectobservations","title":"Agent.CollectObservations()","text":"<p>Agent.CollectObservations() is best used for aspects of the environment which are numerical and non-visual. The Policy class calls the <code>CollectObservations(VectorSensor sensor)</code> method of each Agent. Your implementation of this function must call <code>VectorSensor.AddObservation</code> to add vector observations.</p> <p>The <code>VectorSensor.AddObservation</code> method provides a number of overloads for adding common types of data to your observation vector. You can add Integers and booleans directly to the observation vector, as well as some common Unity data types such as <code>Vector2</code>, <code>Vector3</code>, and <code>Quaternion</code>.</p> <p>For examples of various state observation functions, you can look at the example environments included in the ML-Agents SDK. For instance, the 3DBall example uses the rotation of the platform, the relative position of the ball, and the velocity of the ball as its state observation.</p> <pre><code>public GameObject ball;\n\npublic override void CollectObservations(VectorSensor sensor)\n{\n    // Orientation of the cube (2 floats)\n    sensor.AddObservation(gameObject.transform.rotation.z);\n    sensor.AddObservation(gameObject.transform.rotation.x);\n    // Relative position of the ball to the cube (3 floats)\n    sensor.AddObservation(ball.transform.position - gameObject.transform.position);\n    // Velocity of the ball (3 floats)\n    sensor.AddObservation(m_BallRb.velocity);\n    // 8 floats total\n}\n</code></pre> <p>As an experiment, you can remove the velocity components from the observation and retrain the 3DBall agent. While it will learn to balance the ball reasonably well, the performance of the agent without using velocity is noticeably worse.</p> <p>The observations passed to <code>VectorSensor.AddObservation()</code> must always contain the same number of elements must always be in the same order. If the number of observed entities in an environment can vary, you can pad the calls with zeros for any missing entities in a specific observation, or you can limit an agent's observations to a fixed subset. For example, instead of observing every enemy in an environment, you could only observe the closest five.</p> <p>Additionally, when you set up an Agent's <code>Behavior Parameters</code> in the Unity Editor, you must set the Vector Observations &gt; Space Size to equal the number of floats that are written by <code>CollectObservations()</code>.</p>"},{"location":"Learning-Environment-Design-Agents/#observable-fields-and-properties","title":"Observable Fields and Properties","text":"<p>Another approach is to define the relevant observations as fields or properties on your Agent class, and annotate them with an <code>ObservableAttribute</code>. For example, in the Ball3DHardAgent, the difference between positions could be observed by adding a property to the Agent:</p> <pre><code>using Unity.MLAgents.Sensors.Reflection;\n\npublic class Ball3DHardAgent : Agent {\n\n    [Observable(numStackedObservations: 9)]\n    Vector3 PositionDelta\n    {\n        get\n        {\n            return ball.transform.position - gameObject.transform.position;\n        }\n    }\n}\n</code></pre> <p><code>ObservableAttribute</code> currently supports most basic types (e.g. floats, ints, bools), as well as <code>Vector2</code>, <code>Vector3</code>, <code>Vector4</code>, <code>Quaternion</code>, and enums.</p> <p>The behavior of <code>ObservableAttribute</code>s are controlled by the \"Observable Attribute Handling\" in the Agent's <code>Behavior Parameters</code>. The possible values for this are:  * Ignore (default) - All ObservableAttributes on the Agent will be ignored.   If there are no ObservableAttributes on the Agent, this will result in the   fastest  initialization time.  * Exclude Inherited - Only members on the declared class will be examined;   members that are inherited are ignored. This is a reasonable tradeoff between   performance and flexibility.  * Examine All All members on the class will be examined. This can lead to   slower startup times.</p> <p>\"Exclude Inherited\" is generally sufficient, but if your Agent inherits from another Agent implementation that has Observable members, you will need to use \"Examine All\".</p> <p>Internally, ObservableAttribute uses reflection to determine which members of the Agent have ObservableAttributes, and also uses reflection to access the fields or invoke the properties at runtime. This may be slower than using CollectObservations or an ISensor, although this might not be enough to noticeably affect performance.</p> <p>NOTE: you do not need to adjust the Space Size in the Agent's <code>Behavior Parameters</code> when you add <code>[Observable]</code> fields or properties to an Agent, since their size can be computed before they are used.</p>"},{"location":"Learning-Environment-Design-Agents/#isensor-interface-and-sensorcomponents","title":"ISensor interface and SensorComponents","text":"<p>The <code>ISensor</code> interface is generally intended for advanced users. The <code>Write()</code> method is used to actually generate the observation, but some other methods such as returning the shape of the observations must also be implemented.</p> <p>The <code>SensorComponent</code> abstract class is used to create the actual <code>ISensor</code> at runtime. It must be attached to the same <code>GameObject</code> as the <code>Agent</code>, or to a child <code>GameObject</code>.</p> <p>There are several SensorComponents provided in the API, including: - <code>CameraSensorComponent</code> - Uses images from a <code>Camera</code> as observations. - <code>RenderTextureSensorComponent</code> - Uses the content of a <code>RenderTexture</code> as observations. - <code>RayPerceptionSensorComponent</code> - Uses the information from set of ray casts as observations. - <code>Match3SensorComponent</code> - Uses the board of a Match-3 game as observations. - <code>GridSensorComponent</code> - Uses a set of box queries in a grid shape as observations.</p> <p>NOTE: you do not need to adjust the Space Size in the Agent's <code>Behavior Parameters</code> when using <code>SensorComponents</code>s.</p> <p>Internally, both <code>Agent.CollectObservations</code> and <code>[Observable]</code> attribute use an ISensors to write observations, although this is mostly abstracted from the user.</p>"},{"location":"Learning-Environment-Design-Agents/#vector-observations","title":"Vector Observations","text":"<p>Both <code>Agent.CollectObservations()</code> and <code>ObservableAttribute</code>s produce vector observations, which are represented at lists of <code>float</code>s. <code>ISensor</code>s can produce both vector observations and visual observations, which are multi-dimensional arrays of floats.</p> <p>Below are some additional considerations when dealing with vector observations:</p>"},{"location":"Learning-Environment-Design-Agents/#one-hot-encoding-categorical-information","title":"One-hot encoding categorical information","text":"<p>Type enumerations should be encoded in the one-hot style. That is, add an element to the feature vector for each element of enumeration, setting the element representing the observed member to one and set the rest to zero. For example, if your enumeration contains [Sword, Shield, Bow] and the agent observes that the current item is a Bow, you would add the elements: 0, 0, 1 to the feature vector. The following code example illustrates how to add.</p> <pre><code>enum ItemType { Sword, Shield, Bow, LastItem }\npublic override void CollectObservations(VectorSensor sensor)\n{\n    for (int ci = 0; ci &lt; (int)ItemType.LastItem; ci++)\n    {\n        sensor.AddObservation((int)currentItem == ci ? 1.0f : 0.0f);\n    }\n}\n</code></pre> <p><code>VectorSensor</code> also provides a two-argument function <code>AddOneHotObservation()</code> as a shortcut for one-hot style observations. The following example is identical to the previous one.</p> <pre><code>enum ItemType { Sword, Shield, Bow, LastItem }\nconst int NUM_ITEM_TYPES = (int)ItemType.LastItem + 1;\n\npublic override void CollectObservations(VectorSensor sensor)\n{\n    // The first argument is the selection index; the second is the\n    // number of possibilities\n    sensor.AddOneHotObservation((int)currentItem, NUM_ITEM_TYPES);\n}\n</code></pre> <p><code>ObservableAttribute</code> has built-in support for enums. Note that you don't need the <code>LastItem</code> placeholder in this case:</p> <pre><code>enum ItemType { Sword, Shield, Bow }\n\npublic class HeroAgent : Agent\n{\n    [Observable]\n    ItemType m_CurrentItem;\n}\n</code></pre>"},{"location":"Learning-Environment-Design-Agents/#normalization","title":"Normalization","text":"<p>For the best results when training, you should normalize the components of your feature vector to the range [-1, +1] or [0, 1]. When you normalize the values, the PPO neural network can often converge to a solution faster. Note that it isn't always necessary to normalize to these recommended ranges, but it is considered a best practice when using neural networks. The greater the variation in ranges between the components of your observation, the more likely that training will be affected.</p> <p>To normalize a value to [0, 1], you can use the following formula:</p> <pre><code>normalizedValue = (currentValue - minValue)/(maxValue - minValue)\n</code></pre> <p>:warning: For vectors, you should apply the above formula to each component (x, y, and z). Note that this is not the same as using the <code>Vector3.normalized</code> property or <code>Vector3.Normalize()</code> method in Unity (and similar for <code>Vector2</code>).</p> <p>Rotations and angles should also be normalized. For angles between 0 and 360 degrees, you can use the following formulas:</p> <pre><code>Quaternion rotation = transform.rotation;\nVector3 normalized = rotation.eulerAngles / 180.0f - Vector3.one;  // [-1,1]\nVector3 normalized = rotation.eulerAngles / 360.0f;  // [0,1]\n</code></pre> <p>For angles that can be outside the range [0,360], you can either reduce the angle, or, if the number of turns is significant, increase the maximum value used in your normalization formula.</p>"},{"location":"Learning-Environment-Design-Agents/#stacking","title":"Stacking","text":"<p>Stacking refers to repeating observations from previous steps as part of a larger observation. For example, consider an Agent that generates these observations in four steps</p> <pre><code>step 1: [0.1]\nstep 2: [0.2]\nstep 3: [0.3]\nstep 4: [0.4]\n</code></pre> <p>If we use a stack size of 3, the observations would instead be:</p> <pre><code>step 1: [0.1, 0.0, 0.0]\nstep 2: [0.2, 0.1, 0.0]\nstep 3: [0.3, 0.2, 0.1]\nstep 4: [0.4, 0.3, 0.2]\n</code></pre> <p>(The observations are padded with zeroes for the first <code>stackSize-1</code> steps). This is a simple way to give an Agent limited \"memory\" without the complexity of adding a recurrent neural network (RNN).</p> <p>The steps for enabling stacking depends on how you generate observations: * For Agent.CollectObservations(), set \"Stacked Vectors\" on the Agent's   <code>Behavior Parameters</code> to a value greater than 1. * For ObservableAttribute, set the <code>numStackedObservations</code> parameter in the   constructor, e.g. <code>[Observable(numStackedObservations: 2)]</code>. * For <code>ISensor</code>s, wrap them in a <code>StackingSensor</code> (which is also an <code>ISensor</code>).   Generally, this should happen in the <code>CreateSensor()</code> method of your   <code>SensorComponent</code>.</p>"},{"location":"Learning-Environment-Design-Agents/#vector-observation-summary-best-practices","title":"Vector Observation Summary &amp; Best Practices","text":"<ul> <li>Vector Observations should include all variables relevant for allowing the   agent to take the optimally informed decision, and ideally no extraneous   information.</li> <li>In cases where Vector Observations need to be remembered or compared over   time, either an RNN should be used in the model, or the <code>Stacked Vectors</code>   value in the agent GameObject's <code>Behavior Parameters</code> should be changed.</li> <li>Categorical variables such as type of object (Sword, Shield, Bow) should be   encoded in one-hot fashion (i.e. <code>3</code> -&gt; <code>0, 0, 1</code>). This can be done   automatically using the <code>AddOneHotObservation()</code> method of the <code>VectorSensor</code>,   or using <code>[Observable]</code> on an enum field or property of the Agent.</li> <li>In general, all inputs should be normalized to be in the range 0 to +1 (or -1   to 1). For example, the <code>x</code> position information of an agent where the maximum   possible value is <code>maxValue</code> should be recorded as   <code>VectorSensor.AddObservation(transform.position.x / maxValue);</code> rather than   <code>VectorSensor.AddObservation(transform.position.x);</code>.</li> <li>Positional information of relevant GameObjects should be encoded in relative   coordinates wherever possible. This is often relative to the agent position.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#visual-observations","title":"Visual Observations","text":"<p>Visual observations are generally provided to agent via either a <code>CameraSensor</code> or <code>RenderTextureSensor</code>. These collect image information and transforms it into a 3D Tensor which can be fed into the convolutional neural network (CNN) of the agent policy. For more information on CNNs, see this guide. This allows agents to learn from spatial regularities in the observation images. It is possible to use visual and vector observations with the same agent.</p> <p>Agents using visual observations can capture state of arbitrary complexity and are useful when the state is difficult to describe numerically. However, they are also typically less efficient and slower to train, and sometimes don't succeed at all as compared to vector observations. As such, they should only be used when it is not possible to properly define the problem using vector or ray-cast observations.</p> <p>Visual observations can be derived from Cameras or RenderTextures within your scene. To add a visual observation to an Agent, add either a Camera Sensor Component or RenderTextures Sensor Component to the Agent. Then drag the camera or render texture you want to add to the <code>Camera</code> or <code>RenderTexture</code> field. You can have more than one camera or render texture and even use a combination of both attached to an Agent. For each visual observation, set the width and height of the image (in pixels) and whether or not the observation is color or grayscale.</p> <p></p> <p>or</p> <p></p> <p>Each Agent that uses the same Policy must have the same number of visual observations, and they must all have the same resolutions (including whether or not they are grayscale). Additionally, each Sensor Component on an Agent must have a unique name so that they can be sorted deterministically (the name must be unique for that Agent, but multiple Agents can have a Sensor Component with the same name).</p> <p>Visual observations also support stacking, by specifying <code>Observation Stacks</code> to a value greater than 1. The visual observations from the last <code>stackSize</code> steps will be stacked on the last dimension (channel dimension).</p> <p>When using <code>RenderTexture</code> visual observations, a handy feature for debugging is adding a <code>Canvas</code>, then adding a <code>Raw Image</code> with it's texture set to the Agent's <code>RenderTexture</code>. This will render the agent observation on the game screen.</p> <p></p> <p>The GridWorld environment is an example on how to use a RenderTexture for both debugging and observation. Note that in this example, a Camera is rendered to a RenderTexture, which is then used for observations and debugging. To update the RenderTexture, the Camera must be asked to render every time a decision is requested within the game code. When using Cameras as observations directly, this is done automatically by the Agent.</p> <p></p>"},{"location":"Learning-Environment-Design-Agents/#visual-observation-summary-best-practices","title":"Visual Observation Summary &amp; Best Practices","text":"<ul> <li>To collect visual observations, attach <code>CameraSensor</code> or <code>RenderTextureSensor</code>   components to the agent GameObject.</li> <li>Visual observations should generally only be used when vector observations are   not sufficient.</li> <li>Image size should be kept as small as possible, without the loss of needed   details for decision making.</li> <li>Images should be made grayscale in situations where color information is not   needed for making informed decisions.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#raycast-observations","title":"Raycast Observations","text":"<p>Raycasts are another possible method for providing observations to an agent. This can be easily implemented by adding a <code>RayPerceptionSensorComponent3D</code> (or <code>RayPerceptionSensorComponent2D</code>) to the Agent GameObject.</p> <p>During observations, several rays (or spheres, depending on settings) are cast into the physics world, and the objects that are hit determine the observation vector that is produced.</p> <p></p> <p>Both sensor components have several settings:</p> <ul> <li>Detectable Tags A list of strings corresponding to the types of objects that   the Agent should be able to distinguish between. For example, in the WallJump   example, we use \"wall\", \"goal\", and \"block\" as the list of objects to detect.</li> <li>Rays Per Direction Determines the number of rays that are cast. One ray is   always cast forward, and this many rays are cast to the left and right.</li> <li>Max Ray Degrees The angle (in degrees) for the outermost rays. 90 degrees   corresponds to the left and right of the agent.</li> <li>Sphere Cast Radius The size of the sphere used for sphere casting. If set to   0, rays will be used instead of spheres. Rays may be more efficient,   especially in complex scenes.</li> <li>Ray Length The length of the casts</li> <li>Ray Layer Mask The LayerMask   passed to the raycast or spherecast. This can be used to ignore certain types   of objects when casting.</li> <li>Observation Stacks The number of previous results to \"stack\" with the cast   results. Note that this can be independent of the \"Stacked Vectors\" setting in   <code>Behavior Parameters</code>.</li> <li>Start Vertical Offset (3D only) The vertical offset of the ray start point.</li> <li>End Vertical Offset (3D only) The vertical offset of the ray end point.</li> <li>Alternating Ray Order Alternating is the default, it gives an order of (0,   -delta, delta, -2delta, 2delta, ..., -ndelta, ndelta). If alternating is   disabled the order is left to right (-ndelta, -(n-1)delta, ..., -delta, 0,   delta, ..., (n-1)delta, ndelta). For general usage there is no difference   but if using custom models the left-to-right layout that matches the spatial   structuring can be preferred (e.g. for processing with conv nets).</li> <li>Use Batched Raycasts (3D only) Whether to use batched raycasts. Enable to use batched raycasts and the jobs system.</li> </ul> <p>In the example image above, the Agent has two <code>RayPerceptionSensorComponent3D</code>s. Both use 3 Rays Per Direction and 90 Max Ray Degrees. One of the components had a vertical offset, so the Agent can tell whether it's clear to jump over the wall.</p> <p>The total size of the created observations is</p> <pre><code>(Observation Stacks) * (1 + 2 * Rays Per Direction) * (Num Detectable Tags + 2)\n</code></pre> <p>so the number of rays and tags should be kept as small as possible to reduce the amount of data used. Note that this is separate from the State Size defined in <code>Behavior Parameters</code>, so you don't need to worry about the formula above when setting the State Size.</p>"},{"location":"Learning-Environment-Design-Agents/#raycast-observation-summary-best-practices","title":"RayCast Observation Summary &amp; Best Practices","text":"<ul> <li>Attach <code>RayPerceptionSensorComponent3D</code> or <code>RayPerceptionSensorComponent2D</code> to   use.</li> <li>This observation type is best used when there is relevant spatial information   for the agent that doesn't require a fully rendered image to convey.</li> <li>Use as few rays and tags as necessary to solve the problem in order to improve   learning stability and agent performance.</li> <li>If you run into performance issues, try using batched raycasts by enabling the Use Batched Raycast setting.   (Only available for 3D ray perception sensors.)</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#grid-observations","title":"Grid Observations","text":"<p>Grid-base observations combine the advantages of 2D spatial representation in visual observations, and the flexibility of defining detectable objects in RayCast observations. The sensor uses a set of box queries in a grid shape and gives a top-down 2D view around the agent. This can be implemented by adding a <code>GridSensorComponent</code> to the Agent GameObject.</p> <p>During observations, the sensor detects the presence of detectable objects in each cell and encode that into one-hot representation. The collected information from each cell forms a 3D tensor observation and will be fed into the convolutional neural network (CNN) of the agent policy just like visual observations.</p> <p></p> <p>The sensor component has the following settings: - Cell Scale The scale of each cell in the grid. - Grid Size Number of cells on each side of the grid. - Agent Game Object The Agent that holds the grid sensor. This is used to   disambiguate objects with the same tag as the agent so that the agent doesn't   detect itself. - Rotate With Agent Whether the grid rotates with the Agent. - Detectable Tags A list of strings corresponding to the types of objects that   the Agent should be able to distinguish between. - Collider Mask The LayerMask   passed to the collider detection. This can be used to ignore certain types   of objects. - Initial Collider Buffer Size The initial size of the Collider buffer used   in the non-allocating Physics calls for each cell. - Max Collider Buffer Size The max size of the Collider buffer used in the   non-allocating Physics calls for each cell.</p> <p>The observation for each grid cell is a one-hot encoding of the detected object. The total size of the created observations is</p> <pre><code>GridSize.x * GridSize.z * Num Detectable Tags\n</code></pre> <p>so the number of detectable tags and size of the grid should be kept as small as possible to reduce the amount of data used. This makes a trade-off between the granularity of the observation and training speed.</p> <p>To allow more variety of observations that grid sensor can capture, the <code>GridSensorComponent</code> and the underlying <code>GridSensorBase</code> also provides interfaces that can be overridden to collect customized observation from detected objects. See the doc on extending grid Sensors for more details on custom grid sensors.</p> <p>Note: The <code>GridSensor</code> only works in 3D environments and will not behave properly in 2D environments.</p>"},{"location":"Learning-Environment-Design-Agents/#grid-observation-summary-best-practices","title":"Grid Observation Summary &amp; Best Practices","text":"<ul> <li>Attach <code>GridSensorComponent</code> to use.</li> <li>This observation type is best used when there is relevant non-visual spatial information that   can be best captured in 2D representations.</li> <li>Use as small grid size and as few tags as necessary to solve the problem in order to improve   learning stability and agent performance.</li> <li>Do not use <code>GridSensor</code> in a 2D game.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#variable-length-observations","title":"Variable Length Observations","text":"<p>It is possible for agents to collect observations from a varying number of GameObjects by using a <code>BufferSensor</code>. You can add a <code>BufferSensor</code> to your Agent by adding a <code>BufferSensorComponent</code> to its GameObject. The <code>BufferSensor</code> can be useful in situations in which the Agent must pay attention to a varying number of entities (for example, a varying number of enemies or projectiles). On the trainer side, the <code>BufferSensor</code> is processed using an attention module. More information about attention mechanisms can be found here. Training or doing inference with variable length observations can be slower than using a flat vector observation. However, attention mechanisms enable solving problems that require comparative reasoning between entities in a scene such as our Sorter environment. Note that even though the <code>BufferSensor</code> can process a variable number of entities, you still need to define a maximum number of entities. This is because our network architecture requires to know what the shape of the observations will be. If fewer entities are observed than the maximum, the observation will be padded with zeros and the trainer will ignore the padded observations. Note that attention layers are invariant to the order of the entities, so there is no need to properly \"order\" the entities before feeding them into the <code>BufferSensor</code>.</p> <p>The <code>BufferSensorComponent</code> Editor inspector has two arguments:</p> <ul> <li><code>Observation Size</code> : This is how many floats each entities will be  represented with. This number is fixed and all entities must  have the same representation. For example, if the entities you want to  put into the <code>BufferSensor</code> have for relevant information position and  speed, then the <code>Observation Size</code> should be 6 floats.</li> <li><code>Maximum Number of Entities</code> : This is the maximum number of entities  the <code>BufferSensor</code> will be able to collect.</li> </ul> <p>To add an entity's observations to a <code>BufferSensorComponent</code>, you need to call <code>BufferSensorComponent.AppendObservation()</code> in the Agent.CollectObservations() method with a float array of size <code>Observation Size</code> as argument.</p> <p>Note: Currently, the observations put into the <code>BufferSensor</code> are not normalized, you will need to normalize your observations manually between -1 and 1.</p>"},{"location":"Learning-Environment-Design-Agents/#variable-length-observation-summary-best-practices","title":"Variable Length Observation Summary &amp; Best Practices","text":"<ul> <li>Attach <code>BufferSensorComponent</code> to use.</li> <li>Call <code>BufferSensorComponent.AppendObservation()</code> in the  Agent.CollectObservations() methodto add the observations  of an entity to the <code>BufferSensor</code>.</li> <li>Normalize the entities observations before feeding them into the <code>BufferSensor</code>.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#goal-signal","title":"Goal Signal","text":"<p>It is possible for agents to collect observations that will be treated as \"goal signal\". A goal signal is used to condition the policy of the agent, meaning that if the goal changes, the policy (i.e. the mapping from observations to actions) will change as well. Note that this is true for any observation since all observations influence the policy of the Agent to some degree. But by specifying a goal signal explicitly, we can make this conditioning more important to the agent. This feature can be used in settings where an agent must learn to solve different tasks that are similar by some aspects because the agent will learn to reuse learnings from different tasks to generalize better. In Unity, you can specify that a <code>VectorSensor</code> or a <code>CameraSensor</code> is a goal by attaching a <code>VectorSensorComponent</code> or a <code>CameraSensorComponent</code> to the Agent and selecting <code>Goal Signal</code> as <code>Observation Type</code>. On the trainer side, there are two different ways to condition the policy. This setting is determined by the conditioning_type parameter. If set to <code>hyper</code> (default) a HyperNetwork will be used to generate some of the weights of the policy using the goal observations as input. Note that using a HyperNetwork requires a lot of computations, it is recommended to use a smaller number of hidden units in the policy to alleviate this. If set to <code>none</code> the goal signal will be considered as regular observations. For an example on how to use a goal signal, see the GridWorld example.</p>"},{"location":"Learning-Environment-Design-Agents/#goal-signal-summary-best-practices","title":"Goal Signal Summary &amp; Best Practices","text":"<ul> <li>Attach a <code>VectorSensorComponent</code> or <code>CameraSensorComponent</code> to an agent and  set the observation type to goal to use the feature.</li> <li>Set the conditioning_type parameter in the training configuration.</li> <li>Reduce the number of hidden units in the network when using the HyperNetwork  conditioning type.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#actions-and-actuators","title":"Actions and Actuators","text":"<p>An action is an instruction from the Policy that the agent carries out. The action is passed to the an <code>IActionReceiver</code> (either an <code>Agent</code> or an <code>IActuator</code>) as the <code>ActionBuffers</code> parameter when the Academy invokes the <code>IActionReciever.OnActionReceived()</code> function. There are two types of actions supported: Continuous and Discrete.</p> <p>Neither the Policy nor the training algorithm know anything about what the action values themselves mean. The training algorithm simply tries different values for the action list and observes the affect on the accumulated rewards over time and many training episodes. Thus, the only place actions are defined for an Agent is in the <code>OnActionReceived()</code> function.</p> <p>For example, if you designed an agent to move in two dimensions, you could use either continuous or the discrete actions. In the continuous case, you would set the action size to two (one for each dimension), and the agent's Policy would output an action with two floating point values. In the discrete case, you would use one Branch with a size of four (one for each direction), and the Policy would create an action array containing a single element with a value ranging from zero to three. Alternatively, you could create two branches of size two (one for horizontal movement and one for vertical movement), and the Policy would output an action array containing two elements with values ranging from zero to one. You could alternatively use a combination of continuous and discrete actions e.g., using one continuous action for horizontal movement and a discrete branch of size two for the vertical movement.</p> <p>Note that when you are programming actions for an agent, it is often helpful to test your action logic using the <code>Heuristic()</code> method of the Agent, which lets you map keyboard commands to actions.</p>"},{"location":"Learning-Environment-Design-Agents/#continuous-actions","title":"Continuous Actions","text":"<p>When an Agent's Policy has Continuous actions, the <code>ActionBuffers.ContinuousActions</code> passed to the Agent's <code>OnActionReceived()</code> function is an array with length equal to the <code>Continuous Action Size</code> property value. The individual values in the array have whatever meanings that you ascribe to them. If you assign an element in the array as the speed of an Agent, for example, the training process learns to control the speed of the Agent through this parameter.</p> <p>The 3DBall example uses continuous actions with two control values.</p> <p></p> <p>These control values are applied as rotation to the cube:</p> <pre><code>    public override void OnActionReceived(ActionBuffers actionBuffers)\n    {\n        var actionZ = 2f * Mathf.Clamp(actionBuffers.ContinuousActions[0], -1f, 1f);\n        var actionX = 2f * Mathf.Clamp(actionBuffers.ContinuousActions[1], -1f, 1f);\n\n        gameObject.transform.Rotate(new Vector3(0, 0, 1), actionZ);\n        gameObject.transform.Rotate(new Vector3(1, 0, 0), actionX);\n    }\n</code></pre> <p>By default the output from our provided PPO algorithm pre-clamps the values of <code>ActionBuffers.ContinuousActions</code> into the [-1, 1] range. It is a best practice to manually clip these as well, if you plan to use a 3rd party algorithm with your environment. As shown above, you can scale the control values as needed after clamping them.</p>"},{"location":"Learning-Environment-Design-Agents/#discrete-actions","title":"Discrete Actions","text":"<p>When an Agent's Policy uses discrete actions, the <code>ActionBuffers.DiscreteActions</code> passed to the Agent's <code>OnActionReceived()</code> function is an array of integers with length equal to <code>Discrete Branch Size</code>. When defining the discrete actions, <code>Branches</code> is an array of integers, each value corresponds to the number of possibilities for each branch.</p> <p>For example, if we wanted an Agent that can move in a plane and jump, we could define two branches (one for motion and one for jumping) because we want our agent be able to move and jump concurrently. We define the first branch to have 5 possible actions (don't move, go left, go right, go backward, go forward) and the second one to have 2 possible actions (don't jump, jump). The <code>OnActionReceived()</code> method would look something like:</p> <pre><code>// Get the action index for movement\nint movement = actionBuffers.DiscreteActions[0];\n// Get the action index for jumping\nint jump = actionBuffers.DiscreteActions[1];\n\n// Look up the index in the movement action list:\nif (movement == 1) { directionX = -1; }\nif (movement == 2) { directionX = 1; }\nif (movement == 3) { directionZ = -1; }\nif (movement == 4) { directionZ = 1; }\n// Look up the index in the jump action list:\nif (jump == 1 &amp;&amp; IsGrounded()) { directionY = 1; }\n\n// Apply the action results to move the Agent\ngameObject.GetComponent&lt;Rigidbody&gt;().AddForce(\n    new Vector3(\n        directionX * 40f, directionY * 300f, directionZ * 40f));\n</code></pre>"},{"location":"Learning-Environment-Design-Agents/#masking-discrete-actions","title":"Masking Discrete Actions","text":"<p>When using Discrete Actions, it is possible to specify that some actions are impossible for the next decision. When the Agent is controlled by a neural network, the Agent will be unable to perform the specified action. Note that when the Agent is controlled by its Heuristic, the Agent will still be able to decide to perform the masked action. In order to disallow an action, override the <code>Agent.WriteDiscreteActionMask()</code> virtual method, and call <code>SetActionEnabled()</code> on the provided <code>IDiscreteActionMask</code>:</p> <pre><code>public override void WriteDiscreteActionMask(IDiscreteActionMask actionMask)\n{\n    actionMask.SetActionEnabled(branch, actionIndex, isEnabled);\n}\n</code></pre> <p>Where:</p> <ul> <li><code>branch</code> is the index (starting at 0) of the branch on which you want to allow or disallow the action</li> <li><code>actionIndex</code> is the index of the action that you want to allow or disallow.</li> <li><code>isEnabled</code> is a bool indicating whether the action should be allowed or now.</li> </ul> <p>For example, if you have an Agent with 2 branches and on the first branch (branch 0) there are 4 possible actions : \"do nothing\", \"jump\", \"shoot\" and \"change weapon\". Then with the code bellow, the Agent will either \"do nothing\" or \"change weapon\" for their next decision (since action index 1 and 2 are masked)</p> <pre><code>actionMask.SetActionEnabled(0, 1, false);\nactionMask.SetActionEnabled(0, 2, false);\n</code></pre> <p>Notes:</p> <ul> <li>You can call <code>SetActionEnabled</code> multiple times if you want to put masks on multiple   branches.</li> <li>At each step, the state of an action is reset and enabled by default.</li> <li>You cannot mask all the actions of a branch.</li> <li>You cannot mask actions in continuous control.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#iactuator-interface-and-actuatorcomponents","title":"IActuator interface and ActuatorComponents","text":"<p>The Actuator API allows users to abstract behavior out of Agents and in to components (similar to the ISensor API).  The <code>IActuator</code> interface and <code>Agent</code> class both implement the <code>IActionReceiver</code> interface to allow for backward compatibility with the current <code>Agent.OnActionReceived</code>. This means you will not have to change your code until you decide to use the <code>IActuator</code> API.</p> <p>Like the <code>ISensor</code> interface, the <code>IActuator</code> interface is intended for advanced users.</p> <p>The <code>ActuatorComponent</code> abstract class is used to create the actual <code>IActuator</code> at runtime. It must be attached to the same <code>GameObject</code> as the <code>Agent</code>, or to a child <code>GameObject</code>.  Actuators and all of their data structures are initialized during <code>Agent.Initialize</code>.  This was done to prevent an unexpected allocations at runtime.</p> <p>You can find an example of an <code>IActuator</code> implementation in the <code>Basic</code> example scene. NOTE: you do not need to adjust the Actions in the Agent's <code>Behavior Parameters</code> when using an <code>IActuator</code> and <code>ActuatorComponents</code>.</p> <p>Internally, <code>Agent.OnActionReceived</code> uses an <code>IActuator</code> to send actions to the Agent, although this is mostly abstracted from the user.</p>"},{"location":"Learning-Environment-Design-Agents/#actions-summary-best-practices","title":"Actions Summary &amp; Best Practices","text":"<ul> <li>Agents can use <code>Discrete</code> and/or <code>Continuous</code> actions.</li> <li>Discrete actions can have multiple action branches, and it's possible to mask   certain actions so that they won't be taken.</li> <li>In general, fewer actions will make for easier learning.</li> <li>Be sure to set the Continuous Action Size and Discrete Branch Size to the desired   number for each type of action, and not greater, as doing the latter can interfere with the   efficiency of the training process.</li> <li>Continuous action values should be clipped to an   appropriate range. The provided PPO model automatically clips these values   between -1 and 1, but third party training systems may not do so.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#rewards","title":"Rewards","text":"<p>In reinforcement learning, the reward is a signal that the agent has done something right. The PPO reinforcement learning algorithm works by optimizing the choices an agent makes such that the agent earns the highest cumulative reward over time. The better your reward mechanism, the better your agent will learn.</p> <p>Note: Rewards are not used during inference by an Agent using a trained model and is also not used during imitation learning.</p> <p>Perhaps the best advice is to start simple and only add complexity as needed. In general, you should reward results rather than actions you think will lead to the desired results. You can even use the Agent's Heuristic to control the Agent while watching how it accumulates rewards.</p> <p>Allocate rewards to an Agent by calling the <code>AddReward()</code> or <code>SetReward()</code> methods on the agent. The reward assigned between each decision should be in the range [-1,1]. Values outside this range can lead to unstable training. The <code>reward</code> value is reset to zero when the agent receives a new decision. If there are multiple calls to <code>AddReward()</code> for a single agent decision, the rewards will be summed together to evaluate how good the previous decision was. The <code>SetReward()</code> will override all previous rewards given to an agent since the previous decision.</p>"},{"location":"Learning-Environment-Design-Agents/#examples","title":"Examples","text":"<p>You can examine the <code>OnActionReceived()</code> functions defined in the example environments to see how those projects allocate rewards.</p> <p>The <code>GridAgent</code> class in the GridWorld example uses a very simple reward system:</p> <pre><code>Collider[] hitObjects = Physics.OverlapBox(trueAgent.transform.position,\n                                           new Vector3(0.3f, 0.3f, 0.3f));\nif (hitObjects.Where(col =&gt; col.gameObject.tag == \"goal\").ToArray().Length == 1)\n{\n    AddReward(1.0f);\n    EndEpisode();\n}\nelse if (hitObjects.Where(col =&gt; col.gameObject.tag == \"pit\").ToArray().Length == 1)\n{\n    AddReward(-1f);\n    EndEpisode();\n}\n</code></pre> <p>The agent receives a positive reward when it reaches the goal and a negative reward when it falls into the pit. Otherwise, it gets no rewards. This is an example of a sparse reward system. The agent must explore a lot to find the infrequent reward.</p> <p>In contrast, the <code>AreaAgent</code> in the Area example gets a small negative reward every step. In order to get the maximum reward, the agent must finish its task of reaching the goal square as quickly as possible:</p> <pre><code>AddReward( -0.005f);\nMoveAgent(act);\n\nif (gameObject.transform.position.y &lt; 0.0f ||\n    Mathf.Abs(gameObject.transform.position.x - area.transform.position.x) &gt; 8f ||\n    Mathf.Abs(gameObject.transform.position.z + 5 - area.transform.position.z) &gt; 8)\n{\n    AddReward(-1f);\n    EndEpisode();\n}\n</code></pre> <p>The agent also gets a larger negative penalty if it falls off the playing surface.</p> <p>The <code>Ball3DAgent</code> in the 3DBall takes a similar approach, but allocates a small positive reward as long as the agent balances the ball. The agent can maximize its rewards by keeping the ball on the platform:</p> <pre><code>\nSetReward(0.1f);\n\n// When ball falls mark Agent as finished and give a negative penalty\nif ((ball.transform.position.y - gameObject.transform.position.y) &lt; -2f ||\n    Mathf.Abs(ball.transform.position.x - gameObject.transform.position.x) &gt; 3f ||\n    Mathf.Abs(ball.transform.position.z - gameObject.transform.position.z) &gt; 3f)\n{\n    SetReward(-1f);\n    EndEpisode();\n\n}\n</code></pre> <p>The <code>Ball3DAgent</code> also assigns a negative penalty when the ball falls off the platform.</p> <p>Note that all of these environments make use of the <code>EndEpisode()</code> method, which manually terminates an episode when a termination condition is reached. This can be called independently of the <code>Max Step</code> property.</p>"},{"location":"Learning-Environment-Design-Agents/#rewards-summary-best-practices","title":"Rewards Summary &amp; Best Practices","text":"<ul> <li>Use <code>AddReward()</code> to accumulate rewards between decisions. Use <code>SetReward()</code>   to overwrite any previous rewards accumulate between decisions.</li> <li>The magnitude of any given reward should typically not be greater than 1.0 in   order to ensure a more stable learning process.</li> <li>Positive rewards are often more helpful to shaping the desired behavior of an   agent than negative rewards. Excessive negative rewards can result in the   agent failing to learn any meaningful behavior.</li> <li>For locomotion tasks, a small positive reward (+0.1) for forward velocity is   typically used.</li> <li>If you want the agent to finish a task quickly, it is often helpful to provide   a small penalty every step (-0.05) that the agent does not complete the task.   In this case completion of the task should also coincide with the end of the   episode by calling <code>EndEpisode()</code> on the agent when it has accomplished its   goal.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#agent-properties","title":"Agent Properties","text":"<ul> <li><code>Behavior Parameters</code> - The parameters dictating what Policy the Agent will   receive.</li> <li><code>Behavior Name</code> - The identifier for the behavior. Agents with the same     behavior name will learn the same policy.</li> <li><code>Vector Observation</code><ul> <li><code>Space Size</code> - Length of vector observation for the Agent.</li> <li><code>Stacked Vectors</code> - The number of previous vector observations that will   be stacked and used collectively for decision making. This results in the   effective size of the vector observation being passed to the Policy being:   Space Size x Stacked Vectors.</li> </ul> </li> <li><code>Actions</code><ul> <li><code>Continuous Actions</code> - The number of concurrent continuous actions that  the Agent can take.</li> <li><code>Discrete Branches</code> - An array of integers, defines multiple concurrent   discrete actions. The values in the <code>Discrete Branches</code> array correspond   to the number of possible discrete values for each action branch.</li> </ul> </li> <li><code>Model</code> - The neural network model used for inference (obtained after     training)</li> <li><code>Inference Device</code> - Whether to use CPU or GPU to run the model during     inference</li> <li><code>Behavior Type</code> - Determines whether the Agent will do training, inference,     or use its Heuristic() method:<ul> <li><code>Default</code> - the Agent will train if they connect to a python trainer,   otherwise they will perform inference.</li> <li><code>Heuristic Only</code> - the Agent will always use the <code>Heuristic()</code> method.</li> <li><code>Inference Only</code> - the Agent will always perform inference.</li> </ul> </li> <li><code>Team ID</code> - Used to define the team for self-play</li> <li><code>Use Child Sensors</code> - Whether to use all Sensor components attached to child     GameObjects of this Agent.</li> <li><code>Max Step</code> - The per-agent maximum number of steps. Once this number is   reached, the Agent will be reset.</li> </ul>"},{"location":"Learning-Environment-Design-Agents/#destroying-an-agent","title":"Destroying an Agent","text":"<p>You can destroy an Agent GameObject during the simulation. Make sure that there is always at least one Agent training at all times by either spawning a new Agent every time one is destroyed or by re-spawning new Agents when the whole environment resets.</p>"},{"location":"Learning-Environment-Design-Agents/#defining-multi-agent-scenarios","title":"Defining Multi-agent Scenarios","text":""},{"location":"Learning-Environment-Design-Agents/#teams-for-adversarial-scenarios","title":"Teams for Adversarial Scenarios","text":"<p>Self-play is triggered by including the self-play hyperparameter hierarchy in the trainer configuration. To distinguish opposing agents, set the team ID to different integer values in the behavior parameters script on the agent prefab.</p> <p> </p> <p>Team ID must be 0 or an integer greater than 0.</p> <p>In symmetric games, since all agents (even on opposing teams) will share the same policy, they should have the same 'Behavior Name' in their Behavior Parameters Script. In asymmetric games, they should have a different Behavior Name in their Behavior Parameters script. Note, in asymmetric games, the agents must have both different Behavior Names and different team IDs!</p> <p>For examples of how to use this feature, you can see the trainer configurations and agent prefabs for our Tennis and Soccer environments. Tennis and Soccer provide examples of symmetric games. To train an asymmetric game, specify trainer configurations for each of your behavior names and include the self-play hyperparameter hierarchy in both.</p>"},{"location":"Learning-Environment-Design-Agents/#groups-for-cooperative-scenarios","title":"Groups for Cooperative Scenarios","text":"<p>Cooperative behavior in ML-Agents can be enabled by instantiating a <code>SimpleMultiAgentGroup</code>, typically in an environment controller or similar script, and adding agents to it using the <code>RegisterAgent(Agent agent)</code> method. Note that all agents added to the same <code>SimpleMultiAgentGroup</code> must have the same behavior name and Behavior Parameters. Using <code>SimpleMultiAgentGroup</code> enables the agents within a group to learn how to work together to achieve a common goal (i.e., maximize a group-given reward), even if one or more of the group members are removed before the episode ends. You can then use this group to add/set rewards, end or interrupt episodes at a group level using the <code>AddGroupReward()</code>, <code>SetGroupReward()</code>, <code>EndGroupEpisode()</code>, and <code>GroupEpisodeInterrupted()</code> methods. For example:</p> <pre><code>// Create a Multi Agent Group in Start() or Initialize()\nm_AgentGroup = new SimpleMultiAgentGroup();\n\n// Register agents in group at the beginning of an episode\nfor (var agent in AgentList)\n{\n  m_AgentGroup.RegisterAgent(agent);\n}\n\n// if the team scores a goal\nm_AgentGroup.AddGroupReward(rewardForGoal);\n\n// If the goal is reached and the episode is over\nm_AgentGroup.EndGroupEpisode();\nResetScene();\n\n// If time ran out and we need to interrupt the episode\nm_AgentGroup.GroupEpisodeInterrupted();\nResetScene();\n</code></pre> <p>Multi Agent Groups should be used with the MA-POCA trainer, which is explicitly designed to train cooperative environments. This can be enabled by using the <code>poca</code> trainer - see the training configurations doc for more information on configuring MA-POCA. When using MA-POCA, agents which are deactivated or removed from the Scene during the episode will still learn to contribute to the group's long term rewards, even if they are not active in the scene to experience them.</p> <p>See the Cooperative Push Block environment for an example of how to use Multi Agent Groups, and the Dungeon Escape environment for an example of how the Multi Agent Group can be used with agents that are removed from the scene mid-episode.</p> <p>NOTE: Groups differ from Teams (for competitive settings) in the following way - Agents working together should be added to the same Group, while agents playing against each other should be given different Team Ids. If in the Scene there is one playing field and two teams, there should be two Groups, one for each team, and each team should be assigned a different Team Id. If this playing field is duplicated many times in the Scene (e.g. for training speedup), there should be two Groups per playing field, and two unique Team Ids for the entire Scene. In environments with both Groups and Team Ids configured, MA-POCA and self-play can be used together for training. In the diagram below, there are two agents on each team, and two playing fields where teams are pitted against each other. All the blue agents should share a Team Id (and the orange ones a different ID), and there should be four group managers, one per pair of agents.</p> <p> </p> <p>Please see the SoccerTwos environment for an example.</p>"},{"location":"Learning-Environment-Design-Agents/#cooperative-behaviors-notes-and-best-practices","title":"Cooperative Behaviors Notes and Best Practices","text":"<ul> <li> <p>An agent can only be registered to one MultiAgentGroup at a time. If you want to re-assign an agent from one group to another, you have to unregister it from the current group first.</p> </li> <li> <p>Agents with different behavior names in the same group are not supported.</p> </li> <li> <p>Agents within groups should always set the <code>Max Steps</code> parameter in the Agent script to 0. Instead, handle Max Steps using the MultiAgentGroup by ending the episode for the entire Group using <code>GroupEpisodeInterrupted()</code>.</p> </li> <li> <p><code>EndGroupEpisode</code> and <code>GroupEpisodeInterrupted</code> do the same job in the game, but has slightly different effect on the training. If the episode is completed, you would want to call <code>EndGroupEpisode</code>. But if the episode is not over but it has been running for enough steps, i.e. reaching max step, you would call <code>GroupEpisodeInterrupted</code>.</p> </li> <li> <p>If an agent finished earlier, e.g. completed tasks/be removed/be killed in the game, do not call <code>EndEpisode()</code> on the Agent. Instead, disable the agent and re-enable it when the next episode starts, or destroy the agent entirely. This is because calling <code>EndEpisode()</code> will call <code>OnEpisodeBegin()</code>, which will reset the agent immediately. While it is possible to call <code>EndEpisode()</code> in this way, it is usually not the desired behavior when training groups of agents.</p> </li> <li> <p>If an agent that was disabled in a scene needs to be re-enabled, it must be re-registered to the MultiAgentGroup.</p> </li> <li> <p>Group rewards are meant to reinforce agents to act in the group's best interest instead of individual ones, and are treated differently than individual agent rewards during training. So calling <code>AddGroupReward()</code> is not equivalent to calling agent.AddReward() on each agent in the group.</p> </li> <li> <p>You can still add incremental rewards to agents using <code>Agent.AddReward()</code> if they are in a Group. These rewards will only be given to those agents and are received when the Agent is active.</p> </li> <li> <p>Environments which use Multi Agent Groups can be trained using PPO or SAC, but agents will not be able to learn from group rewards after deactivation/removal, nor will they behave as cooperatively.</p> </li> </ul>"},{"location":"Learning-Environment-Design-Agents/#recording-demonstrations","title":"Recording Demonstrations","text":"<p>In order to record demonstrations from an agent, add the <code>Demonstration Recorder</code> component to a GameObject in the scene which contains an <code>Agent</code> component. Once added, it is possible to name the demonstration that will be recorded from the agent.</p> <p> </p> <p>When <code>Record</code> is checked, a demonstration will be created whenever the scene is played from the Editor. Depending on the complexity of the task, anywhere from a few minutes or a few hours of demonstration data may be necessary to be useful for imitation learning. To specify an exact number of steps you want to record use the <code>Num Steps To Record</code> field and the editor will end your play session automatically once that many steps are recorded. If you set <code>Num Steps To Record</code> to <code>0</code> then recording will continue until you manually end the play session. Once the play session ends a <code>.demo</code> file will be created in the <code>Assets/Demonstrations</code> folder (by default). This file contains the demonstrations. Clicking on the file will provide metadata about the demonstration in the inspector.</p> <p> </p> <p>You can then specify the path to this file in your training configurations.</p>"},{"location":"Learning-Environment-Design/","title":"Designing a Learning Environment","text":"<p>This page contains general advice on how to design your learning environment, in addition to overviewing aspects of the ML-Agents Unity SDK that pertain to setting up your scene and simulation as opposed to designing your agents within the scene. We have a dedicated page on Designing Agents which includes how to instrument observations, actions and rewards, define teams for multi-agent scenarios and record agent demonstrations for imitation learning.</p> <p>To help on-board to the entire set of functionality provided by the ML-Agents Toolkit, we recommend exploring our API documentation. Additionally, our example environments are a great resource as they provide sample usage of almost all of our features.</p>"},{"location":"Learning-Environment-Design/#the-simulation-and-training-process","title":"The Simulation and Training Process","text":"<p>Training and simulation proceed in steps orchestrated by the ML-Agents Academy class. The Academy works with Agent objects in the scene to step through the simulation.</p> <p>During training, the external Python training process communicates with the Academy to run a series of episodes while it collects data and optimizes its neural network model. When training is completed successfully, you can add the trained model file to your Unity project for later use.</p> <p>The ML-Agents Academy class orchestrates the agent simulation loop as follows:</p> <ol> <li>Calls your Academy's <code>OnEnvironmentReset</code> delegate.</li> <li>Calls the <code>OnEpisodeBegin()</code> function for each Agent in the scene.</li> <li>Gathers information about the scene. This is done by calling the   <code>CollectObservations(VectorSensor sensor)</code> function for each Agent in the   scene, as well as updating their sensor and collecting the resulting   observations.</li> <li>Uses each Agent's Policy to decide on the Agent's next action.</li> <li>Calls the <code>OnActionReceived()</code> function for each Agent in the scene, passing    in the action chosen by the Agent's Policy.</li> <li>Calls the Agent's <code>OnEpisodeBegin()</code> function if the Agent has reached its    <code>Max Step</code> count or has otherwise marked itself as <code>EndEpisode()</code>.</li> </ol> <p>To create a training environment, extend the Agent class to implement the above methods whether you need to implement them or not depends on your specific scenario.</p>"},{"location":"Learning-Environment-Design/#organizing-the-unity-scene","title":"Organizing the Unity Scene","text":"<p>To train and use the ML-Agents Toolkit in a Unity scene, the scene as many Agent subclasses as you need. Agent instances should be attached to the GameObject representing that Agent.</p>"},{"location":"Learning-Environment-Design/#academy","title":"Academy","text":"<p>The Academy is a singleton which orchestrates Agents and their decision making processes. Only a single Academy exists at a time.</p>"},{"location":"Learning-Environment-Design/#academy-resetting","title":"Academy resetting","text":"<p>To alter the environment at the start of each episode, add your method to the Academy's OnEnvironmentReset action.</p> <pre><code>public class MySceneBehavior : MonoBehaviour\n{\n    public void Awake()\n    {\n        Academy.Instance.OnEnvironmentReset += EnvironmentReset;\n    }\n\n    void EnvironmentReset()\n    {\n        // Reset the scene here\n    }\n}\n</code></pre> <p>For example, you might want to reset an Agent to its starting position or move a goal to a random position. An environment resets when the <code>reset()</code> method is called on the Python <code>UnityEnvironment</code>.</p> <p>When you reset an environment, consider the factors that should change so that training is generalizable to different conditions. For example, if you were training a maze-solving agent, you would probably want to change the maze itself for each training episode. Otherwise, the agent would probably on learn to solve one, particular maze, not mazes in general.</p>"},{"location":"Learning-Environment-Design/#multiple-areas","title":"Multiple Areas","text":"<p>In many of the example environments, many copies of the training area are instantiated in the scene. This generally speeds up training, allowing the environment to gather many experiences in parallel. This can be achieved simply by instantiating many Agents with the same Behavior Name. If possible, consider designing your scene to support multiple areas.</p> <p>Check out our example environments to see examples of multiple areas. Additionally, the Making a New Learning Environment guide demonstrates this option.</p>"},{"location":"Learning-Environment-Design/#environments","title":"Environments","text":"<p>When you create a training environment in Unity, you must set up the scene so that it can be controlled by the external training process. Considerations include:</p> <ul> <li>The training scene must start automatically when your Unity application is   launched by the training process.</li> <li>The Academy must reset the scene to a valid starting point for each episode of   training.</li> <li>A training episode must have a definite end \u2014 either using <code>Max Steps</code> or by   each Agent ending its episode manually with <code>EndEpisode()</code>.</li> </ul>"},{"location":"Learning-Environment-Design/#environment-parameters","title":"Environment Parameters","text":"<p>Curriculum learning and environment parameter randomization are two training methods that control specific parameters in your environment. As such, it is important to ensure that your environment parameters are updated at each step to the correct values. To enable this, we expose a <code>EnvironmentParameters</code> C# class that you can use to retrieve the values of the parameters defined in the training configurations for both of those features. Please see our documentation for curriculum learning and environment parameter randomization for details.</p> <p>We recommend modifying the environment from the Agent's <code>OnEpisodeBegin()</code> function by leveraging <code>Academy.Instance.EnvironmentParameters</code>. See the WallJump example environment for a sample usage (specifically, WallJumpAgent.cs ).</p>"},{"location":"Learning-Environment-Design/#agent","title":"Agent","text":"<p>The Agent class represents an actor in the scene that collects observations and carries out actions. The Agent class is typically attached to the GameObject in the scene that otherwise represents the actor \u2014 for example, to a player object in a football game or a car object in a vehicle simulation. Every Agent must have appropriate <code>Behavior Parameters</code>.</p> <p>Generally, when creating an Agent, you should extend the Agent class and implement the <code>CollectObservations(VectorSensor sensor)</code> and <code>OnActionReceived()</code> methods:</p> <ul> <li><code>CollectObservations(VectorSensor sensor)</code> \u2014 Collects the Agent's observation   of its environment.</li> <li><code>OnActionReceived()</code> \u2014 Carries out the action chosen by the Agent's Policy and   assigns a reward to the current state.</li> </ul> <p>Your implementations of these functions determine how the Behavior Parameters assigned to this Agent must be set.</p> <p>You must also determine how an Agent finishes its task or times out. You can manually terminate an Agent episode in your <code>OnActionReceived()</code> function when the Agent has finished (or irrevocably failed) its task by calling the <code>EndEpisode()</code> function. You can also set the Agent's <code>Max Steps</code> property to a positive value and the Agent will consider the episode over after it has taken that many steps. You can use the <code>Agent.OnEpisodeBegin()</code> function to prepare the Agent to start again.</p> <p>See Agents for detailed information about programming your own Agents.</p>"},{"location":"Learning-Environment-Design/#recording-statistics","title":"Recording Statistics","text":"<p>We offer developers a mechanism to record statistics from within their Unity environments. These statistics are aggregated and generated during the training process. To record statistics, see the <code>StatsRecorder</code> C# class.</p> <p>See the FoodCollector example environment for a sample usage (specifically, FoodCollectorSettings.cs ).</p>"},{"location":"Learning-Environment-Examples/","title":"Example Learning Environments","text":"<p>The Unity ML-Agents Toolkit includes an expanding set of example environments that highlight the various features of the toolkit. These environments can also serve as templates for new environments or as ways to test new ML algorithms. Environments are located in <code>Project/Assets/ML-Agents/Examples</code> and summarized below.</p> <p>For the environments that highlight specific features of the toolkit, we provide the pre-trained model files and the training config file that enables you to train the scene yourself. The environments that are designed to serve as challenges for researchers do not have accompanying pre-trained model files or training configs and are marked as Optional below.</p> <p>This page only overviews the example environments we provide. To learn more on how to design and build your own environments see our Making a New Learning Environment page. If you would like to contribute environments, please see our contribution guidelines page.</p>"},{"location":"Learning-Environment-Examples/#basic","title":"Basic","text":"<ul> <li>Set-up: A linear movement task where the agent must move left or right to   rewarding states.</li> <li>Goal: Move to the most reward state.</li> <li>Agents: The environment contains one agent.</li> <li>Agent Reward Function:</li> <li>-0.01 at each step</li> <li>+0.1 for arriving at suboptimal state.</li> <li>+1.0 for arriving at optimal state.</li> <li>Behavior Parameters:</li> <li>Vector Observation space: One variable corresponding to current state.</li> <li>Actions: 1 discrete action branch with 3 actions (Move left, do nothing, move     right).</li> <li>Visual Observations: None</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward: 0.93</li> </ul>"},{"location":"Learning-Environment-Examples/#3dball-3d-balance-ball","title":"3DBall: 3D Balance Ball","text":"<ul> <li>Set-up: A balance-ball task, where the agent balances the ball on it's head.</li> <li>Goal: The agent must balance the ball on it's head for as long as possible.</li> <li>Agents: The environment contains 12 agents of the same kind, all using the   same Behavior Parameters.</li> <li>Agent Reward Function:</li> <li>+0.1 for every step the ball remains on it's head.</li> <li>-1.0 if the ball falls off.</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 8 variables corresponding to rotation of the agent     cube, and position and velocity of ball.</li> <li>Vector Observation space (Hard Version): 5 variables corresponding to     rotation of the agent cube and position of ball.</li> <li>Actions: 2 continuous actions, with one value corresponding to     X-rotation, and the other to Z-rotation.</li> <li>Visual Observations: Third-person view from the upper-front of the agent. Use     <code>Visual3DBall</code> scene.</li> <li>Float Properties: Three</li> <li>scale: Specifies the scale of the ball in the 3 dimensions (equal across the     three dimensions)<ul> <li>Default: 1</li> <li>Recommended Minimum: 0.2</li> <li>Recommended Maximum: 5</li> </ul> </li> <li>gravity: Magnitude of gravity<ul> <li>Default: 9.81</li> <li>Recommended Minimum: 4</li> <li>Recommended Maximum: 105</li> </ul> </li> <li>mass: Specifies mass of the ball<ul> <li>Default: 1</li> <li>Recommended Minimum: 0.1</li> <li>Recommended Maximum: 20</li> </ul> </li> <li>Benchmark Mean Reward: 100</li> </ul>"},{"location":"Learning-Environment-Examples/#gridworld","title":"GridWorld","text":"<ul> <li>Set-up: A multi-goal version of the grid-world task. Scene contains agent, goal,   and obstacles.</li> <li>Goal: The agent must navigate the grid to the appropriate goal while   avoiding the obstacles.</li> <li>Agents: The environment contains nine agents with the same Behavior   Parameters.</li> <li>Agent Reward Function:</li> <li>-0.01 for every step.</li> <li>+1.0 if the agent navigates to the correct goal (episode ends).</li> <li>-1.0 if the agent navigates to an incorrect goal (episode ends).</li> <li>Behavior Parameters:</li> <li>Vector Observation space: None</li> <li>Actions: 1 discrete action branch with 5 actions, corresponding to movement in     cardinal directions or not moving. Note that for this environment,     action masking     is turned on by default (this option can be toggled using the <code>Mask Actions</code>     checkbox within the <code>trueAgent</code> GameObject). The trained model file provided     was generated with action masking turned on.</li> <li>Visual Observations: One corresponding to top-down view of GridWorld.</li> <li>Goal Signal : A one hot vector corresponding to which color is the correct goal   for the Agent</li> <li>Float Properties: Three, corresponding to grid size, number of green goals, and   number of red goals.</li> <li>Benchmark Mean Reward: 0.8</li> </ul>"},{"location":"Learning-Environment-Examples/#push-block","title":"Push Block","text":"<ul> <li>Set-up: A platforming environment where the agent can push a block around.</li> <li>Goal: The agent must push the block to the goal.</li> <li>Agents: The environment contains one agent.</li> <li>Agent Reward Function:</li> <li>-0.0025 for every step.</li> <li>+1.0 if the block touches the goal.</li> <li>Behavior Parameters:</li> <li>Vector Observation space: (Continuous) 70 variables corresponding to 14     ray-casts each detecting one of three possible objects (wall, goal, or     block).</li> <li>Actions: 1 discrete action branch with 7 actions, corresponding to turn clockwise     and counterclockwise, move along four different face directions, or do nothing.</li> <li>Float Properties: Four</li> <li>block_scale: Scale of the block along the x and z dimensions<ul> <li>Default: 2</li> <li>Recommended Minimum: 0.5</li> <li>Recommended Maximum: 4</li> </ul> </li> <li>dynamic_friction: Coefficient of friction for the ground material acting on     moving objects<ul> <li>Default: 0</li> <li>Recommended Minimum: 0</li> <li>Recommended Maximum: 1</li> </ul> </li> <li>static_friction: Coefficient of friction for the ground material acting on     stationary objects<ul> <li>Default: 0</li> <li>Recommended Minimum: 0</li> <li>Recommended Maximum: 1</li> </ul> </li> <li>block_drag: Effect of air resistance on block<ul> <li>Default: 0.5</li> <li>Recommended Minimum: 0</li> <li>Recommended Maximum: 2000</li> </ul> </li> <li>Benchmark Mean Reward: 4.5</li> </ul>"},{"location":"Learning-Environment-Examples/#wall-jump","title":"Wall Jump","text":"<ul> <li>Set-up: A platforming environment where the agent can jump over a wall.</li> <li>Goal: The agent must use the block to scale the wall and reach the goal.</li> <li>Agents: The environment contains one agent linked to two different Models. The   Policy the agent is linked to changes depending on the height of the wall. The   change of Policy is done in the WallJumpAgent class.</li> <li>Agent Reward Function:</li> <li>-0.0005 for every step.</li> <li>+1.0 if the agent touches the goal.</li> <li>-1.0 if the agent falls off the platform.</li> <li>Behavior Parameters:</li> <li>Vector Observation space: Size of 74, corresponding to 14 ray casts each     detecting 4 possible objects. plus the global position of the agent and     whether or not the agent is grounded.</li> <li>Actions: 4 discrete action branches:<ul> <li>Forward Motion (3 possible actions: Forward, Backwards, No Action)</li> <li>Rotation (3 possible actions: Rotate Left, Rotate Right, No Action)</li> <li>Side Motion (3 possible actions: Left, Right, No Action)</li> <li>Jump (2 possible actions: Jump, No Action)</li> </ul> </li> <li>Visual Observations: None</li> <li>Float Properties: Four</li> <li>Benchmark Mean Reward (Big &amp; Small Wall): 0.8</li> </ul>"},{"location":"Learning-Environment-Examples/#crawler","title":"Crawler","text":"<ul> <li>Set-up: A creature with 4 arms and 4 forearms.</li> <li>Goal: The agents must move its body toward the goal direction without falling.</li> <li>Agents: The environment contains 10 agents with same Behavior Parameters.</li> <li>Agent Reward Function (independent):   The reward function is now geometric meaning the reward each step is a product   of all the rewards instead of a sum, this helps the agent try to maximize all   rewards instead of the easiest rewards.</li> <li>Body velocity matches goal velocity. (normalized between (0,1))</li> <li>Head direction alignment with goal direction. (normalized between (0,1))</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 172 variables corresponding to position, rotation,     velocity, and angular velocities of each limb plus the acceleration and     angular acceleration of the body.</li> <li>Actions: 20 continuous actions, corresponding to target     rotations for joints.</li> <li>Visual Observations: None</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward: 3000</li> </ul>"},{"location":"Learning-Environment-Examples/#worm","title":"Worm","text":"<ul> <li>Set-up: A worm with a head and 3 body segments.</li> <li>Goal: The agents must move its body toward the goal direction.</li> <li>Agents: The environment contains 10 agents with same Behavior Parameters.</li> <li>Agent Reward Function (independent):   The reward function is now geometric meaning the reward each step is a product   of all the rewards instead of a sum, this helps the agent try to maximize all   rewards instead of the easiest rewards.</li> <li>Body velocity matches goal velocity. (normalized between (0,1))</li> <li>Body direction alignment with goal direction. (normalized between (0,1))</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 64 variables corresponding to position, rotation,     velocity, and angular velocities of each limb plus the acceleration and     angular acceleration of the body.</li> <li>Actions: 9 continuous actions, corresponding to target     rotations for joints.</li> <li>Visual Observations: None</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward: 800</li> </ul>"},{"location":"Learning-Environment-Examples/#food-collector","title":"Food Collector","text":"<ul> <li>Set-up: A multi-agent environment where agents compete to collect food.</li> <li>Goal: The agents must learn to collect as many green food spheres as possible   while avoiding red spheres.</li> <li>Agents: The environment contains 5 agents with same Behavior Parameters.</li> <li>Agent Reward Function (independent):</li> <li>+1 for interaction with green spheres</li> <li>-1 for interaction with red spheres</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 53 corresponding to velocity of agent (2), whether     agent is frozen and/or shot its laser (2), plus grid based perception of     objects around agent's forward direction (40 by 40 with 6 different categories).</li> <li>Actions:<ul> <li>3 continuous actions correspond to Forward Motion, Side Motion and Rotation</li> <li>1 discrete action branch for Laser with 2 possible actions corresponding to   Shoot Laser or No Action</li> </ul> </li> <li>Visual Observations (Optional): First-person camera per-agent, plus one vector     flag representing the frozen state of the agent. This scene uses a combination     of vector and visual observations and the training will not succeed without     the frozen vector flag. Use <code>VisualFoodCollector</code> scene.</li> <li>Float Properties: Two</li> <li>laser_length: Length of the laser used by the agent<ul> <li>Default: 1</li> <li>Recommended Minimum: 0.2</li> <li>Recommended Maximum: 7</li> </ul> </li> <li>agent_scale: Specifies the scale of the agent in the 3 dimensions (equal     across the three dimensions)<ul> <li>Default: 1</li> <li>Recommended Minimum: 0.5</li> <li>Recommended Maximum: 5</li> </ul> </li> <li>Benchmark Mean Reward: 10</li> </ul>"},{"location":"Learning-Environment-Examples/#hallway","title":"Hallway","text":"<ul> <li>Set-up: Environment where the agent needs to find information in a room,   remember it, and use it to move to the correct goal.</li> <li>Goal: Move to the goal which corresponds to the color of the block in the   room.</li> <li>Agents: The environment contains one agent.</li> <li>Agent Reward Function (independent):</li> <li>+1 For moving to correct goal.</li> <li>-0.1 For moving to incorrect goal.</li> <li>-0.0003 Existential penalty.</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 30 corresponding to local ray-casts detecting     objects, goals, and walls.</li> <li>Actions: 1 discrete action Branch, with 4 actions corresponding to agent     rotation and forward/backward movement.</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward: 0.7</li> <li>To train this environment, you can enable curiosity by adding the <code>curiosity</code> reward signal     in <code>config/ppo/Hallway.yaml</code></li> </ul>"},{"location":"Learning-Environment-Examples/#soccer-twos","title":"Soccer Twos","text":"<ul> <li>Set-up: Environment where four agents compete in a 2 vs 2 toy soccer game.</li> <li>Goal:</li> <li>Get the ball into the opponent's goal while preventing the ball from     entering own goal.</li> <li>Agents: The environment contains two different Multi Agent Groups with two agents in each.   Parameters : SoccerTwos.</li> <li>Agent Reward Function (dependent):</li> <li>(1 - <code>accumulated time penalty</code>) When ball enters opponent's goal     <code>accumulated time penalty</code> is incremented by (1 / <code>MaxStep</code>) every fixed     update and is reset to 0 at the beginning of an episode.</li> <li>-1 When ball enters team's goal.</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 336 corresponding to 11 ray-casts forward     distributed over 120 degrees and 3 ray-casts backward distributed over 90     degrees each detecting 6 possible object types, along with the object's     distance. The forward ray-casts contribute 264 state dimensions and backward     72 state dimensions over three observation stacks.</li> <li>Actions: 3 discrete branched actions corresponding to     forward, backward, sideways movement, as well as rotation.</li> <li>Visual Observations: None</li> <li>Float Properties: Two</li> <li>ball_scale: Specifies the scale of the ball in the 3 dimensions (equal     across the three dimensions)<ul> <li>Default: 7.5</li> <li>Recommended minimum: 4</li> <li>Recommended maximum: 10</li> </ul> </li> <li>gravity: Magnitude of the gravity<ul> <li>Default: 9.81</li> <li>Recommended minimum: 6</li> <li>Recommended maximum: 20</li> </ul> </li> </ul>"},{"location":"Learning-Environment-Examples/#strikers-vs-goalie","title":"Strikers Vs. Goalie","text":"<ul> <li>Set-up: Environment where two agents compete in a 2 vs 1 soccer variant.</li> <li>Goal:</li> <li>Striker: Get the ball into the opponent's goal.</li> <li>Goalie: Keep the ball out of the goal.</li> <li>Agents: The environment contains two different Multi Agent Groups. One with two Strikers and the other one Goalie.   Behavior Parameters : Striker, Goalie.</li> <li>Striker Agent Reward Function (dependent):</li> <li>+1 When ball enters opponent's goal.</li> <li>-0.001 Existential penalty.</li> <li>Goalie Agent Reward Function (dependent):</li> <li>-1 When ball enters goal.</li> <li>0.001 Existential bonus.</li> <li>Behavior Parameters:</li> <li>Striker Vector Observation space: 294 corresponding to 11 ray-casts forward     distributed over 120 degrees and 3 ray-casts backward distributed over 90     degrees each detecting 5 possible object types, along with the object's     distance. The forward ray-casts contribute 231 state dimensions and backward     63 state dimensions over three observation stacks.</li> <li>Striker Actions: 3 discrete branched actions corresponding     to forward, backward, sideways movement, as well as rotation.</li> <li>Goalie Vector Observation space: 738 corresponding to 41 ray-casts     distributed over 360 degrees each detecting 4 possible object types, along     with the object's distance and 3 observation stacks.</li> <li>Goalie Actions: 3 discrete branched actions corresponding     to forward, backward, sideways movement, as well as rotation.</li> <li>Visual Observations: None</li> <li>Float Properties: Two</li> <li>ball_scale: Specifies the scale of the ball in the 3 dimensions (equal     across the three dimensions)<ul> <li>Default: 7.5</li> <li>Recommended minimum: 4</li> <li>Recommended maximum: 10</li> </ul> </li> <li>gravity: Magnitude of the gravity<ul> <li>Default: 9.81</li> <li>Recommended minimum: 6</li> <li>Recommended maximum: 20</li> </ul> </li> </ul>"},{"location":"Learning-Environment-Examples/#walker","title":"Walker","text":"<ul> <li>Set-up: Physics-based Humanoid agents with 26 degrees of freedom. These DOFs   correspond to articulation of the following body-parts: hips, chest, spine,   head, thighs, shins, feet, arms, forearms and hands.</li> <li>Goal: The agents must move its body toward the goal direction without falling.</li> <li>Agents: The environment contains 10 independent agents with same Behavior   Parameters.</li> <li>Agent Reward Function (independent):   The reward function is now geometric meaning the reward each step is a product   of all the rewards instead of a sum, this helps the agent try to maximize all   rewards instead of the easiest rewards.</li> <li>Body velocity matches goal velocity. (normalized between (0,1))</li> <li>Head direction alignment with goal direction. (normalized between (0,1))</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 243 variables corresponding to position, rotation,     velocity, and angular velocities of each limb, along with goal direction.</li> <li>Actions: 39 continuous actions, corresponding to target     rotations and strength applicable to the joints.</li> <li>Visual Observations: None</li> <li>Float Properties: Four</li> <li>gravity: Magnitude of gravity<ul> <li>Default: 9.81</li> <li>Recommended Minimum:</li> <li>Recommended Maximum:</li> </ul> </li> <li>hip_mass: Mass of the hip component of the walker<ul> <li>Default: 8</li> <li>Recommended Minimum: 7</li> <li>Recommended Maximum: 28</li> </ul> </li> <li>chest_mass: Mass of the chest component of the walker<ul> <li>Default: 8</li> <li>Recommended Minimum: 3</li> <li>Recommended Maximum: 20</li> </ul> </li> <li>spine_mass: Mass of the spine component of the walker<ul> <li>Default: 8</li> <li>Recommended Minimum: 3</li> <li>Recommended Maximum: 20</li> </ul> </li> <li>Benchmark Mean Reward : 2500</li> </ul>"},{"location":"Learning-Environment-Examples/#pyramids","title":"Pyramids","text":"<ul> <li>Set-up: Environment where the agent needs to press a button to spawn a   pyramid, then navigate to the pyramid, knock it over, and move to the gold   brick at the top.</li> <li>Goal: Move to the golden brick on top of the spawned pyramid.</li> <li>Agents: The environment contains one agent.</li> <li>Agent Reward Function (independent):</li> <li>+2 For moving to golden brick (minus 0.001 per step).</li> <li>Behavior Parameters:</li> <li>Vector Observation space: 148 corresponding to local ray-casts detecting     switch, bricks, golden brick, and walls, plus variable indicating switch     state.</li> <li>Actions: 1 discrete action branch, with 4 actions corresponding to agent rotation and     forward/backward movement.</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward: 1.75</li> </ul>"},{"location":"Learning-Environment-Examples/#match-3","title":"Match 3","text":"<ul> <li>Set-up: Simple match-3 game. Matched pieces are removed, and remaining pieces drop down. New pieces are spawned randomly at the top, with a chance of being \"special\".</li> <li>Goal: Maximize score from matching pieces.</li> <li>Agents: The environment contains several independent Agents.</li> <li>Agent Reward Function (independent):</li> <li>.01 for each normal piece cleared. Special pieces are worth 2x or 3x.</li> <li>Behavior Parameters:</li> <li>None</li> <li>Observations and actions are defined with a sensor and actuator respectively.</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward:</li> <li>39.5 for visual observations</li> <li>38.5 for vector observations</li> <li>34.2 for simple heuristic (pick a random valid move)</li> <li>37.0 for greedy heuristic (pick the highest-scoring valid move)</li> </ul>"},{"location":"Learning-Environment-Examples/#sorter","title":"Sorter","text":"<ul> <li>Set-up: The Agent is in a circular room with numbered tiles. The values of the  tiles are random between 1 and 20. The tiles present in the room are randomized  at each episode. When the Agent visits a tile, it turns green.</li> <li>Goal: Visit all the tiles in ascending order.</li> <li>Agents: The environment contains a single Agent</li> <li>Agent Reward Function:</li> <li>-.0002 Existential penalty.</li> <li>+1 For visiting the right tile</li> <li>-1 For visiting the wrong tile</li> <li>BehaviorParameters:</li> <li>Vector Observations : 4 : 2 floats for Position and 2 floats for orientation</li> <li>Variable Length Observations : Between 1 and 20 entities (one for each tile)   each with 22 observations, the first 20 are one hot encoding of the value of the tile,   the 21st and 22nd represent the position of the tile relative to the Agent and the 23rd   is <code>1</code> if the tile was visited and <code>0</code> otherwise.</li> <li>Actions: 3 discrete branched actions corresponding to forward, backward,   sideways movement, as well as rotation.</li> <li>Float Properties: One<ul> <li>num_tiles: The maximum number of tiles to sample.</li> <li>Default: 2</li> <li>Recommended Minimum: 1</li> <li>Recommended Maximum: 20</li> </ul> </li> <li>Benchmark Mean Reward: Depends on the number of tiles.</li> </ul>"},{"location":"Learning-Environment-Examples/#cooperative-push-block","title":"Cooperative Push Block","text":"<ul> <li>Set-up: Similar to Push Block, the agents are in an area with blocks that need to be pushed into a goal. Small blocks can be pushed by one agents and are worth +1 value, medium blocks require two agents to push in and are worth +2, and large blocks require all 3 agents to push and are worth +3.</li> <li>Goal: Push all blocks into the goal.</li> <li>Agents: The environment contains three Agents in a Multi Agent Group.</li> <li>Agent Reward Function:</li> <li>-0.0001 Existential penalty, as a group reward.</li> <li>+1, +2, or +3 for pushing in a block, added as a group reward.</li> <li>Behavior Parameters:</li> <li>Observation space: A single Grid Sensor with separate tags for each block size,     the goal, the walls, and other agents.</li> <li>Actions: 1 discrete action branch with 7 actions, corresponding to turn clockwise     and counterclockwise, move along four different face directions, or do nothing.</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward: 11 (Group Reward)</li> </ul>"},{"location":"Learning-Environment-Examples/#dungeon-escape","title":"Dungeon Escape","text":"<ul> <li>Set-up: Agents are trapped in a dungeon with a dragon, and must work together to escape.   To retrieve the key, one of the agents must find and slay the dragon, sacrificing itself   to do so. The dragon will drop a key for the others to use. The other agents can then pick   up this key and unlock the dungeon door. If the agents take too long, the dragon will escape   through a portal and the environment resets.</li> <li>Goal: Unlock the dungeon door and leave.</li> <li>Agents: The environment contains three Agents in a Multi Agent Group and one Dragon, which   moves in a predetermined pattern.</li> <li>Agent Reward Function:</li> <li>+1 group reward if any agent successfully unlocks the door and leaves the dungeon.</li> <li>Behavior Parameters:</li> <li>Observation space: A Ray Perception Sensor with separate tags for the walls, other agents,     the door, key, the dragon, and the dragon's portal. A single Vector Observation which indicates     whether the agent is holding a key.</li> <li>Actions: 1 discrete action branch with 7 actions, corresponding to turn clockwise     and counterclockwise, move along four different face directions, or do nothing.</li> <li>Float Properties: None</li> <li>Benchmark Mean Reward: 1.0 (Group Reward)</li> </ul>"},{"location":"Learning-Environment-Executable/","title":"Using an Environment Executable","text":"<p>This section will help you create and use built environments rather than the Editor to interact with an environment. Using an executable has some advantages over using the Editor:</p> <ul> <li>You can exchange executable with other people without having to share your   entire repository.</li> <li>You can put your executable on a remote machine for faster training.</li> <li>You can use <code>Server Build</code> (<code>Headless</code>) mode for faster training (as long as the executable does not need rendering).</li> <li>You can keep using the Unity Editor for other tasks while the agents are   training.</li> </ul>"},{"location":"Learning-Environment-Executable/#building-the-3dball-environment","title":"Building the 3DBall environment","text":"<p>The first step is to open the Unity scene containing the 3D Balance Ball environment:</p> <ol> <li>Launch Unity.</li> <li>On the Projects dialog, choose the Open option at the top of the window.</li> <li>Using the file dialog that opens, locate the <code>Project</code> folder within the    ML-Agents project and click Open.</li> <li>In the Project window, navigate to the folder    <code>Assets/ML-Agents/Examples/3DBall/Scenes/</code>.</li> <li>Double-click the <code>3DBall</code> file to load the scene containing the Balance Ball    environment.</li> </ol> <p></p> <p>Next, we want the set up scene to play correctly when the training process launches our environment executable. This means:</p> <ul> <li>The environment application runs in the background.</li> <li>No dialogs require interaction.</li> <li> <p>The correct scene loads automatically.</p> </li> <li> <p>Open Player Settings (menu: Edit &gt; Project Settings &gt; Player).</p> </li> <li>Under Resolution and Presentation:</li> <li>Ensure that Run in Background is Checked.</li> <li>Ensure that Display Resolution Dialog is set to Disabled. (Note: this     setting may not be available in newer versions of the editor.)</li> <li>Open the Build Settings window (menu:File &gt; Build Settings).</li> <li>Choose your target platform.</li> <li>(optional) Select \u201cDevelopment Build\u201d to      log debug messages.</li> <li>If any scenes are shown in the Scenes in Build list, make sure that the    3DBall Scene is the only one checked. (If the list is empty, then only the    current scene is included in the build).</li> <li>Click Build:</li> <li>In the File dialog, navigate to your ML-Agents directory.</li> <li>Assign a file name and click Save.</li> <li>(For Windows\uff09With Unity 2018.1, it will ask you to select a folder instead      of a file name. Create a subfolder within the root directory and select      that folder to build. In the following steps you will refer to this      subfolder's name as <code>env_name</code>. You cannot create builds in the Assets      folder</li> </ul> <p></p> <p>Now that we have a Unity executable containing the simulation environment, we can interact with it.</p>"},{"location":"Learning-Environment-Executable/#interacting-with-the-environment","title":"Interacting with the Environment","text":"<p>If you want to use the Python API to interact with your executable, you can pass the name of the executable with the argument 'file_name' of the <code>UnityEnvironment</code>. For instance:</p> <pre><code>from mlagents_envs.environment import UnityEnvironment\nenv = UnityEnvironment(file_name=&lt;env_name&gt;)\n</code></pre>"},{"location":"Learning-Environment-Executable/#training-the-environment","title":"Training the Environment","text":"<ol> <li>Open a command or terminal window.</li> <li>Navigate to the folder where you installed the ML-Agents Toolkit. If you    followed the default installation, then navigate to the    <code>ml-agents/</code> folder.</li> <li>Run    <code>mlagents-learn &lt;trainer-config-file&gt; --env=&lt;env_name&gt; --run-id=&lt;run-identifier&gt;</code>    Where:</li> <li><code>&lt;trainer-config-file&gt;</code> is the file path of the trainer configuration yaml</li> <li><code>&lt;env_name&gt;</code> is the name and path to the executable you exported from Unity      (without extension)</li> <li><code>&lt;run-identifier&gt;</code> is a string used to separate the results of different      training runs</li> </ol> <p>For example, if you are training with a 3DBall executable, and you saved it to the directory where you installed the ML-Agents Toolkit, run:</p> <pre><code>mlagents-learn config/ppo/3DBall.yaml --env=3DBall --run-id=firstRun\n</code></pre> <p>And you should see something like</p> <pre><code>ml-agents$ mlagents-learn config/ppo/3DBall.yaml --env=3DBall --run-id=first-run\n\n\n                        \u2584\u2584\u2584\u2593\u2593\u2593\u2593\n                   \u2553\u2593\u2593\u2593\u2593\u2593\u2593\u2588\u2593\u2593\u2593\u2593\u2593\n              ,\u2584\u2584\u2584m\u2580\u2580\u2580'  ,\u2593\u2593\u2593\u2580\u2593\u2593\u2584                           \u2593\u2593\u2593  \u2593\u2593\u258c\n            \u2584\u2593\u2593\u2593\u2580'      \u2584\u2593\u2593\u2580  \u2593\u2593\u2593      \u2584\u2584     \u2584\u2584 ,\u2584\u2584 \u2584\u2584\u2584\u2584   ,\u2584\u2584 \u2584\u2593\u2593\u258c\u2584 \u2584\u2584\u2584    ,\u2584\u2584\n          \u2584\u2593\u2593\u2593\u2580        \u2584\u2593\u2593\u2580   \u2590\u2593\u2593\u258c     \u2593\u2593\u258c   \u2590\u2593\u2593 \u2590\u2593\u2593\u2593\u2580\u2580\u2580\u2593\u2593\u258c \u2593\u2593\u2593 \u2580\u2593\u2593\u258c\u2580 ^\u2593\u2593\u258c  \u2552\u2593\u2593\u258c\n        \u2584\u2593\u2593\u2593\u2593\u2593\u2584\u2584\u2584\u2584\u2584\u2584\u2584\u2584\u2593\u2593\u2593      \u2593\u2580      \u2593\u2593\u258c   \u2590\u2593\u2593 \u2590\u2593\u2593    \u2593\u2593\u2593 \u2593\u2593\u2593  \u2593\u2593\u258c   \u2590\u2593\u2593\u2584 \u2593\u2593\u258c\n        \u2580\u2593\u2593\u2593\u2593\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2593\u2593\u2584     \u2593\u2593      \u2593\u2593\u258c   \u2590\u2593\u2593 \u2590\u2593\u2593    \u2593\u2593\u2593 \u2593\u2593\u2593  \u2593\u2593\u258c    \u2590\u2593\u2593\u2590\u2593\u2593\n          ^\u2588\u2593\u2593\u2593        \u2580\u2593\u2593\u2584   \u2590\u2593\u2593\u258c     \u2593\u2593\u2593\u2593\u2584\u2593\u2593\u2593\u2593 \u2590\u2593\u2593    \u2593\u2593\u2593 \u2593\u2593\u2593  \u2593\u2593\u2593\u2584    \u2593\u2593\u2593\u2593`\n            '\u2580\u2593\u2593\u2593\u2584      ^\u2593\u2593\u2593  \u2593\u2593\u2593       \u2514\u2580\u2580\u2580\u2580 \u2580\u2580 ^\u2580\u2580    `\u2580\u2580 `\u2580\u2580   '\u2580\u2580    \u2590\u2593\u2593\u258c\n               \u2580\u2580\u2580\u2580\u2593\u2584\u2584\u2584   \u2593\u2593\u2593\u2593\u2593\u2593,                                      \u2593\u2593\u2593\u2593\u2580\n                   `\u2580\u2588\u2593\u2593\u2593\u2593\u2593\u2593\u2593\u2593\u2593\u258c\n                        \u00ac`\u2580\u2580\u2580\u2588\u2593\n\n</code></pre> <p>Note: If you're using Anaconda, don't forget to activate the ml-agents environment first.</p> <p>If <code>mlagents-learn</code> runs correctly and starts training, you should see something like this:</p> <pre><code>CrashReporter: initialized\nMono path[0] = '/Users/dericp/workspace/ml-agents/3DBall.app/Contents/Resources/Data/Managed'\nMono config path = '/Users/dericp/workspace/ml-agents/3DBall.app/Contents/MonoBleedingEdge/etc'\nINFO:mlagents_envs:\n'Ball3DAcademy' started successfully!\nUnity Academy name: Ball3DAcademy\n\nINFO:mlagents_envs:Connected new brain:\nUnity brain name: Ball3DLearning\n        Number of Visual Observations (per agent): 0\n        Vector Observation space size (per agent): 8\n        Number of stacked Vector Observation: 1\nINFO:mlagents_envs:Hyperparameters for the PPO Trainer of brain Ball3DLearning:\n        batch_size:          64\n        beta:                0.001\n        buffer_size:         12000\n        epsilon:             0.2\n        gamma:               0.995\n        hidden_units:        128\n        lambd:               0.99\n        learning_rate:       0.0003\n        max_steps:           5.0e4\n        normalize:           True\n        num_epoch:           3\n        num_layers:          2\n        time_horizon:        1000\n        sequence_length:     64\n        summary_freq:        1000\n        use_recurrent:       False\n        memory_size:         256\n        use_curiosity:       False\n        curiosity_strength:  0.01\n        curiosity_enc_size:  128\n        output_path: ./results/first-run-0/Ball3DLearning\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 1000. Mean Reward: 1.242. Std of Reward: 0.746. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 2000. Mean Reward: 1.319. Std of Reward: 0.693. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 3000. Mean Reward: 1.804. Std of Reward: 1.056. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 4000. Mean Reward: 2.151. Std of Reward: 1.432. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 5000. Mean Reward: 3.175. Std of Reward: 2.250. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 6000. Mean Reward: 4.898. Std of Reward: 4.019. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 7000. Mean Reward: 6.716. Std of Reward: 5.125. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 8000. Mean Reward: 12.124. Std of Reward: 11.929. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 9000. Mean Reward: 18.151. Std of Reward: 16.871. Training.\nINFO:mlagents.trainers: first-run-0: Ball3DLearning: Step: 10000. Mean Reward: 27.284. Std of Reward: 28.667. Training.\n</code></pre> <p>You can press Ctrl+C to stop the training, and your trained model will be at <code>results/&lt;run-identifier&gt;/&lt;behavior_name&gt;.onnx</code>, which corresponds to your model's latest checkpoint. (Note: There is a known bug on Windows that causes the saving of the model to fail when you early terminate the training, it's recommended to wait until Step has reached the max_steps parameter you set in your config YAML.) You can now embed this trained model into your Agent by following the steps below:</p> <ol> <li>Move your model file into    <code>Project/Assets/ML-Agents/Examples/3DBall/TFModels/</code>.</li> <li>Open the Unity Editor, and select the 3DBall scene as described above.</li> <li>Select the 3DBall prefab from the Project window and select Agent.</li> <li>Drag the <code>&lt;behavior_name&gt;.onnx</code> file from the Project window of the Editor to    the Model placeholder in the Ball3DAgent inspector window.</li> <li>Press the Play button at the top of the Editor.</li> </ol>"},{"location":"Learning-Environment-Executable/#training-on-headless-server","title":"Training on Headless Server","text":"<p>To run training on headless server with no graphics rendering support, you need to turn off graphics display in the Unity executable. There are two ways to achieve this: 1. Pass <code>--no-graphics</code> option to mlagents-learn training command. This is equivalent to    adding <code>-nographics -batchmode</code> to the Unity executable's commandline. 2. Build your Unity executable with Server Build. You can find this setting in Build Settings    in the Unity Editor.</p> <p>If you want to train with graphics (for example, using camera and visual observations), you'll need to set up display rendering support (e.g. xvfb) on you server machine. In our Colab Notebook Tutorials, the Setup section has examples of setting up xvfb on servers.</p>"},{"location":"Limitations/","title":"Limitations","text":"<p>See the package-specific Limitations pages:</p> <ul> <li><code>com.unity.mlagents</code> Unity package</li> <li><code>mlagents</code> Python package</li> <li><code>mlagents_envs</code> Python package</li> </ul>"},{"location":"ML-Agents-Envs-README/","title":"Unity ML-Agents Python Interface","text":"<p>The <code>mlagents_envs</code> Python package is part of the ML-Agents Toolkit. <code>mlagents_envs</code> provides three Python APIs that allows direct interaction with the Unity game engine: - A single agent API (Gym API) - A gym-like multi-agent API (PettingZoo API) - A low-level API (LLAPI)</p> <p>The LLAPI is used by the trainer implementation in <code>mlagents</code>. <code>mlagents_envs</code> can be used independently of <code>mlagents</code> for Python communication.</p>"},{"location":"ML-Agents-Envs-README/#installation","title":"Installation","text":"<p>Install the <code>mlagents_envs</code> package with:</p> <pre><code>python -m pip install mlagents_envs==1.0.0\n</code></pre>"},{"location":"ML-Agents-Envs-README/#usage-more-information","title":"Usage &amp; More Information","text":"<p>See - Gym API Guide - PettingZoo API Guide - Python API Guide</p> <p>for more information on how to use the API to interact with a Unity environment.</p> <p>For more information on the ML-Agents Toolkit and how to instrument a Unity scene with the ML-Agents SDK, check out the main ML-Agents Toolkit documentation.</p>"},{"location":"ML-Agents-Envs-README/#limitations","title":"Limitations","text":"<ul> <li><code>mlagents_envs</code> uses localhost ports to exchange data between Unity and   Python. As such, multiple instances can have their ports collide, leading to   errors. Make sure to use a different port if you are using multiple instances   of <code>UnityEnvironment</code>.</li> <li>Communication between Unity and the Python <code>UnityEnvironment</code> is not secure.</li> <li>On Linux, ports are not released immediately after the communication closes.   As such, you cannot reuse ports right after closing a <code>UnityEnvironment</code>.</li> </ul>"},{"location":"ML-Agents-Overview/","title":"ML-Agents Toolkit Overview","text":"<p>Table of Contents</p> <ul> <li>Running Example: Training NPC Behaviors</li> <li>Key Components</li> <li>Training Modes</li> <li>Built-in Training and Inference<ul> <li>Cross-Platform Inference</li> </ul> </li> <li>Custom Training and Inference</li> <li>Flexible Training Scenarios</li> <li>Training Methods: Environment-agnostic</li> <li>A Quick Note on Reward Signals</li> <li>Deep Reinforcement Learning<ul> <li>Curiosity for Sparse-reward Environments</li> <li>RND for Sparse-reward Environments</li> </ul> </li> <li>Imitation Learning<ul> <li>GAIL (Generative Adversarial Imitation Learning)</li> <li>Behavioral Cloning (BC)</li> <li>Recording Demonstrations</li> </ul> </li> <li>Summary</li> <li>Training Methods: Environment-specific</li> <li>Training in Competitive Multi-Agent Environments with Self-Play</li> <li>Training in Cooperative Multi-Agent Environments with MA-POCA</li> <li>Solving Complex Tasks using Curriculum Learning</li> <li>Training Robust Agents using Environment Parameter Randomization</li> <li>Model Types</li> <li>Learning from Vector Observations</li> <li>Learning from Cameras using Convolutional Neural Networks</li> <li>Learning from Variable Length Observations using Attention</li> <li>Memory-enhanced Agents using Recurrent Neural Networks</li> <li>Additional Features</li> <li>Summary and Next Steps</li> </ul> <p>The Unity Machine Learning Agents Toolkit (ML-Agents Toolkit) is an open-source project that enables games and simulations to serve as environments for training intelligent agents. Agents can be trained using reinforcement learning, imitation learning, neuroevolution, or other machine learning methods through a simple-to-use Python API. We also provide implementations (based on PyTorch) of state-of-the-art algorithms to enable game developers and hobbyists to easily train intelligent agents for 2D, 3D and VR/AR games. These trained agents can be used for multiple purposes, including controlling NPC behavior (in a variety of settings such as multi-agent and adversarial), automated testing of game builds and evaluating different game design decisions pre-release. The ML-Agents Toolkit is mutually beneficial for both game developers and AI researchers as it provides a central platform where advances in AI can be evaluated on Unity\u2019s rich environments and then made accessible to the wider research and game developer communities.</p> <p>Depending on your background (i.e. researcher, game developer, hobbyist), you may have very different questions on your mind at the moment. To make your transition to the ML-Agents Toolkit easier, we provide several background pages that include overviews and helpful resources on the Unity Engine, machine learning and PyTorch. We strongly recommend browsing the relevant background pages if you're not familiar with a Unity scene, basic machine learning concepts or have not previously heard of PyTorch.</p> <p>The remainder of this page contains a deep dive into ML-Agents, its key components, different training modes and scenarios. By the end of it, you should have a good sense of what the ML-Agents Toolkit allows you to do. The subsequent documentation pages provide examples of how to use ML-Agents. To get started, watch this demo video of ML-Agents in action.</p>"},{"location":"ML-Agents-Overview/#running-example-training-npc-behaviors","title":"Running Example: Training NPC Behaviors","text":"<p>To help explain the material and terminology in this page, we'll use a hypothetical, running example throughout. We will explore the problem of training the behavior of a non-playable character (NPC) in a game. (An NPC is a game character that is never controlled by a human player and its behavior is pre-defined by the game developer.) More specifically, let's assume we're building a multi-player, war-themed game in which players control the soldiers. In this game, we have a single NPC who serves as a medic, finding and reviving wounded players. Lastly, let us assume that there are two teams, each with five players and one NPC medic.</p> <p>The behavior of a medic is quite complex. It first needs to avoid getting injured, which requires detecting when it is in danger and moving to a safe location. Second, it needs to be aware of which of its team members are injured and require assistance. In the case of multiple injuries, it needs to assess the degree of injury and decide who to help first. Lastly, a good medic will always place itself in a position where it can quickly help its team members. Factoring in all of these traits means that at every instance, the medic needs to measure several attributes of the environment (e.g. position of team members, position of enemies, which of its team members are injured and to what degree) and then decide on an action (e.g. hide from enemy fire, move to help one of its members). Given the large number of settings of the environment and the large number of actions that the medic can take, defining and implementing such complex behaviors by hand is challenging and prone to errors.</p> <p>With ML-Agents, it is possible to train the behaviors of such NPCs (called Agents) using a variety of methods. The basic idea is quite simple. We need to define three entities at every moment of the game (called environment):</p> <ul> <li>Observations - what the medic perceives about the environment.   Observations can be numeric and/or visual. Numeric observations measure   attributes of the environment from the point of view of the agent. For our   medic this would be attributes of the battlefield that are visible to it. For   most interesting environments, an agent will require several continuous   numeric observations. Visual observations, on the other hand, are images   generated from the cameras attached to the agent and represent what the agent   is seeing at that point in time. It is common to confuse an agent's   observation with the environment (or game) state. The environment state   represents information about the entire scene containing all the game   characters. The agents observation, however, only contains information that   the agent is aware of and is typically a subset of the environment state. For   example, the medic observation cannot include information about an enemy in   hiding that the medic is unaware of.</li> <li>Actions - what actions the medic can take. Similar to observations,   actions can either be continuous or discrete depending on the complexity of   the environment and agent. In the case of the medic, if the environment is a   simple grid world where only their location matters, then a discrete action   taking on one of four values (north, south, east, west) suffices. However, if   the environment is more complex and the medic can move freely then using two   continuous actions (one for direction and another for speed) is more   appropriate.</li> <li>Reward signals - a scalar value indicating how well the medic is doing.   Note that the reward signal need not be provided at every moment, but only   when the medic performs an action that is good or bad. For example, it can   receive a large negative reward if it dies, a modest positive reward whenever   it revives a wounded team member, and a modest negative reward when a wounded   team member dies due to lack of assistance. Note that the reward signal is how   the objectives of the task are communicated to the agent, so they need to be   set up in a manner where maximizing reward generates the desired optimal   behavior.</li> </ul> <p>After defining these three entities (the building blocks of a reinforcement learning task), we can now train the medic's behavior. This is achieved by simulating the environment for many trials where the medic, over time, learns what is the optimal action to take for every observation it measures by maximizing its future reward. The key is that by learning the actions that maximize its reward, the medic is learning the behaviors that make it a good medic (i.e. one who saves the most number of lives). In reinforcement learning terminology, the behavior that is learned is called a policy, which is essentially a (optimal) mapping from observations to actions. Note that the process of learning a policy through running simulations is called the training phase, while playing the game with an NPC that is using its learned policy is called the inference phase.</p> <p>The ML-Agents Toolkit provides all the necessary tools for using Unity as the simulation engine for learning the policies of different objects in a Unity environment. In the next few sections, we discuss how the ML-Agents Toolkit achieves this and what features it provides.</p>"},{"location":"ML-Agents-Overview/#key-components","title":"Key Components","text":"<p>The ML-Agents Toolkit contains five high-level components:</p> <ul> <li>Learning Environment - which contains the Unity scene and all the game   characters. The Unity scene provides the environment in which agents observe,   act, and learn. How you set up the Unity scene to serve as a learning   environment really depends on your goal. You may be trying to solve a specific   reinforcement learning problem of limited scope, in which case you can use the   same scene for both training and for testing trained agents. Or, you may be   training agents to operate in a complex game or simulation. In this case, it   might be more efficient and practical to create a purpose-built training   scene. The ML-Agents Toolkit includes an ML-Agents Unity SDK   (<code>com.unity.ml-agents</code> package) that enables you to transform any Unity scene   into a learning environment by defining the agents and their behaviors.</li> <li>Python Low-Level API - which contains a low-level Python interface for   interacting and manipulating a learning environment. Note that, unlike the   Learning Environment, the Python API is not part of Unity, but lives outside   and communicates with Unity through the Communicator. This API is contained in   a dedicated <code>mlagents_envs</code> Python package and is used by the Python training   process to communicate with and control the Academy during training. However,   it can be used for other purposes as well. For example, you could use the API   to use Unity as the simulation engine for your own machine learning   algorithms. See Python API for more information.</li> <li>External Communicator - which connects the Learning Environment with the   Python Low-Level API. It lives within the Learning Environment.</li> <li>Python Trainers which contains all the machine learning algorithms that   enable training agents. The algorithms are implemented in Python and are part   of their own <code>mlagents</code> Python package. The package exposes a single   command-line utility <code>mlagents-learn</code> that supports all the training methods   and options outlined in this document. The Python Trainers interface solely   with the Python Low-Level API.</li> <li>Gym Wrapper (not pictured). A common way in which machine learning   researchers interact with simulation environments is via a wrapper provided by   OpenAI called gym. We provide a gym wrapper   in the <code>ml-agents-envs</code> package and instructions for using   it with existing machine learning algorithms which utilize gym.</li> <li>PettingZoo Wrapper (not pictured) PettingZoo is python API for   interacting with multi-agent simulation environments that provides a   gym-like interface. We provide a PettingZoo wrapper for Unity ML-Agents   environments in the <code>ml-agents-envs</code> package and   instructions for using it with machine learning   algorithms.</li> </ul> <p> </p> <p>Simplified block diagram of ML-Agents.</p> <p>The Learning Environment contains two Unity Components that help organize the Unity scene:</p> <ul> <li>Agents - which is attached to a Unity GameObject (any character within a   scene) and handles generating its observations, performing the actions it   receives and assigning a reward (positive / negative) when appropriate. Each   Agent is linked to a Behavior.</li> <li>Behavior - defines specific attributes of the agent such as the number of   actions that agent can take. Each Behavior is uniquely identified by a   <code>Behavior Name</code> field. A Behavior can be thought as a function that receives   observations and rewards from the Agent and returns actions. A Behavior can be   of one of three types: Learning, Heuristic or Inference. A Learning Behavior   is one that is not, yet, defined but about to be trained. A Heuristic Behavior   is one that is defined by a hard-coded set of rules implemented in code. An   Inference Behavior is one that includes a trained Neural Network file. In   essence, after a Learning Behavior is trained, it becomes an Inference   Behavior.</li> </ul> <p>Every Learning Environment will always have one Agent for every character in the scene. While each Agent must be linked to a Behavior, it is possible for Agents that have similar observations and actions to have the same Behavior. In our sample game, we have two teams each with their own medic. Thus we will have two Agents in our Learning Environment, one for each medic, but both of these medics can have the same Behavior. This does not mean that at each instance they will have identical observation and action values.</p> <p> </p> <p>Example block diagram of ML-Agents Toolkit for our sample game.</p> <p>Note that in a single environment, there can be multiple Agents and multiple Behaviors at the same time. For example, if we expanded our game to include tank driver NPCs, then the Agent attached to those characters cannot share its Behavior with the Agent linked to the medics (medics and drivers have different actions). The Learning Environment through the Academy (not represented in the diagram) ensures that all the Agents are in sync in addition to controlling environment-wide settings.</p> <p>Lastly, it is possible to exchange data between Unity and Python outside of the machine learning loop through Side Channels. One example of using Side Channels is to exchange data with Python about Environment Parameters. The following diagram illustrates the above.</p> <p> </p>"},{"location":"ML-Agents-Overview/#training-modes","title":"Training Modes","text":"<p>Given the flexibility of ML-Agents, there are a few ways in which training and inference can proceed.</p>"},{"location":"ML-Agents-Overview/#built-in-training-and-inference","title":"Built-in Training and Inference","text":"<p>As mentioned previously, the ML-Agents Toolkit ships with several implementations of state-of-the-art algorithms for training intelligent agents. More specifically, during training, all the medics in the scene send their observations to the Python API through the External Communicator. The Python API processes these observations and sends back actions for each medic to take. During training these actions are mostly exploratory to help the Python API learn the best policy for each medic. Once training concludes, the learned policy for each medic can be exported as a model file. Then during the inference phase, the medics still continue to generate their observations, but instead of being sent to the Python API, they will be fed into their (internal, embedded) model to generate the optimal action for each medic to take at every point in time.</p> <p>The Getting Started Guide tutorial covers this training mode with the 3D Balance Ball sample environment.</p>"},{"location":"ML-Agents-Overview/#cross-platform-inference","title":"Cross-Platform Inference","text":"<p>It is important to note that the ML-Agents Toolkit leverages the Sentis to run the models within a Unity scene such that an agent can take the optimal action at each step. Given that Sentis support most platforms that Unity does, this means that any model you train with the ML-Agents Toolkit can be embedded into your Unity application that runs on any platform. See our dedicated blog post for additional information.</p>"},{"location":"ML-Agents-Overview/#custom-training-and-inference","title":"Custom Training and Inference","text":"<p>In the previous mode, the Agents were used for training to generate a PyTorch model that the Agents can later use. However, any user of the ML-Agents Toolkit can leverage their own algorithms for training. In this case, the behaviors of all the Agents in the scene will be controlled within Python. You can even turn your environment into a gym.</p> <p>We do not currently have a tutorial highlighting this mode, but you can learn more about the Python API here.</p>"},{"location":"ML-Agents-Overview/#flexible-training-scenarios","title":"Flexible Training Scenarios","text":"<p>While the discussion so-far has mostly focused on training a single agent, with ML-Agents, several training scenarios are possible. We are excited to see what kinds of novel and fun environments the community creates. For those new to training intelligent agents, below are a few examples that can serve as inspiration:</p> <ul> <li>Single-Agent. A single agent, with its own reward signal. The traditional way   of training an agent. An example is any single-player game, such as Chicken.</li> <li>Simultaneous Single-Agent. Multiple independent agents with independent reward   signals with same <code>Behavior Parameters</code>. A parallelized version of the   traditional training scenario, which can speed-up and stabilize the training   process. Helpful when you have multiple versions of the same character in an   environment who should learn similar behaviors. An example might be training a   dozen robot-arms to each open a door simultaneously.</li> <li>Adversarial Self-Play. Two interacting agents with inverse reward signals. In   two-player games, adversarial self-play can allow an agent to become   increasingly more skilled, while always having the perfectly matched opponent:   itself. This was the strategy employed when training AlphaGo, and more   recently used by OpenAI to train a human-beating 1-vs-1 Dota 2 agent.</li> <li>Cooperative Multi-Agent. Multiple interacting agents with a shared reward   signal with same or different <code>Behavior Parameters</code>. In this scenario, all   agents must work together to accomplish a task that cannot be done alone.   Examples include environments where each agent only has access to partial   information, which needs to be shared in order to accomplish the task or   collaboratively solve a puzzle.</li> <li>Competitive Multi-Agent. Multiple interacting agents with inverse reward   signals with same or different <code>Behavior Parameters</code>. In this scenario, agents   must compete with one another to either win a competition, or obtain some   limited set of resources. All team sports fall into this scenario.</li> <li>Ecosystem. Multiple interacting agents with independent reward signals with   same or different <code>Behavior Parameters</code>. This scenario can be thought of as   creating a small world in which animals with different goals all interact,   such as a savanna in which there might be zebras, elephants and giraffes, or   an autonomous driving simulation within an urban environment.</li> </ul>"},{"location":"ML-Agents-Overview/#training-methods-environment-agnostic","title":"Training Methods: Environment-agnostic","text":"<p>The remaining sections overview the various state-of-the-art machine learning algorithms that are part of the ML-Agents Toolkit. If you aren't studying machine and reinforcement learning as a subject and just want to train agents to accomplish tasks, you can treat these algorithms as black boxes. There are a few training-related parameters to adjust inside Unity as well as on the Python training side, but you do not need in-depth knowledge of the algorithms themselves to successfully create and train agents. Step-by-step procedures for running the training process are provided in the Training ML-Agents page.</p> <p>This section specifically focuses on the training methods that are available regardless of the specifics of your learning environment.</p>"},{"location":"ML-Agents-Overview/#a-quick-note-on-reward-signals","title":"A Quick Note on Reward Signals","text":"<p>In this section we introduce the concepts of intrinsic and extrinsic rewards, which helps explain some of the training methods.</p> <p>In reinforcement learning, the end goal for the Agent is to discover a behavior (a Policy) that maximizes a reward. You will need to provide the agent one or more reward signals to use during training. Typically, a reward is defined by your environment, and corresponds to reaching some goal. These are what we refer to as extrinsic rewards, as they are defined external of the learning algorithm.</p> <p>Rewards, however, can be defined outside of the environment as well, to encourage the agent to behave in certain ways, or to aid the learning of the true extrinsic reward. We refer to these rewards as intrinsic reward signals. The total reward that the agent will learn to maximize can be a mix of extrinsic and intrinsic reward signals.</p> <p>The ML-Agents Toolkit allows reward signals to be defined in a modular way, and we provide four reward signals that can the mixed and matched to help shape your agent's behavior:</p> <ul> <li><code>extrinsic</code>: represents the rewards defined in your environment, and is   enabled by default</li> <li><code>gail</code>: represents an intrinsic reward signal that is defined by GAIL (see   below)</li> <li><code>curiosity</code>: represents an intrinsic reward signal that encourages exploration   in sparse-reward environments that is defined by the Curiosity module (see   below).</li> <li><code>rnd</code>: represents an intrinsic reward signal that encourages exploration   in sparse-reward environments that is defined by the Curiosity module (see   below).</li> </ul>"},{"location":"ML-Agents-Overview/#deep-reinforcement-learning","title":"Deep Reinforcement Learning","text":"<p>ML-Agents provide an implementation of two reinforcement learning algorithms:</p> <ul> <li>Proximal Policy Optimization (PPO)</li> <li>Soft Actor-Critic (SAC)</li> </ul> <p>The default algorithm is PPO. This is a method that has been shown to be more general purpose and stable than many other RL algorithms.</p> <p>In contrast with PPO, SAC is off-policy, which means it can learn from experiences collected at any time during the past. As experiences are collected, they are placed in an experience replay buffer and randomly drawn during training. This makes SAC significantly more sample-efficient, often requiring 5-10 times less samples to learn the same task as PPO. However, SAC tends to require more model updates. SAC is a good choice for heavier or slower environments (about 0.1 seconds per step or more). SAC is also a \"maximum entropy\" algorithm, and enables exploration in an intrinsic way. Read more about maximum entropy RL here.</p>"},{"location":"ML-Agents-Overview/#curiosity-for-sparse-reward-environments","title":"Curiosity for Sparse-reward Environments","text":"<p>In environments where the agent receives rare or infrequent rewards (i.e. sparse-reward), an agent may never receive a reward signal on which to bootstrap its training process. This is a scenario where the use of an intrinsic reward signals can be valuable. Curiosity is one such signal which can help the agent explore when extrinsic rewards are sparse.</p> <p>The <code>curiosity</code> Reward Signal enables the Intrinsic Curiosity Module. This is an implementation of the approach described in Curiosity-driven Exploration by Self-supervised Prediction by Pathak, et al. It trains two networks:</p> <ul> <li>an inverse model, which takes the current and next observation of the agent,   encodes them, and uses the encoding to predict the action that was taken   between the observations</li> <li>a forward model, which takes the encoded current observation and action, and   predicts the next encoded observation.</li> </ul> <p>The loss of the forward model (the difference between the predicted and actual encoded observations) is used as the intrinsic reward, so the more surprised the model is, the larger the reward will be.</p> <p>For more information, see our dedicated blog post on the Curiosity module.</p>"},{"location":"ML-Agents-Overview/#rnd-for-sparse-reward-environments","title":"RND for Sparse-reward Environments","text":"<p>Similarly to Curiosity, Random Network Distillation (RND) is useful in sparse or rare reward environments as it helps the Agent explore. The RND Module is implemented following the paper Exploration by Random Network Distillation. RND uses two networks:</p> <ul> <li>The first is a network with fixed random weights that takes observations as inputs and  generates an encoding</li> <li>The second is a network with similar architecture that is trained to predict the  outputs of the first network and uses the observations the Agent collects as training data.</li> </ul> <p>The loss (the squared difference between the predicted and actual encoded observations) of the trained model is used as intrinsic reward. The more an Agent visits a state, the more accurate the predictions and the lower the rewards which encourages the Agent to explore new states with higher prediction errors.</p>"},{"location":"ML-Agents-Overview/#imitation-learning","title":"Imitation Learning","text":"<p>It is often more intuitive to simply demonstrate the behavior we want an agent to perform, rather than attempting to have it learn via trial-and-error methods. For example, instead of indirectly training a medic with the help of a reward function, we can give the medic real world examples of observations from the game and actions from a game controller to guide the medic's behavior. Imitation Learning uses pairs of observations and actions from a demonstration to learn a policy. See this video demo of imitation learning .</p> <p>Imitation learning can either be used alone or in conjunction with reinforcement learning. If used alone it can provide a mechanism for learning a specific type of behavior (i.e. a specific style of solving the task). If used in conjunction with reinforcement learning it can dramatically reduce the time the agent takes to solve the environment. This can be especially pronounced in sparse-reward environments. For instance, on the Pyramids environment, using 6 episodes of demonstrations can reduce training steps by more than 4 times. See Behavioral Cloning + GAIL + Curiosity + RL below.</p> <p> </p> <p>The ML-Agents Toolkit provides a way to learn directly from demonstrations, as well as use them to help speed up reward-based training (RL). We include two algorithms called Behavioral Cloning (BC) and Generative Adversarial Imitation Learning (GAIL). In most scenarios, you can combine these two features:</p> <ul> <li>If you want to help your agents learn (especially with environments that have   sparse rewards) using pre-recorded demonstrations, you can generally enable   both GAIL and Behavioral Cloning at low strengths in addition to having an   extrinsic reward. An example of this is provided for the PushBlock example   environment in <code>config/imitation/PushBlock.yaml</code>.</li> <li>If you want to train purely from demonstrations with GAIL and BC without an   extrinsic reward signal, please see the CrawlerStatic example environment under   in <code>config/imitation/CrawlerStatic.yaml</code>.</li> </ul> <p>Note: GAIL introduces a survivor bias to the learning process. That is, by giving positive rewards based on similarity to the expert, the agent is incentivized to remain alive for as long as possible. This can directly conflict with goal-oriented tasks like our PushBlock or Pyramids example environments where an agent must reach a goal state thus ending the episode as quickly as possible. In these cases, we strongly recommend that you use a low strength GAIL reward signal and a sparse extrinsic signal when the agent achieves the task. This way, the GAIL reward signal will guide the agent until it discovers the extrinsic signal and will not overpower it. If the agent appears to be ignoring the extrinsic reward signal, you should reduce the strength of GAIL.</p>"},{"location":"ML-Agents-Overview/#gail-generative-adversarial-imitation-learning","title":"GAIL (Generative Adversarial Imitation Learning)","text":"<p>GAIL, or Generative Adversarial Imitation Learning, uses an adversarial approach to reward your Agent for behaving similar to a set of demonstrations. GAIL can be used with or without environment rewards, and works well when there are a limited number of demonstrations. In this framework, a second neural network, the discriminator, is taught to distinguish whether an observation/action is from a demonstration or produced by the agent. This discriminator can then examine a new observation/action and provide it a reward based on how close it believes this new observation/action is to the provided demonstrations.</p> <p>At each training step, the agent tries to learn how to maximize this reward. Then, the discriminator is trained to better distinguish between demonstrations and agent state/actions. In this way, while the agent gets better and better at mimicking the demonstrations, the discriminator keeps getting stricter and stricter and the agent must try harder to \"fool\" it.</p> <p>This approach learns a policy that produces states and actions similar to the demonstrations, requiring fewer demonstrations than direct cloning of the actions. In addition to learning purely from demonstrations, the GAIL reward signal can be mixed with an extrinsic reward signal to guide the learning process.</p>"},{"location":"ML-Agents-Overview/#behavioral-cloning-bc","title":"Behavioral Cloning (BC)","text":"<p>BC trains the Agent's policy to exactly mimic the actions shown in a set of demonstrations. The BC feature can be enabled on the PPO or SAC trainers. As BC cannot generalize past the examples shown in the demonstrations, BC tends to work best when there exists demonstrations for nearly all of the states that the agent can experience, or in conjunction with GAIL and/or an extrinsic reward.</p>"},{"location":"ML-Agents-Overview/#recording-demonstrations","title":"Recording Demonstrations","text":"<p>Demonstrations of agent behavior can be recorded from the Unity Editor or build, and saved as assets. These demonstrations contain information on the observations, actions, and rewards for a given agent during the recording session. They can be managed in the Editor, as well as used for training with BC and GAIL. See the Designing Agents page for more information on how to record demonstrations for your agent.</p>"},{"location":"ML-Agents-Overview/#summary","title":"Summary","text":"<p>To summarize, we provide 3 training methods: BC, GAIL and RL (PPO or SAC) that can be used independently or together:</p> <ul> <li>BC can be used on its own or as a pre-training step before GAIL and/or RL</li> <li>GAIL can be used with or without extrinsic rewards</li> <li>RL can be used on its own (either PPO or SAC) or in conjunction with BC and/or   GAIL.</li> </ul> <p>Leveraging either BC or GAIL requires recording demonstrations to be provided as input to the training algorithms.</p>"},{"location":"ML-Agents-Overview/#training-methods-environment-specific","title":"Training Methods: Environment-specific","text":"<p>In addition to the three environment-agnostic training methods introduced in the previous section, the ML-Agents Toolkit provides additional methods that can aid in training behaviors for specific types of environments.</p>"},{"location":"ML-Agents-Overview/#training-in-competitive-multi-agent-environments-with-self-play","title":"Training in Competitive Multi-Agent Environments with Self-Play","text":"<p>ML-Agents provides the functionality to train both symmetric and asymmetric adversarial games with Self-Play. A symmetric game is one in which opposing agents are equal in form, function and objective. Examples of symmetric games are our Tennis and Soccer example environments. In reinforcement learning, this means both agents have the same observation and actions and learn from the same reward function and so they can share the same policy. In asymmetric games, this is not the case. An example of an asymmetric games are Hide and Seek. Agents in these types of games do not always have the same observation or actions and so sharing policy networks is not necessarily ideal.</p> <p>With self-play, an agent learns in adversarial games by competing against fixed, past versions of its opponent (which could be itself as in symmetric games) to provide a more stable, stationary learning environment. This is compared to competing against the current, best opponent in every episode, which is constantly changing (because it's learning).</p> <p>Self-play can be used with our implementations of both Proximal Policy Optimization (PPO) and Soft Actor-Critic (SAC). However, from the perspective of an individual agent, these scenarios appear to have non-stationary dynamics because the opponent is often changing. This can cause significant issues in the experience replay mechanism used by SAC. Thus, we recommend that users use PPO. For further reading on this issue in particular, see the paper Stabilising Experience Replay for Deep Multi-Agent Reinforcement Learning.</p> <p>See our Designing Agents page for more information on setting up teams in your Unity scene. Also, read our blog post on self-play for additional information. Additionally, check ELO Rating System the method we use to calculate the relative skill level between two players.</p>"},{"location":"ML-Agents-Overview/#training-in-cooperative-multi-agent-environments-with-ma-poca","title":"Training In Cooperative Multi-Agent Environments with MA-POCA","text":"<p>ML-Agents provides the functionality for training cooperative behaviors - i.e., groups of agents working towards a common goal, where the success of the individual is linked to the success of the whole group. In such a scenario, agents typically receive rewards as a group. For instance, if a team of agents wins a game against an opposing team, everyone is rewarded - even agents who did not directly contribute to the win. This makes learning what to do as an individual difficult - you may get a win for doing nothing, and a loss for doing your best.</p> <p>In ML-Agents, we provide MA-POCA (MultiAgent POsthumous Credit Assignment), which is a novel multi-agent trainer that trains a centralized critic, a neural network that acts as a \"coach\" for a whole group of agents. You can then give rewards to the team as a whole, and the agents will learn how best to contribute to achieving that reward. Agents can also be given rewards individually, and the team will work together to help the individual achieve those goals. During an episode, agents can be added or removed from the group, such as when agents spawn or die in a game. If agents are removed mid-episode (e.g., if teammates die or are removed from the game), they will still learn whether their actions contributed to the team winning later, enabling agents to take group-beneficial actions even if they result in the individual being removed from the game (i.e., self-sacrifice). MA-POCA can also be combined with self-play to train teams of agents to play against each other.</p> <p>To learn more about enabling cooperative behaviors for agents in an ML-Agents environment, check out this page.</p> <p>To learn more about MA-POCA, please see our paper On the Use and Misuse of Absorbing States in Multi-Agent Reinforcement Learning. For further reading, MA-POCA builds on previous work in multi-agent cooperative learning (Lowe et al., Foerster et al., among others) to enable the above use-cases.</p>"},{"location":"ML-Agents-Overview/#solving-complex-tasks-using-curriculum-learning","title":"Solving Complex Tasks using Curriculum Learning","text":"<p>Curriculum learning is a way of training a machine learning model where more difficult aspects of a problem are gradually introduced in such a way that the model is always optimally challenged. This idea has been around for a long time, and it is how we humans typically learn. If you imagine any childhood primary school education, there is an ordering of classes and topics. Arithmetic is taught before algebra, for example. Likewise, algebra is taught before calculus. The skills and knowledge learned in the earlier subjects provide a scaffolding for later lessons. The same principle can be applied to machine learning, where training on easier tasks can provide a scaffolding for harder tasks in the future.</p> <p>Imagine training the medic to scale a wall to arrive at a wounded team member. The starting point when training a medic to accomplish this task will be a random policy. That starting policy will have the medic running in circles, and will likely never, or very rarely scale the wall properly to revive their team member (and achieve the reward). If we start with a simpler task, such as moving toward an unobstructed team member, then the medic can easily learn to accomplish the task. From there, we can slowly add to the difficulty of the task by increasing the size of the wall until the medic can complete the initially near-impossible task of scaling the wall. We have included an environment to demonstrate this with ML-Agents, called Wall Jump.</p> <p></p> <p>Demonstration of a hypothetical curriculum training scenario in which a progressively taller wall obstructs the path to the goal.</p> <p>[Note: The example provided above is for instructional purposes, and was based on an early version of the Wall Jump example environment. As such, it is not possible to directly replicate the results here using that environment.]</p> <p>The ML-Agents Toolkit supports modifying custom environment parameters during the training process to aid in learning. This allows elements of the environment related to difficulty or complexity to be dynamically adjusted based on training progress. The Training ML-Agents page has more information on defining training curriculums.</p>"},{"location":"ML-Agents-Overview/#training-robust-agents-using-environment-parameter-randomization","title":"Training Robust Agents using Environment Parameter Randomization","text":"<p>An agent trained on a specific environment, may be unable to generalize to any tweaks or variations in the environment (in machine learning this is referred to as overfitting). This becomes problematic in cases where environments are instantiated with varying objects or properties. One mechanism to alleviate this and train more robust agents that can generalize to unseen variations of the environment is to expose them to these variations during training. Similar to Curriculum Learning, where environments become more difficult as the agent learns, the ML-Agents Toolkit provides a way to randomly sample parameters of the environment during training. We refer to this approach as Environment Parameter Randomization. For those familiar with Reinforcement Learning research, this approach is based on the concept of Domain Randomization. By using parameter randomization during training, the agent can be better suited to adapt (with higher performance) to future unseen variations of the environment.</p> Ball scale of 0.5 Ball scale of 4 <p>Example of variations of the 3D Ball environment. The environment parameters are <code>gravity</code>, <code>ball_mass</code> and <code>ball_scale</code>.</p>"},{"location":"ML-Agents-Overview/#model-types","title":"Model Types","text":"<p>Regardless of the training method deployed, there are a few model types that users can train using the ML-Agents Toolkit. This is due to the flexibility in defining agent observations, which include vector, ray cast and visual observations. You can learn more about how to instrument an agent's observation in the Designing Agents guide.</p>"},{"location":"ML-Agents-Overview/#learning-from-vector-observations","title":"Learning from Vector Observations","text":"<p>Whether an agent's observations are ray cast or vector, the ML-Agents Toolkit provides a fully connected neural network model to learn from those observations. At training time you can configure different aspects of this model such as the number of hidden units and number of layers.</p>"},{"location":"ML-Agents-Overview/#learning-from-cameras-using-convolutional-neural-networks","title":"Learning from Cameras using Convolutional Neural Networks","text":"<p>Unlike other platforms, where the agent\u2019s observation might be limited to a single vector or image, the ML-Agents Toolkit allows multiple cameras to be used for observations per agent. This enables agents to learn to integrate information from multiple visual streams. This can be helpful in several scenarios such as training a self-driving car which requires multiple cameras with different viewpoints, or a navigational agent which might need to integrate aerial and first-person visuals. You can learn more about adding visual observations to an agent here.</p> <p>When visual observations are utilized, the ML-Agents Toolkit leverages convolutional neural networks (CNN) to learn from the input images. We offer three network architectures:</p> <ul> <li>a simple encoder which consists of two convolutional layers</li> <li>the implementation proposed by   Mnih et al., consisting of   three convolutional layers,</li> <li>the IMPALA Resnet consisting of three   stacked layers, each with two residual blocks, making a much larger network   than the other two.</li> </ul> <p>The choice of the architecture depends on the visual complexity of the scene and the available computational resources.</p>"},{"location":"ML-Agents-Overview/#learning-from-variable-length-observations-using-attention","title":"Learning from Variable Length Observations using Attention","text":"<p>Using the ML-Agents Toolkit, it is possible to have agents learn from a varying number of inputs. To do so, each agent can keep track of a buffer of vector observations. At each step, the agent will go through all the elements in the buffer and extract information but the elements in the buffer can change at every step. This can be useful in scenarios in which the agents must keep track of a varying number of elements throughout the episode. For example in a game where an agent must learn to avoid projectiles, but the projectiles can vary in numbers.</p> <p></p> <p>You can learn more about variable length observations here. When variable length observations are utilized, the ML-Agents Toolkit leverages attention networks to learn from a varying number of entities. Agents using attention will ignore entities that are deemed not relevant and pay special attention to entities relevant to the current situation based on context.</p>"},{"location":"ML-Agents-Overview/#memory-enhanced-agents-using-recurrent-neural-networks","title":"Memory-enhanced Agents using Recurrent Neural Networks","text":"<p>Have you ever entered a room to get something and immediately forgot what you were looking for? Don't let that happen to your agents.</p> <p></p> <p>In some scenarios, agents must learn to remember the past in order to take the best decision. When an agent only has partial observability of the environment, keeping track of past observations can help the agent learn. Deciding what the agents should remember in order to solve a task is not easy to do by hand, but our training algorithms can learn to keep track of what is important to remember with LSTM.</p>"},{"location":"ML-Agents-Overview/#additional-features","title":"Additional Features","text":"<p>Beyond the flexible training scenarios available, the ML-Agents Toolkit includes additional features which improve the flexibility and interpretability of the training process.</p> <ul> <li>Concurrent Unity Instances - We enable developers to run concurrent,   parallel instances of the Unity executable during training. For certain   scenarios, this should speed up training. Check out our dedicated page on   creating a Unity executable and the   Training ML-Agents   page for instructions on how to set the number of concurrent instances.</li> <li>Recording Statistics from Unity - We enable developers to   record statistics from   within their Unity environments. These statistics are aggregated and generated   during the training process.</li> <li>Custom Side Channels - We enable developers to   create custom side channels to manage data transfer   between Unity and Python that is unique to their training workflow and/or   environment.</li> <li>Custom Samplers - We enable developers to   create custom sampling methods   for Environment Parameter Randomization. This enables users to customize this   training method for their particular environment.</li> </ul>"},{"location":"ML-Agents-Overview/#summary-and-next-steps","title":"Summary and Next Steps","text":"<p>To briefly summarize: The ML-Agents Toolkit enables games and simulations built in Unity to serve as the platform for training intelligent agents. It is designed to enable a large variety of training modes and scenarios and comes packed with several features to enable researchers and developers to leverage (and enhance) machine learning within Unity.</p> <p>In terms of next steps:</p> <ul> <li>For a walkthrough of running ML-Agents with a simple scene, check out the   Getting Started guide.</li> <li>For a \"Hello World\" introduction to creating your own Learning Environment,   check out the   Making a New Learning Environment page.</li> <li>For an overview on the more complex example environments that are provided in   this toolkit, check out the   Example Environments page.</li> <li>For more information on the various training options available, check out the   Training ML-Agents page.</li> </ul>"},{"location":"ML-Agents-README/","title":"Unity ML-Agents Trainers","text":"<p>The <code>mlagents</code> Python package is part of the ML-Agents Toolkit. <code>mlagents</code> provides a set of reinforcement and imitation learning algorithms designed to be used with Unity environments. The algorithms interface with the Python API provided by the <code>mlagents_envs</code> package. See here for more information on <code>mlagents_envs</code>.</p> <p>The algorithms can be accessed using the: <code>mlagents-learn</code> access point. See here for more information on using this package.</p>"},{"location":"ML-Agents-README/#installation","title":"Installation","text":"<p>Install the <code>mlagents</code> package with:</p> <pre><code>python -m pip install mlagents==1.0.0\n</code></pre>"},{"location":"ML-Agents-README/#usage-more-information","title":"Usage &amp; More Information","text":"<p>For more information on the ML-Agents Toolkit and how to instrument a Unity scene with the ML-Agents SDK, check out the main ML-Agents Toolkit documentation.</p>"},{"location":"ML-Agents-README/#limitations","title":"Limitations","text":"<ul> <li>Resuming self-play from a checkpoint resets the reported ELO to the default   value.</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/","title":"Unity ML-Agents Toolkit Documentation","text":""},{"location":"ML-Agents-Toolkit-Documentation/#installation-set-up","title":"Installation &amp; Set-up","text":"<ul> <li>Installation</li> <li>Using Virtual Environment</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#getting-started","title":"Getting Started","text":"<ul> <li>Getting Started Guide</li> <li>ML-Agents Toolkit Overview</li> <li>Background: Unity</li> <li>Background: Machine Learning</li> <li>Background: PyTorch</li> <li>Example Environments</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#creating-learning-environments","title":"Creating Learning Environments","text":"<ul> <li>Making a New Learning Environment</li> <li>Designing a Learning Environment</li> <li>Designing Agents</li> <li>Using an Executable Environment</li> <li>ML-Agents Package Settings</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#training-inference","title":"Training &amp; Inference","text":"<ul> <li>Training ML-Agents</li> <li>Training Configuration File</li> <li>Using TensorBoard to Observe Training</li> <li>Profiling Trainers</li> <li>Sentis</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#extending-ml-agents","title":"Extending ML-Agents","text":"<ul> <li>Creating Custom Side Channels</li> <li>Creating Custom Samplers for Environment Parameter Randomization</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#hugging-face-integration","title":"Hugging Face Integration","text":"<ul> <li>Using Hugging Face to download and upload trained models</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#python-tutorial-with-google-colab","title":"Python Tutorial with Google Colab","text":"<ul> <li>Using a UnityEnvironment</li> <li>Q-Learning with a UnityEnvironment</li> <li>Using Side Channels on a UnityEnvironment</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#help","title":"Help","text":"<ul> <li>Migrating from earlier versions of ML-Agents</li> <li>Frequently Asked Questions</li> <li>ML-Agents Glossary</li> <li>Limitations</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#api-docs","title":"API Docs","text":"<ul> <li>API Reference</li> <li>Python API Documentation</li> <li>How to use the Python API</li> <li>How to use the Unity Environment Registry</li> <li>Wrapping Learning Environment as a Gym (+Baselines/Dopamine Integration)</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#translations","title":"Translations","text":"<p>To make the Unity ML-Agents Toolkit accessible to the global research and Unity developer communities, we're attempting to create and maintain translations of our documentation. We've started with translating a subset of the documentation to one language (Chinese), but we hope to continue translating more pages and to other languages. Consequently, we welcome any enhancements and improvements from the community.</p> <ul> <li>Chinese</li> <li>Korean</li> </ul>"},{"location":"ML-Agents-Toolkit-Documentation/#deprecated-docs","title":"Deprecated Docs","text":"<p>We no longer use them ourselves and so they may not be up-to-date. We've decided to keep them up just in case they are helpful to you.</p> <ul> <li>Windows Anaconda Installation</li> <li>Using Docker</li> <li>Training on the Cloud with Amazon Web Services</li> <li>Training on the Cloud with Microsoft Azure</li> <li>Using the Video Recorder</li> </ul>"},{"location":"Migrating/","title":"Upgrading","text":""},{"location":"Migrating/#migrating","title":"Migrating","text":""},{"location":"Migrating/#migrating-to-the-ml-agents-envs-0300-package","title":"Migrating to the ml-agents-envs 0.30.0 package","text":"<ul> <li>Python 3.10.12 is now the minimum version of python supported due to python3.6 EOL.   Please update your python installation to 3.10.12 or higher.</li> <li>The <code>gym-unity</code> package has been refactored into the <code>ml-agents-envs</code> package. Please update your imports accordingly.</li> <li>Example:</li> <li>Before     <code>python     from gym_unity.unity_gym_env import UnityToGymWrapper</code></li> <li>After:     <code>python     from mlagents_envs.envs.unity_gym_env import UnityToGymWrapper</code></li> </ul>"},{"location":"Migrating/#migrating-the-package-to-version-3x","title":"Migrating the package to version 3.x","text":"<ul> <li>The official version of Unity ML-Agents supports is now 2023.2. If you run   into issues, please consider deleting your project's Library folder and reopening your   project.</li> </ul>"},{"location":"Migrating/#migrating-the-package-to-version-2x","title":"Migrating the package to version 2.x","text":"<ul> <li>The official version of Unity ML-Agents supports is now 2022.3 LTS. If you run   into issues, please consider deleting your project's Library folder and reopening your   project.</li> <li>If you used any of the APIs that were deprecated before version 2.0, you need to use their replacement. These deprecated APIs have been removed. See the migration steps bellow for specific API replacements.</li> </ul>"},{"location":"Migrating/#deprecated-methods-removed","title":"Deprecated methods removed","text":"Deprecated API Suggested Replacement <code>IActuator ActuatorComponent.CreateActuator()</code> <code>IActuator[] ActuatorComponent.CreateActuators()</code> <code>IActionReceiver.PackActions(in float[] destination)</code> none <code>Agent.CollectDiscreteActionMasks(DiscreteActionMasker actionMasker)</code> <code>Agent.WriteDiscreteActionMask(IDiscreteActionMask actionMask)</code> <code>Agent.Heuristic(float[] actionsOut)</code> <code>Agent.Heuristic(in ActionBuffers actionsOut)</code> <code>Agent.OnActionReceived(float[] vectorAction)</code> <code>Agent.OnActionReceived(ActionBuffers actions)</code> <code>Agent.GetAction()</code> <code>Agent.GetStoredActionBuffers()</code> <code>BrainParameters.SpaceType</code>, <code>VectorActionSize</code>, <code>VectorActionSpaceType</code>, and <code>NumActions</code> <code>BrainParameters.ActionSpec</code> <code>ObservationWriter.AddRange(IEnumerable&lt;float&gt; data, int writeOffset = 0)</code> <code>ObservationWriter. AddList(IList&lt;float&gt; data, int writeOffset = 0</code> <code>SensorComponent.IsVisual()</code> and <code>IsVector()</code> none <code>VectorSensor.AddObservation(IEnumerable&lt;float&gt; observation)</code> <code>VectorSensor.AddObservation(IList&lt;float&gt; observation)</code> <code>SideChannelsManager</code> <code>SideChannelManager</code>"},{"location":"Migrating/#idiscreteactionmask-changes","title":"IDiscreteActionMask changes","text":"<ul> <li>The interface for disabling specific discrete actions has changed. <code>IDiscreteActionMask.WriteMask()</code> was removed, and replaced with <code>SetActionEnabled()</code>. Instead of returning an IEnumerable with indices to disable, you can now call <code>SetActionEnabled</code> for each index to disable (or enable). As an example, if you overrode <code>Agent.WriteDiscreteActionMask()</code> with something that looked like:</li> </ul> <pre><code>public override void WriteDiscreteActionMask(IDiscreteActionMask actionMask)\n{\n    var branch = 2;\n    var actionsToDisable = new[] {1, 3};\n    actionMask.WriteMask(branch, actionsToDisable);\n}\n</code></pre> <p>the equivalent code would now be</p> <pre><code>public override void WriteDiscreteActionMask(IDiscreteActionMask actionMask)\n{\n    var branch = 2;\n    actionMask.SetActionEnabled(branch, 1, false);\n    actionMask.SetActionEnabled(branch, 3, false);\n}\n</code></pre>"},{"location":"Migrating/#iactuator-changes","title":"IActuator changes","text":"<ul> <li>The <code>IActuator</code> interface now implements <code>IHeuristicProvider</code>.  Please add the corresponding <code>Heuristic(in ActionBuffers)</code> method to your custom Actuator classes.</li> </ul>"},{"location":"Migrating/#isensor-and-sensorcomponent-changes","title":"ISensor and SensorComponent changes","text":"<ul> <li>The <code>ISensor.GetObservationShape()</code> method and <code>ITypedSensor</code> and <code>IDimensionPropertiesSensor</code> interfaces were removed, and <code>GetObservationSpec()</code> was added. You can use <code>ObservationSpec.Vector()</code> or <code>ObservationSpec.Visual()</code> to generate <code>ObservationSpec</code>s that are equivalent to the previous shape. For example, if your old ISensor looked like:</li> </ul> <pre><code>public override int[] GetObservationShape()\n{\n    return new[] { m_Height, m_Width, m_NumChannels };\n}\n</code></pre> <p>the equivalent code would now be</p> <pre><code>public override ObservationSpec GetObservationSpec()\n{\n    return ObservationSpec.Visual(m_Height, m_Width, m_NumChannels);\n}\n</code></pre> <ul> <li>The <code>ISensor.GetCompressionType()</code> method and <code>ISparseChannelSensor</code> interface was removed, and <code>GetCompressionSpec()</code> was added. You can use <code>CompressionSpec.Default()</code> or <code>CompressionSpec.Compressed()</code> to generate <code>CompressionSpec</code>s that are  equivalent to  the previous values. For example, if your old ISensor looked like:  ```csharp public virtual SensorCompressionType GetCompressionType() {     return SensorCompressionType.None; }</li> </ul> <pre><code>\nthe equivalent code would now be\n\n```csharp\npublic CompressionSpec GetCompressionSpec()\n{\n    return CompressionSpec.Default();\n}\n</code></pre> <ul> <li>The abstract method <code>SensorComponent.GetObservationShape()</code> was removed.</li> <li>The abstract method <code>SensorComponent.CreateSensor()</code> was replaced with <code>CreateSensors()</code>, which returns an <code>ISensor[]</code>.</li> </ul>"},{"location":"Migrating/#match3-integration-changes","title":"Match3 integration changes","text":"<p>The Match-3 integration utilities were moved from <code>com.unity.ml-agents.extensions</code> to <code>com.unity.ml-agents</code>.</p> <p>The <code>AbstractBoard</code> interface was changed: * <code>AbstractBoard</code> no longer contains <code>Rows</code>, <code>Columns</code>, <code>NumCellTypes</code>, and <code>NumSpecialTypes</code> fields. * <code>public abstract BoardSize GetMaxBoardSize()</code> was added as an abstract method. <code>BoardSize</code> is a new struct that contains <code>Rows</code>, <code>Columns</code>, <code>NumCellTypes</code>, and <code>NumSpecialTypes</code> fields, with the same meanings as the old <code>AbstractBoard</code> fields. * <code>public virtual BoardSize GetCurrentBoardSize()</code> is an optional method; by default it returns <code>GetMaxBoardSize()</code>. If you wish to use a single behavior to work with multiple board sizes, override <code>GetCurrentBoardSize()</code> to return the current <code>BoardSize</code>. The values returned by <code>GetCurrentBoardSize()</code> must be less than or equal to the corresponding values from <code>GetMaxBoardSize()</code>.</p>"},{"location":"Migrating/#gridsensor-changes","title":"GridSensor changes","text":"<p>The sensor configuration has changed: * The sensor implementation has been refactored and existing GridSensor created from extension package will not work in newer version. Some errors might show up when loading the old sensor in the scene. You'll need to remove the old sensor and create a new GridSensor. * These parameters names have changed but still refer to the same concept in the sensor: <code>GridNumSide</code> -&gt; <code>GridSize</code>, <code>RotateToAgent</code> -&gt; <code>RotateWithAgent</code>, <code>ObserveMask</code> -&gt; <code>ColliderMask</code>, <code>DetectableObjects</code> -&gt; <code>DetectableTags</code> * <code>DepthType</code> (<code>ChanelBase</code>/<code>ChannelHot</code>) option and <code>ChannelDepth</code> are removed. Now the default is one-hot encoding for detected tag. If you were using original GridSensor without overriding any method, switching to new GridSensor will produce similar effect for training although the actual observations will be slightly different.</p> <p>For creating your GridSensor implementation with custom data: * To create custom GridSensor, derive from <code>GridSensorBase</code> instead of <code>GridSensor</code>. Besides overriding <code>GetObjectData()</code>, you will also need to consider override <code>GetCellObservationSize()</code>, <code>IsDataNormalized()</code> and <code>GetProcessCollidersMethod()</code> according to the data you collect. Also you'll need to override <code>GridSensorComponent.GetGridSensors()</code> and return your custom GridSensor. * The input argument <code>tagIndex</code> in <code>GetObjectData()</code> has changed from 1-indexed to 0-indexed and the data type changed from <code>float</code> to <code>int</code>. The index of first detectable tag will be 0 instead of 1. <code>normalizedDistance</code> was removed from input. * The observation data should be written to the input <code>dataBuffer</code> instead of creating and returning a new array. * Removed the constraint of all data required to be normalized. You should specify it in <code>IsDataNormalized()</code>. Sensors with non-normalized data cannot use PNG compression type. * The sensor will not further encode the data received from <code>GetObjectData()</code> anymore. The values received from <code>GetObjectData()</code> will be the observation sent to the trainer.</p>"},{"location":"Migrating/#lstm-models-from-previous-releases-no-longer-supported","title":"LSTM models from previous releases no longer supported","text":"<p>The way that Sentis processes LSTM (recurrent neural networks) has changed. As a result, models trained with previous versions of ML-Agents will not be usable at inference if they were trained with a <code>memory</code> setting in the <code>.yaml</code> config file. If you want to use a model that has a recurrent neural network in this release of ML-Agents, you need to train the model using the python trainer from this release.</p>"},{"location":"Migrating/#migrating-to-release-13","title":"Migrating to Release 13","text":""},{"location":"Migrating/#implementing-iheuristic-in-your-iactuator-implementations","title":"Implementing IHeuristic in your IActuator implementations","text":"<ul> <li>If you have any custom actuators, you can now implement the <code>IHeuristicProvider</code> interface to have your actuator   handle the generation of actions when an Agent is running in heuristic mode.</li> <li><code>VectorSensor.AddObservation(IEnumerable&lt;float&gt;)</code> is deprecated. Use <code>VectorSensor.AddObservation(IList&lt;float&gt;)</code>   instead.</li> <li><code>ObservationWriter.AddRange()</code> is deprecated. Use <code>ObservationWriter.AddList()</code> instead.</li> <li><code>ActuatorComponent.CreateActuator()</code> is deprecated.  Please use override <code>ActuatorComponent.CreateActuators</code>   instead.  Since <code>ActuatorComponent.CreateActuator()</code> is abstract, you will still need to override it in your   class until it is removed.  It is only ever called if you don't override <code>ActuatorComponent.CreateActuators</code>.   You can suppress the warnings by surrounding the method with the following pragma:     <code>c#     #pragma warning disable 672     public IActuator CreateActuator() { ... }     #pragma warning restore 672</code></li> </ul>"},{"location":"Migrating/#migrating_1","title":"Migrating","text":""},{"location":"Migrating/#migrating-to-release-11","title":"Migrating to Release 11","text":""},{"location":"Migrating/#agent-virtual-method-deprecation","title":"Agent virtual method deprecation","text":"<ul> <li><code>Agent.CollectDiscreteActionMasks()</code> was deprecated and should be replaced with <code>Agent.WriteDiscreteActionMask()</code></li> <li><code>Agent.Heuristic(float[])</code> was deprecated and should be replaced with <code>Agent.Heuristic(ActionBuffers)</code>.</li> <li><code>Agent.OnActionReceived(float[])</code> was deprecated and should be replaced with <code>Agent.OnActionReceived(ActionBuffers)</code>.</li> <li><code>Agent.GetAction()</code> was deprecated and should be replaced with <code>Agent.GetStoredActionBuffers()</code>.</li> </ul> <p>The default implementation of these will continue to call the deprecated versions where appropriate. However, the deprecated versions may not be compatible with continuous and discrete actions on the same Agent.</p>"},{"location":"Migrating/#brainparameters-field-and-method-deprecation","title":"BrainParameters field and method deprecation","text":"<ul> <li><code>BrainParameters.VectorActionSize</code> was deprecated; you can now set <code>BrainParameters.ActionSpec.NumContinuousActions</code>  or <code>BrainParameters.ActionSpec.BranchSizes</code> instead.</li> <li><code>BrainParameters.VectorActionSpaceType</code> was deprecated, since both continuous and discrete actions can now be used.</li> <li><code>BrainParameters.NumActions()</code> was deprecated. Use  <code>BrainParameters.ActionSpec.NumContinuousActions</code> and  <code>BrainParameters.ActionSpec.NumDiscreteActions</code> instead.</li> </ul>"},{"location":"Migrating/#migrating-from-release-7-to-latest","title":"Migrating from Release 7 to latest","text":""},{"location":"Migrating/#important-changes","title":"Important changes","text":"<ul> <li>Some trainer files were moved. If you were using the <code>TrainerFactory</code> class, it was moved to the <code>trainers/trainer</code> folder.</li> <li>The <code>components</code> folder containing <code>bc</code> and <code>reward_signals</code> code was moved to the <code>trainers/tf</code> folder</li> </ul>"},{"location":"Migrating/#steps-to-migrate","title":"Steps to Migrate","text":"<ul> <li>Replace calls to <code>from mlagents.trainers.trainer_util import TrainerFactory</code> to <code>from mlagents.trainers.trainer import TrainerFactory</code></li> <li>Replace calls to <code>from mlagents.trainers.trainer_util import handle_existing_directories</code> to <code>from mlagents.trainers.directory_utils import validate_existing_directories</code></li> <li>Replace <code>mlagents.trainers.components</code> with <code>mlagents.trainers.tf.components</code> in your import statements.</li> </ul>"},{"location":"Migrating/#migrating-from-release-3-to-release-7","title":"Migrating from Release 3 to Release 7","text":""},{"location":"Migrating/#important-changes_1","title":"Important changes","text":"<ul> <li>The Parameter Randomization feature has been merged with the Curriculum feature. It is now possible to specify a sampler in the lesson of a Curriculum. Curriculum has been refactored and is now specified at the level of the parameter, not the behavior. More information here.(#4160)</li> </ul>"},{"location":"Migrating/#steps-to-migrate_1","title":"Steps to Migrate","text":"<ul> <li>The configuration format for curriculum and parameter randomization has changed. To upgrade your configuration files, an upgrade script has been provided. Run <code>python -m mlagents.trainers.upgrade_config -h</code> to see the script usage. Note that you will have had to upgrade to/install the current version of ML-Agents before running the script. To update manually:</li> <li>If your config file used a <code>parameter_randomization</code> section, rename that section to <code>environment_parameters</code></li> <li>If your config file used a <code>curriculum</code> section, you will need to rewrite your curriculum with this format.</li> </ul>"},{"location":"Migrating/#migrating-from-release-1-to-release-3","title":"Migrating from Release 1 to Release 3","text":""},{"location":"Migrating/#important-changes_2","title":"Important changes","text":"<ul> <li>Training artifacts (trained models, summaries) are now found under <code>results/</code>   instead of <code>summaries/</code> and <code>models/</code>.</li> <li>Trainer configuration, curriculum configuration, and parameter randomization   configuration have all been moved to a single YAML file. (#3791)</li> <li>Trainer configuration format has changed, and using a \"default\" behavior name has   been deprecated. (#3936)</li> <li><code>max_step</code> in the <code>TerminalStep</code> and <code>TerminalSteps</code> objects was renamed <code>interrupted</code>.</li> <li>On the UnityEnvironment API, <code>get_behavior_names()</code> and <code>get_behavior_specs()</code> methods were combined into the property <code>behavior_specs</code> that contains a mapping from behavior names to behavior spec.</li> <li><code>use_visual</code> and <code>allow_multiple_visual_obs</code> in the <code>UnityToGymWrapper</code> constructor were replaced by <code>allow_multiple_obs</code> which allows one or more visual observations and vector observations to be used simultaneously.</li> <li><code>--save-freq</code> has been removed from the CLI and is now configurable in the trainer configuration   file.</li> <li><code>--lesson</code> has been removed from the CLI. Lessons will resume when using <code>--resume</code>.   To start at a different lesson, modify your Curriculum configuration.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_2","title":"Steps to Migrate","text":"<ul> <li>To upgrade your configuration files, an upgrade script has been provided. Run   <code>python -m mlagents.trainers.upgrade_config -h</code> to see the script usage. Note that you will have   had to upgrade to/install the current version of ML-Agents before running the script.</li> </ul> <p>To do it manually, copy your <code>&lt;BehaviorName&gt;</code> sections from <code>trainer_config.yaml</code> into a separate trainer configuration file, under a <code>behaviors</code> section.   The <code>default</code> section is no longer needed. This new file should be specific to your environment, and not contain   configurations for multiple environments (unless they have the same Behavior Names).   - You will need to reformat your trainer settings as per the example.   - If your training uses curriculum, move those configurations under a <code>curriculum</code> section.   - If your training uses parameter randomization, move   the contents of the sampler config to <code>parameter_randomization</code> in the main trainer configuration. - If you are using <code>UnityEnvironment</code> directly, replace <code>max_step</code> with <code>interrupted</code>  in the <code>TerminalStep</code> and <code>TerminalSteps</code> objects.  - Replace usage of <code>get_behavior_names()</code> and <code>get_behavior_specs()</code> in UnityEnvironment with <code>behavior_specs</code>.  - If you use the <code>UnityToGymWrapper</code>, remove <code>use_visual</code> and <code>allow_multiple_visual_obs</code>  from the constructor and add <code>allow_multiple_obs = True</code> if the environment contains either  both visual and vector observations or multiple visual observations.  - If you were setting <code>--save-freq</code> in the CLI, add a <code>checkpoint_interval</code> value in your   trainer configuration, and set it equal to <code>save-freq * n_agents_in_scene</code>.</p>"},{"location":"Migrating/#migrating-from-015-to-release-1","title":"Migrating from 0.15 to Release 1","text":""},{"location":"Migrating/#important-changes_3","title":"Important changes","text":"<ul> <li>The <code>MLAgents</code> C# namespace was renamed to <code>Unity.MLAgents</code>, and other nested   namespaces were similarly renamed (#3843).</li> <li>The <code>--load</code> and <code>--train</code> command-line flags have been deprecated and   replaced with <code>--resume</code> and <code>--inference</code>.</li> <li>Running with the same <code>--run-id</code> twice will now throw an error.</li> <li>The <code>play_against_current_self_ratio</code> self-play trainer hyperparameter has   been renamed to <code>play_against_latest_model_ratio</code></li> <li>Removed the multi-agent gym option from the gym wrapper. For multi-agent   scenarios, use the Low Level Python API.</li> <li>The low level Python API has changed. You can look at the document   Low Level Python API documentation for more information. If   you use <code>mlagents-learn</code> for training, this should be a transparent change.</li> <li>The obsolete <code>Agent</code> methods <code>GiveModel</code>, <code>Done</code>, <code>InitializeAgent</code>,   <code>AgentAction</code> and <code>AgentReset</code> have been removed.</li> <li>The signature of <code>Agent.Heuristic()</code> was changed to take a <code>float[]</code> as a   parameter, instead of returning the array. This was done to prevent a common   source of error where users would return arrays of the wrong size.</li> <li>The SideChannel API has changed (#3833, #3660) :</li> <li>Introduced the <code>SideChannelManager</code> to register, unregister and access side     channels.</li> <li><code>EnvironmentParameters</code> replaces the default <code>FloatProperties</code>. You can     access the <code>EnvironmentParameters</code> with     <code>Academy.Instance.EnvironmentParameters</code> on C#. If you were previously     creating a <code>UnityEnvironment</code> in python and passing it a     <code>FloatPropertiesChannel</code>, create an <code>EnvironmentParametersChannel</code> instead.</li> <li><code>SideChannel.OnMessageReceived</code> is now a protected method (was public)</li> <li>SideChannel IncomingMessages methods now take an optional default argument,     which is used when trying to read more data than the message contains.</li> <li>Added a feature to allow sending stats from C# environments to TensorBoard     (and other python StatsWriters). To do this from your code, use     <code>Academy.Instance.StatsRecorder.Add(key, value)</code>(#3660)</li> <li><code>num_updates</code> and <code>train_interval</code> for SAC have been replaced with   <code>steps_per_update</code>.</li> <li>The <code>UnityEnv</code> class from the <code>gym-unity</code> package was renamed   <code>UnityToGymWrapper</code> and no longer creates the <code>UnityEnvironment</code>. Instead, the   <code>UnityEnvironment</code> must be passed as input to the constructor of   <code>UnityToGymWrapper</code></li> <li>Public fields and properties on several classes were renamed to follow Unity's   C# style conventions. All public fields and properties now use \"PascalCase\"   instead of \"camelCase\"; for example, <code>Agent.maxStep</code> was renamed to   <code>Agent.MaxStep</code>. For a full list of changes, see the pull request. (#3828)</li> <li><code>WriteAdapter</code> was renamed to <code>ObservationWriter</code>. (#3834)</li> </ul>"},{"location":"Migrating/#steps-to-migrate_3","title":"Steps to Migrate","text":"<ul> <li>In C# code, replace <code>using MLAgents</code> with <code>using Unity.MLAgents</code>. Replace   other nested namespaces such as <code>using MLAgents.Sensors</code> with   <code>using Unity.MLAgents.Sensors</code></li> <li>Replace the <code>--load</code> flag with <code>--resume</code> when calling <code>mlagents-learn</code>, and   don't use the <code>--train</code> flag as training will happen by default. To run   without training, use <code>--inference</code>.</li> <li>To force-overwrite files from a pre-existing run, add the <code>--force</code>   command-line flag.</li> <li>The Jupyter notebooks have been removed from the repository.</li> <li>If your Agent class overrides <code>Heuristic()</code>, change the signature to   <code>public override void Heuristic(float[] actionsOut)</code> and assign values to   <code>actionsOut</code> instead of returning an array.</li> <li>If you used <code>SideChannels</code> you must:</li> <li>Replace <code>Academy.FloatProperties</code> with     <code>Academy.Instance.EnvironmentParameters</code>.</li> <li><code>Academy.RegisterSideChannel</code> and <code>Academy.UnregisterSideChannel</code> were     removed. Use <code>SideChannelManager.RegisterSideChannel</code> and     <code>SideChannelManager.UnregisterSideChannel</code> instead.</li> <li>Set <code>steps_per_update</code> to be around equal to the number of agents in your   environment, times <code>num_updates</code> and divided by <code>train_interval</code>.</li> <li>Replace <code>UnityEnv</code> with <code>UnityToGymWrapper</code> in your code. The constructor no   longer takes a file name as input but a fully constructed <code>UnityEnvironment</code>   instead.</li> <li>Update uses of \"camelCase\" fields and properties to \"PascalCase\".</li> </ul>"},{"location":"Migrating/#migrating-from-014-to-015","title":"Migrating from 0.14 to 0.15","text":""},{"location":"Migrating/#important-changes_4","title":"Important changes","text":"<ul> <li>The <code>Agent.CollectObservations()</code> virtual method now takes as input a   <code>VectorSensor</code> sensor as argument. The <code>Agent.AddVectorObs()</code> methods were   removed.</li> <li>The <code>SetMask</code> was renamed to <code>SetMask</code> method must now be called on the   <code>DiscreteActionMasker</code> argument of the <code>CollectDiscreteActionMasks</code> virtual   method.</li> <li>We consolidated our API for <code>DiscreteActionMasker</code>. <code>SetMask</code> takes two   arguments : the branch index and the list of masked actions for that branch.</li> <li>The <code>Monitor</code> class has been moved to the Examples Project. (It was prone to   errors during testing)</li> <li>The <code>MLAgents.Sensors</code> namespace has been introduced. All sensors classes are   part of the <code>MLAgents.Sensors</code> namespace.</li> <li>The <code>MLAgents.SideChannels</code> namespace has been introduced. All side channel   classes are part of the <code>MLAgents.SideChannels</code> namespace.</li> <li>The interface for <code>RayPerceptionSensor.PerceiveStatic()</code> was changed to take   an input class and write to an output class, and the method was renamed to   <code>Perceive()</code>.</li> <li>The <code>SetMask</code> method must now be called on the <code>DiscreteActionMasker</code> argument   of the <code>CollectDiscreteActionMasks</code> method.</li> <li>The method <code>GetStepCount()</code> on the Agent class has been replaced with the   property getter <code>StepCount</code></li> <li>The <code>--multi-gpu</code> option has been removed temporarily.</li> <li><code>AgentInfo.actionMasks</code> has been renamed to <code>AgentInfo.discreteActionMasks</code>.</li> <li><code>BrainParameters</code> and <code>SpaceType</code> have been removed from the public API</li> <li><code>BehaviorParameters</code> have been removed from the public API.</li> <li><code>DecisionRequester</code> has been made internal (you can still use the   DecisionRequesterComponent from the inspector). <code>RepeatAction</code> was renamed   <code>TakeActionsBetweenDecisions</code> for clarity.</li> <li>The following methods in the <code>Agent</code> class have been renamed. The original   method names will be removed in a later release:</li> <li><code>InitializeAgent()</code> was renamed to <code>Initialize()</code></li> <li><code>AgentAction()</code> was renamed to <code>OnActionReceived()</code></li> <li><code>AgentReset()</code> was renamed to <code>OnEpisodeBegin()</code></li> <li><code>Done()</code> was renamed to <code>EndEpisode()</code></li> <li><code>GiveModel()</code> was renamed to <code>SetModel()</code></li> <li>The <code>IFloatProperties</code> interface has been removed.</li> <li>The interface for SideChannels was changed:</li> <li>In C#, <code>OnMessageReceived</code> now takes a <code>IncomingMessage</code> argument, and     <code>QueueMessageToSend</code> takes an <code>OutgoingMessage</code> argument.</li> <li>In python, <code>on_message_received</code> now takes a <code>IncomingMessage</code> argument, and     <code>queue_message_to_send</code> takes an <code>OutgoingMessage</code> argument.</li> <li>Automatic stepping for Academy is now controlled from the     AutomaticSteppingEnabled property.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_4","title":"Steps to Migrate","text":"<ul> <li>Add the <code>using MLAgents.Sensors;</code> in addition to <code>using MLAgents;</code> on top of   your Agent's script.</li> <li>Replace your Agent's implementation of <code>CollectObservations()</code> with   <code>CollectObservations(VectorSensor sensor)</code>. In addition, replace all calls to   <code>AddVectorObs()</code> with <code>sensor.AddObservation()</code> or   <code>sensor.AddOneHotObservation()</code> on the <code>VectorSensor</code> passed as argument.</li> <li>Replace your calls to <code>SetActionMask</code> on your Agent to   <code>DiscreteActionMasker.SetActionMask</code> in <code>CollectDiscreteActionMasks</code>.</li> <li>If you call <code>RayPerceptionSensor.PerceiveStatic()</code> manually, add your inputs   to a <code>RayPerceptionInput</code>. To get the previous float array output, iterate   through <code>RayPerceptionOutput.rayOutputs</code> and call   <code>RayPerceptionOutput.RayOutput.ToFloatArray()</code>.</li> <li>Replace all calls to <code>Agent.GetStepCount()</code> with <code>Agent.StepCount</code></li> <li>We strongly recommend replacing the following methods with their new   equivalent as they will be removed in a later release:</li> <li><code>InitializeAgent()</code> to <code>Initialize()</code></li> <li><code>AgentAction()</code> to <code>OnActionReceived()</code></li> <li><code>AgentReset()</code> to <code>OnEpisodeBegin()</code></li> <li><code>Done()</code> to <code>EndEpisode()</code></li> <li><code>GiveModel()</code> to <code>SetModel()</code></li> <li>Replace <code>IFloatProperties</code> variables with <code>FloatPropertiesChannel</code> variables.</li> <li>If you implemented custom <code>SideChannels</code>, update the signatures of your   methods, and add your data to the <code>OutgoingMessage</code> or read it from the   <code>IncomingMessage</code>.</li> <li>Replace calls to Academy.EnableAutomaticStepping()/DisableAutomaticStepping()   with Academy.AutomaticSteppingEnabled = true/false.</li> </ul>"},{"location":"Migrating/#migrating-from-013-to-014","title":"Migrating from 0.13 to 0.14","text":""},{"location":"Migrating/#important-changes_5","title":"Important changes","text":"<ul> <li>The <code>UnitySDK</code> folder has been split into a Unity Package   (<code>com.unity.ml-agents</code>) and an examples project (<code>Project</code>). Please follow the   Installation Guide to get up and running with this new repo   structure.</li> <li>Several changes were made to how agents are reset and marked as done:</li> <li>Calling <code>Done()</code> on the Agent will now reset it immediately and call the     <code>AgentReset</code> virtual method. (This is to simplify the previous logic in     which the Agent had to wait for the next <code>EnvironmentStep</code> to reset)</li> <li>The \"Reset on Done\" setting in AgentParameters was removed; this is now     effectively always true. <code>AgentOnDone</code> virtual method on the Agent has been     removed.</li> <li>The <code>Decision Period</code> and <code>On Demand decision</code> checkbox have been removed from   the Agent. On demand decision is now the default (calling <code>RequestDecision</code> on   the Agent manually.)</li> <li>The Academy class was changed to a singleton, and its virtual methods were   removed.</li> <li>Trainer steps are now counted per-Agent, not per-environment as in previous   versions. For instance, if you have 10 Agents in the scene, 20 environment   steps now corresponds to 200 steps as printed in the terminal and in   Tensorboard.</li> <li>Curriculum config files are now YAML formatted and all curricula for a   training run are combined into a single file.</li> <li>The <code>--num-runs</code> command-line option has been removed from <code>mlagents-learn</code>.</li> <li>Several fields on the Agent were removed or made private in order to simplify   the interface.</li> <li>The <code>agentParameters</code> field of the Agent has been removed. (Contained only     <code>maxStep</code> information)</li> <li><code>maxStep</code> is now a public field on the Agent. (Was moved from     <code>agentParameters</code>)</li> <li>The <code>Info</code> field of the Agent has been made private. (Was only used     internally and not meant to be modified outside of the Agent)</li> <li>The <code>GetReward()</code> method on the Agent has been removed. (It was being     confused with <code>GetCumulativeReward()</code>)</li> <li>The <code>AgentAction</code> struct no longer contains a <code>value</code> field. (Value     estimates were not set during inference)</li> <li>The <code>GetValueEstimate()</code> method on the Agent has been removed.</li> <li>The <code>UpdateValueAction()</code> method on the Agent has been removed.</li> <li>The deprecated <code>RayPerception3D</code> and <code>RayPerception2D</code> classes were removed,   and the <code>legacyHitFractionBehavior</code> argument was removed from   <code>RayPerceptionSensor.PerceiveStatic()</code>.</li> <li>RayPerceptionSensor was inconsistent in how it handle scale on the Agent's   transform. It now scales the ray length and sphere size for casting as the   transform's scale changes.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_5","title":"Steps to Migrate","text":"<ul> <li>Follow the instructions on how to install the <code>com.unity.ml-agents</code> package   into your project in the Installation Guide.</li> <li>If your Agent implemented <code>AgentOnDone</code> and did not have the checkbox   <code>Reset On Done</code> checked in the inspector, you must call the code that was in   <code>AgentOnDone</code> manually.</li> <li>If you give your Agent a reward or penalty at the end of an episode (e.g. for   reaching a goal or falling off of a platform), make sure you call   <code>AddReward()</code> or <code>SetReward()</code> before calling <code>Done()</code>. Previously, the   order didn't matter.</li> <li>If you were not using <code>On Demand Decision</code> for your Agent, you must add a   <code>DecisionRequester</code> component to your Agent GameObject and set its   <code>Decision Period</code> field to the old <code>Decision Period</code> of the Agent.</li> <li>If you have a class that inherits from Academy:</li> <li>If the class didn't override any of the virtual methods and didn't store any     additional data, you can just remove the old script from the scene.</li> <li>If the class had additional data, create a new MonoBehaviour and store the     data in the new MonoBehaviour instead.</li> <li>If the class overrode the virtual methods, create a new MonoBehaviour and     move the logic to it:<ul> <li>Move the InitializeAcademy code to MonoBehaviour.Awake</li> <li>Move the AcademyStep code to MonoBehaviour.FixedUpdate</li> <li>Move the OnDestroy code to MonoBehaviour.OnDestroy.</li> <li>Move the AcademyReset code to a new method and add it to the   Academy.OnEnvironmentReset action.</li> </ul> </li> <li>Multiply <code>max_steps</code> and <code>summary_freq</code> in your <code>trainer_config.yaml</code> by the   number of Agents in the scene.</li> <li>Combine curriculum configs into a single file. See   the WallJump curricula for an example of   the new curriculum config format. A tool like https://www.json2yaml.com may be   useful to help with the conversion.</li> <li>If you have a model trained which uses RayPerceptionSensor and has non-1.0   scale in the Agent's transform, it must be retrained.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v0120-to-v0130","title":"Migrating from ML-Agents Toolkit v0.12.0 to v0.13.0","text":""},{"location":"Migrating/#important-changes_6","title":"Important changes","text":"<ul> <li>The low level Python API has changed. You can look at the document   Low Level Python API documentation for more information. This   should only affect you if you're writing a custom trainer; if you use   <code>mlagents-learn</code> for training, this should be a transparent change.</li> <li><code>reset()</code> on the Low-Level Python API no longer takes a <code>train_mode</code>     argument. To modify the performance/speed of the engine, you must use an     <code>EngineConfigurationChannel</code></li> <li><code>reset()</code> on the Low-Level Python API no longer takes a <code>config</code> argument.     <code>UnityEnvironment</code> no longer has a <code>reset_parameters</code> field. To modify float     properties in the environment, you must use a <code>FloatPropertiesChannel</code>. For     more information, refer to the     Low Level Python API documentation</li> <li><code>CustomResetParameters</code> are now removed.</li> <li>The Academy no longer has a <code>Training Configuration</code> nor   <code>Inference Configuration</code> field in the inspector. To modify the configuration   from the Low-Level Python API, use an <code>EngineConfigurationChannel</code>. To modify   it during training, use the new command line arguments <code>--width</code>, <code>--height</code>,   <code>--quality-level</code>, <code>--time-scale</code> and <code>--target-frame-rate</code> in   <code>mlagents-learn</code>.</li> <li>The Academy no longer has a <code>Default Reset Parameters</code> field in the inspector.   The Academy class no longer has a <code>ResetParameters</code>. To access shared float   properties with Python, use the new <code>FloatProperties</code> field on the Academy.</li> <li>Offline Behavioral Cloning has been removed. To learn from demonstrations, use   the GAIL and Behavioral Cloning features with either PPO or SAC.</li> <li><code>mlagents.envs</code> was renamed to <code>mlagents_envs</code>. The previous repo layout   depended on PEP420, which caused   problems with some of our tooling such as mypy and pylint.</li> <li>The official version of Unity ML-Agents supports is now 2022.3 LTS. If you run   into issues, please consider deleting your library folder and reopening your   projects. You will need to install the Sentis package into your project in   order to ML-Agents to compile correctly.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_6","title":"Steps to Migrate","text":"<ul> <li>If you had a custom <code>Training Configuration</code> in the Academy inspector, you   will need to pass your custom configuration at every training run using the   new command line arguments <code>--width</code>, <code>--height</code>, <code>--quality-level</code>,   <code>--time-scale</code> and <code>--target-frame-rate</code>.</li> <li>If you were using <code>--slow</code> in <code>mlagents-learn</code>, you will need to pass your old   <code>Inference Configuration</code> of the Academy inspector with the new command line   arguments <code>--width</code>, <code>--height</code>, <code>--quality-level</code>, <code>--time-scale</code> and   <code>--target-frame-rate</code> instead.</li> <li>Any imports from <code>mlagents.envs</code> should be replaced with <code>mlagents_envs</code>.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v0110-to-v0120","title":"Migrating from ML-Agents Toolkit v0.11.0 to v0.12.0","text":""},{"location":"Migrating/#important-changes_7","title":"Important Changes","text":"<ul> <li>Text actions and observations, and custom action and observation protos have   been removed.</li> <li>RayPerception3D and RayPerception2D are marked deprecated, and will be removed   in a future release. They can be replaced by RayPerceptionSensorComponent3D   and RayPerceptionSensorComponent2D.</li> <li>The <code>Use Heuristic</code> checkbox in Behavior Parameters has been replaced with a   <code>Behavior Type</code> dropdown menu. This has the following options:</li> <li><code>Default</code> corresponds to the previous unchecked behavior, meaning that     Agents will train if they connect to a python trainer, otherwise they will     perform inference.</li> <li><code>Heuristic Only</code> means the Agent will always use the <code>Heuristic()</code> method.     This corresponds to having \"Use Heuristic\" selected in 0.11.0.</li> <li><code>Inference Only</code> means the Agent will always perform inference.</li> <li>ML-Agents was upgraded to use Sentis 1.2.0-exp.2 and is installed via the package manager.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_7","title":"Steps to Migrate","text":"<ul> <li>We fixed a bug in   <code>RayPerception3d.Perceive()</code> that was causing the <code>endOffset</code> to be used   incorrectly. However this may produce different behavior from previous   versions if you use a non-zero <code>startOffset</code>. To reproduce the old behavior,   you should increase the value of <code>endOffset</code> by <code>startOffset</code>. You can   verify your raycasts are performing as expected in scene view using the debug   rays.</li> <li>If you use RayPerception3D, replace it with RayPerceptionSensorComponent3D   (and similarly for 2D). The settings, such as ray angles and detectable tags,   are configured on the component now. RayPerception3D would contribute   <code>(# of rays) * (# of tags + 2)</code> to the State Size in Behavior Parameters, but   this is no longer necessary, so you should reduce the State Size by this   amount. Making this change will require retraining your model, since the   observations that RayPerceptionSensorComponent3D produces are different from   the old behavior.</li> <li>If you see messages such as   <code>The type or namespace 'Sentis' could not be found</code> or   <code>The type or namespace 'Google' could not be found</code>, you will need to   install the Sentis preview package.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v010-to-v0110","title":"Migrating from ML-Agents Toolkit v0.10 to v0.11.0","text":""},{"location":"Migrating/#important-changes_8","title":"Important Changes","text":"<ul> <li>The definition of the gRPC service has changed.</li> <li>The online BC training feature has been removed.</li> <li>The BroadcastHub has been deprecated. If there is a training Python process,   all LearningBrains in the scene will automatically be trained. If there is no   Python process, inference will be used.</li> <li>The Brain ScriptableObjects have been deprecated. The Brain Parameters are now   on the Agent and are referred to as Behavior Parameters. Make sure the   Behavior Parameters is attached to the Agent GameObject.</li> <li>To use a heuristic behavior, implement the <code>Heuristic()</code> method in the Agent   class and check the <code>use heuristic</code> checkbox in the Behavior Parameters.</li> <li>Several changes were made to the setup for visual observations (i.e. using   Cameras or RenderTextures):</li> <li>Camera resolutions are no longer stored in the Brain Parameters.</li> <li>AgentParameters no longer stores lists of Cameras and RenderTextures</li> <li>To add visual observations to an Agent, you must now attach a     CameraSensorComponent or RenderTextureComponent to the agent. The     corresponding Camera or RenderTexture can be added to these in the editor,     and the resolution and color/grayscale is configured on the component     itself.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_8","title":"Steps to Migrate","text":"<ul> <li>In order to be able to train, make sure both your ML-Agents Python package and   UnitySDK code come from the v0.11 release. Training will not work, for   example, if you update the ML-Agents Python package, and only update the API   Version in UnitySDK.</li> <li>If your Agents used visual observations, you must add a CameraSensorComponent   corresponding to each old Camera in the Agent's camera list (and similarly for   RenderTextures).</li> <li>Since Brain ScriptableObjects have been removed, you will need to delete all   the Brain ScriptableObjects from your <code>Assets</code> folder. Then, add a   <code>Behavior Parameters</code> component to each <code>Agent</code> GameObject. You will then need   to complete the fields on the new <code>Behavior Parameters</code> component with the   BrainParameters of the old Brain.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v09-to-v010","title":"Migrating from ML-Agents Toolkit v0.9 to v0.10","text":""},{"location":"Migrating/#important-changes_9","title":"Important Changes","text":"<ul> <li>We have updated the C# code in our repository to be in line with Unity Coding   Conventions. This has changed the name of some public facing classes and   enums.</li> <li>The example environments have been updated. If you were using these   environments to benchmark your training, please note that the resulting   rewards may be slightly different in v0.10.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_9","title":"Steps to Migrate","text":"<ul> <li><code>UnitySDK/Assets/ML-Agents/Scripts/Communicator.cs</code> and its class   <code>Communicator</code> have been renamed to   <code>UnitySDK/Assets/ML-Agents/Scripts/ICommunicator.cs</code> and <code>ICommunicator</code>   respectively.</li> <li>The <code>SpaceType</code> Enums <code>discrete</code>, and <code>continuous</code> have been renamed to   <code>Discrete</code> and <code>Continuous</code>.</li> <li>We have removed the <code>Done</code> call as well as the capacity to set <code>Max Steps</code> on   the Academy. Therefore an AcademyReset will never be triggered from C# (only   from Python). If you want to reset the simulation after a fixed number of   steps, or when an event in the simulation occurs, we recommend looking at our   multi-agent example environments (such as FoodCollector). In our examples,   groups of Agents can be reset through an \"Area\" that can reset groups of   Agents.</li> <li>The import for <code>mlagents.envs.UnityEnvironment</code> was removed. If you are using   the Python API, change <code>from mlagents_envs import UnityEnvironment</code> to   <code>from mlagents_envs.environment import UnityEnvironment</code>.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v08-to-v09","title":"Migrating from ML-Agents Toolkit v0.8 to v0.9","text":""},{"location":"Migrating/#important-changes_10","title":"Important Changes","text":"<ul> <li>We have changed the way reward signals (including Curiosity) are defined in   the <code>trainer_config.yaml</code>.</li> <li>When using multiple environments, every \"step\" is recorded in TensorBoard.</li> <li>The steps in the command line console corresponds to a single step of a single   environment. Previously, each step corresponded to one step for all   environments (i.e., <code>num_envs</code> steps).</li> </ul>"},{"location":"Migrating/#steps-to-migrate_10","title":"Steps to Migrate","text":"<ul> <li>If you were overriding any of these following parameters in your config file,   remove them from the top-level config and follow the steps below:</li> <li><code>gamma</code>: Define a new <code>extrinsic</code> reward signal and set it's <code>gamma</code> to your     new gamma.</li> <li><code>use_curiosity</code>, <code>curiosity_strength</code>, <code>curiosity_enc_size</code>: Define a     <code>curiosity</code> reward signal and set its <code>strength</code> to <code>curiosity_strength</code>,     and <code>encoding_size</code> to <code>curiosity_enc_size</code>. Give it the same <code>gamma</code> as     your <code>extrinsic</code> signal to mimic previous behavior.</li> <li>TensorBoards generated when running multiple environments in v0.8 are not   comparable to those generated in v0.9 in terms of step count. Multiply your   v0.8 step count by <code>num_envs</code> for an approximate comparison. You may need to   change <code>max_steps</code> in your config as appropriate as well.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v07-to-v08","title":"Migrating from ML-Agents Toolkit v0.7 to v0.8","text":""},{"location":"Migrating/#important-changes_11","title":"Important Changes","text":"<ul> <li>We have split the Python packages into two separate packages <code>ml-agents</code> and   <code>ml-agents-envs</code>.</li> <li><code>--worker-id</code> option of <code>learn.py</code> has been removed, use <code>--base-port</code> instead   if you'd like to run multiple instances of <code>learn.py</code>.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_11","title":"Steps to Migrate","text":"<ul> <li>If you are installing via PyPI, there is no change.</li> <li>If you intend to make modifications to <code>ml-agents</code> or <code>ml-agents-envs</code> please   check the Installing for Development in the   Installation documentation.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v06-to-v07","title":"Migrating from ML-Agents Toolkit v0.6 to v0.7","text":""},{"location":"Migrating/#important-changes_12","title":"Important Changes","text":"<ul> <li>We no longer support TFS and are now using the   Sentis</li> </ul>"},{"location":"Migrating/#steps-to-migrate_12","title":"Steps to Migrate","text":"<ul> <li>Make sure to remove the <code>ENABLE_TENSORFLOW</code> flag in your Unity Project   settings</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v05-to-v06","title":"Migrating from ML-Agents Toolkit v0.5 to v0.6","text":""},{"location":"Migrating/#important-changes_13","title":"Important Changes","text":"<ul> <li>Brains are now Scriptable Objects instead of MonoBehaviors.</li> <li> <p>You can no longer modify the type of a Brain. If you want to switch between   <code>PlayerBrain</code> and <code>LearningBrain</code> for multiple agents, you will need to assign   a new Brain to each agent separately. Note: You can pass the same Brain to   multiple agents in a scene by leveraging Unity's prefab system or look for all   the agents in a scene using the search bar of the <code>Hierarchy</code> window with the   word <code>Agent</code>.</p> </li> <li> <p>We replaced the Internal and External Brain with Learning Brain.   When you need to train a model, you need to drag it into the <code>Broadcast Hub</code>   inside the <code>Academy</code> and check the <code>Control</code> checkbox.</p> </li> <li>We removed the <code>Broadcast</code> checkbox of the Brain, to use the broadcast   functionality, you need to drag the Brain into the <code>Broadcast Hub</code>.</li> <li>When training multiple Brains at the same time, each model is now stored into   a separate model file rather than in the same file under different graph   scopes.</li> <li>The Learning Brain graph scope, placeholder names, output names and custom   placeholders can no longer be modified.</li> </ul>"},{"location":"Migrating/#steps-to-migrate_13","title":"Steps to Migrate","text":"<ul> <li>To update a scene from v0.5 to v0.6, you must:</li> <li>Remove the <code>Brain</code> GameObjects in the scene. (Delete all of the Brain     GameObjects under Academy in the scene.)</li> <li>Create new <code>Brain</code> Scriptable Objects using <code>Assets -&gt; Create -&gt; ML-Agents</code>     for each type of the Brain you plan to use, and put the created files under     a folder called Brains within your project.</li> <li>Edit their <code>Brain Parameters</code> to be the same as the parameters used in the     <code>Brain</code> GameObjects.</li> <li>Agents have a <code>Brain</code> field in the Inspector, you need to drag the     appropriate Brain ScriptableObject in it.</li> <li>The Academy has a <code>Broadcast Hub</code> field in the inspector, which is list of     brains used in the scene. To train or control your Brain from the     <code>mlagents-learn</code> Python script, you need to drag the relevant     <code>LearningBrain</code> ScriptableObjects used in your scene into entries into this     list.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v04-to-v05","title":"Migrating from ML-Agents Toolkit v0.4 to v0.5","text":""},{"location":"Migrating/#important","title":"Important","text":"<ul> <li>The Unity project <code>unity-environment</code> has been renamed <code>UnitySDK</code>.</li> <li>The <code>python</code> folder has been renamed to <code>ml-agents</code>. It now contains two   packages, <code>mlagents.env</code> and <code>mlagents.trainers</code>. <code>mlagents.env</code> can be used   to interact directly with a Unity environment, while <code>mlagents.trainers</code>   contains the classes for training agents.</li> <li>The supported Unity version has changed from <code>2017.1 or later</code> to   <code>2017.4 or later</code>. 2017.4 is an LTS (Long Term Support) version that helps us   maintain good quality and support. Earlier versions of Unity might still work,   but you may encounter an   error listed here.</li> </ul>"},{"location":"Migrating/#unity-api","title":"Unity API","text":"<ul> <li>Discrete Actions now use branches. You can   now specify concurrent discrete actions. You will need to update the Brain   Parameters in the Brain Inspector in all your environments that use discrete   actions. Refer to the   discrete action documentation   for more information.</li> </ul>"},{"location":"Migrating/#python-api","title":"Python API","text":"<ul> <li>In order to run a training session, you can now use the command   <code>mlagents-learn</code> instead of <code>python3 learn.py</code> after installing the <code>mlagents</code>   packages. This change is documented   here. For example, if we   previously ran</li> </ul> <p><code>sh   python3 learn.py 3DBall --train</code></p> <p>from the <code>python</code> subdirectory (which is changed to <code>ml-agents</code> subdirectory   in v0.5), we now run</p> <p><code>sh   mlagents-learn config/trainer_config.yaml --env=3DBall --train</code></p> <p>from the root directory where we installed the ML-Agents Toolkit.</p> <ul> <li>It is now required to specify the path to the yaml trainer configuration file   when running <code>mlagents-learn</code>. For an example trainer configuration file, see   trainer_config.yaml. An example of passing a   trainer configuration to <code>mlagents-learn</code> is shown above.</li> <li>The environment name is now passed through the <code>--env</code> option.</li> <li>Curriculum learning has been changed. In summary:</li> <li>Curriculum files for the same environment must now be placed into a folder.     Each curriculum file should be named after the Brain whose curriculum it     specifies.</li> <li><code>min_lesson_length</code> now specifies the minimum number of episodes in a lesson     and affects reward thresholding.</li> <li>It is no longer necessary to specify the <code>Max Steps</code> of the Academy to use     curriculum learning.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v03-to-v04","title":"Migrating from ML-Agents Toolkit v0.3 to v0.4","text":""},{"location":"Migrating/#unity-api_1","title":"Unity API","text":"<ul> <li><code>using MLAgents;</code> needs to be added in all of the C# scripts that use   ML-Agents.</li> </ul>"},{"location":"Migrating/#python-api_1","title":"Python API","text":"<ul> <li>We've changed some of the Python packages dependencies in requirement.txt   file. Make sure to run <code>pip3 install -e .</code> within your <code>ml-agents/python</code>   folder to update your Python packages.</li> </ul>"},{"location":"Migrating/#migrating-from-ml-agents-toolkit-v02-to-v03","title":"Migrating from ML-Agents Toolkit v0.2 to v0.3","text":"<p>There are a large number of new features and improvements in the ML-Agents toolkit v0.3 which change both the training process and Unity API in ways which will cause incompatibilities with environments made using older versions. This page is designed to highlight those changes for users familiar with v0.1 or v0.2 in order to ensure a smooth transition.</p>"},{"location":"Migrating/#important_1","title":"Important","text":"<ul> <li>The ML-Agents Toolkit is no longer compatible with Python 2.</li> </ul>"},{"location":"Migrating/#python-training","title":"Python Training","text":"<ul> <li>The training script <code>ppo.py</code> and <code>PPO.ipynb</code> Python notebook have been   replaced with a single <code>learn.py</code> script as the launching point for training   with ML-Agents. For more information on using <code>learn.py</code>, see   here.</li> <li>Hyperparameters for training Brains are now stored in the   <code>trainer_config.yaml</code> file. For more information on using this file, see   here.</li> </ul>"},{"location":"Migrating/#unity-api_2","title":"Unity API","text":"<ul> <li>Modifications to an Agent's rewards must now be done using either   <code>AddReward()</code> or <code>SetReward()</code>.</li> <li>Setting an Agent to done now requires the use of the <code>Done()</code> method.</li> <li><code>CollectStates()</code> has been replaced by <code>CollectObservations()</code>, which now no   longer returns a list of floats.</li> <li>To collect observations, call <code>AddVectorObs()</code> within <code>CollectObservations()</code>.   Note that you can call <code>AddVectorObs()</code> with floats, integers, lists and   arrays of floats, Vector3 and Quaternions.</li> <li><code>AgentStep()</code> has been replaced by <code>AgentAction()</code>.</li> <li><code>WaitTime()</code> has been removed.</li> <li>The <code>Frame Skip</code> field of the Academy is replaced by the Agent's   <code>Decision Frequency</code> field, enabling the Agent to make decisions at different   frequencies.</li> <li>The names of the inputs in the Internal Brain have been changed. You must   replace <code>state</code> with <code>vector_observation</code> and <code>observation</code> with   <code>visual_observation</code>. In addition, you must remove the <code>epsilon</code> placeholder.</li> </ul>"},{"location":"Migrating/#semantics","title":"Semantics","text":"<p>In order to more closely align with the terminology used in the Reinforcement Learning field, and to be more descriptive, we have changed the names of some of the concepts used in ML-Agents. The changes are highlighted in the table below.</p> Old - v0.2 and earlier New - v0.3 and later State Vector Observation Observation Visual Observation Action Vector Action N/A Text Observation N/A Text Action"},{"location":"Package-Settings/","title":"ML-Agents Package Settings","text":"<p>ML-Agents Package Settings contains settings that apply to the whole project. It allows you to configure ML-Agents-specific settings in the Editor.  These settings are available for use in both the Editor and Player.</p> <p>You can find them at <code>Edit</code> &gt; <code>Project Settings...</code> &gt; <code>ML-Agents</code>. It lists out all the available settings and their default values.</p>"},{"location":"Package-Settings/#create-custom-settings","title":"Create Custom Settings","text":"<p>In order to to use your own settings for your project, you'll need to create a settings asset.</p> <p>You can do this by clicking the <code>Create Settings Asset</code> button or clicking the gear on the top right and select <code>New Settings Asset...</code>. The asset file can be placed anywhere in the <code>Asset/</code> folder in your project. After Creating the settings asset, you'll be able to modify the settings for your project and your settings will be saved in the asset.</p> <p></p>"},{"location":"Package-Settings/#multiple-custom-settings-for-different-scenarios","title":"Multiple Custom Settings for Different Scenarios","text":"<p>You can create multiple settings assets in one project.</p> <p>By clicking the gear on the top right you'll see all available settings listed in the drop-down menu to choose from.</p> <p>This allows you to create different settings for different scenarios. For example, you can create two separate settings for training and inference, and specify which one you want to use according to what you're currently running.</p> <p></p>"},{"location":"Profiling-Python/","title":"Profiling in Python","text":"<p>As part of the ML-Agents Toolkit, we provide a lightweight profiling system, in order to identity hotspots in the training process and help spot regressions from changes.</p> <p>Timers are hierarchical, meaning that the time tracked in a block of code can be further split into other blocks if desired. This also means that a function that is called from multiple places in the code will appear in multiple places in the timing output.</p> <p>All timers operate using a \"global\" instance by default, but this can be overridden if necessary (mainly for testing).</p>"},{"location":"Profiling-Python/#adding-profiling","title":"Adding Profiling","text":"<p>There are two ways to indicate code should be included in profiling. The simplest way is to add the <code>@timed</code> decorator to a function or method of interested.</p> <pre><code>class TrainerController:\n    # ....\n    @timed\n    def advance(self, env: EnvManager) -&gt; int:\n        # do stuff\n</code></pre> <p>You can also used the <code>hierarchical_timer</code> context manager.</p> <pre><code>with hierarchical_timer(\"communicator.exchange\"):\n    outputs = self.communicator.exchange(step_input)\n</code></pre> <p>The context manager may be easier than the <code>@timed</code> decorator for profiling different parts of a large function, or profiling calls to abstract methods that might not use decorator.</p>"},{"location":"Profiling-Python/#output","title":"Output","text":"<p>By default, at the end of training, timers are collected and written in json format to <code>{summaries_dir}/{run_id}_timers.json</code>. The output consists of node objects with the following keys:</p> <ul> <li>total (float): The total time in seconds spent in the block, including child   calls.</li> <li>count (int): The number of times the block was called.</li> <li>self (float): The total time in seconds spent in the block, excluding child   calls.</li> <li>children (dictionary): A dictionary of child nodes, keyed by the node name.</li> <li>is_parallel (bool): Indicates that the block of code was executed in multiple   threads or processes (see below). This is optional and defaults to false.</li> </ul>"},{"location":"Profiling-Python/#parallel-execution","title":"Parallel execution","text":""},{"location":"Profiling-Python/#subprocesses","title":"Subprocesses","text":"<p>For code that executes in multiple processes (for example, SubprocessEnvManager), we periodically send the timer information back to the \"main\" process, aggregate the timers there, and flush them in the subprocess. Note that (depending on the number of processes) this can result in timers where the total time may exceed the parent's total time. This is analogous to the difference between \"real\" and \"user\" values reported from the unix <code>time</code> command. In the timer output, blocks that were run in parallel are indicated by the <code>is_parallel</code> flag.</p>"},{"location":"Profiling-Python/#threads","title":"Threads","text":"<p>Timers currently use <code>time.perf_counter()</code> to track time spent, which may not give accurate results for multiple threads. If this is problematic, set <code>threaded: false</code> in your trainer configuration.</p>"},{"location":"Python-Custom-Trainer-Plugin/","title":"Unity Ml-Agents Custom trainers Plugin","text":"<p>As an attempt to bring a wider variety of reinforcement learning algorithms to our users, we have added custom trainers capabilities. we introduce an extensible plugin system to define new trainers based on the High level trainer API in <code>Ml-agents</code> Package. This will allow rerouting <code>mlagents-learn</code> CLI to custom trainers and extending the config files with hyper-parameters specific to your new trainers. We will expose a high-level extensible trainer (both on-policy, and off-policy trainers) optimizer and hyperparameter classes with documentation for the use of this plugin. For more information on how python plugin system works see Plugin interfaces.</p>"},{"location":"Python-Custom-Trainer-Plugin/#overview","title":"Overview","text":"<p>Model-free RL algorithms generally fall into two broad categories: on-policy and off-policy. On-policy algorithms perform updates based on data gathered from the current policy. Off-policy algorithms learn a Q function from a buffer of previous data, then use this Q function to make decisions. Off-policy algorithms have three key benefits in the context of ML-Agents: They tend to use fewer samples than on-policy as they can pull and re-use data from the buffer many times. They allow player demonstrations to be inserted in-line with RL data into the buffer, enabling new ways of doing imitation learning by streaming player data.</p> <p>To add new custom trainers to ML-agents, you would need to create a new python package. To give you an idea of how to structure your package, we have created a mlagents_trainer_plugin package ourselves as an example, with implementation of <code>A2c</code> and <code>DQN</code> algorithms. You would need a <code>setup.py</code> file to list extra requirements and register the new RL algorithm in ml-agents ecosystem and be able to call <code>mlagents-learn</code> CLI with your customized configuration.</p> <pre><code>\u251c\u2500\u2500 mlagents_trainer_plugin\n\u2502    \u251c\u2500\u2500 __init__.py\n\u2502    \u251c\u2500\u2500 a2c\n\u2502    \u2502    \u251c\u2500\u2500 __init__.py\n\u2502    \u2502    \u251c\u2500\u2500 a2c_3DBall.yaml\n\u2502    \u2502    \u251c\u2500\u2500 a2c_optimizer.py\n\u2502    \u2502    \u2514\u2500\u2500 a2c_trainer.py\n\u2502    \u2514\u2500\u2500 dqn\n\u2502        \u251c\u2500\u2500 __init__.py\n\u2502        \u251c\u2500\u2500 dqn_basic.yaml\n\u2502        \u251c\u2500\u2500 dqn_optimizer.py\n\u2502        \u2514\u2500\u2500 dqn_trainer.py\n\u2514\u2500\u2500 setup.py\n</code></pre>"},{"location":"Python-Custom-Trainer-Plugin/#installation-and-execution","title":"Installation and Execution","text":"<p>If you haven't already, follow the installation instructions. Once you have the <code>ml-agents-env</code> and <code>ml-agents</code> packages you can install the plugin package. From the repository's root directory install <code>ml-agents-trainer-plugin</code> (or replace with the name of your plugin folder).</p> <pre><code>pip3 install -e &lt;./ml-agents-trainer-plugin&gt;\n</code></pre> <p>Following the previous installations your package is added as an entrypoint and you can use a config file with new trainers:</p> <pre><code>mlagents-learn ml-agents-trainer-plugin/mlagents_trainer_plugin/a2c/a2c_3DBall.yaml --run-id &lt;run-id-name&gt;\n--env &lt;env-executable&gt;\n</code></pre>"},{"location":"Python-Custom-Trainer-Plugin/#tutorial","title":"Tutorial","text":"<p>Here\u2019s a step-by-step tutorial on how to write a setup file and extend ml-agents trainers, optimizers, and hyperparameter settings.To extend ML-agents classes see references on trainers and Optimizer.</p>"},{"location":"Python-Gym-API-Documentation/","title":"Table of Contents","text":"<ul> <li>mlagents_envs.envs.unity_gym_env</li> <li>UnityGymException</li> <li>UnityToGymWrapper<ul> <li>__init__</li> <li>reset</li> <li>step</li> <li>render</li> <li>close</li> <li>seed</li> </ul> </li> <li>ActionFlattener<ul> <li>__init__</li> <li>lookup_action</li> </ul> </li> </ul>"},{"location":"Python-Gym-API-Documentation/#mlagents_envsenvsunity_gym_env","title":"mlagents_envs.envs.unity_gym_env","text":""},{"location":"Python-Gym-API-Documentation/#unitygymexception-objects","title":"UnityGymException Objects","text":"<pre><code>class UnityGymException(error.Error)\n</code></pre> <p>Any error related to the gym wrapper of ml-agents.</p> <p></p>"},{"location":"Python-Gym-API-Documentation/#unitytogymwrapper-objects","title":"UnityToGymWrapper Objects","text":"<pre><code>class UnityToGymWrapper(gym.Env)\n</code></pre> <p>Provides Gym wrapper for Unity Learning Environments.</p> <p></p>"},{"location":"Python-Gym-API-Documentation/#__init__","title":"__init__","text":"<pre><code> | __init__(unity_env: BaseEnv, uint8_visual: bool = False, flatten_branched: bool = False, allow_multiple_obs: bool = False, action_space_seed: Optional[int] = None)\n</code></pre> <p>Environment initialization</p> <p>Arguments:</p> <ul> <li><code>unity_env</code>: The Unity BaseEnv to be wrapped in the gym. Will be closed when the UnityToGymWrapper closes.</li> <li><code>uint8_visual</code>: Return visual observations as uint8 (0-255) matrices instead of float (0.0-1.0).</li> <li><code>flatten_branched</code>: If True, turn branched discrete action spaces into a Discrete space rather than     MultiDiscrete.</li> <li><code>allow_multiple_obs</code>: If True, return a list of np.ndarrays as observations with the first elements     containing the visual observations and the last element containing the array of vector observations.     If False, returns a single np.ndarray containing either only a single visual observation or the array of     vector observations.</li> <li><code>action_space_seed</code>: If non-None, will be used to set the random seed on created gym.Space instances.</li> </ul> <p></p>"},{"location":"Python-Gym-API-Documentation/#reset","title":"reset","text":"<pre><code> | reset() -&gt; Union[List[np.ndarray], np.ndarray]\n</code></pre> <p>Resets the state of the environment and returns an initial observation. Returns: observation (object/list): the initial observation of the space.</p> <p></p>"},{"location":"Python-Gym-API-Documentation/#step","title":"step","text":"<pre><code> | step(action: List[Any]) -&gt; GymStepResult\n</code></pre> <p>Run one timestep of the environment's dynamics. When end of episode is reached, you are responsible for calling <code>reset()</code> to reset this environment's state. Accepts an action and returns a tuple (observation, reward, done, info).</p> <p>Arguments:</p> <ul> <li><code>action</code> object/list - an action provided by the environment</li> </ul> <p>Returns:</p> <ul> <li><code>observation</code> object/list - agent's observation of the current environment   reward (float/list) : amount of reward returned after previous action</li> <li><code>done</code> boolean/list - whether the episode has ended.</li> <li><code>info</code> dict - contains auxiliary diagnostic information.</li> </ul> <p></p>"},{"location":"Python-Gym-API-Documentation/#render","title":"render","text":"<pre><code> | render(mode=\"rgb_array\")\n</code></pre> <p>Return the latest visual observations. Note that it will not render a new frame of the environment.</p> <p></p>"},{"location":"Python-Gym-API-Documentation/#close","title":"close","text":"<pre><code> | close() -&gt; None\n</code></pre> <p>Override _close in your subclass to perform any necessary cleanup. Environments will automatically close() themselves when garbage collected or when the program exits.</p> <p></p>"},{"location":"Python-Gym-API-Documentation/#seed","title":"seed","text":"<pre><code> | seed(seed: Any = None) -&gt; None\n</code></pre> <p>Sets the seed for this env's random number generator(s). Currently not implemented.</p> <p></p>"},{"location":"Python-Gym-API-Documentation/#actionflattener-objects","title":"ActionFlattener Objects","text":"<pre><code>class ActionFlattener()\n</code></pre> <p>Flattens branched discrete action spaces into single-branch discrete action spaces.</p> <p></p>"},{"location":"Python-Gym-API-Documentation/#__init___1","title":"__init__","text":"<pre><code> | __init__(branched_action_space)\n</code></pre> <p>Initialize the flattener.</p> <p>Arguments:</p> <ul> <li><code>branched_action_space</code>: A List containing the sizes of each branch of the action space, e.g. [2,3,3] for three branches with size 2, 3, and 3 respectively.</li> </ul> <p></p>"},{"location":"Python-Gym-API-Documentation/#lookup_action","title":"lookup_action","text":"<pre><code> | lookup_action(action)\n</code></pre> <p>Convert a scalar discrete action into a unique set of branched actions.</p> <p>Arguments:</p> <ul> <li><code>action</code>: A scalar value representing one of the discrete actions.</li> </ul> <p>Returns:</p> <p>The List containing the branched actions.</p>"},{"location":"Python-Gym-API/","title":"Unity ML-Agents Gym Wrapper","text":"<p>A common way in which machine learning researchers interact with simulation environments is via a wrapper provided by OpenAI called <code>gym</code>. For more information on the gym interface, see here.</p> <p>We provide a gym wrapper and instructions for using it with existing machine learning algorithms which utilize gym. Our wrapper provides interfaces on top of our <code>UnityEnvironment</code> class, which is the default way of interfacing with a Unity environment via Python.</p>"},{"location":"Python-Gym-API/#installation","title":"Installation","text":"<p>The gym wrapper is part of the <code>mlagents_envs</code> package. Please refer to the mlagents_envs installation instructions.</p>"},{"location":"Python-Gym-API/#using-the-gym-wrapper","title":"Using the Gym Wrapper","text":"<p>The gym interface is available from <code>gym_unity.envs</code>. To launch an environment from the root of the project repository use:</p> <pre><code>from mlagents_envs.envs.unity_gym_env import UnityToGymWrapper\n\nenv = UnityToGymWrapper(unity_env, uint8_visual, flatten_branched, allow_multiple_obs)\n</code></pre> <ul> <li> <p><code>unity_env</code> refers to the Unity environment to be wrapped.</p> </li> <li> <p><code>uint8_visual</code> refers to whether to output visual observations as <code>uint8</code>   values (0-255). Many common Gym environments (e.g. Atari) do this. By default   they will be floats (0.0-1.0). Defaults to <code>False</code>.</p> </li> <li> <p><code>flatten_branched</code> will flatten a branched discrete action space into a Gym   Discrete. Otherwise, it will be converted into a MultiDiscrete. Defaults to   <code>False</code>.</p> </li> <li> <p><code>allow_multiple_obs</code> will return a list of observations. The first elements   contain the visual observations and the last element contains the array of   vector observations. If False the environment returns a single array (containing   a single visual observations, if present, otherwise the vector observation).   Defaults to <code>False</code>.</p> </li> <li> <p><code>action_space_seed</code> is the optional seed for action sampling. If non-None, will   be used to set the random seed on created gym.Space instances.</p> </li> </ul> <p>The returned environment <code>env</code> will function as a gym.</p>"},{"location":"Python-Gym-API/#limitations","title":"Limitations","text":"<ul> <li>It is only possible to use an environment with a single Agent.</li> <li>By default, the first visual observation is provided as the <code>observation</code>, if   present. Otherwise, vector observations are provided. You can receive all   visual and vector observations by using the <code>allow_multiple_obs=True</code> option in   the gym parameters. If set to <code>True</code>, you will receive a list of <code>observation</code>   instead of only one.</li> <li>The <code>TerminalSteps</code> or <code>DecisionSteps</code> output from the environment can still   be accessed from the <code>info</code> provided by <code>env.step(action)</code>.</li> <li>Stacked vector observations are not supported.</li> <li>Environment registration for use with <code>gym.make()</code> is currently not supported.</li> <li>Calling env.render() will not render a new frame of the environment. It will   return the latest visual observation if using visual observations.</li> </ul>"},{"location":"Python-Gym-API/#running-openai-baselines-algorithms","title":"Running OpenAI Baselines Algorithms","text":"<p>OpenAI provides a set of open-source maintained and tested Reinforcement Learning algorithms called the Baselines.</p> <p>Using the provided Gym wrapper, it is possible to train ML-Agents environments using these algorithms. This requires the creation of custom training scripts to launch each algorithm. In most cases these scripts can be created by making slight modifications to the ones provided for Atari and Mujoco environments.</p> <p>These examples were tested with baselines version 0.1.6.</p>"},{"location":"Python-Gym-API/#example-dqn-baseline","title":"Example - DQN Baseline","text":"<p>In order to train an agent to play the <code>GridWorld</code> environment using the Baselines DQN algorithm, you first need to install the baselines package using pip:</p> <pre><code>pip install git+git://github.com/openai/baselines\n</code></pre> <p>Next, create a file called <code>train_unity.py</code>. Then create an <code>/envs/</code> directory and build the environment to that directory. For more information on building Unity environments, see here. Note that because of limitations of the DQN baseline, the environment must have a single visual observation, a single discrete action and a single Agent in the scene. Add the following code to the <code>train_unity.py</code> file:</p> <pre><code>import gym\n\nfrom baselines import deepq\nfrom baselines import logger\n\nfrom mlagents_envs.environment import UnityEnvironment\nfrom mlagents_envs.envs.unity_gym_env import UnityToGymWrapper\n\n\ndef main():\n  unity_env = UnityEnvironment( &lt; path - to - environment &gt;)\n  env = UnityToGymWrapper(unity_env, uint8_visual=True)\n  logger.configure('./logs')  # Change to log in a different directory\n  act = deepq.learn(\n    env,\n    \"cnn\",  # For visual inputs\n    lr=2.5e-4,\n    total_timesteps=1000000,\n    buffer_size=50000,\n    exploration_fraction=0.05,\n    exploration_final_eps=0.1,\n    print_freq=20,\n    train_freq=5,\n    learning_starts=20000,\n    target_network_update_freq=50,\n    gamma=0.99,\n    prioritized_replay=False,\n    checkpoint_freq=1000,\n    checkpoint_path='./logs',  # Change to save model in a different directory\n    dueling=True\n  )\n  print(\"Saving model to unity_model.pkl\")\n  act.save(\"unity_model.pkl\")\n\n\nif __name__ == '__main__':\n  main()\n</code></pre> <p>To start the training process, run the following from the directory containing <code>train_unity.py</code>:</p> <pre><code>python -m train_unity\n</code></pre>"},{"location":"Python-Gym-API/#other-algorithms","title":"Other Algorithms","text":"<p>Other algorithms in the Baselines repository can be run using scripts similar to the examples from the baselines package. In most cases, the primary changes needed to use a Unity environment are to import <code>UnityToGymWrapper</code>, and to replace the environment creation code, typically <code>gym.make()</code>, with a call to <code>UnityToGymWrapper(unity_environment)</code> passing the environment as input.</p> <p>A typical rule of thumb is that for vision-based environments, modification should be done to Atari training scripts, and for vector observation environments, modification should be done to Mujoco scripts.</p> <p>Some algorithms will make use of <code>make_env()</code> or <code>make_mujoco_env()</code> functions. You can define a similar function for Unity environments. An example of such a method using the PPO2 baseline:</p> <pre><code>from mlagents_envs.environment import UnityEnvironment\nfrom mlagents_envs.envs import UnityToGymWrapper\nfrom baselines.common.vec_env.subproc_vec_env import SubprocVecEnv\nfrom baselines.common.vec_env.dummy_vec_env import DummyVecEnv\nfrom baselines.bench import Monitor\nfrom baselines import logger\nimport baselines.ppo2.ppo2 as ppo2\n\nimport os\n\ntry:\n  from mpi4py import MPI\nexcept ImportError:\n  MPI = None\n\n\ndef make_unity_env(env_directory, num_env, visual, start_index=0):\n  \"\"\"\n  Create a wrapped, monitored Unity environment.\n  \"\"\"\n\n  def make_env(rank, use_visual=True):  # pylint: disable=C0111\n    def _thunk():\n      unity_env = UnityEnvironment(env_directory, base_port=5000 + rank)\n      env = UnityToGymWrapper(unity_env, uint8_visual=True)\n      env = Monitor(env, logger.get_dir() and os.path.join(logger.get_dir(), str(rank)))\n      return env\n\n    return _thunk\n\n  if visual:\n    return SubprocVecEnv([make_env(i + start_index) for i in range(num_env)])\n  else:\n    rank = MPI.COMM_WORLD.Get_rank() if MPI else 0\n    return DummyVecEnv([make_env(rank, use_visual=False)])\n\n\ndef main():\n  env = make_unity_env( &lt; path - to - environment &gt;, 4, True)\n  ppo2.learn(\n    network=\"mlp\",\n    env=env,\n    total_timesteps=100000,\n    lr=1e-3,\n  )\n\n\nif __name__ == '__main__':\n  main()\n</code></pre>"},{"location":"Python-Gym-API/#run-google-dopamine-algorithms","title":"Run Google Dopamine Algorithms","text":"<p>Google provides a framework Dopamine, and implementations of algorithms, e.g. DQN, Rainbow, and the C51 variant of Rainbow. Using the Gym wrapper, we can run Unity environments using Dopamine.</p> <p>First, after installing the Gym wrapper, clone the Dopamine repository.</p> <pre><code>git clone https://github.com/google/dopamine\n</code></pre> <p>Then, follow the appropriate install instructions as specified on Dopamine's homepage. Note that the Dopamine guide specifies using a virtualenv. If you choose to do so, make sure your unity_env package is also installed within the same virtualenv as Dopamine.</p>"},{"location":"Python-Gym-API/#adapting-dopamines-scripts","title":"Adapting Dopamine's Scripts","text":"<p>First, open <code>dopamine/atari/run_experiment.py</code>. Alternatively, copy the entire <code>atari</code> folder, and name it something else (e.g. <code>unity</code>). If you choose the copy approach, be sure to change the package names in the import statements in <code>train.py</code> to your new directory.</p> <p>Within <code>run_experiment.py</code>, we will need to make changes to which environment is instantiated, just as in the Baselines example. At the top of the file, insert</p> <pre><code>from mlagents_envs.environment import UnityEnvironment\nfrom mlagents_envs.envs import UnityToGymWrapper\n</code></pre> <p>to import the Gym Wrapper. Navigate to the <code>create_atari_environment</code> method in the same file, and switch to instantiating a Unity environment by replacing the method with the following code.</p> <pre><code>    game_version = 'v0' if sticky_actions else 'v4'\n    full_game_name = '{}NoFrameskip-{}'.format(game_name, game_version)\n    unity_env = UnityEnvironment(&lt;path-to-environment&gt;)\n    env = UnityToGymWrapper(unity_env, uint8_visual=True)\n    return env\n</code></pre> <p><code>&lt;path-to-environment&gt;</code> is the path to your built Unity executable. For more information on building Unity environments, see here, and note the Limitations section below.</p> <p>Note that we are not using the preprocessor from Dopamine, as it uses many Atari-specific calls. Furthermore, frame-skipping can be done from within Unity, rather than on the Python side.</p>"},{"location":"Python-Gym-API/#limitations_1","title":"Limitations","text":"<p>Since Dopamine is designed around variants of DQN, it is only compatible with discrete action spaces, and specifically the Discrete Gym space. For environments that use branched discrete action spaces, you can enable the <code>flatten_branched</code> parameter in <code>UnityToGymWrapper</code>, which treats each combination of branched actions as separate actions.</p> <p>Furthermore, when building your environments, ensure that your Agent is using visual observations with greyscale enabled, and that the dimensions of the visual observations is 84 by 84 (matches the parameter found in <code>dqn_agent.py</code> and <code>rainbow_agent.py</code>). Dopamine's agents currently do not automatically adapt to the observation dimensions or number of channels.</p>"},{"location":"Python-Gym-API/#hyperparameters","title":"Hyperparameters","text":"<p>The hyperparameters provided by Dopamine are tailored to the Atari games, and you will likely need to adjust them for ML-Agents environments. Here is a sample <code>dopamine/agents/rainbow/configs/rainbow.gin</code> file that is known to work with a simple GridWorld.</p> <pre><code>import dopamine.agents.rainbow.rainbow_agent\nimport dopamine.unity.run_experiment\nimport dopamine.replay_memory.prioritized_replay_buffer\nimport gin.tf.external_configurables\n\nRainbowAgent.num_atoms = 51\nRainbowAgent.stack_size = 1\nRainbowAgent.vmax = 10.\nRainbowAgent.gamma = 0.99\nRainbowAgent.update_horizon = 3\nRainbowAgent.min_replay_history = 20000  # agent steps\nRainbowAgent.update_period = 5\nRainbowAgent.target_update_period = 50  # agent steps\nRainbowAgent.epsilon_train = 0.1\nRainbowAgent.epsilon_eval = 0.01\nRainbowAgent.epsilon_decay_period = 50000  # agent steps\nRainbowAgent.replay_scheme = 'prioritized'\nRainbowAgent.tf_device = '/cpu:0'  # use '/cpu:*' for non-GPU version\nRainbowAgent.optimizer = @tf.train.AdamOptimizer()\n\ntf.train.AdamOptimizer.learning_rate = 0.00025\ntf.train.AdamOptimizer.epsilon = 0.0003125\n\nRunner.game_name = \"Unity\" # any name can be used here\nRunner.sticky_actions = False\nRunner.num_iterations = 200\nRunner.training_steps = 10000  # agent steps\nRunner.evaluation_steps = 500  # agent steps\nRunner.max_steps_per_episode = 27000  # agent steps\n\nWrappedPrioritizedReplayBuffer.replay_capacity = 1000000\nWrappedPrioritizedReplayBuffer.batch_size = 32\n</code></pre> <p>This example assumed you copied <code>atari</code> to a separate folder named <code>unity</code>. Replace <code>unity</code> in <code>import dopamine.unity.run_experiment</code> with the folder you copied your <code>run_experiment.py</code> and <code>trainer.py</code> files to. If you directly modified the existing files, then use <code>atari</code> here.</p>"},{"location":"Python-Gym-API/#starting-a-run","title":"Starting a Run","text":"<p>You can now run Dopamine as you would normally:</p> <pre><code>python -um dopamine.unity.train \\\n  --agent_name=rainbow \\\n  --base_dir=/tmp/dopamine \\\n  --gin_files='dopamine/agents/rainbow/configs/rainbow.gin'\n</code></pre> <p>Again, we assume that you've copied <code>atari</code> into a separate folder. Remember to replace <code>unity</code> with the directory you copied your files into. If you edited the Atari files directly, this should be <code>atari</code>.</p>"},{"location":"Python-Gym-API/#example-gridworld","title":"Example: GridWorld","text":"<p>As a baseline, here are rewards over time for the three algorithms provided with Dopamine as run on the GridWorld example environment. All Dopamine (DQN, Rainbow, C51) runs were done with the same epsilon, epsilon decay, replay history, training steps, and buffer settings as specified above. Note that the first 20000 steps are used to pre-fill the training buffer, and no learning happens.</p> <p>We provide results from our PPO implementation and the DQN from Baselines as reference. Note that all runs used the same greyscale GridWorld as Dopamine. For PPO, <code>num_layers</code> was set to 2, and all other hyperparameters are the default for GridWorld in <code>config/ppo/GridWorld.yaml</code>. For Baselines DQN, the provided hyperparameters in the previous section are used. Note that Baselines implements certain features (e.g. dueling-Q) that are not enabled in Dopamine DQN.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/","title":"Table of Contents","text":"<ul> <li>mlagents_envs.base_env</li> <li>DecisionStep</li> <li>DecisionSteps<ul> <li>agent_id_to_index</li> <li>__getitem__</li> <li>empty</li> </ul> </li> <li>TerminalStep</li> <li>TerminalSteps<ul> <li>agent_id_to_index</li> <li>__getitem__</li> <li>empty</li> </ul> </li> <li>ActionTuple<ul> <li>discrete_dtype</li> </ul> </li> <li>ActionSpec<ul> <li>is_discrete</li> <li>is_continuous</li> <li>discrete_size</li> <li>empty_action</li> <li>random_action</li> <li>create_continuous</li> <li>create_discrete</li> <li>create_hybrid</li> </ul> </li> <li>DimensionProperty<ul> <li>UNSPECIFIED</li> <li>NONE</li> <li>TRANSLATIONAL_EQUIVARIANCE</li> <li>VARIABLE_SIZE</li> </ul> </li> <li>ObservationType<ul> <li>DEFAULT</li> <li>GOAL_SIGNAL</li> </ul> </li> <li>ObservationSpec</li> <li>BehaviorSpec</li> <li>BaseEnv<ul> <li>step</li> <li>reset</li> <li>close</li> <li>behavior_specs</li> <li>set_actions</li> <li>set_action_for_agent</li> <li>get_steps</li> </ul> </li> <li>mlagents_envs.environment</li> <li>UnityEnvironment<ul> <li>__init__</li> <li>close</li> </ul> </li> <li>mlagents_envs.registry</li> <li>mlagents_envs.registry.unity_env_registry</li> <li>UnityEnvRegistry<ul> <li>register</li> <li>register_from_yaml</li> <li>clear</li> <li>__getitem__</li> </ul> </li> <li>mlagents_envs.side_channel</li> <li>mlagents_envs.side_channel.raw_bytes_channel</li> <li>RawBytesChannel<ul> <li>on_message_received</li> <li>get_and_clear_received_messages</li> <li>send_raw_data</li> </ul> </li> <li>mlagents_envs.side_channel.outgoing_message</li> <li>OutgoingMessage<ul> <li>__init__</li> <li>write_bool</li> <li>write_int32</li> <li>write_float32</li> <li>write_float32_list</li> <li>write_string</li> <li>set_raw_bytes</li> </ul> </li> <li>mlagents_envs.side_channel.engine_configuration_channel</li> <li>EngineConfigurationChannel<ul> <li>on_message_received</li> <li>set_configuration_parameters</li> <li>set_configuration</li> </ul> </li> <li>mlagents_envs.side_channel.side_channel_manager</li> <li>SideChannelManager<ul> <li>process_side_channel_message</li> <li>generate_side_channel_messages</li> </ul> </li> <li>mlagents_envs.side_channel.stats_side_channel</li> <li>StatsSideChannel<ul> <li>on_message_received</li> <li>get_and_reset_stats</li> </ul> </li> <li>mlagents_envs.side_channel.incoming_message</li> <li>IncomingMessage<ul> <li>__init__</li> <li>read_bool</li> <li>read_int32</li> <li>read_float32</li> <li>read_float32_list</li> <li>read_string</li> <li>get_raw_bytes</li> </ul> </li> <li>mlagents_envs.side_channel.float_properties_channel</li> <li>FloatPropertiesChannel<ul> <li>on_message_received</li> <li>set_property</li> <li>get_property</li> <li>list_properties</li> <li>get_property_dict_copy</li> </ul> </li> <li>mlagents_envs.side_channel.environment_parameters_channel</li> <li>EnvironmentParametersChannel<ul> <li>set_float_parameter</li> <li>set_uniform_sampler_parameters</li> <li>set_gaussian_sampler_parameters</li> <li>set_multirangeuniform_sampler_parameters</li> </ul> </li> <li>mlagents_envs.side_channel.side_channel</li> <li>SideChannel<ul> <li>queue_message_to_send</li> <li>on_message_received</li> <li>channel_id</li> </ul> </li> </ul>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsbase_env","title":"mlagents_envs.base_env","text":"<p>Python Environment API for the ML-Agents Toolkit The aim of this API is to expose Agents evolving in a simulation to perform reinforcement learning on. This API supports multi-agent scenarios and groups similar Agents (same observations, actions spaces and behavior) together. These groups of Agents are identified by their BehaviorName. For performance reasons, the data of each group of agents is processed in a batched manner. Agents are identified by a unique AgentId identifier that allows tracking of Agents across simulation steps. Note that there is no guarantee that the number or order of the Agents in the state will be consistent across simulation steps. A simulation steps corresponds to moving the simulation forward until at least one agent in the simulation sends its observations to Python again. Since Agents can request decisions at different frequencies, a simulation step does not necessarily correspond to a fixed simulation time increment.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#decisionstep-objects","title":"DecisionStep Objects","text":"<pre><code>class DecisionStep(NamedTuple)\n</code></pre> <p>Contains the data a single Agent collected since the last simulation step.  - obs is a list of numpy arrays observations collected by the agent.  - reward is a float. Corresponds to the rewards collected by the agent  since the last simulation step.  - agent_id is an int and an unique identifier for the corresponding Agent.  - action_mask is an optional list of one dimensional array of booleans.  Only available when using multi-discrete actions.  Each array corresponds to an action branch. Each array contains a mask  for each action of the branch. If true, the action is not available for  the agent during this simulation step.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#decisionsteps-objects","title":"DecisionSteps Objects","text":"<pre><code>class DecisionSteps(Mapping)\n</code></pre> <p>Contains the data a batch of similar Agents collected since the last simulation step. Note that all Agents do not necessarily have new information to send at each simulation step. Therefore, the ordering of agents and the batch size of the DecisionSteps are not fixed across simulation steps.  - obs is a list of numpy arrays observations collected by the batch of  agent. Each obs has one extra dimension compared to DecisionStep: the  first dimension of the array corresponds to the batch size of the batch.  - reward is a float vector of length batch size. Corresponds to the  rewards collected by each agent since the last simulation step.  - agent_id is an int vector of length batch size containing unique  identifier for the corresponding Agent. This is used to track Agents  across simulation steps.  - action_mask is an optional list of two dimensional array of booleans.  Only available when using multi-discrete actions.  Each array corresponds to an action branch. The first dimension of each  array is the batch size and the second contains a mask for each action of  the branch. If true, the action is not available for the agent during  this simulation step.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#agent_id_to_index","title":"agent_id_to_index","text":"<pre><code> | @property\n | agent_id_to_index() -&gt; Dict[AgentId, int]\n</code></pre> <p>Returns:</p> <p>A Dict that maps agent_id to the index of those agents in this DecisionSteps.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#__getitem__","title":"__getitem__","text":"<pre><code> | __getitem__(agent_id: AgentId) -&gt; DecisionStep\n</code></pre> <p>returns the DecisionStep for a specific agent.</p> <p>Arguments:</p> <ul> <li><code>agent_id</code>: The id of the agent</li> </ul> <p>Returns:</p> <p>The DecisionStep</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#empty","title":"empty","text":"<pre><code> | @staticmethod\n | empty(spec: \"BehaviorSpec\") -&gt; \"DecisionSteps\"\n</code></pre> <p>Returns an empty DecisionSteps.</p> <p>Arguments:</p> <ul> <li><code>spec</code>: The BehaviorSpec for the DecisionSteps</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#terminalstep-objects","title":"TerminalStep Objects","text":"<pre><code>class TerminalStep(NamedTuple)\n</code></pre> <p>Contains the data a single Agent collected when its episode ended.  - obs is a list of numpy arrays observations collected by the agent.  - reward is a float. Corresponds to the rewards collected by the agent  since the last simulation step.  - interrupted is a bool. Is true if the Agent was interrupted since the last  decision step. For example, if the Agent reached the maximum number of steps for  the episode.  - agent_id is an int and an unique identifier for the corresponding Agent.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#terminalsteps-objects","title":"TerminalSteps Objects","text":"<pre><code>class TerminalSteps(Mapping)\n</code></pre> <p>Contains the data a batch of Agents collected when their episode terminated. All Agents present in the TerminalSteps have ended their episode.  - obs is a list of numpy arrays observations collected by the batch of  agent. Each obs has one extra dimension compared to DecisionStep: the  first dimension of the array corresponds to the batch size of the batch.  - reward is a float vector of length batch size. Corresponds to the  rewards collected by each agent since the last simulation step.  - interrupted is an array of booleans of length batch size. Is true if the  associated Agent was interrupted since the last decision step. For example, if the  Agent reached the maximum number of steps for the episode.  - agent_id is an int vector of length batch size containing unique  identifier for the corresponding Agent. This is used to track Agents  across simulation steps.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#agent_id_to_index_1","title":"agent_id_to_index","text":"<pre><code> | @property\n | agent_id_to_index() -&gt; Dict[AgentId, int]\n</code></pre> <p>Returns:</p> <p>A Dict that maps agent_id to the index of those agents in this TerminalSteps.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#__getitem___1","title":"__getitem__","text":"<pre><code> | __getitem__(agent_id: AgentId) -&gt; TerminalStep\n</code></pre> <p>returns the TerminalStep for a specific agent.</p> <p>Arguments:</p> <ul> <li><code>agent_id</code>: The id of the agent</li> </ul> <p>Returns:</p> <p>obs, reward, done, agent_id and optional action mask for a specific agent</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#empty_1","title":"empty","text":"<pre><code> | @staticmethod\n | empty(spec: \"BehaviorSpec\") -&gt; \"TerminalSteps\"\n</code></pre> <p>Returns an empty TerminalSteps.</p> <p>Arguments:</p> <ul> <li><code>spec</code>: The BehaviorSpec for the TerminalSteps</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#actiontuple-objects","title":"ActionTuple Objects","text":"<pre><code>class ActionTuple(_ActionTupleBase)\n</code></pre> <p>An object whose fields correspond to actions of different types. Continuous and discrete actions are numpy arrays of type float32 and int32, respectively and are type checked on construction. Dimensions are of (n_agents, continuous_size) and (n_agents, discrete_size), respectively. Note, this also holds when continuous or discrete size is zero.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#discrete_dtype","title":"discrete_dtype","text":"<pre><code> | @property\n | discrete_dtype() -&gt; np.dtype\n</code></pre> <p>The dtype of a discrete action.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#actionspec-objects","title":"ActionSpec Objects","text":"<pre><code>class ActionSpec(NamedTuple)\n</code></pre> <p>A NamedTuple containing utility functions and information about the action spaces for a group of Agents under the same behavior. - num_continuous_actions is an int corresponding to the number of floats which constitute the action. - discrete_branch_sizes is a Tuple of int where each int corresponds to the number of discrete actions available to the agent on an independent action branch.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#is_discrete","title":"is_discrete","text":"<pre><code> | is_discrete() -&gt; bool\n</code></pre> <p>Returns true if this Behavior uses discrete actions</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#is_continuous","title":"is_continuous","text":"<pre><code> | is_continuous() -&gt; bool\n</code></pre> <p>Returns true if this Behavior uses continuous actions</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#discrete_size","title":"discrete_size","text":"<pre><code> | @property\n | discrete_size() -&gt; int\n</code></pre> <p>Returns a an int corresponding to the number of discrete branches.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#empty_action","title":"empty_action","text":"<pre><code> | empty_action(n_agents: int) -&gt; ActionTuple\n</code></pre> <p>Generates ActionTuple corresponding to an empty action (all zeros) for a number of agents.</p> <p>Arguments:</p> <ul> <li><code>n_agents</code>: The number of agents that will have actions generated</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#random_action","title":"random_action","text":"<pre><code> | random_action(n_agents: int) -&gt; ActionTuple\n</code></pre> <p>Generates ActionTuple corresponding to a random action (either discrete or continuous) for a number of agents.</p> <p>Arguments:</p> <ul> <li><code>n_agents</code>: The number of agents that will have actions generated</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#create_continuous","title":"create_continuous","text":"<pre><code> | @staticmethod\n | create_continuous(continuous_size: int) -&gt; \"ActionSpec\"\n</code></pre> <p>Creates an ActionSpec that is homogenously continuous</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#create_discrete","title":"create_discrete","text":"<pre><code> | @staticmethod\n | create_discrete(discrete_branches: Tuple[int]) -&gt; \"ActionSpec\"\n</code></pre> <p>Creates an ActionSpec that is homogenously discrete</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#create_hybrid","title":"create_hybrid","text":"<pre><code> | @staticmethod\n | create_hybrid(continuous_size: int, discrete_branches: Tuple[int]) -&gt; \"ActionSpec\"\n</code></pre> <p>Creates a hybrid ActionSpace</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#dimensionproperty-objects","title":"DimensionProperty Objects","text":"<pre><code>class DimensionProperty(IntFlag)\n</code></pre> <p>The dimension property of a dimension of an observation.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#unspecified","title":"UNSPECIFIED","text":"<p>No properties specified.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#none","title":"NONE","text":"<p>No Property of the observation in that dimension. Observation can be processed with Fully connected networks.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#translational_equivariance","title":"TRANSLATIONAL_EQUIVARIANCE","text":"<p>Means it is suitable to do a convolution in this dimension.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#variable_size","title":"VARIABLE_SIZE","text":"<p>Means that there can be a variable number of observations in this dimension. The observations are unordered.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#observationtype-objects","title":"ObservationType Objects","text":"<pre><code>class ObservationType(Enum)\n</code></pre> <p>An Enum which defines the type of information carried in the observation of the agent.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#default","title":"DEFAULT","text":"<p>Observation information is generic.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#goal_signal","title":"GOAL_SIGNAL","text":"<p>Observation contains goal information for current task.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#observationspec-objects","title":"ObservationSpec Objects","text":"<pre><code>class ObservationSpec(NamedTuple)\n</code></pre> <p>A NamedTuple containing information about the observation of Agents. - shape is a Tuple of int : It corresponds to the shape of an observation's dimensions. - dimension_property is a Tuple of DimensionProperties flag, one flag for each dimension. - observation_type is an enum of ObservationType.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#behaviorspec-objects","title":"BehaviorSpec Objects","text":"<pre><code>class BehaviorSpec(NamedTuple)\n</code></pre> <p>A NamedTuple containing information about the observation and action spaces for a group of Agents under the same behavior. - observation_specs is a List of ObservationSpec NamedTuple containing information about the information of the Agent's observations such as their shapes. The order of the ObservationSpec is the same as the order of the observations of an agent. - action_spec is an ActionSpec NamedTuple.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#baseenv-objects","title":"BaseEnv Objects","text":"<pre><code>class BaseEnv(ABC)\n</code></pre>"},{"location":"Python-LLAPI-Documentation/#step","title":"step","text":"<pre><code> | @abstractmethod\n | step() -&gt; None\n</code></pre> <p>Signals the environment that it must move the simulation forward by one step.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#reset","title":"reset","text":"<pre><code> | @abstractmethod\n | reset() -&gt; None\n</code></pre> <p>Signals the environment that it must reset the simulation.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#close","title":"close","text":"<pre><code> | @abstractmethod\n | close() -&gt; None\n</code></pre> <p>Signals the environment that it must close.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#behavior_specs","title":"behavior_specs","text":"<pre><code> | @property\n | @abstractmethod\n | behavior_specs() -&gt; MappingType[str, BehaviorSpec]\n</code></pre> <p>Returns a Mapping from behavior names to behavior specs. Agents grouped under the same behavior name have the same action and observation specs, and are expected to behave similarly in the environment. Note that new keys can be added to this mapping as new policies are instantiated.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_actions","title":"set_actions","text":"<pre><code> | @abstractmethod\n | set_actions(behavior_name: BehaviorName, action: ActionTuple) -&gt; None\n</code></pre> <p>Sets the action for all of the agents in the simulation for the next step. The Actions must be in the same order as the order received in the DecisionSteps.</p> <p>Arguments:</p> <ul> <li><code>behavior_name</code>: The name of the behavior the agents are part of</li> <li><code>action</code>: ActionTuple tuple of continuous and/or discrete action. Actions are np.arrays with dimensions  (n_agents, continuous_size) and (n_agents, discrete_size), respectively.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_action_for_agent","title":"set_action_for_agent","text":"<pre><code> | @abstractmethod\n | set_action_for_agent(behavior_name: BehaviorName, agent_id: AgentId, action: ActionTuple) -&gt; None\n</code></pre> <p>Sets the action for one of the agents in the simulation for the next step.</p> <p>Arguments:</p> <ul> <li><code>behavior_name</code>: The name of the behavior the agent is part of</li> <li><code>agent_id</code>: The id of the agent the action is set for</li> <li><code>action</code>: ActionTuple tuple of continuous and/or discrete action Actions are np.arrays with dimensions  (1, continuous_size) and (1, discrete_size), respectively. Note, this initial dimensions of 1 is because this action is meant for a single agent.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#get_steps","title":"get_steps","text":"<pre><code> | @abstractmethod\n | get_steps(behavior_name: BehaviorName) -&gt; Tuple[DecisionSteps, TerminalSteps]\n</code></pre> <p>Retrieves the steps of the agents that requested a step in the simulation.</p> <p>Arguments:</p> <ul> <li><code>behavior_name</code>: The name of the behavior the agents are part of</li> </ul> <p>Returns:</p> <p>A tuple containing : - A DecisionSteps NamedTuple containing the observations, the rewards, the agent ids and the action masks for the Agents of the specified behavior. These Agents need an action this step. - A TerminalSteps NamedTuple containing the observations, rewards, agent ids and interrupted flags of the agents that had their episode terminated last step.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsenvironment","title":"mlagents_envs.environment","text":""},{"location":"Python-LLAPI-Documentation/#unityenvironment-objects","title":"UnityEnvironment Objects","text":"<pre><code>class UnityEnvironment(BaseEnv)\n</code></pre>"},{"location":"Python-LLAPI-Documentation/#__init__","title":"__init__","text":"<pre><code> | __init__(file_name: Optional[str] = None, worker_id: int = 0, base_port: Optional[int] = None, seed: int = 0, no_graphics: bool = False, no_graphics_monitor: bool = False, timeout_wait: int = 60, additional_args: Optional[List[str]] = None, side_channels: Optional[List[SideChannel]] = None, log_folder: Optional[str] = None, num_areas: int = 1)\n</code></pre> <p>Starts a new unity environment and establishes a connection with the environment. Notice: Currently communication between Unity and Python takes place over an open socket without authentication. Ensure that the network where training takes place is secure.</p> <p>:string file_name: Name of Unity environment binary. :int base_port: Baseline port number to connect to Unity environment over. worker_id increments over this. If no environment is specified (i.e. file_name is None), the DEFAULT_EDITOR_PORT will be used. :int worker_id: Offset from base_port. Used for training multiple environments simultaneously. :bool no_graphics: Whether to run the Unity simulator in no-graphics mode :bool no_graphics_monitor: Whether to run the main worker in graphics mode, with the remaining in no-graphics mode :int timeout_wait: Time (in seconds) to wait for connection from environment. :list args: Addition Unity command line arguments :list side_channels: Additional side channel for no-rl communication with Unity :str log_folder: Optional folder to write the Unity Player log file into.  Requires absolute path.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#close_1","title":"close","text":"<pre><code> | close()\n</code></pre> <p>Sends a shutdown signal to the unity environment, and closes the socket connection.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsregistry","title":"mlagents_envs.registry","text":""},{"location":"Python-LLAPI-Documentation/#mlagents_envsregistryunity_env_registry","title":"mlagents_envs.registry.unity_env_registry","text":""},{"location":"Python-LLAPI-Documentation/#unityenvregistry-objects","title":"UnityEnvRegistry Objects","text":"<pre><code>class UnityEnvRegistry(Mapping)\n</code></pre>"},{"location":"Python-LLAPI-Documentation/#unityenvregistry","title":"UnityEnvRegistry","text":"<p>Provides a library of Unity environments that can be launched without the need of downloading the Unity Editor. The UnityEnvRegistry implements a Map, to access an entry of the Registry, use:</p> <pre><code>registry = UnityEnvRegistry()\nentry = registry[&lt;environment_identifier&gt;]\n</code></pre> <p>An entry has the following properties :  * <code>identifier</code> : Uniquely identifies this environment  * <code>expected_reward</code> : Corresponds to the reward an agent must obtained for the task  to be considered completed.  * <code>description</code> : A human readable description of the environment.</p> <p>To launch a Unity environment from a registry entry, use the <code>make</code> method:</p> <pre><code>registry = UnityEnvRegistry()\nenv = registry[&lt;environment_identifier&gt;].make()\n</code></pre> <p></p>"},{"location":"Python-LLAPI-Documentation/#register","title":"register","text":"<pre><code> | register(new_entry: BaseRegistryEntry) -&gt; None\n</code></pre> <p>Registers a new BaseRegistryEntry to the registry. The BaseRegistryEntry.identifier value will be used as indexing key. If two are more environments are registered under the same key, the most recentry added will replace the others.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#register_from_yaml","title":"register_from_yaml","text":"<pre><code> | register_from_yaml(path_to_yaml: str) -&gt; None\n</code></pre> <p>Registers the environments listed in a yaml file (either local or remote). Note that the entries are registered lazily: the registration will only happen when an environment is accessed. The yaml file must have the following format :</p> <pre><code>environments:\n- &lt;identifier of the first environment&gt;:\n    expected_reward: &lt;expected reward of the environment&gt;\n    description: | &lt;a multi line description of the environment&gt;\n      &lt;continued multi line description&gt;\n    linux_url: &lt;The url for the Linux executable zip file&gt;\n    darwin_url: &lt;The url for the OSX executable zip file&gt;\n    win_url: &lt;The url for the Windows executable zip file&gt;\n\n- &lt;identifier of the second environment&gt;:\n    expected_reward: &lt;expected reward of the environment&gt;\n    description: | &lt;a multi line description of the environment&gt;\n      &lt;continued multi line description&gt;\n    linux_url: &lt;The url for the Linux executable zip file&gt;\n    darwin_url: &lt;The url for the OSX executable zip file&gt;\n    win_url: &lt;The url for the Windows executable zip file&gt;\n\n- ...\n</code></pre> <p>Arguments:</p> <ul> <li><code>path_to_yaml</code>: A local path or url to the yaml file</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#clear","title":"clear","text":"<pre><code> | clear() -&gt; None\n</code></pre> <p>Deletes all entries in the registry.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#__getitem___2","title":"__getitem__","text":"<pre><code> | __getitem__(identifier: str) -&gt; BaseRegistryEntry\n</code></pre> <p>Returns the BaseRegistryEntry with the provided identifier. BaseRegistryEntry can then be used to make a Unity Environment.</p> <p>Arguments:</p> <ul> <li><code>identifier</code>: The identifier of the BaseRegistryEntry</li> </ul> <p>Returns:</p> <p>The associated BaseRegistryEntry</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channel","title":"mlagents_envs.side_channel","text":""},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelraw_bytes_channel","title":"mlagents_envs.side_channel.raw_bytes_channel","text":""},{"location":"Python-LLAPI-Documentation/#rawbyteschannel-objects","title":"RawBytesChannel Objects","text":"<pre><code>class RawBytesChannel(SideChannel)\n</code></pre> <p>This is an example of what the SideChannel for raw bytes exchange would look like. Is meant to be used for general research purpose.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#on_message_received","title":"on_message_received","text":"<pre><code> | on_message_received(msg: IncomingMessage) -&gt; None\n</code></pre> <p>Is called by the environment to the side channel. Can be called multiple times per step if multiple messages are meant for that SideChannel.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#get_and_clear_received_messages","title":"get_and_clear_received_messages","text":"<pre><code> | get_and_clear_received_messages() -&gt; List[bytes]\n</code></pre> <p>returns a list of bytearray received from the environment.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#send_raw_data","title":"send_raw_data","text":"<pre><code> | send_raw_data(data: bytearray) -&gt; None\n</code></pre> <p>Queues a message to be sent by the environment at the next call to step.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channeloutgoing_message","title":"mlagents_envs.side_channel.outgoing_message","text":""},{"location":"Python-LLAPI-Documentation/#outgoingmessage-objects","title":"OutgoingMessage Objects","text":"<pre><code>class OutgoingMessage()\n</code></pre> <p>Utility class for forming the message that is written to a SideChannel. All data is written in little-endian format using the struct module.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#__init___1","title":"__init__","text":"<pre><code> | __init__()\n</code></pre> <p>Create an OutgoingMessage with an empty buffer.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#write_bool","title":"write_bool","text":"<pre><code> | write_bool(b: bool) -&gt; None\n</code></pre> <p>Append a boolean value.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#write_int32","title":"write_int32","text":"<pre><code> | write_int32(i: int) -&gt; None\n</code></pre> <p>Append an integer value.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#write_float32","title":"write_float32","text":"<pre><code> | write_float32(f: float) -&gt; None\n</code></pre> <p>Append a float value. It will be truncated to 32-bit precision.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#write_float32_list","title":"write_float32_list","text":"<pre><code> | write_float32_list(float_list: List[float]) -&gt; None\n</code></pre> <p>Append a list of float values. They will be truncated to 32-bit precision.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#write_string","title":"write_string","text":"<pre><code> | write_string(s: str) -&gt; None\n</code></pre> <p>Append a string value. Internally, it will be encoded to ascii, and the encoded length will also be written to the message.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_raw_bytes","title":"set_raw_bytes","text":"<pre><code> | set_raw_bytes(buffer: bytearray) -&gt; None\n</code></pre> <p>Set the internal buffer to a new bytearray. This will overwrite any existing data.</p> <p>Arguments:</p> <ul> <li><code>buffer</code>:</li> </ul> <p>Returns:</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelengine_configuration_channel","title":"mlagents_envs.side_channel.engine_configuration_channel","text":""},{"location":"Python-LLAPI-Documentation/#engineconfigurationchannel-objects","title":"EngineConfigurationChannel Objects","text":"<pre><code>class EngineConfigurationChannel(SideChannel)\n</code></pre> <p>This is the SideChannel for engine configuration exchange. The data in the engine configuration is as follows :  - int width;  - int height;  - int qualityLevel;  - float timeScale;  - int targetFrameRate;  - int captureFrameRate;</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#on_message_received_1","title":"on_message_received","text":"<pre><code> | on_message_received(msg: IncomingMessage) -&gt; None\n</code></pre> <p>Is called by the environment to the side channel. Can be called multiple times per step if multiple messages are meant for that SideChannel. Note that Python should never receive an engine configuration from Unity</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_configuration_parameters","title":"set_configuration_parameters","text":"<pre><code> | set_configuration_parameters(width: Optional[int] = None, height: Optional[int] = None, quality_level: Optional[int] = None, time_scale: Optional[float] = None, target_frame_rate: Optional[int] = None, capture_frame_rate: Optional[int] = None) -&gt; None\n</code></pre> <p>Sets the engine configuration. Takes as input the configurations of the engine.</p> <p>Arguments:</p> <ul> <li><code>width</code>: Defines the width of the display. (Must be set alongside height)</li> <li><code>height</code>: Defines the height of the display. (Must be set alongside width)</li> <li><code>quality_level</code>: Defines the quality level of the simulation.</li> <li><code>time_scale</code>: Defines the multiplier for the deltatime in the simulation. If set to a higher value, time will pass faster in the simulation but the physics might break.</li> <li><code>target_frame_rate</code>: Instructs simulation to try to render at a specified frame rate.</li> <li><code>capture_frame_rate</code>: Instructs the simulation to consider time between updates to always be constant, regardless of the actual frame rate.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_configuration","title":"set_configuration","text":"<pre><code> | set_configuration(config: EngineConfig) -&gt; None\n</code></pre> <p>Sets the engine configuration. Takes as input an EngineConfig.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelside_channel_manager","title":"mlagents_envs.side_channel.side_channel_manager","text":""},{"location":"Python-LLAPI-Documentation/#sidechannelmanager-objects","title":"SideChannelManager Objects","text":"<pre><code>class SideChannelManager()\n</code></pre>"},{"location":"Python-LLAPI-Documentation/#process_side_channel_message","title":"process_side_channel_message","text":"<pre><code> | process_side_channel_message(data: bytes) -&gt; None\n</code></pre> <p>Separates the data received from Python into individual messages for each registered side channel and calls on_message_received on them.</p> <p>Arguments:</p> <ul> <li><code>data</code>: The packed message sent by Unity</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#generate_side_channel_messages","title":"generate_side_channel_messages","text":"<pre><code> | generate_side_channel_messages() -&gt; bytearray\n</code></pre> <p>Gathers the messages that the registered side channels will send to Unity and combines them into a single message ready to be sent.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelstats_side_channel","title":"mlagents_envs.side_channel.stats_side_channel","text":""},{"location":"Python-LLAPI-Documentation/#statssidechannel-objects","title":"StatsSideChannel Objects","text":"<pre><code>class StatsSideChannel(SideChannel)\n</code></pre> <p>Side channel that receives (string, float) pairs from the environment, so that they can eventually be passed to a StatsReporter.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#on_message_received_2","title":"on_message_received","text":"<pre><code> | on_message_received(msg: IncomingMessage) -&gt; None\n</code></pre> <p>Receive the message from the environment, and save it for later retrieval.</p> <p>Arguments:</p> <ul> <li><code>msg</code>:</li> </ul> <p>Returns:</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#get_and_reset_stats","title":"get_and_reset_stats","text":"<pre><code> | get_and_reset_stats() -&gt; EnvironmentStats\n</code></pre> <p>Returns the current stats, and resets the internal storage of the stats.</p> <p>Returns:</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelincoming_message","title":"mlagents_envs.side_channel.incoming_message","text":""},{"location":"Python-LLAPI-Documentation/#incomingmessage-objects","title":"IncomingMessage Objects","text":"<pre><code>class IncomingMessage()\n</code></pre> <p>Utility class for reading the message written to a SideChannel. Values must be read in the order they were written.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#__init___2","title":"__init__","text":"<pre><code> | __init__(buffer: bytes, offset: int = 0)\n</code></pre> <p>Create a new IncomingMessage from the bytes.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#read_bool","title":"read_bool","text":"<pre><code> | read_bool(default_value: bool = False) -&gt; bool\n</code></pre> <p>Read a boolean value from the message buffer.</p> <p>Arguments:</p> <ul> <li><code>default_value</code>: Default value to use if the end of the message is reached.</li> </ul> <p>Returns:</p> <p>The value read from the message, or the default value if the end was reached.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#read_int32","title":"read_int32","text":"<pre><code> | read_int32(default_value: int = 0) -&gt; int\n</code></pre> <p>Read an integer value from the message buffer.</p> <p>Arguments:</p> <ul> <li><code>default_value</code>: Default value to use if the end of the message is reached.</li> </ul> <p>Returns:</p> <p>The value read from the message, or the default value if the end was reached.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#read_float32","title":"read_float32","text":"<pre><code> | read_float32(default_value: float = 0.0) -&gt; float\n</code></pre> <p>Read a float value from the message buffer.</p> <p>Arguments:</p> <ul> <li><code>default_value</code>: Default value to use if the end of the message is reached.</li> </ul> <p>Returns:</p> <p>The value read from the message, or the default value if the end was reached.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#read_float32_list","title":"read_float32_list","text":"<pre><code> | read_float32_list(default_value: List[float] = None) -&gt; List[float]\n</code></pre> <p>Read a list of float values from the message buffer.</p> <p>Arguments:</p> <ul> <li><code>default_value</code>: Default value to use if the end of the message is reached.</li> </ul> <p>Returns:</p> <p>The value read from the message, or the default value if the end was reached.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#read_string","title":"read_string","text":"<pre><code> | read_string(default_value: str = \"\") -&gt; str\n</code></pre> <p>Read a string value from the message buffer.</p> <p>Arguments:</p> <ul> <li><code>default_value</code>: Default value to use if the end of the message is reached.</li> </ul> <p>Returns:</p> <p>The value read from the message, or the default value if the end was reached.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#get_raw_bytes","title":"get_raw_bytes","text":"<pre><code> | get_raw_bytes() -&gt; bytes\n</code></pre> <p>Get a copy of the internal bytes used by the message.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelfloat_properties_channel","title":"mlagents_envs.side_channel.float_properties_channel","text":""},{"location":"Python-LLAPI-Documentation/#floatpropertieschannel-objects","title":"FloatPropertiesChannel Objects","text":"<pre><code>class FloatPropertiesChannel(SideChannel)\n</code></pre> <p>This is the SideChannel for float properties shared with Unity. You can modify the float properties of an environment with the commands set_property, get_property and list_properties.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#on_message_received_3","title":"on_message_received","text":"<pre><code> | on_message_received(msg: IncomingMessage) -&gt; None\n</code></pre> <p>Is called by the environment to the side channel. Can be called multiple times per step if multiple messages are meant for that SideChannel.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_property","title":"set_property","text":"<pre><code> | set_property(key: str, value: float) -&gt; None\n</code></pre> <p>Sets a property in the Unity Environment.</p> <p>Arguments:</p> <ul> <li><code>key</code>: The string identifier of the property.</li> <li><code>value</code>: The float value of the property.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#get_property","title":"get_property","text":"<pre><code> | get_property(key: str) -&gt; Optional[float]\n</code></pre> <p>Gets a property in the Unity Environment. If the property was not found, will return None.</p> <p>Arguments:</p> <ul> <li><code>key</code>: The string identifier of the property.</li> </ul> <p>Returns:</p> <p>The float value of the property or None.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#list_properties","title":"list_properties","text":"<pre><code> | list_properties() -&gt; List[str]\n</code></pre> <p>Returns a list of all the string identifiers of the properties currently present in the Unity Environment.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#get_property_dict_copy","title":"get_property_dict_copy","text":"<pre><code> | get_property_dict_copy() -&gt; Dict[str, float]\n</code></pre> <p>Returns a copy of the float properties.</p> <p>Returns:</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelenvironment_parameters_channel","title":"mlagents_envs.side_channel.environment_parameters_channel","text":""},{"location":"Python-LLAPI-Documentation/#environmentparameterschannel-objects","title":"EnvironmentParametersChannel Objects","text":"<pre><code>class EnvironmentParametersChannel(SideChannel)\n</code></pre> <p>This is the SideChannel for sending environment parameters to Unity. You can send parameters to an environment with the command set_float_parameter.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_float_parameter","title":"set_float_parameter","text":"<pre><code> | set_float_parameter(key: str, value: float) -&gt; None\n</code></pre> <p>Sets a float environment parameter in the Unity Environment.</p> <p>Arguments:</p> <ul> <li><code>key</code>: The string identifier of the parameter.</li> <li><code>value</code>: The float value of the parameter.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_uniform_sampler_parameters","title":"set_uniform_sampler_parameters","text":"<pre><code> | set_uniform_sampler_parameters(key: str, min_value: float, max_value: float, seed: int) -&gt; None\n</code></pre> <p>Sets a uniform environment parameter sampler.</p> <p>Arguments:</p> <ul> <li><code>key</code>: The string identifier of the parameter.</li> <li><code>min_value</code>: The minimum of the sampling distribution.</li> <li><code>max_value</code>: The maximum of the sampling distribution.</li> <li><code>seed</code>: The random seed to initialize the sampler.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_gaussian_sampler_parameters","title":"set_gaussian_sampler_parameters","text":"<pre><code> | set_gaussian_sampler_parameters(key: str, mean: float, st_dev: float, seed: int) -&gt; None\n</code></pre> <p>Sets a gaussian environment parameter sampler.</p> <p>Arguments:</p> <ul> <li><code>key</code>: The string identifier of the parameter.</li> <li><code>mean</code>: The mean of the sampling distribution.</li> <li><code>st_dev</code>: The standard deviation of the sampling distribution.</li> <li><code>seed</code>: The random seed to initialize the sampler.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#set_multirangeuniform_sampler_parameters","title":"set_multirangeuniform_sampler_parameters","text":"<pre><code> | set_multirangeuniform_sampler_parameters(key: str, intervals: List[Tuple[float, float]], seed: int) -&gt; None\n</code></pre> <p>Sets a multirangeuniform environment parameter sampler.</p> <p>Arguments:</p> <ul> <li><code>key</code>: The string identifier of the parameter.</li> <li><code>intervals</code>: The lists of min and max that define each uniform distribution.</li> <li><code>seed</code>: The random seed to initialize the sampler.</li> </ul> <p></p>"},{"location":"Python-LLAPI-Documentation/#mlagents_envsside_channelside_channel","title":"mlagents_envs.side_channel.side_channel","text":""},{"location":"Python-LLAPI-Documentation/#sidechannel-objects","title":"SideChannel Objects","text":"<pre><code>class SideChannel(ABC)\n</code></pre> <p>The side channel just get access to a bytes buffer that will be shared between C# and Python. For example, We will create a specific side channel for properties that will be a list of string (fixed size) to float number, that can be modified by both C# and Python. All side channels are passed to the Env object at construction.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#queue_message_to_send","title":"queue_message_to_send","text":"<pre><code> | queue_message_to_send(msg: OutgoingMessage) -&gt; None\n</code></pre> <p>Queues a message to be sent by the environment at the next call to step.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#on_message_received_4","title":"on_message_received","text":"<pre><code> | @abstractmethod\n | on_message_received(msg: IncomingMessage) -&gt; None\n</code></pre> <p>Is called by the environment to the side channel. Can be called multiple times per step if multiple messages are meant for that SideChannel.</p> <p></p>"},{"location":"Python-LLAPI-Documentation/#channel_id","title":"channel_id","text":"<pre><code> | @property\n | channel_id() -&gt; uuid.UUID\n</code></pre> <p>Returns:</p> <p>The type of side channel used. Will influence how the data is processed in the environment.</p>"},{"location":"Python-LLAPI/","title":"Unity ML-Agents Python Low Level API","text":"<p>The <code>mlagents</code> Python package contains two components: a low level API which allows you to interact directly with a Unity Environment (<code>mlagents_envs</code>) and an entry point to train (<code>mlagents-learn</code>) which allows you to train agents in Unity Environments using our implementations of reinforcement learning or imitation learning. This document describes how to use the <code>mlagents_envs</code> API. For information on using <code>mlagents-learn</code>, see here. For Python Low Level API documentation, see here.</p> <p>The Python Low Level API can be used to interact directly with your Unity learning environment. As such, it can serve as the basis for developing and evaluating new learning algorithms.</p>"},{"location":"Python-LLAPI/#mlagents_envs","title":"mlagents_envs","text":"<p>The ML-Agents Toolkit Low Level API is a Python API for controlling the simulation loop of an environment or game built with Unity. This API is used by the training algorithms inside the ML-Agent Toolkit, but you can also write your own Python programs using this API.</p> <p>The key objects in the Python API include:</p> <ul> <li>UnityEnvironment \u2014 the main interface between the Unity application and   your code. Use UnityEnvironment to start and control a simulation or training   session.</li> <li>BehaviorName - is a string that identifies a behavior in the simulation.</li> <li>AgentId - is an <code>int</code> that serves as unique identifier for Agents in the   simulation.</li> <li>DecisionSteps \u2014 contains the data from Agents belonging to the same   \"Behavior\" in the simulation, such as observations and rewards. Only Agents   that requested a decision since the last call to <code>env.step()</code> are in the   DecisionSteps object.</li> <li>TerminalSteps \u2014 contains the data from Agents belonging to the same   \"Behavior\" in the simulation, such as observations and rewards. Only Agents   whose episode ended since the last call to <code>env.step()</code> are in the   TerminalSteps object.</li> <li>BehaviorSpec \u2014 describes the shape of the observation data inside   DecisionSteps and TerminalSteps as well as the expected action shapes.</li> </ul> <p>These classes are all defined in the base_env script.</p> <p>An Agent \"Behavior\" is a group of Agents identified by a <code>BehaviorName</code> that share the same observations and action types (described in their <code>BehaviorSpec</code>). You can think about Agent Behavior as a group of agents that will share the same policy. All Agents with the same behavior have the same goal and reward signals.</p> <p>To communicate with an Agent in a Unity environment from a Python program, the Agent in the simulation must have <code>Behavior Parameters</code> set to communicate. You must set the <code>Behavior Type</code> to <code>Default</code> and give it a <code>Behavior Name</code>.</p> <p>Notice: Currently communication between Unity and Python takes place over an open socket without authentication. As such, please make sure that the network where training takes place is secure. This will be addressed in a future release.</p>"},{"location":"Python-LLAPI/#loading-a-unity-environment","title":"Loading a Unity Environment","text":"<p>Python-side communication happens through <code>UnityEnvironment</code> which is located in <code>environment.py</code>. To load a Unity environment from a built binary file, put the file in the same directory as <code>envs</code>. For example, if the filename of your Unity environment is <code>3DBall</code>, in python, run:</p> <pre><code>from mlagents_envs.environment import UnityEnvironment\n# This is a non-blocking call that only loads the environment.\nenv = UnityEnvironment(file_name=\"3DBall\", seed=1, side_channels=[])\n# Start interacting with the environment.\nenv.reset()\nbehavior_names = env.behavior_specs.keys()\n...\n</code></pre> <p>NOTE: Please read Interacting with a Unity Environment to read more about how you can interact with the Unity environment from Python.</p> <ul> <li><code>file_name</code> is the name of the environment binary (located in the root   directory of the python project).</li> <li><code>worker_id</code> indicates which port to use for communication with the   environment. For use in parallel training regimes such as A3C.</li> <li><code>seed</code> indicates the seed to use when generating random numbers during the   training process. In environments which are stochastic, setting the seed   enables reproducible experimentation by ensuring that the environment and   trainers utilize the same random seed.</li> <li><code>side_channels</code> provides a way to exchange data with the Unity simulation that   is not related to the reinforcement learning loop. For example: configurations   or properties. More on them in the Side Channels doc.</li> </ul> <p>If you want to directly interact with the Editor, you need to use <code>file_name=None</code>, then press the Play button in the Editor when the message \"Start training by pressing the Play button in the Unity Editor\" is displayed on the screen</p>"},{"location":"Python-LLAPI/#interacting-with-a-unity-environment","title":"Interacting with a Unity Environment","text":""},{"location":"Python-LLAPI/#the-baseenv-interface","title":"The BaseEnv interface","text":"<p>A <code>BaseEnv</code> has the following methods:</p> <ul> <li>Reset : <code>env.reset()</code> Sends a signal to reset the environment. Returns   None.</li> <li>Step : <code>env.step()</code> Sends a signal to step the environment. Returns None.   Note that a \"step\" for Python does not correspond to either Unity <code>Update</code> nor   <code>FixedUpdate</code>. When <code>step()</code> or <code>reset()</code> is called, the Unity simulation will   move forward until an Agent in the simulation needs a input from Python to   act.</li> <li>Close : <code>env.close()</code> Sends a shutdown signal to the environment and   terminates the communication.</li> <li>Behavior Specs : <code>env.behavior_specs</code> Returns a Mapping of   <code>BehaviorName</code> to <code>BehaviorSpec</code> objects (read only).   A <code>BehaviorSpec</code> contains the observation shapes and the   <code>ActionSpec</code> (which defines the action shape). Note that   the <code>BehaviorSpec</code> for a specific group is fixed throughout the simulation.   The number of entries in the Mapping can change over time in the simulation   if new Agent behaviors are created in the simulation.</li> <li>Get Steps : <code>env.get_steps(behavior_name: str)</code> Returns a tuple   <code>DecisionSteps, TerminalSteps</code> corresponding to the behavior_name given as   input. The <code>DecisionSteps</code> contains information about the state of the agents   that need an action this step and have the behavior behavior_name. The   <code>TerminalSteps</code> contains information about the state of the agents whose   episode ended and have the behavior behavior_name. Both <code>DecisionSteps</code> and   <code>TerminalSteps</code> contain information such as the observations, the rewards and   the agent identifiers. <code>DecisionSteps</code> also contains action masks for the next   action while <code>TerminalSteps</code> contains the reason for termination (did the   Agent reach its maximum step and was interrupted). The data is in <code>np.array</code>   of which the first dimension is always the number of agents note that the   number of agents is not guaranteed to remain constant during the simulation   and it is not unusual to have either <code>DecisionSteps</code> or <code>TerminalSteps</code>   contain no Agents at all.</li> <li>Set Actions :<code>env.set_actions(behavior_name: str, action: ActionTuple)</code> Sets   the actions for a whole agent group. <code>action</code> is an <code>ActionTuple</code>, which   is made up of a 2D <code>np.array</code> of <code>dtype=np.int32</code> for discrete actions, and   <code>dtype=np.float32</code> for continuous actions. The first dimension of <code>np.array</code>   in the tuple is the number of agents that requested a decision since the   last call to <code>env.step()</code>. The second dimension is the number of discrete or   continuous actions for the corresponding array.</li> <li>Set Action for Agent :   <code>env.set_action_for_agent(agent_group: str, agent_id: int, action: ActionTuple)</code>   Sets the action for a specific Agent in an agent group. <code>agent_group</code> is the   name of the group the Agent belongs to and <code>agent_id</code> is the integer   identifier of the Agent. <code>action</code> is an <code>ActionTuple</code> as described above. Note: If no action is provided for an agent group between two calls to <code>env.step()</code> then the default action will be all zeros.</li> </ul>"},{"location":"Python-LLAPI/#decisionsteps-and-decisionstep","title":"DecisionSteps and DecisionStep","text":"<p><code>DecisionSteps</code> (with <code>s</code>) contains information about a whole batch of Agents while <code>DecisionStep</code> (no <code>s</code>) only contains information about a single Agent.</p> <p>A <code>DecisionSteps</code> has the following fields :</p> <ul> <li><code>obs</code> is a list of numpy arrays observations collected by the group of agent.   The first dimension of the array corresponds to the batch size of the group   (number of agents requesting a decision since the last call to <code>env.step()</code>).</li> <li><code>reward</code> is a float vector of length batch size. Corresponds to the rewards   collected by each agent since the last simulation step.</li> <li><code>agent_id</code> is an int vector of length batch size containing unique identifier   for the corresponding Agent. This is used to track Agents across simulation   steps.</li> <li><code>action_mask</code> is an optional list of two dimensional arrays of booleans which is only   available when using multi-discrete actions. Each array corresponds to an   action branch. The first dimension of each array is the batch size and the   second contains a mask for each action of the branch. If true, the action is   not available for the agent during this simulation step.</li> </ul> <p>It also has the two following methods:</p> <ul> <li><code>len(DecisionSteps)</code> Returns the number of agents requesting a decision since   the last call to <code>env.step()</code>.</li> <li><code>DecisionSteps[agent_id]</code> Returns a <code>DecisionStep</code> for the Agent with the   <code>agent_id</code> unique identifier.</li> </ul> <p>A <code>DecisionStep</code> has the following fields:</p> <ul> <li><code>obs</code> is a list of numpy arrays observations collected by the agent. (Each   array has one less dimension than the arrays in <code>DecisionSteps</code>)</li> <li><code>reward</code> is a float. Corresponds to the rewards collected by the agent since   the last simulation step.</li> <li><code>agent_id</code> is an int and an unique identifier for the corresponding Agent.</li> <li><code>action_mask</code> is an optional list of one dimensional arrays of booleans which is only   available when using multi-discrete actions. Each array corresponds to an   action branch. Each array contains a mask for each action of the branch. If   true, the action is not available for the agent during this simulation step.</li> </ul>"},{"location":"Python-LLAPI/#terminalsteps-and-terminalstep","title":"TerminalSteps and TerminalStep","text":"<p>Similarly to <code>DecisionSteps</code> and <code>DecisionStep</code>, <code>TerminalSteps</code> (with <code>s</code>) contains information about a whole batch of Agents while <code>TerminalStep</code> (no <code>s</code>) only contains information about a single Agent.</p> <p>A <code>TerminalSteps</code> has the following fields :</p> <ul> <li><code>obs</code> is a list of numpy arrays observations collected by the group of agent.   The first dimension of the array corresponds to the batch size of the group   (number of agents requesting a decision since the last call to <code>env.step()</code>).</li> <li><code>reward</code> is a float vector of length batch size. Corresponds to the rewards   collected by each agent since the last simulation step.</li> <li><code>agent_id</code> is an int vector of length batch size containing unique identifier   for the corresponding Agent. This is used to track Agents across simulation   steps.</li> <li><code>interrupted</code> is an array of booleans of length batch size. Is true if the  associated Agent was interrupted since the last decision step. For example,  if the Agent reached the maximum number of steps for the episode.</li> </ul> <p>It also has the two following methods:</p> <ul> <li><code>len(TerminalSteps)</code> Returns the number of agents requesting a decision since   the last call to <code>env.step()</code>.</li> <li><code>TerminalSteps[agent_id]</code> Returns a <code>TerminalStep</code> for the Agent with the   <code>agent_id</code> unique identifier.</li> </ul> <p>A <code>TerminalStep</code> has the following fields:</p> <ul> <li><code>obs</code> is a list of numpy arrays observations collected by the agent. (Each   array has one less dimension than the arrays in <code>TerminalSteps</code>)</li> <li><code>reward</code> is a float. Corresponds to the rewards collected by the agent since   the last simulation step.</li> <li><code>agent_id</code> is an int and an unique identifier for the corresponding Agent.</li> <li><code>interrupted</code> is a bool. Is true if the Agent was interrupted since the last  decision step. For example, if the Agent reached the maximum number of steps for  the episode.</li> </ul>"},{"location":"Python-LLAPI/#behaviorspec","title":"BehaviorSpec","text":"<p>A <code>BehaviorSpec</code> has the following fields :</p> <ul> <li><code>observation_specs</code> is a List of <code>ObservationSpec</code> objects : Each <code>ObservationSpec</code>   corresponds to an observation's properties: <code>shape</code> is a tuple of ints that   corresponds to the shape of the observation (without the number of agents dimension).   <code>dimension_property</code> is a tuple of flags containing extra information about how the   data should be processed in the corresponding dimension. <code>observation_type</code> is an enum   corresponding to what type of observation is generating the data (i.e., default, goal,   etc). Note that the <code>ObservationSpec</code> have the same ordering as the ordering of observations   in the DecisionSteps, DecisionStep, TerminalSteps and TerminalStep.</li> <li><code>action_spec</code> is an <code>ActionSpec</code> namedtuple that defines the number and types   of actions for the Agent.</li> </ul> <p>An <code>ActionSpec</code> has the following fields and properties: - <code>continuous_size</code> is the number of floats that constitute the continuous actions. - <code>discrete_size</code> is the number of branches (the number of independent actions) that   constitute the multi-discrete actions. - <code>discrete_branches</code> is a Tuple of ints. Each int corresponds to the number of   different options for each branch of the action. For example:   In a game direction input (no movement, left, right) and   jump input (no jump, jump) there will be two branches (direction and jump),   the first one with 3 options and the second with 2 options. (<code>discrete_size = 2</code>   and <code>discrete_action_branches = (3,2,)</code>)</p>"},{"location":"Python-LLAPI/#communicating-additional-information-with-the-environment","title":"Communicating additional information with the Environment","text":"<p>In addition to the means of communicating between Unity and python described above, we also provide methods for sharing agent-agnostic information. These additional methods are referred to as side channels. ML-Agents includes two ready-made side channels, described below. It is also possible to create custom side channels to communicate any additional data between a Unity environment and Python. Instructions for creating custom side channels can be found here.</p> <p>Side channels exist as separate classes which are instantiated, and then passed as list to the <code>side_channels</code> argument of the constructor of the <code>UnityEnvironment</code> class.</p> <pre><code>channel = MyChannel()\n\nenv = UnityEnvironment(side_channels = [channel])\n</code></pre> <p>Note : A side channel will only send/receive messages when <code>env.step</code> or <code>env.reset()</code> is called.</p>"},{"location":"Python-LLAPI/#engineconfigurationchannel","title":"EngineConfigurationChannel","text":"<p>The <code>EngineConfiguration</code> side channel allows you to modify the time-scale, resolution, and graphics quality of the environment. This can be useful for adjusting the environment to perform better during training, or be more interpretable during inference.</p> <p><code>EngineConfigurationChannel</code> has two methods :</p> <ul> <li><code>set_configuration_parameters</code> which takes the following arguments:</li> <li><code>width</code>: Defines the width of the display. (Must be set alongside height)</li> <li><code>height</code>: Defines the height of the display. (Must be set alongside width)</li> <li><code>quality_level</code>: Defines the quality level of the simulation.</li> <li><code>time_scale</code>: Defines the multiplier for the deltatime in the simulation. If     set to a higher value, time will pass faster in the simulation but the     physics may perform unpredictably.</li> <li><code>target_frame_rate</code>: Instructs simulation to try to render at a specified     frame rate.</li> <li><code>capture_frame_rate</code> Instructs the simulation to consider time between     updates to always be constant, regardless of the actual frame rate.</li> <li><code>set_configuration</code> with argument config which is an <code>EngineConfig</code> NamedTuple   object.</li> </ul> <p>For example, the following code would adjust the time-scale of the simulation to be 2x realtime.</p> <pre><code>from mlagents_envs.environment import UnityEnvironment\nfrom mlagents_envs.side_channel.engine_configuration_channel import EngineConfigurationChannel\n\nchannel = EngineConfigurationChannel()\n\nenv = UnityEnvironment(side_channels=[channel])\n\nchannel.set_configuration_parameters(time_scale = 2.0)\n\ni = env.reset()\n...\n</code></pre>"},{"location":"Python-LLAPI/#environmentparameters","title":"EnvironmentParameters","text":"<p>The <code>EnvironmentParameters</code> will allow you to get and set pre-defined numerical values in the environment. This can be useful for adjusting environment-specific settings, or for reading non-agent related information from the environment. You can call <code>get_property</code> and <code>set_property</code> on the side channel to read and write properties.</p> <p><code>EnvironmentParametersChannel</code> has one methods:</p> <ul> <li><code>set_float_parameter</code> Sets a float parameter in the Unity Environment.</li> <li>key: The string identifier of the property.</li> <li>value: The float value of the property.</li> </ul> <pre><code>from mlagents_envs.environment import UnityEnvironment\nfrom mlagents_envs.side_channel.environment_parameters_channel import EnvironmentParametersChannel\n\nchannel = EnvironmentParametersChannel()\n\nenv = UnityEnvironment(side_channels=[channel])\n\nchannel.set_float_parameter(\"parameter_1\", 2.0)\n\ni = env.reset()\n...\n</code></pre> <p>Once a property has been modified in Python, you can access it in C# after the next call to <code>step</code> as follows:</p> <pre><code>var envParameters = Academy.Instance.EnvironmentParameters;\nfloat property1 = envParameters.GetWithDefault(\"parameter_1\", 0.0f);\n</code></pre>"},{"location":"Python-LLAPI/#custom-side-channels","title":"Custom side channels","text":"<p>For information on how to make custom side channels for sending additional data types, see the documentation here.</p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/","title":"Table of Contents","text":"<ul> <li>mlagents.trainers.trainer.on_policy_trainer</li> <li>OnPolicyTrainer<ul> <li>__init__</li> <li>add_policy</li> </ul> </li> <li>mlagents.trainers.trainer.off_policy_trainer</li> <li>OffPolicyTrainer<ul> <li>__init__</li> <li>save_model</li> <li>save_replay_buffer</li> <li>load_replay_buffer</li> <li>add_policy</li> </ul> </li> <li>mlagents.trainers.trainer.rl_trainer</li> <li>RLTrainer<ul> <li>end_episode</li> <li>create_optimizer</li> <li>save_model</li> <li>advance</li> </ul> </li> <li>mlagents.trainers.trainer.trainer</li> <li>Trainer<ul> <li>__init__</li> <li>stats_reporter</li> <li>parameters</li> <li>get_max_steps</li> <li>get_step</li> <li>threaded</li> <li>should_still_train</li> <li>reward_buffer</li> <li>save_model</li> <li>end_episode</li> <li>create_policy</li> <li>add_policy</li> <li>get_policy</li> <li>advance</li> <li>publish_policy_queue</li> <li>subscribe_trajectory_queue</li> </ul> </li> <li>mlagents.trainers.settings</li> <li>deep_update_dict</li> <li>RewardSignalSettings<ul> <li>structure</li> </ul> </li> <li>ParameterRandomizationSettings<ul> <li>__str__</li> <li>structure</li> <li>unstructure</li> <li>apply</li> </ul> </li> <li>ConstantSettings<ul> <li>__str__</li> <li>apply</li> </ul> </li> <li>UniformSettings<ul> <li>__str__</li> <li>apply</li> </ul> </li> <li>GaussianSettings<ul> <li>__str__</li> <li>apply</li> </ul> </li> <li>MultiRangeUniformSettings<ul> <li>__str__</li> <li>apply</li> </ul> </li> <li>CompletionCriteriaSettings<ul> <li>need_increment</li> </ul> </li> <li>Lesson</li> <li>EnvironmentParameterSettings<ul> <li>structure</li> </ul> </li> <li>TrainerSettings<ul> <li>structure</li> </ul> </li> <li>CheckpointSettings<ul> <li>prioritize_resume_init</li> </ul> </li> <li>RunOptions<ul> <li>from_argparse</li> </ul> </li> </ul>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#mlagentstrainerstraineron_policy_trainer","title":"mlagents.trainers.trainer.on_policy_trainer","text":""},{"location":"Python-On-Off-Policy-Trainer-Documentation/#onpolicytrainer-objects","title":"OnPolicyTrainer Objects","text":"<pre><code>class OnPolicyTrainer(RLTrainer)\n</code></pre> <p>The PPOTrainer is an implementation of the PPO algorithm.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__init__","title":"__init__","text":"<pre><code> | __init__(behavior_name: str, reward_buff_cap: int, trainer_settings: TrainerSettings, training: bool, load: bool, seed: int, artifact_path: str)\n</code></pre> <p>Responsible for collecting experiences and training an on-policy model.</p> <p>Arguments:</p> <ul> <li><code>behavior_name</code>: The name of the behavior associated with trainer config</li> <li><code>reward_buff_cap</code>: Max reward history to track in the reward buffer</li> <li><code>trainer_settings</code>: The parameters for the trainer.</li> <li><code>training</code>: Whether the trainer is set for training.</li> <li><code>load</code>: Whether the model should be loaded.</li> <li><code>seed</code>: The seed the model will be initialized with</li> <li><code>artifact_path</code>: The directory within which to store artifacts from this trainer.</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#add_policy","title":"add_policy","text":"<pre><code> | add_policy(parsed_behavior_id: BehaviorIdentifiers, policy: Policy) -&gt; None\n</code></pre> <p>Adds policy to trainer.</p> <p>Arguments:</p> <ul> <li><code>parsed_behavior_id</code>: Behavior identifiers that the policy should belong to.</li> <li><code>policy</code>: Policy to associate with name_behavior_id.</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#mlagentstrainerstraineroff_policy_trainer","title":"mlagents.trainers.trainer.off_policy_trainer","text":""},{"location":"Python-On-Off-Policy-Trainer-Documentation/#offpolicytrainer-objects","title":"OffPolicyTrainer Objects","text":"<pre><code>class OffPolicyTrainer(RLTrainer)\n</code></pre> <p>The SACTrainer is an implementation of the SAC algorithm, with support for discrete actions and recurrent networks.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__init___1","title":"__init__","text":"<pre><code> | __init__(behavior_name: str, reward_buff_cap: int, trainer_settings: TrainerSettings, training: bool, load: bool, seed: int, artifact_path: str)\n</code></pre> <p>Responsible for collecting experiences and training an off-policy model.</p> <p>Arguments:</p> <ul> <li><code>behavior_name</code>: The name of the behavior associated with trainer config</li> <li><code>reward_buff_cap</code>: Max reward history to track in the reward buffer</li> <li><code>trainer_settings</code>: The parameters for the trainer.</li> <li><code>training</code>: Whether the trainer is set for training.</li> <li><code>load</code>: Whether the model should be loaded.</li> <li><code>seed</code>: The seed the model will be initialized with</li> <li><code>artifact_path</code>: The directory within which to store artifacts from this trainer.</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#save_model","title":"save_model","text":"<pre><code> | save_model() -&gt; None\n</code></pre> <p>Saves the final training model to memory Overrides the default to save the replay buffer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#save_replay_buffer","title":"save_replay_buffer","text":"<pre><code> | save_replay_buffer() -&gt; None\n</code></pre> <p>Save the training buffer's update buffer to a pickle file.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#load_replay_buffer","title":"load_replay_buffer","text":"<pre><code> | load_replay_buffer() -&gt; None\n</code></pre> <p>Loads the last saved replay buffer from a file.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#add_policy_1","title":"add_policy","text":"<pre><code> | add_policy(parsed_behavior_id: BehaviorIdentifiers, policy: Policy) -&gt; None\n</code></pre> <p>Adds policy to trainer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#mlagentstrainerstrainerrl_trainer","title":"mlagents.trainers.trainer.rl_trainer","text":""},{"location":"Python-On-Off-Policy-Trainer-Documentation/#rltrainer-objects","title":"RLTrainer Objects","text":"<pre><code>class RLTrainer(Trainer)\n</code></pre> <p>This class is the base class for trainers that use Reward Signals.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#end_episode","title":"end_episode","text":"<pre><code> | end_episode() -&gt; None\n</code></pre> <p>A signal that the Episode has ended. The buffer must be reset. Get only called when the academy resets.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#create_optimizer","title":"create_optimizer","text":"<pre><code> | @abc.abstractmethod\n | create_optimizer() -&gt; TorchOptimizer\n</code></pre> <p>Creates an Optimizer object</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#save_model_1","title":"save_model","text":"<pre><code> | save_model() -&gt; None\n</code></pre> <p>Saves the policy associated with this trainer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#advance","title":"advance","text":"<pre><code> | advance() -&gt; None\n</code></pre> <p>Steps the trainer, taking in trajectories and updates if ready. Will block and wait briefly if there are no trajectories.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#mlagentstrainerstrainertrainer","title":"mlagents.trainers.trainer.trainer","text":""},{"location":"Python-On-Off-Policy-Trainer-Documentation/#trainer-objects","title":"Trainer Objects","text":"<pre><code>class Trainer(abc.ABC)\n</code></pre> <p>This class is the base class for the mlagents_envs.trainers</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__init___2","title":"__init__","text":"<pre><code> | __init__(brain_name: str, trainer_settings: TrainerSettings, training: bool, load: bool, artifact_path: str, reward_buff_cap: int = 1)\n</code></pre> <p>Responsible for collecting experiences and training a neural network model.</p> <p>Arguments:</p> <ul> <li><code>brain_name</code>: Brain name of brain to be trained.</li> <li><code>trainer_settings</code>: The parameters for the trainer (dictionary).</li> <li><code>training</code>: Whether the trainer is set for training.</li> <li><code>artifact_path</code>: The directory within which to store artifacts from this trainer</li> <li><code>reward_buff_cap</code>:</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#stats_reporter","title":"stats_reporter","text":"<pre><code> | @property\n | stats_reporter()\n</code></pre> <p>Returns the stats reporter associated with this Trainer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#parameters","title":"parameters","text":"<pre><code> | @property\n | parameters() -&gt; TrainerSettings\n</code></pre> <p>Returns the trainer parameters of the trainer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#get_max_steps","title":"get_max_steps","text":"<pre><code> | @property\n | get_max_steps() -&gt; int\n</code></pre> <p>Returns the maximum number of steps. Is used to know when the trainer should be stopped.</p> <p>Returns:</p> <p>The maximum number of steps of the trainer</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#get_step","title":"get_step","text":"<pre><code> | @property\n | get_step() -&gt; int\n</code></pre> <p>Returns the number of steps the trainer has performed</p> <p>Returns:</p> <p>the step count of the trainer</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#threaded","title":"threaded","text":"<pre><code> | @property\n | threaded() -&gt; bool\n</code></pre> <p>Whether or not to run the trainer in a thread. True allows the trainer to update the policy while the environment is taking steps. Set to False to enforce strict on-policy updates (i.e. don't update the policy when taking steps.)</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#should_still_train","title":"should_still_train","text":"<pre><code> | @property\n | should_still_train() -&gt; bool\n</code></pre> <p>Returns whether or not the trainer should train. A Trainer could stop training if it wasn't training to begin with, or if max_steps is reached.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#reward_buffer","title":"reward_buffer","text":"<pre><code> | @property\n | reward_buffer() -&gt; Deque[float]\n</code></pre> <p>Returns the reward buffer. The reward buffer contains the cumulative rewards of the most recent episodes completed by agents using this trainer.</p> <p>Returns:</p> <p>the reward buffer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#save_model_2","title":"save_model","text":"<pre><code> | @abc.abstractmethod\n | save_model() -&gt; None\n</code></pre> <p>Saves model file(s) for the policy or policies associated with this trainer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#end_episode_1","title":"end_episode","text":"<pre><code> | @abc.abstractmethod\n | end_episode()\n</code></pre> <p>A signal that the Episode has ended. The buffer must be reset. Get only called when the academy resets.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#create_policy","title":"create_policy","text":"<pre><code> | @abc.abstractmethod\n | create_policy(parsed_behavior_id: BehaviorIdentifiers, behavior_spec: BehaviorSpec) -&gt; Policy\n</code></pre> <p>Creates a Policy object</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#add_policy_2","title":"add_policy","text":"<pre><code> | @abc.abstractmethod\n | add_policy(parsed_behavior_id: BehaviorIdentifiers, policy: Policy) -&gt; None\n</code></pre> <p>Adds policy to trainer.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#get_policy","title":"get_policy","text":"<pre><code> | get_policy(name_behavior_id: str) -&gt; Policy\n</code></pre> <p>Gets policy associated with name_behavior_id</p> <p>Arguments:</p> <ul> <li><code>name_behavior_id</code>: Fully qualified behavior name</li> </ul> <p>Returns:</p> <p>Policy associated with name_behavior_id</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#advance_1","title":"advance","text":"<pre><code> | @abc.abstractmethod\n | advance() -&gt; None\n</code></pre> <p>Advances the trainer. Typically, this means grabbing trajectories from all subscribed trajectory queues (self.trajectory_queues), and updating a policy using the steps in them, and if needed pushing a new policy onto the right policy queues (self.policy_queues).</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#publish_policy_queue","title":"publish_policy_queue","text":"<pre><code> | publish_policy_queue(policy_queue: AgentManagerQueue[Policy]) -&gt; None\n</code></pre> <p>Adds a policy queue to the list of queues to publish to when this Trainer makes a policy update</p> <p>Arguments:</p> <ul> <li><code>policy_queue</code>: Policy queue to publish to.</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#subscribe_trajectory_queue","title":"subscribe_trajectory_queue","text":"<pre><code> | subscribe_trajectory_queue(trajectory_queue: AgentManagerQueue[Trajectory]) -&gt; None\n</code></pre> <p>Adds a trajectory queue to the list of queues for the trainer to ingest Trajectories from.</p> <p>Arguments:</p> <ul> <li><code>trajectory_queue</code>: Trajectory queue to read from.</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#mlagentstrainerssettings","title":"mlagents.trainers.settings","text":""},{"location":"Python-On-Off-Policy-Trainer-Documentation/#deep_update_dict","title":"deep_update_dict","text":"<pre><code>deep_update_dict(d: Dict, update_d: Mapping) -&gt; None\n</code></pre> <p>Similar to dict.update(), but works for nested dicts of dicts as well.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#rewardsignalsettings-objects","title":"RewardSignalSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass RewardSignalSettings()\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#structure","title":"structure","text":"<pre><code> | @staticmethod\n | structure(d: Mapping, t: type) -&gt; Any\n</code></pre> <p>Helper method to structure a Dict of RewardSignalSettings class. Meant to be registered with cattr.register_structure_hook() and called with cattr.structure(). This is needed to handle the special Enum selection of RewardSignalSettings classes.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#parameterrandomizationsettings-objects","title":"ParameterRandomizationSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass ParameterRandomizationSettings(abc.ABC)\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__str__","title":"__str__","text":"<pre><code> | __str__() -&gt; str\n</code></pre> <p>Helper method to output sampler stats to console.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#structure_1","title":"structure","text":"<pre><code> | @staticmethod\n | structure(d: Union[Mapping, float], t: type) -&gt; \"ParameterRandomizationSettings\"\n</code></pre> <p>Helper method to a ParameterRandomizationSettings class. Meant to be registered with cattr.register_structure_hook() and called with cattr.structure(). This is needed to handle the special Enum selection of ParameterRandomizationSettings classes.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#unstructure","title":"unstructure","text":"<pre><code> | @staticmethod\n | unstructure(d: \"ParameterRandomizationSettings\") -&gt; Mapping\n</code></pre> <p>Helper method to a ParameterRandomizationSettings class. Meant to be registered with cattr.register_unstructure_hook() and called with cattr.unstructure().</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#apply","title":"apply","text":"<pre><code> | @abc.abstractmethod\n | apply(key: str, env_channel: EnvironmentParametersChannel) -&gt; None\n</code></pre> <p>Helper method to send sampler settings over EnvironmentParametersChannel Calls the appropriate sampler type set method.</p> <p>Arguments:</p> <ul> <li><code>key</code>: environment parameter to be sampled</li> <li><code>env_channel</code>: The EnvironmentParametersChannel to communicate sampler settings to environment</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#constantsettings-objects","title":"ConstantSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass ConstantSettings(ParameterRandomizationSettings)\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__str___1","title":"__str__","text":"<pre><code> | __str__() -&gt; str\n</code></pre> <p>Helper method to output sampler stats to console.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#apply_1","title":"apply","text":"<pre><code> | apply(key: str, env_channel: EnvironmentParametersChannel) -&gt; None\n</code></pre> <p>Helper method to send sampler settings over EnvironmentParametersChannel Calls the constant sampler type set method.</p> <p>Arguments:</p> <ul> <li><code>key</code>: environment parameter to be sampled</li> <li><code>env_channel</code>: The EnvironmentParametersChannel to communicate sampler settings to environment</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#uniformsettings-objects","title":"UniformSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass UniformSettings(ParameterRandomizationSettings)\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__str___2","title":"__str__","text":"<pre><code> | __str__() -&gt; str\n</code></pre> <p>Helper method to output sampler stats to console.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#apply_2","title":"apply","text":"<pre><code> | apply(key: str, env_channel: EnvironmentParametersChannel) -&gt; None\n</code></pre> <p>Helper method to send sampler settings over EnvironmentParametersChannel Calls the uniform sampler type set method.</p> <p>Arguments:</p> <ul> <li><code>key</code>: environment parameter to be sampled</li> <li><code>env_channel</code>: The EnvironmentParametersChannel to communicate sampler settings to environment</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#gaussiansettings-objects","title":"GaussianSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass GaussianSettings(ParameterRandomizationSettings)\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__str___3","title":"__str__","text":"<pre><code> | __str__() -&gt; str\n</code></pre> <p>Helper method to output sampler stats to console.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#apply_3","title":"apply","text":"<pre><code> | apply(key: str, env_channel: EnvironmentParametersChannel) -&gt; None\n</code></pre> <p>Helper method to send sampler settings over EnvironmentParametersChannel Calls the gaussian sampler type set method.</p> <p>Arguments:</p> <ul> <li><code>key</code>: environment parameter to be sampled</li> <li><code>env_channel</code>: The EnvironmentParametersChannel to communicate sampler settings to environment</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#multirangeuniformsettings-objects","title":"MultiRangeUniformSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass MultiRangeUniformSettings(ParameterRandomizationSettings)\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#__str___4","title":"__str__","text":"<pre><code> | __str__() -&gt; str\n</code></pre> <p>Helper method to output sampler stats to console.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#apply_4","title":"apply","text":"<pre><code> | apply(key: str, env_channel: EnvironmentParametersChannel) -&gt; None\n</code></pre> <p>Helper method to send sampler settings over EnvironmentParametersChannel Calls the multirangeuniform sampler type set method.</p> <p>Arguments:</p> <ul> <li><code>key</code>: environment parameter to be sampled</li> <li><code>env_channel</code>: The EnvironmentParametersChannel to communicate sampler settings to environment</li> </ul> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#completioncriteriasettings-objects","title":"CompletionCriteriaSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass CompletionCriteriaSettings()\n</code></pre> <p>CompletionCriteriaSettings contains the information needed to figure out if the next lesson must start.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#need_increment","title":"need_increment","text":"<pre><code> | need_increment(progress: float, reward_buffer: List[float], smoothing: float) -&gt; Tuple[bool, float]\n</code></pre> <p>Given measures, this method returns a boolean indicating if the lesson needs to change now, and a float corresponding to the new smoothed value.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#lesson-objects","title":"Lesson Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass Lesson()\n</code></pre> <p>Gathers the data of one lesson for one environment parameter including its name, the condition that must be fulfilled for the lesson to be completed and a sampler for the environment parameter. If the completion_criteria is None, then this is the last lesson in the curriculum.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#environmentparametersettings-objects","title":"EnvironmentParameterSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass EnvironmentParameterSettings()\n</code></pre> <p>EnvironmentParameterSettings is an ordered list of lessons for one environment parameter.</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#structure_2","title":"structure","text":"<pre><code> | @staticmethod\n | structure(d: Mapping, t: type) -&gt; Dict[str, \"EnvironmentParameterSettings\"]\n</code></pre> <p>Helper method to structure a Dict of EnvironmentParameterSettings class. Meant to be registered with cattr.register_structure_hook() and called with cattr.structure().</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#trainersettings-objects","title":"TrainerSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass TrainerSettings(ExportableSettings)\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#structure_3","title":"structure","text":"<pre><code> | @staticmethod\n | structure(d: Mapping, t: type) -&gt; Any\n</code></pre> <p>Helper method to structure a TrainerSettings class. Meant to be registered with cattr.register_structure_hook() and called with cattr.structure().</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#checkpointsettings-objects","title":"CheckpointSettings Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass CheckpointSettings()\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#prioritize_resume_init","title":"prioritize_resume_init","text":"<pre><code> | prioritize_resume_init() -&gt; None\n</code></pre> <p>Prioritize explicit command line resume/init over conflicting yaml options. if both resume/init are set at one place use resume</p> <p></p>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#runoptions-objects","title":"RunOptions Objects","text":"<pre><code>@attr.s(auto_attribs=True)\nclass RunOptions(ExportableSettings)\n</code></pre>"},{"location":"Python-On-Off-Policy-Trainer-Documentation/#from_argparse","title":"from_argparse","text":"<pre><code> | @staticmethod\n | from_argparse(args: argparse.Namespace) -&gt; \"RunOptions\"\n</code></pre> <p>Takes an argparse.Namespace as specified in <code>parse_command_line</code>, loads input configuration files from file paths, and converts to a RunOptions instance.</p> <p>Arguments:</p> <ul> <li><code>args</code>: collection of command-line parameters passed to mlagents-learn</li> </ul> <p>Returns:</p> <p>RunOptions representing the passed in arguments, with trainer config, curriculum and sampler configs loaded from files.</p>"},{"location":"Python-Optimizer-Documentation/","title":"Table of Contents","text":"<ul> <li>mlagents.trainers.optimizer.torch_optimizer</li> <li>TorchOptimizer<ul> <li>create_reward_signals</li> <li>get_trajectory_value_estimates</li> </ul> </li> <li>mlagents.trainers.optimizer.optimizer</li> <li>Optimizer<ul> <li>update</li> </ul> </li> </ul>"},{"location":"Python-Optimizer-Documentation/#mlagentstrainersoptimizertorch_optimizer","title":"mlagents.trainers.optimizer.torch_optimizer","text":""},{"location":"Python-Optimizer-Documentation/#torchoptimizer-objects","title":"TorchOptimizer Objects","text":"<pre><code>class TorchOptimizer(Optimizer)\n</code></pre>"},{"location":"Python-Optimizer-Documentation/#create_reward_signals","title":"create_reward_signals","text":"<pre><code> | create_reward_signals(reward_signal_configs: Dict[RewardSignalType, RewardSignalSettings]) -&gt; None\n</code></pre> <p>Create reward signals</p> <p>Arguments:</p> <ul> <li><code>reward_signal_configs</code>: Reward signal config.</li> </ul> <p></p>"},{"location":"Python-Optimizer-Documentation/#get_trajectory_value_estimates","title":"get_trajectory_value_estimates","text":"<pre><code> | get_trajectory_value_estimates(batch: AgentBuffer, next_obs: List[np.ndarray], done: bool, agent_id: str = \"\") -&gt; Tuple[Dict[str, np.ndarray], Dict[str, float], Optional[AgentBufferField]]\n</code></pre> <p>Get value estimates and memories for a trajectory, in batch form.</p> <p>Arguments:</p> <ul> <li><code>batch</code>: An AgentBuffer that consists of a trajectory.</li> <li><code>next_obs</code>: the next observation (after the trajectory). Used for bootstrapping     if this is not a terminal trajectory.</li> <li><code>done</code>: Set true if this is a terminal trajectory.</li> <li><code>agent_id</code>: Agent ID of the agent that this trajectory belongs to.</li> </ul> <p>Returns:</p> <p>A Tuple of the Value Estimates as a Dict of [name, np.ndarray(trajectory_len)],     the final value estimate as a Dict of [name, float], and optionally (if using memories)     an AgentBufferField of initial critic memories to be used during update.</p> <p></p>"},{"location":"Python-Optimizer-Documentation/#mlagentstrainersoptimizeroptimizer","title":"mlagents.trainers.optimizer.optimizer","text":""},{"location":"Python-Optimizer-Documentation/#optimizer-objects","title":"Optimizer Objects","text":"<pre><code>class Optimizer(abc.ABC)\n</code></pre> <p>Creates loss functions and auxillary networks (e.g. Q or Value) needed for training. Provides methods to update the Policy.</p> <p></p>"},{"location":"Python-Optimizer-Documentation/#update","title":"update","text":"<pre><code> | @abc.abstractmethod\n | update(batch: AgentBuffer, num_sequences: int) -&gt; Dict[str, float]\n</code></pre> <p>Update the Policy based on the batch that was passed in.</p> <p>Arguments:</p> <ul> <li><code>batch</code>: AgentBuffer that contains the minibatch of data used for this update.</li> <li><code>num_sequences</code>: Number of recurrent sequences found in the minibatch.</li> </ul> <p>Returns:</p> <p>A Dict containing statistics (name, value) from the update (e.g. loss)</p>"},{"location":"Python-PettingZoo-API-Documentation/","title":"Table of Contents","text":"<ul> <li>mlagents_envs.envs.pettingzoo_env_factory</li> <li>PettingZooEnvFactory<ul> <li>env</li> </ul> </li> <li>mlagents_envs.envs.unity_aec_env</li> <li>UnityAECEnv<ul> <li>__init__</li> <li>step</li> <li>observe</li> <li>last</li> </ul> </li> <li>mlagents_envs.envs.unity_parallel_env</li> <li>UnityParallelEnv<ul> <li>__init__</li> <li>reset</li> </ul> </li> <li>mlagents_envs.envs.unity_pettingzoo_base_env</li> <li>UnityPettingzooBaseEnv<ul> <li>observation_spaces</li> <li>observation_space</li> <li>action_spaces</li> <li>action_space</li> <li>side_channel</li> <li>reset</li> <li>seed</li> <li>render</li> <li>close</li> </ul> </li> </ul>"},{"location":"Python-PettingZoo-API-Documentation/#mlagents_envsenvspettingzoo_env_factory","title":"mlagents_envs.envs.pettingzoo_env_factory","text":""},{"location":"Python-PettingZoo-API-Documentation/#pettingzooenvfactory-objects","title":"PettingZooEnvFactory Objects","text":"<pre><code>class PettingZooEnvFactory()\n</code></pre>"},{"location":"Python-PettingZoo-API-Documentation/#env","title":"env","text":"<pre><code> | env(seed: Optional[int] = None, **kwargs: Union[List, int, bool, None]) -&gt; UnityAECEnv\n</code></pre> <p>Creates the environment with env_id from unity's default_registry and wraps it in a UnityToPettingZooWrapper</p> <p>Arguments:</p> <ul> <li><code>seed</code>: The seed for the action spaces of the agents.</li> <li><code>kwargs</code>: Any argument accepted by <code>UnityEnvironment</code>class except file_name</li> </ul> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#mlagents_envsenvsunity_aec_env","title":"mlagents_envs.envs.unity_aec_env","text":""},{"location":"Python-PettingZoo-API-Documentation/#unityaecenv-objects","title":"UnityAECEnv Objects","text":"<pre><code>class UnityAECEnv(UnityPettingzooBaseEnv,  AECEnv)\n</code></pre> <p>Unity AEC (PettingZoo) environment wrapper.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#__init__","title":"__init__","text":"<pre><code> | __init__(env: BaseEnv, seed: Optional[int] = None)\n</code></pre> <p>Initializes a Unity AEC environment wrapper.</p> <p>Arguments:</p> <ul> <li><code>env</code>: The UnityEnvironment that is being wrapped.</li> <li><code>seed</code>: The seed for the action spaces of the agents.</li> </ul> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#step","title":"step","text":"<pre><code> | step(action: Any) -&gt; None\n</code></pre> <p>Sets the action of the active agent and get the observation, reward, done and info of the next agent.</p> <p>Arguments:</p> <ul> <li><code>action</code>: The action for the active agent</li> </ul> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#observe","title":"observe","text":"<pre><code> | observe(agent_id)\n</code></pre> <p>Returns the observation an agent currently can make. <code>last()</code> calls this function.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#last","title":"last","text":"<pre><code> | last(observe=True)\n</code></pre> <p>returns observation, cumulative reward, done, info for the current agent (specified by self.agent_selection)</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#mlagents_envsenvsunity_parallel_env","title":"mlagents_envs.envs.unity_parallel_env","text":""},{"location":"Python-PettingZoo-API-Documentation/#unityparallelenv-objects","title":"UnityParallelEnv Objects","text":"<pre><code>class UnityParallelEnv(UnityPettingzooBaseEnv,  ParallelEnv)\n</code></pre> <p>Unity Parallel (PettingZoo) environment wrapper.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#__init___1","title":"__init__","text":"<pre><code> | __init__(env: BaseEnv, seed: Optional[int] = None)\n</code></pre> <p>Initializes a Unity Parallel environment wrapper.</p> <p>Arguments:</p> <ul> <li><code>env</code>: The UnityEnvironment that is being wrapped.</li> <li><code>seed</code>: The seed for the action spaces of the agents.</li> </ul> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#reset","title":"reset","text":"<pre><code> | reset() -&gt; Dict[str, Any]\n</code></pre> <p>Resets the environment.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#mlagents_envsenvsunity_pettingzoo_base_env","title":"mlagents_envs.envs.unity_pettingzoo_base_env","text":""},{"location":"Python-PettingZoo-API-Documentation/#unitypettingzoobaseenv-objects","title":"UnityPettingzooBaseEnv Objects","text":"<pre><code>class UnityPettingzooBaseEnv()\n</code></pre> <p>Unity Petting Zoo base environment.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#observation_spaces","title":"observation_spaces","text":"<pre><code> | @property\n | observation_spaces() -&gt; Dict[str, spaces.Space]\n</code></pre> <p>Return the observation spaces of all the agents.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#observation_space","title":"observation_space","text":"<pre><code> | observation_space(agent: str) -&gt; Optional[spaces.Space]\n</code></pre> <p>The observation space of the current agent.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#action_spaces","title":"action_spaces","text":"<pre><code> | @property\n | action_spaces() -&gt; Dict[str, spaces.Space]\n</code></pre> <p>Return the action spaces of all the agents.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#action_space","title":"action_space","text":"<pre><code> | action_space(agent: str) -&gt; Optional[spaces.Space]\n</code></pre> <p>The action space of the current agent.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#side_channel","title":"side_channel","text":"<pre><code> | @property\n | side_channel() -&gt; Dict[str, Any]\n</code></pre> <p>The side channels of the environment. You can access the side channels of an environment with <code>env.side_channel[&lt;name-of-channel&gt;]</code>.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#reset_1","title":"reset","text":"<pre><code> | reset()\n</code></pre> <p>Resets the environment.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#seed","title":"seed","text":"<pre><code> | seed(seed=None)\n</code></pre> <p>Reseeds the environment (making the resulting environment deterministic). <code>reset()</code> must be called after <code>seed()</code>, and before <code>step()</code>.</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#render","title":"render","text":"<pre><code> | render(mode=\"human\")\n</code></pre> <p>NOT SUPPORTED.</p> <p>Displays a rendered frame from the environment, if supported. Alternate render modes in the default environments are <code>'rgb_array'</code> which returns a numpy array and is supported by all environments outside of classic, and <code>'ansi'</code> which returns the strings printed (specific to classic environments).</p> <p></p>"},{"location":"Python-PettingZoo-API-Documentation/#close","title":"close","text":"<pre><code> | close() -&gt; None\n</code></pre> <p>Close the environment.</p>"},{"location":"Python-PettingZoo-API/","title":"Unity ML-Agents PettingZoo Wrapper","text":"<p>With the increasing interest in multi-agent training with a gym-like API, we provide a PettingZoo Wrapper around the Petting Zoo API. Our wrapper provides interfaces on top of our <code>UnityEnvironment</code> class, which is the default way of interfacing with a Unity environment via Python.</p>"},{"location":"Python-PettingZoo-API/#installation-and-examples","title":"Installation and Examples","text":"<p>The PettingZoo wrapper is part of the <code>mlagents_envs</code> package. Please refer to the mlagents_envs installation instructions.</p> <p>[Colab] PettingZoo Wrapper Example</p> <p>This colab notebook demonstrates the example usage of the wrapper, including installation, basic usages, and an example with our Striker vs Goalie environment which is a multi-agents environment with multiple different behavior names.</p>"},{"location":"Python-PettingZoo-API/#api-interface","title":"API interface","text":"<p>This wrapper is compatible with PettingZoo API. Please check out PettingZoo API page for more details. Here's an example of interacting with wrapped environment:</p> <pre><code>from mlagents_envs.environment import UnityEnvironment\nfrom mlagents_envs.envs import UnityToPettingZooWrapper\n\nunity_env = UnityEnvironment(\"StrikersVsGoalie\")\nenv = UnityToPettingZooWrapper(unity_env)\nenv.reset()\nfor agent in env.agent_iter():\n    observation, reward, done, info = env.last()\n    action = policy(observation, agent)\n    env.step(action)\n</code></pre>"},{"location":"Python-PettingZoo-API/#notes","title":"Notes","text":"<ul> <li>There is support for both AEC   and Parallel PettingZoo APIs.</li> <li>The AEC wrapper is compatible with PettingZoo (PZ) API interface but works in a slightly   different way under the hood. For the AEC API, Instead of stepping the environment in every <code>env.step(action)</code>,   the PZ wrapper will store the action, and will only perform environment stepping when all the   agents requesting for actions in the current step have been assigned an action. This is for   performance, considering that the communication between Unity and python is more efficient   when data are sent in batches.</li> <li>Since the actions for the AEC wrapper are stored without applying them to the environment until   all the actions are queued, some components of the API might behave in unexpected way. For example, a call   to <code>env.reward</code> should return the instantaneous reward for that particular step, but the true   reward would only be available when an actual environment step is performed. It's recommended that   you follow the API definition for training (access rewards from <code>env.last()</code> instead of   <code>env.reward</code>) and the underlying mechanism shouldn't affect training results.</li> <li>The environments will automatically reset when it's done, so <code>env.agent_iter(max_step)</code> will   keep going on until the specified max step is reached (default: <code>2**63</code>). There is no need to   call <code>env.reset()</code> except for the very beginning of instantiating an environment.</li> </ul>"},{"location":"Readme/","title":"Unity ML-Agents Toolkit","text":"<p>(latest release) (all releases)</p> <p>The Unity Machine Learning Agents Toolkit (ML-Agents) is an open-source project that enables games and simulations to serve as environments for training intelligent agents. We provide implementations (based on PyTorch) of state-of-the-art algorithms to enable game developers and hobbyists to easily train intelligent agents for 2D, 3D and VR/AR games. Researchers can also use the provided simple-to-use Python API to train Agents using reinforcement learning, imitation learning, neuroevolution, or any other methods. These trained agents can be used for multiple purposes, including controlling NPC behavior (in a variety of settings such as multi-agent and adversarial), automated testing of game builds and evaluating different game design decisions pre-release. The ML-Agents Toolkit is mutually beneficial for both game developers and AI researchers as it provides a central platform where advances in AI can be evaluated on Unity\u2019s rich environments and then made accessible to the wider research and game developer communities.</p>"},{"location":"Readme/#features","title":"Features","text":"<ul> <li>17+ example Unity environments</li> <li>Support for multiple environment configurations and training scenarios</li> <li>Flexible Unity SDK that can be integrated into your game or custom Unity scene</li> <li>Support for training single-agent, multi-agent cooperative, and multi-agent   competitive scenarios via several Deep Reinforcement Learning algorithms (PPO, SAC, MA-POCA, self-play).</li> <li>Support for learning from demonstrations through two Imitation Learning algorithms (BC and GAIL).</li> <li>Quickly and easily add your own custom training algorithm and/or components.</li> <li>Easily definable Curriculum Learning scenarios for complex tasks</li> <li>Train robust agents using environment randomization</li> <li>Flexible agent control with On Demand Decision Making</li> <li>Train using multiple concurrent Unity environment instances</li> <li>Utilizes the Sentis to   provide native cross-platform support</li> <li>Unity environment control from Python</li> <li>Wrap Unity learning environments as a gym environment</li> <li>Wrap Unity learning environments as a PettingZoo environment</li> </ul> <p>See our ML-Agents Overview page for detailed descriptions of all these features. Or go straight to our web docs.</p>"},{"location":"Readme/#releases-documentation","title":"Releases &amp; Documentation","text":"<p>Our latest, stable release is <code>Release 21</code>. Click here to get started with the latest release of ML-Agents.</p> <p>You can also check out our new web docs!</p> <p>The table below lists all our releases, including our <code>main</code> branch which is under active development and may be unstable. A few helpful guidelines:</p> <ul> <li>The Versioning page overviews how we manage our GitHub   releases and the versioning process for each of the ML-Agents components.</li> <li>The Releases page   contains details of the changes between releases.</li> <li>The Migration page contains details on how to upgrade   from earlier releases of the ML-Agents Toolkit.</li> <li>The Documentation links in the table below include installation and usage   instructions specific to each release. Remember to always use the   documentation that corresponds to the release version you're using.</li> <li>The <code>com.unity.ml-agents</code> package is verified   for Unity 2020.1 and later. Verified packages releases are numbered 1.0.x.</li> </ul> Version Release Date Source Documentation Download Python Package Unity Package develop (unstable) -- source docs download -- -- Release 21 October 9, 2023 source docs download 1.0.0 3.0.0 <p>If you are a researcher interested in a discussion of Unity as an AI platform, see a pre-print of our reference paper on Unity and the ML-Agents Toolkit.</p> <p>If you use Unity or the ML-Agents Toolkit to conduct research, we ask that you cite the following paper as a reference:</p> <pre><code>@article{juliani2020,\n  title={Unity: A general platform for intelligent agents},\n  author={Juliani, Arthur and Berges, Vincent-Pierre and Teng, Ervin and Cohen, Andrew and Harper, Jonathan and Elion, Chris and Goy, Chris and Gao, Yuan and Henry, Hunter and Mattar, Marwan and Lange, Danny},\n  journal={arXiv preprint arXiv:1809.02627},\n  url={https://arxiv.org/pdf/1809.02627.pdf},\n  year={2020}\n}\n</code></pre> <p>Additionally, if you use the MA-POCA trainer in your research, we ask that you cite the following paper as a reference:</p> <pre><code>@article{cohen2022,\n  title={On the Use and Misuse of Absorbing States in Multi-agent Reinforcement Learning},\n  author={Cohen, Andrew and Teng, Ervin and Berges, Vincent-Pierre and Dong, Ruo-Ping and Henry, Hunter and Mattar, Marwan and Zook, Alexander and Ganguly, Sujoy},\n  journal={RL in Games Workshop AAAI 2022},\n  url={http://aaai-rlg.mlanctot.info/papers/AAAI22-RLG_paper_32.pdf},\n  year={2022}\n}\n</code></pre>"},{"location":"Readme/#additional-resources","title":"Additional Resources","text":"<p>We have a Unity Learn course, ML-Agents: Hummingbirds, that provides a gentle introduction to Unity and the ML-Agents Toolkit.</p> <p>We've also partnered with CodeMonkeyUnity to create a series of tutorial videos on how to implement and use the ML-Agents Toolkit.</p> <p>We have also published a series of blog posts that are relevant for ML-Agents:</p> <ul> <li>(July 12, 2021)   ML-Agents plays Dodgeball</li> <li>(May 5, 2021)   ML-Agents v2.0 release: Now supports training complex cooperative behaviors</li> <li>(December 28, 2020)   Happy holidays from the Unity ML-Agents team!</li> <li>(November 20, 2020)   How Eidos-Montr\u00e9al created Grid Sensors to improve observations for training agents</li> <li>(November 11, 2020)   2020 AI@Unity interns shoutout</li> <li>(May 12, 2020)   Announcing ML-Agents Unity Package v1.0!</li> <li>(February 28, 2020)   Training intelligent adversaries using self-play with ML-Agents</li> <li>(November 11, 2019)   Training your agents 7 times faster with ML-Agents</li> <li>(October 21, 2019)   The AI@Unity interns help shape the world</li> <li>(April 15, 2019)   Unity ML-Agents Toolkit v0.8: Faster training on real games</li> <li>(March 1, 2019)   Unity ML-Agents Toolkit v0.7: A leap towards cross-platform inference</li> <li>(December 17, 2018)   ML-Agents Toolkit v0.6: Improved usability of Brains and Imitation Learning</li> <li>(October 2, 2018)   Puppo, The Corgi: Cuteness Overload with the Unity ML-Agents Toolkit</li> <li>(September 11, 2018)   ML-Agents Toolkit v0.5, new resources for AI researchers available now</li> <li>(June 26, 2018)   Solving sparse-reward tasks with Curiosity</li> <li>(June 19, 2018)   Unity ML-Agents Toolkit v0.4 and Udacity Deep Reinforcement Learning Nanodegree</li> <li>(May 24, 2018)   Imitation Learning in Unity: The Workflow</li> <li>(March 15, 2018)   ML-Agents Toolkit v0.3 Beta released: Imitation Learning, feedback-driven features, and more</li> <li>(December 11, 2017)   Using Machine Learning Agents in a real game: a beginner\u2019s guide</li> <li>(December 8, 2017)   Introducing ML-Agents Toolkit v0.2: Curriculum Learning, new environments, and more</li> <li>(September 19, 2017)   Introducing: Unity Machine Learning Agents Toolkit</li> <li>Overviewing reinforcement learning concepts   (multi-armed bandit   and   Q-learning)</li> </ul>"},{"location":"Readme/#more-from-unity","title":"More from Unity","text":"<ul> <li>Unity Sentis</li> <li>Introducing Unity Muse and Sentis</li> </ul>"},{"location":"Readme/#community-and-feedback","title":"Community and Feedback","text":"<p>The ML-Agents Toolkit is an open-source project and we encourage and welcome contributions. If you wish to contribute, be sure to review our contribution guidelines and code of conduct.</p> <p>For problems with the installation and setup of the ML-Agents Toolkit, or discussions about how to best setup or train your agents, please create a new thread on the Unity ML-Agents forum and make sure to include as much detail as possible. If you run into any other problems using the ML-Agents Toolkit or have a specific feature request, please submit a GitHub issue.</p> <p>Please tell us which samples you would like to see shipped with the ML-Agents Unity package by replying to this forum thread.</p> <p>Your opinion matters a great deal to us. Only by hearing your thoughts on the Unity ML-Agents Toolkit can we continue to improve and grow. Please take a few minutes to let us know about it.</p> <p>For any other questions or feedback, connect directly with the ML-Agents team at ml-agents@unity3d.com.</p>"},{"location":"Readme/#privacy","title":"Privacy","text":"<p>In order to improve the developer experience for Unity ML-Agents Toolkit, we have added in-editor analytics. Please refer to \"Information that is passively collected by Unity\" in the Unity Privacy Policy.</p>"},{"location":"Sentis/","title":"Sentis","text":"<p>The ML-Agents Toolkit allows you to use pre-trained neural network models inside your Unity games. This support is possible thanks to the Sentis (codenamed Sentis). Sentis uses compute shaders to run the neural network within Unity.</p>"},{"location":"Sentis/#supported-devices","title":"Supported devices","text":"<p>See the Sentis documentation for a list of the supported platforms.</p> <p>Scripting Backends : Sentis is generally faster with IL2CPP than with Mono for Standalone builds. In the Editor, It is not possible to use Sentis with GPU device selected when Editor Graphics Emulation is set to OpenGL(ES) 3.0 or 2.0 emulation. Also there might be non-fatal build time errors when target platform includes Graphics API that does not support Unity Compute Shaders.</p>"},{"location":"Sentis/#using-sentis","title":"Using Sentis","text":"<p>When using a model, drag the model file into the Model field in the Inspector of the Agent. Select the Inference Device : CPU or GPU you want to use for Inference.</p> <p>Note: For most of the models generated with the ML-Agents Toolkit, CPU will be faster than GPU. You should use the GPU only if you use the ResNet visual encoder or have a large number of agents with visual observations.</p>"},{"location":"Sentis/#unsupported-use-cases","title":"Unsupported use cases","text":""},{"location":"Sentis/#externally-trained-models","title":"Externally trained models","text":"<p>The ML-Agents Toolkit only supports the models created with our trainers. Model loading expects certain conventions for constants and tensor names. While it is possible to construct a model that follows these conventions, we don't provide any additional help for this. More details can be found in TensorNames.cs and SentisModelParamLoader.cs.</p> <p>If you wish to run inference on an externally trained model, you should use Sentis directly, instead of trying to run it through ML-Agents.</p>"},{"location":"Sentis/#model-inference-outside-of-unity","title":"Model inference outside of Unity","text":"<p>We do not provide support for inference anywhere outside of Unity. The <code>.onnx</code> files produced by training use the open format ONNX; if you wish to convert a <code>.onnx</code> file to another format or run inference with them, refer to their documentation.</p>"},{"location":"Training-Configuration-File/","title":"Training Configuration File","text":"<p>Table of Contents</p> <ul> <li>Common Trainer Configurations</li> <li>Trainer-specific Configurations</li> <li>PPO-specific Configurations</li> <li>SAC-specific Configurations</li> <li>Reward Signals</li> <li>Extrinsic Rewards</li> <li>Curiosity Intrinsic Reward</li> <li>GAIL Intrinsic Reward</li> <li>RND Intrinsic Reward</li> <li>Reward Signal Settings for SAC</li> <li>Behavioral Cloning</li> <li>Memory-enhanced Agents using Recurrent Neural Networks</li> <li>Self-Play</li> <li>Note on Reward Signals</li> <li>Note on Swap Steps</li> </ul>"},{"location":"Training-Configuration-File/#common-trainer-configurations","title":"Common Trainer Configurations","text":"<p>One of the first decisions you need to make regarding your training run is which trainer to use: PPO, SAC, or POCA. There are some training configurations that are common to both trainers (which we review now) and others that depend on the choice of the trainer (which we review on subsequent sections).</p> Setting Description <code>trainer_type</code> (default = <code>ppo</code>) The type of trainer to use: <code>ppo</code>,  <code>sac</code>, or <code>poca</code>. <code>summary_freq</code> (default = <code>50000</code>) Number of experiences that needs to be collected before generating and displaying training statistics. This determines the granularity of the graphs in Tensorboard. <code>time_horizon</code> (default = <code>64</code>) How many steps of experience to collect per-agent before adding it to the experience buffer. When this limit is reached before the end of an episode, a value estimate is used to predict the overall expected reward from the agent's current state. As such, this parameter trades off between a less biased, but higher variance estimate (long time horizon) and more biased, but less varied estimate (short time horizon). In cases where there are frequent rewards within an episode, or episodes are prohibitively large, a smaller number can be more ideal. This number should be large enough to capture all the important behavior within a sequence of an agent's actions.  Typical range: <code>32</code> - <code>2048</code> <code>max_steps</code> (default = <code>500000</code>) Total number of steps (i.e., observation collected and action taken) that must be taken in the environment (or across all environments if using multiple in parallel) before ending the training process. If you have multiple agents with the same behavior name within your environment, all steps taken by those agents will contribute to the same <code>max_steps</code> count. Typical range: <code>5e5</code> - <code>1e7</code> <code>keep_checkpoints</code> (default = <code>5</code>) The maximum number of model checkpoints to keep. Checkpoints are saved after the number of steps specified by the checkpoint_interval option. Once the maximum number of checkpoints has been reached, the oldest checkpoint is deleted when saving a new checkpoint. <code>even_checkpoints</code> (default = <code>false</code>) If set to true, ignores <code>checkpoint_interval</code> and evenly distributes checkpoints throughout training based on <code>keep_checkpoints</code>and <code>max_steps</code>, i.e. <code>checkpoint_interval = max_steps / keep_checkpoints</code>. Useful for cataloging agent behavior throughout training. <code>checkpoint_interval</code> (default = <code>500000</code>) The number of experiences collected between each checkpoint by the trainer. A maximum of <code>keep_checkpoints</code> checkpoints are saved before old ones are deleted. Each checkpoint saves the <code>.onnx</code> files in <code>results/</code> folder. <code>init_path</code> (default = None) Initialize trainer from a previously saved model. Note that the prior run should have used the same trainer configurations as the current run, and have been saved with the same version of ML-Agents. You can provide either the file name or the full path to the checkpoint, e.g. <code>{checkpoint_name.pt}</code> or <code>./models/{run-id}/{behavior_name}/{checkpoint_name.pt}</code>. This option is provided in case you want to initialize different behaviors from different runs or initialize from an older checkpoint; in most cases, it is sufficient to use the <code>--initialize-from</code> CLI parameter to initialize all models from the same run. <code>threaded</code> (default = <code>false</code>) Allow environments to step while updating the model. This might result in a training speedup, especially when using SAC. For best performance, leave setting to <code>false</code> when using self-play. <code>hyperparameters -&gt; learning_rate</code> (default = <code>3e-4</code>) Initial learning rate for gradient descent. Corresponds to the strength of each gradient descent update step. This should typically be decreased if training is unstable, and the reward does not consistently increase. Typical range: <code>1e-5</code> - <code>1e-3</code> <code>hyperparameters -&gt; batch_size</code> Number of experiences in each iteration of gradient descent. This should always be multiple times smaller than <code>buffer_size</code>. If you are using continuous actions, this value should be large (on the order of 1000s). If you are using only discrete actions, this value should be smaller (on the order of 10s).  Typical range: (Continuous - PPO): <code>512</code> - <code>5120</code>; (Continuous - SAC): <code>128</code> - <code>1024</code>; (Discrete, PPO &amp; SAC): <code>32</code> - <code>512</code>. <code>hyperparameters -&gt; buffer_size</code> (default = <code>10240</code> for PPO and <code>50000</code> for SAC) PPO: Number of experiences to collect before updating the policy model. Corresponds to how many experiences should be collected before we do any learning or updating of the model. This should be multiple times larger than <code>batch_size</code>. Typically a larger <code>buffer_size</code> corresponds to more stable training updates.  SAC: The max size of the experience buffer - on the order of thousands of times longer than your episodes, so that SAC can learn from old as well as new experiences. Typical range: PPO: <code>2048</code> - <code>409600</code>; SAC: <code>50000</code> - <code>1000000</code> <code>hyperparameters -&gt; learning_rate_schedule</code> (default = <code>linear</code> for PPO and <code>constant</code> for SAC) Determines how learning rate changes over time. For PPO, we recommend decaying learning rate until max_steps so learning converges more stably. However, for some cases (e.g. training for an unknown amount of time) this feature can be disabled. For SAC, we recommend holding learning rate constant so that the agent can continue to learn until its Q function converges naturally. <code>linear</code> decays the learning_rate linearly, reaching 0 at max_steps, while <code>constant</code> keeps the learning rate constant for the entire training run. <code>network_settings -&gt; hidden_units</code> (default = <code>128</code>) Number of units in the hidden layers of the neural network. Correspond to how many units are in each fully connected layer of the neural network. For simple problems where the correct action is a straightforward combination of the observation inputs, this should be small. For problems where the action is a very complex interaction between the observation variables, this should be larger.  Typical range: <code>32</code> - <code>512</code> <code>network_settings -&gt; num_layers</code> (default = <code>2</code>) The number of hidden layers in the neural network. Corresponds to how many hidden layers are present after the observation input, or after the CNN encoding of the visual observation. For simple problems, fewer layers are likely to train faster and more efficiently. More layers may be necessary for more complex control problems.  Typical range: <code>1</code> - <code>3</code> <code>network_settings -&gt; normalize</code> (default = <code>false</code>) Whether normalization is applied to the vector observation inputs. This normalization is based on the running average and variance of the vector observation. Normalization can be helpful in cases with complex continuous control problems, but may be harmful with simpler discrete control problems. <code>network_settings -&gt; vis_encode_type</code> (default = <code>simple</code>) Encoder type for encoding visual observations.  <code>simple</code> (default) uses a simple encoder which consists of two convolutional layers, <code>nature_cnn</code> uses the CNN implementation proposed by Mnih et al., consisting of three convolutional layers, and <code>resnet</code> uses the IMPALA Resnet consisting of three stacked layers, each with two residual blocks, making a much larger network than the other two. <code>match3</code> is a smaller CNN (Gudmundsoon et al.) that can capture more granular spatial relationships and is optimized for board games. <code>fully_connected</code> uses a single fully connected dense layer as encoder without any convolutional layers.  Due to the size of convolution kernel, there is a minimum observation size limitation that each encoder type can handle - <code>simple</code>: 20x20, <code>nature_cnn</code>: 36x36, <code>resnet</code>: 15 x 15, <code>match3</code>: 5x5.  <code>fully_connected</code> doesn't have convolutional layers and thus no size limits, but since it has less representation power it should be reserved for very small inputs. Note that using the <code>match3</code> CNN with very large visual input might result in a huge observation encoding and thus potentially slow down training or cause memory issues. <code>network_settings -&gt; conditioning_type</code> (default = <code>hyper</code>) Conditioning type for the policy using goal observations.  <code>none</code> treats the goal observations as regular observations, <code>hyper</code> (default) uses a HyperNetwork with goal observations as input to generate some of the weights of the policy. Note that when using <code>hyper</code> the number of parameters of the network increases greatly. Therefore, it is recommended to reduce the number of <code>hidden_units</code> when using this <code>conditioning_type</code>"},{"location":"Training-Configuration-File/#trainer-specific-configurations","title":"Trainer-specific Configurations","text":"<p>Depending on your choice of a trainer, there are additional trainer-specific configurations. We present them below in two separate tables, but keep in mind that you only need to include the configurations for the trainer selected (i.e. the <code>trainer</code> setting above).</p>"},{"location":"Training-Configuration-File/#ppo-specific-configurations","title":"PPO-specific Configurations","text":"Setting Description <code>hyperparameters -&gt; beta</code> (default = <code>5.0e-3</code>) Strength of the entropy regularization, which makes the policy \"more random.\" This ensures that agents properly explore the action space during training. Increasing this will ensure more random actions are taken. This should be adjusted such that the entropy (measurable from TensorBoard) slowly decreases alongside increases in reward. If entropy drops too quickly, increase beta. If entropy drops too slowly, decrease <code>beta</code>. Typical range: <code>1e-4</code> - <code>1e-2</code> <code>hyperparameters -&gt; epsilon</code> (default = <code>0.2</code>) Influences how rapidly the policy can evolve during training. Corresponds to the acceptable threshold of divergence between the old and new policies during gradient descent updating. Setting this value small will result in more stable updates, but will also slow the training process. Typical range: <code>0.1</code> - <code>0.3</code> <code>hyperparameters -&gt; beta_schedule</code> (default = <code>learning_rate_schedule</code>) Determines how beta changes over time. <code>linear</code> decays beta linearly, reaching 0 at max_steps, while <code>constant</code> keeps beta constant for the entire training run. If not explicitly set, the default beta schedule will be set to <code>hyperparameters -&gt; learning_rate_schedule</code>. <code>hyperparameters -&gt; epsilon_schedule</code> (default = <code>learning_rate_schedule</code>) Determines how epsilon changes over time (PPO only). <code>linear</code> decays epsilon linearly, reaching 0 at max_steps, while <code>constant</code> keeps the epsilon constant for the entire training run. If not explicitly set, the default epsilon schedule will be set to <code>hyperparameters -&gt; learning_rate_schedule</code>. <code>hyperparameters -&gt; lambd</code> (default = <code>0.95</code>) Regularization parameter (lambda) used when calculating the Generalized Advantage Estimate (GAE). This can be thought of as how much the agent relies on its current value estimate when calculating an updated value estimate. Low values correspond to relying more on the current value estimate (which can be high bias), and high values correspond to relying more on the actual rewards received in the environment (which can be high variance). The parameter provides a trade-off between the two, and the right value can lead to a more stable training process. Typical range: <code>0.9</code> - <code>0.95</code> <code>hyperparameters -&gt; num_epoch</code> (default = <code>3</code>) Number of passes to make through the experience buffer when performing gradient descent optimization.The larger the batch_size, the larger it is acceptable to make this. Decreasing this will ensure more stable updates, at the cost of slower learning. Typical range: <code>3</code> - <code>10</code> <code>hyperparameters -&gt; shared_critic</code> (default = <code>False</code>) Whether or not the policy and value function networks share a backbone. It may be useful to use a shared backbone when learning from image observations."},{"location":"Training-Configuration-File/#sac-specific-configurations","title":"SAC-specific Configurations","text":"Setting Description <code>hyperparameters -&gt; buffer_init_steps</code> (default = <code>0</code>) Number of experiences to collect into the buffer before updating the policy model. As the untrained policy is fairly random, pre-filling the buffer with random actions is useful for exploration. Typically, at least several episodes of experiences should be pre-filled. Typical range: <code>1000</code> - <code>10000</code> <code>hyperparameters -&gt; init_entcoef</code> (default = <code>1.0</code>) How much the agent should explore in the beginning of training. Corresponds to the initial entropy coefficient set at the beginning of training. In SAC, the agent is incentivized to make its actions entropic to facilitate better exploration. The entropy coefficient weighs the true reward with a bonus entropy reward. The entropy coefficient is automatically adjusted to a preset target entropy, so the <code>init_entcoef</code> only corresponds to the starting value of the entropy bonus. Increase init_entcoef to explore more in the beginning, decrease to converge to a solution faster. Typical range: (Continuous): <code>0.5</code> - <code>1.0</code>; (Discrete): <code>0.05</code> - <code>0.5</code> <code>hyperparameters -&gt; save_replay_buffer</code> (default = <code>false</code>) Whether to save and load the experience replay buffer as well as the model when quitting and re-starting training. This may help resumes go more smoothly, as the experiences collected won't be wiped. Note that replay buffers can be very large, and will take up a considerable amount of disk space. For that reason, we disable this feature by default. <code>hyperparameters -&gt; tau</code> (default = <code>0.005</code>) How aggressively to update the target network used for bootstrapping value estimation in SAC. Corresponds to the magnitude of the target Q update during the SAC model update. In SAC, there are two neural networks: the target and the policy. The target network is used to bootstrap the policy's estimate of the future rewards at a given state, and is fixed while the policy is being updated. This target is then slowly updated according to tau. Typically, this value should be left at 0.005. For simple problems, increasing tau to 0.01 might reduce the time it takes to learn, at the cost of stability. Typical range: <code>0.005</code> - <code>0.01</code> <code>hyperparameters -&gt; steps_per_update</code> (default = <code>1</code>) Average ratio of agent steps (actions) taken to updates made of the agent's policy. In SAC, a single \"update\" corresponds to grabbing a batch of size <code>batch_size</code> from the experience replay buffer, and using this mini batch to update the models. Note that it is not guaranteed that after exactly <code>steps_per_update</code> steps an update will be made, only that the ratio will hold true over many steps. Typically, <code>steps_per_update</code> should be greater than or equal to 1. Note that setting <code>steps_per_update</code> lower will improve sample efficiency (reduce the number of steps required to train) but increase the CPU time spent performing updates. For most environments where steps are fairly fast (e.g. our example environments) <code>steps_per_update</code> equal to the number of agents in the scene is a good balance. For slow environments (steps take 0.1 seconds or more) reducing <code>steps_per_update</code> may improve training speed. We can also change <code>steps_per_update</code> to lower than 1 to update more often than once per step, though this will usually result in a slowdown unless the environment is very slow. Typical range: <code>1</code> - <code>20</code> <code>hyperparameters -&gt; reward_signal_num_update</code> (default = <code>steps_per_update</code>) Number of steps per mini batch sampled and used for updating the reward signals. By default, we update the reward signals once every time the main policy is updated. However, to imitate the training procedure in certain imitation learning papers (e.g. Kostrikov et. al, Blond\u00e9 et. al), we may want to update the reward signal (GAIL) M times for every update of the policy. We can change <code>steps_per_update</code> of SAC to N, as well as <code>reward_signal_steps_per_update</code> under <code>reward_signals</code> to N / M to accomplish this. By default, <code>reward_signal_steps_per_update</code> is set to <code>steps_per_update</code>."},{"location":"Training-Configuration-File/#ma-poca-specific-configurations","title":"MA-POCA-specific Configurations","text":"<p>MA-POCA uses the same configurations as PPO, and there are no additional POCA-specific parameters.</p> <p>NOTE: Reward signals other than Extrinsic Rewards have not been extensively tested with MA-POCA, though they can still be added and used for training on a your-mileage-may-vary basis.</p>"},{"location":"Training-Configuration-File/#reward-signals","title":"Reward Signals","text":"<p>The <code>reward_signals</code> section enables the specification of settings for both extrinsic (i.e. environment-based) and intrinsic reward signals (e.g. curiosity and GAIL). Each reward signal should define at least two parameters, <code>strength</code> and <code>gamma</code>, in addition to any class-specific hyperparameters. Note that to remove a reward signal, you should delete its entry entirely from <code>reward_signals</code>. At least one reward signal should be left defined at all times. Provide the following configurations to design the reward signal for your training run.</p>"},{"location":"Training-Configuration-File/#extrinsic-rewards","title":"Extrinsic Rewards","text":"<p>Enable these settings to ensure that your training run incorporates your environment-based reward signal:</p> Setting Description <code>extrinsic -&gt; strength</code> (default = <code>1.0</code>) Factor by which to multiply the reward given by the environment. Typical ranges will vary depending on the reward signal. Typical range: <code>1.00</code> <code>extrinsic -&gt; gamma</code> (default = <code>0.99</code>) Discount factor for future rewards coming from the environment. This can be thought of as how far into the future the agent should care about possible rewards. In situations when the agent should be acting in the present in order to prepare for rewards in the distant future, this value should be large. In cases when rewards are more immediate, it can be smaller. Must be strictly smaller than 1. Typical range: <code>0.8</code> - <code>0.995</code>"},{"location":"Training-Configuration-File/#curiosity-intrinsic-reward","title":"Curiosity Intrinsic Reward","text":"<p>To enable curiosity, provide these settings:</p> Setting Description <code>curiosity -&gt; strength</code> (default = <code>1.0</code>) Magnitude of the curiosity reward generated by the intrinsic curiosity module. This should be scaled in order to ensure it is large enough to not be overwhelmed by extrinsic reward signals in the environment. Likewise it should not be too large to overwhelm the extrinsic reward signal. Typical range: <code>0.001</code> - <code>0.1</code> <code>curiosity -&gt; gamma</code> (default = <code>0.99</code>) Discount factor for future rewards. Typical range: <code>0.8</code> - <code>0.995</code> <code>curiosity -&gt; network_settings</code> Please see the documentation for <code>network_settings</code> under Common Trainer Configurations. The network specs used by the intrinsic curiosity model. The value should of <code>hidden_units</code> should be small enough to encourage the ICM to compress the original observation, but also not too small to prevent it from learning to differentiate between expected and actual observations. Typical range: <code>64</code> - <code>256</code> <code>curiosity -&gt; learning_rate</code> (default = <code>3e-4</code>) Learning rate used to update the intrinsic curiosity module. This should typically be decreased if training is unstable, and the curiosity loss is unstable. Typical range: <code>1e-5</code> - <code>1e-3</code>"},{"location":"Training-Configuration-File/#gail-intrinsic-reward","title":"GAIL Intrinsic Reward","text":"<p>To enable GAIL (assuming you have recorded demonstrations), provide these settings:</p> Setting Description <code>gail -&gt; strength</code> (default = <code>1.0</code>) Factor by which to multiply the raw reward. Note that when using GAIL with an Extrinsic Signal, this value should be set lower if your demonstrations are suboptimal (e.g. from a human), so that a trained agent will focus on receiving extrinsic rewards instead of exactly copying the demonstrations. Keep the strength below about 0.1 in those cases. Typical range: <code>0.01</code> - <code>1.0</code> <code>gail -&gt; gamma</code> (default = <code>0.99</code>) Discount factor for future rewards. Typical range: <code>0.8</code> - <code>0.9</code> <code>gail -&gt; demo_path</code> (Required, no default) The path to your .demo file or directory of .demo files. <code>gail -&gt; network_settings</code> Please see the documentation for <code>network_settings</code> under Common Trainer Configurations. The network specs for the GAIL discriminator. The value of <code>hidden_units</code> should be small enough to encourage the discriminator to compress the original observation, but also not too small to prevent it from learning to differentiate between demonstrated and actual behavior. Dramatically increasing this size will also negatively affect training times. Typical range: <code>64</code> - <code>256</code> <code>gail -&gt; learning_rate</code> (Optional, default = <code>3e-4</code>) Learning rate used to update the discriminator. This should typically be decreased if training is unstable, and the GAIL loss is unstable. Typical range: <code>1e-5</code> - <code>1e-3</code> <code>gail -&gt; use_actions</code> (default = <code>false</code>) Determines whether the discriminator should discriminate based on both observations and actions, or just observations. Set to True if you want the agent to mimic the actions from the demonstrations, and False if you'd rather have the agent visit the same states as in the demonstrations but with possibly different actions. Setting to False is more likely to be stable, especially with imperfect demonstrations, but may learn slower. <code>gail -&gt; use_vail</code> (default = <code>false</code>) Enables a variational bottleneck within the GAIL discriminator. This forces the discriminator to learn a more general representation and reduces its tendency to be \"too good\" at discriminating, making learning more stable. However, it does increase training time. Enable this if you notice your imitation learning is unstable, or unable to learn the task at hand."},{"location":"Training-Configuration-File/#rnd-intrinsic-reward","title":"RND Intrinsic Reward","text":"<p>Random Network Distillation (RND) is only available for the PyTorch trainers. To enable RND, provide these settings:</p> Setting Description <code>rnd -&gt; strength</code> (default = <code>1.0</code>) Magnitude of the curiosity reward generated by the intrinsic rnd module. This should be scaled in order to ensure it is large enough to not be overwhelmed by extrinsic reward signals in the environment. Likewise it should not be too large to overwhelm the extrinsic reward signal. Typical range: <code>0.001</code> - <code>0.01</code> <code>rnd -&gt; gamma</code> (default = <code>0.99</code>) Discount factor for future rewards. Typical range: <code>0.8</code> - <code>0.995</code> <code>rnd -&gt; network_settings</code> Please see the documentation for <code>network_settings</code> under Common Trainer Configurations. The network specs for the RND model. <code>curiosity -&gt; learning_rate</code> (default = <code>3e-4</code>) Learning rate used to update the RND module. This should be large enough for the RND module to quickly learn the state representation, but small enough to allow for stable learning. Typical range: <code>1e-5</code> - <code>1e-3</code>"},{"location":"Training-Configuration-File/#behavioral-cloning","title":"Behavioral Cloning","text":"<p>To enable Behavioral Cloning as a pre-training option (assuming you have recorded demonstrations), provide the following configurations under the <code>behavioral_cloning</code> section:</p> Setting Description <code>demo_path</code> (Required, no default) The path to your .demo file or directory of .demo files. <code>strength</code> (default = <code>1.0</code>) Learning rate of the imitation relative to the learning rate of PPO, and roughly corresponds to how strongly we allow BC to influence the policy. Typical range: <code>0.1</code> - <code>0.5</code> <code>steps</code> (default = <code>0</code>) During BC, it is often desirable to stop using demonstrations after the agent has \"seen\" rewards, and allow it to optimize past the available demonstrations and/or generalize outside of the provided demonstrations. steps corresponds to the training steps over which BC is active. The learning rate of BC will anneal over the steps. Set the steps to 0 for constant imitation over the entire training run. <code>batch_size</code> (default = <code>batch_size</code> of trainer) Number of demonstration experiences used for one iteration of a gradient descent update. If not specified, it will default to the <code>batch_size</code> of the trainer. Typical range: (Continuous): <code>512</code> - <code>5120</code>; (Discrete): <code>32</code> - <code>512</code> <code>num_epoch</code> (default = <code>num_epoch</code> of trainer) Number of passes through the experience buffer during gradient descent. If not specified, it will default to the number of epochs set for PPO. Typical range: <code>3</code> - <code>10</code> <code>samples_per_update</code> (default = <code>0</code>) Maximum number of samples to use during each imitation update. You may want to lower this if your demonstration dataset is very large to avoid overfitting the policy on demonstrations. Set to 0 to train over all of the demonstrations at each update step. Typical range: <code>buffer_size</code>"},{"location":"Training-Configuration-File/#memory-enhanced-agents-using-recurrent-neural-networks","title":"Memory-enhanced Agents using Recurrent Neural Networks","text":"<p>You can enable your agents to use memory by adding a <code>memory</code> section under <code>network_settings</code>, and setting <code>memory_size</code> and <code>sequence_length</code>:</p> Setting Description <code>network_settings -&gt; memory -&gt; memory_size</code> (default = <code>128</code>) Size of the memory an agent must keep. In order to use a LSTM, training requires a sequence of experiences instead of single experiences. Corresponds to the size of the array of floating point numbers used to store the hidden state of the recurrent neural network of the policy. This value must be a multiple of 2, and should scale with the amount of information you expect the agent will need to remember in order to successfully complete the task. Typical range: <code>32</code> - <code>256</code> <code>network_settings -&gt; memory -&gt; sequence_length</code> (default = <code>64</code>) Defines how long the sequences of experiences must be while training. Note that if this number is too small, the agent will not be able to remember things over longer periods of time. If this number is too large, the neural network will take longer to train. Typical range: <code>4</code> - <code>128</code> <p>A few considerations when deciding to use memory:</p> <ul> <li>LSTM does not work well with continuous actions. Please use   discrete actions for better results.</li> <li>Adding a recurrent layer increases the complexity of the neural network, it is   recommended to decrease <code>num_layers</code> when using recurrent.</li> <li>It is required that <code>memory_size</code> be divisible by 2.</li> </ul>"},{"location":"Training-Configuration-File/#self-play","title":"Self-Play","text":"<p>Training with self-play adds additional confounding factors to the usual issues faced by reinforcement learning. In general, the tradeoff is between the skill level and generality of the final policy and the stability of learning. Training against a set of slowly or unchanging adversaries with low diversity results in a more stable learning process than training against a set of quickly changing adversaries with high diversity. With this context, this guide discusses the exposed self-play hyperparameters and intuitions for tuning them.</p> <p>If your environment contains multiple agents that are divided into teams, you can leverage our self-play training option by providing these configurations for each Behavior:</p> Setting Description <code>save_steps</code> (default = <code>20000</code>) Number of trainer steps between snapshots. For example, if <code>save_steps=10000</code> then a snapshot of the current policy will be saved every <code>10000</code> trainer steps. Note, trainer steps are counted per agent. For more information, please see the migration doc after v0.13. A larger value of <code>save_steps</code> will yield a set of opponents that cover a wider range of skill levels and possibly play styles since the policy receives more training. As a result, the agent trains against a wider variety of opponents. Learning a policy to defeat more diverse opponents is a harder problem and so may require more overall training steps but also may lead to more general and robust policy at the end of training. This value is also dependent on how intrinsically difficult the environment is for the agent.  Typical range: <code>10000</code> - <code>100000</code> <code>team_change</code> (default = <code>5 * save_steps</code>) Number of trainer_steps between switching the learning team. This is the number of trainer steps the teams associated with a specific ghost trainer will train before a different team becomes the new learning team. It is possible that, in asymmetric games, opposing teams require fewer trainer steps to make similar performance gains. This enables users to train a more complicated team of agents for more trainer steps than a simpler team of agents per team switch. A larger value of <code>team-change</code> will allow the agent to train longer against it's opponents. The longer an agent trains against the same set of opponents the more able it will be to defeat them. However, training against them for too long may result in overfitting to the particular opponent strategies and so the agent may fail against the next batch of opponents.  The value of <code>team-change</code> will determine how many snapshots of the agent's policy are saved to be used as opponents for the other team. So, we recommend setting this value as a function of the <code>save_steps</code> parameter discussed previously.  Typical range: 4x-10x where x=<code>save_steps</code> <code>swap_steps</code> (default = <code>10000</code>) Number of ghost steps (not trainer steps) between swapping the opponents policy with a different snapshot. A 'ghost step' refers to a step taken by an agent that is following a fixed policy and not learning. The reason for this distinction is that in asymmetric games, we may have teams with an unequal number of agents e.g. a 2v1 scenario like our Strikers Vs Goalie example environment. The team with two agents collects twice as many agent steps per environment step as the team with one agent. Thus, these two values will need to be distinct to ensure that the same number of trainer steps corresponds to the same number of opponent swaps for each team. The formula for <code>swap_steps</code> if a user desires <code>x</code> swaps of a team with <code>num_agents</code> agents against an opponent team with <code>num_opponent_agents</code> agents during <code>team-change</code> total steps is: <code>(num_agents / num_opponent_agents) * (team_change / x)</code>  Typical range: <code>10000</code> - <code>100000</code> <code>play_against_latest_model_ratio</code> (default = <code>0.5</code>) Probability an agent will play against the latest opponent policy. With probability 1 - <code>play_against_latest_model_ratio</code>, the agent will play against a snapshot of its opponent from a past iteration.  A larger value of <code>play_against_latest_model_ratio</code> indicates that an agent will be playing against the current opponent more often. Since the agent is updating it's policy, the opponent will be different from iteration to iteration. This can lead to an unstable learning environment, but poses the agent with an auto-curricula of more increasingly challenging situations which may lead to a stronger final policy.  Typical range: <code>0.0</code> - <code>1.0</code> <code>window</code> (default = <code>10</code>) Size of the sliding window of past snapshots from which the agent's opponents are sampled. For example, a <code>window</code> size of 5 will save the last 5 snapshots taken. Each time a new snapshot is taken, the oldest is discarded. A larger value of <code>window</code> means that an agent's pool of opponents will contain a larger diversity of behaviors since it will contain policies from earlier in the training run. Like in the <code>save_steps</code> hyperparameter, the agent trains against a wider variety of opponents. Learning a policy to defeat more diverse opponents is a harder problem and so may require more overall training steps but also may lead to more general and robust policy at the end of training.  Typical range: <code>5</code> - <code>30</code>"},{"location":"Training-Configuration-File/#note-on-reward-signals","title":"Note on Reward Signals","text":"<p>We make the assumption that the final reward in a trajectory corresponds to the outcome of an episode. A final reward of +1 indicates winning, -1 indicates losing and 0 indicates a draw. The ELO calculation (discussed below) depends on this final reward being either +1, 0, -1.</p> <p>The reward signal should still be used as described in the documentation for the other trainers. However, we encourage users to be a bit more conservative when shaping reward functions due to the instability and non-stationarity of learning in adversarial games. Specifically, we encourage users to begin with the simplest possible reward function (+1 winning, -1 losing) and to allow for more iterations of training to compensate for the sparsity of reward.</p>"},{"location":"Training-Configuration-File/#note-on-swap-steps","title":"Note on Swap Steps","text":"<p>As an example, in a 2v1 scenario, if we want the swap to occur x=4 times during team-change=200000 steps, the swap_steps for the team of one agent is:</p> <p>swap_steps = (1 / 2) * (200000 / 4) = 25000 The swap_steps for the team of two agents is:</p> <p>swap_steps = (2 / 1) * (200000 / 4) = 100000 Note, with equal team sizes, the first term is equal to 1 and swap_steps can be calculated by just dividing the total steps by the desired number of swaps.</p> <p>A larger value of swap_steps means that an agent will play against the same fixed opponent for a longer number of training iterations. This results in a more stable training scenario, but leaves the agent open to the risk of overfitting it's behavior for this particular opponent. Thus, when a new opponent is swapped, the agent may lose more often than expected.</p>"},{"location":"Training-ML-Agents/","title":"Training ML-Agents","text":"<p>Table of Contents</p> <ul> <li>Training ML-Agents</li> <li>Training with mlagents-learn<ul> <li>Starting Training</li> <li>Observing Training</li> <li>Stopping and Resuming Training</li> <li>Loading an Existing Model</li> </ul> </li> <li>Training Configurations<ul> <li>Adding CLI Arguments to the Training Configuration file</li> <li>Environment settings</li> <li>Engine settings</li> <li>Checkpoint settings</li> <li>Torch settings:</li> <li>Behavior Configurations</li> <li>Default Behavior Settings</li> <li>Environment Parameters</li> <li>Environment Parameter Randomization<ul> <li>Supported Sampler Types</li> <li>Training with Environment Parameter Randomization</li> </ul> </li> <li>Curriculum<ul> <li>Training with a Curriculum</li> </ul> </li> <li>Training Using Concurrent Unity Instances</li> </ul> </li> </ul> <p>For a broad overview of reinforcement learning, imitation learning and all the training scenarios, methods and options within the ML-Agents Toolkit, see ML-Agents Toolkit Overview.</p> <p>Once your learning environment has been created and is ready for training, the next step is to initiate a training run. Training in the ML-Agents Toolkit is powered by a dedicated Python package, <code>mlagents</code>. This package exposes a command <code>mlagents-learn</code> that is the single entry point for all training workflows (e.g. reinforcement leaning, imitation learning, curriculum learning). Its implementation can be found at ml-agents/mlagents/trainers/learn.py.</p>"},{"location":"Training-ML-Agents/#training-with-mlagents-learn","title":"Training with mlagents-learn","text":""},{"location":"Training-ML-Agents/#starting-training","title":"Starting Training","text":"<p><code>mlagents-learn</code> is the main training utility provided by the ML-Agents Toolkit. It accepts a number of CLI options in addition to a YAML configuration file that contains all the configurations and hyperparameters to be used during training. The set of configurations and hyperparameters to include in this file depend on the agents in your environment and the specific training method you wish to utilize. Keep in mind that the hyperparameter values can have a big impact on the training performance (i.e. your agent's ability to learn a policy that solves the task). In this page, we will review all the hyperparameters for all training methods and provide guidelines and advice on their values.</p> <p>To view a description of all the CLI options accepted by <code>mlagents-learn</code>, use the <code>--help</code>:</p> <pre><code>mlagents-learn --help\n</code></pre> <p>The basic command for training is:</p> <pre><code>mlagents-learn &lt;trainer-config-file&gt; --env=&lt;env_name&gt; --run-id=&lt;run-identifier&gt;\n</code></pre> <p>where</p> <ul> <li><code>&lt;trainer-config-file&gt;</code> is the file path of the trainer configuration YAML.   This contains all the hyperparameter values. We offer a detailed guide on the   structure of this file and the meaning of the hyperparameters (and advice on   how to set them) in the dedicated   Training Configurations section below.</li> <li><code>&lt;env_name&gt;</code>(Optional) is the name (including path) of your   Unity executable containing the agents   to be trained. If <code>&lt;env_name&gt;</code> is not passed, the training will happen in the   Editor. Press the Play button in Unity when the message \"Start training   by pressing the Play button in the Unity Editor\" is displayed on the screen.</li> <li><code>&lt;run-identifier&gt;</code> is a unique name you can use to identify the results of   your training runs.</li> </ul> <p>See the Getting Started Guide for a sample execution of the <code>mlagents-learn</code> command.</p>"},{"location":"Training-ML-Agents/#observing-training","title":"Observing Training","text":"<p>Regardless of which training methods, configurations or hyperparameters you provide, the training process will always generate three artifacts, all found in the <code>results/&lt;run-identifier&gt;</code> folder:</p> <ol> <li>Summaries: these are training metrics that    are updated throughout the training process. They are helpful to monitor your    training performance and may help inform how to update your hyperparameter    values. See Using TensorBoard for more details on how    to visualize the training metrics.</li> <li>Models: these contain the model checkpoints that    are updated throughout training and the final model file (<code>.onnx</code>). This final    model file is generated once either when training completes or is    interrupted.</li> <li>Timers file (under <code>results/&lt;run-identifier&gt;/run_logs</code>): this contains aggregated    metrics on your training process, including time spent on specific code    blocks. See Profiling in Python for more information    on the timers generated.</li> </ol> <p>These artifacts are updated throughout the training process and finalized when training is completed or is interrupted.</p>"},{"location":"Training-ML-Agents/#stopping-and-resuming-training","title":"Stopping and Resuming Training","text":"<p>To interrupt training and save the current progress, hit <code>Ctrl+C</code> once and wait for the model(s) to be saved out.</p> <p>To resume a previously interrupted or completed training run, use the <code>--resume</code> flag and make sure to specify the previously used run ID.</p> <p>If you would like to re-run a previously interrupted or completed training run and re-use the same run ID (in this case, overwriting the previously generated artifacts), then use the <code>--force</code> flag.</p>"},{"location":"Training-ML-Agents/#loading-an-existing-model","title":"Loading an Existing Model","text":"<p>You can also use this mode to run inference of an already-trained model in Python by using both the <code>--resume</code> and <code>--inference</code> flags. Note that if you want to run inference in Unity, you should use the Sentis.</p> <p>Additionally, if the network architecture changes, you may still load an existing model, but ML-Agents will only load the parts of the model it can load and ignore all others. For instance, if you add a new reward signal, the existing model will load but the new reward signal will be initialized from scratch. If you have a model with a visual encoder (CNN) but change the <code>hidden_units</code>, the CNN will be loaded but the body of the network will be initialized from scratch.</p> <p>Alternatively, you might want to start a new training run but initialize it using an already-trained model. You may want to do this, for instance, if your environment changed and you want a new model, but the old behavior is still better than random. You can do this by specifying <code>--initialize-from=&lt;run-identifier&gt;</code>, where <code>&lt;run-identifier&gt;</code> is the old run ID.</p>"},{"location":"Training-ML-Agents/#training-configurations","title":"Training Configurations","text":"<p>The Unity ML-Agents Toolkit provides a wide range of training scenarios, methods and options. As such, specific training runs may require different training configurations and may generate different artifacts and TensorBoard statistics. This section offers a detailed guide into how to manage the different training set-ups withing the toolkit.</p> <p>More specifically, this section offers a detailed guide on the command-line flags for <code>mlagents-learn</code> that control the training configurations:</p> <ul> <li><code>&lt;trainer-config-file&gt;</code>: defines the training hyperparameters for each   Behavior in the scene, and the set-ups for the environment parameters   (Curriculum Learning and Environment Parameter Randomization)</li> </ul> <p>It is important to highlight that successfully training a Behavior in the ML-Agents Toolkit involves tuning the training hyperparameters and configuration. This guide contains some best practices for tuning the training process when the default parameters don't seem to be giving the level of performance you would like. We provide sample configuration files for our example environments in the config/ directory. The <code>config/ppo/3DBall.yaml</code> was used to train the 3D Balance Ball in the Getting Started guide. That configuration file uses the PPO trainer, but we also have configuration files for SAC and GAIL.</p> <p>Additionally, the set of configurations you provide depend on the training functionalities you use (see ML-Agents Toolkit Overview for a description of all the training functionalities). Each functionality you add typically has its own training configurations. For instance:</p> <ul> <li>Use PPO or SAC?</li> <li>Use Recurrent Neural Networks for adding memory to your agents?</li> <li>Use the intrinsic curiosity module?</li> <li>Ignore the environment reward signal?</li> <li>Pre-train using behavioral cloning? (Assuming you have recorded   demonstrations.)</li> <li>Include the GAIL intrinsic reward signals? (Assuming you have recorded   demonstrations.)</li> <li>Use self-play? (Assuming your environment includes multiple agents.)</li> </ul> <p>The trainer config file, <code>&lt;trainer-config-file&gt;</code>, determines the features you will use during training, and the answers to the above questions will dictate its contents. The rest of this guide breaks down the different sub-sections of the trainer config file and explains the possible settings for each. If you need a list of all the trainer configurations, please see Training Configuration File.</p> <p>NOTE: The configuration file format has been changed between 0.17.0 and 0.18.0 and between 0.18.0 and onwards. To convert an old set of configuration files (trainer config, curriculum, and sampler files) to the new format, a script has been provided. Run <code>python -m mlagents.trainers.upgrade_config -h</code> in your console to see the script's usage.</p>"},{"location":"Training-ML-Agents/#adding-cli-arguments-to-the-training-configuration-file","title":"Adding CLI Arguments to the Training Configuration file","text":"<p>Additionally, within the training configuration YAML file, you can also add the CLI arguments (such as <code>--num-envs</code>).</p> <p>Reminder that a detailed description of all the CLI arguments can be found by using the help utility:</p> <pre><code>mlagents-learn --help\n</code></pre> <p>These additional CLI arguments are grouped into environment, engine, checkpoint and torch. The available settings and example values are shown below.</p>"},{"location":"Training-ML-Agents/#environment-settings","title":"Environment settings","text":"<pre><code>env_settings:\n  env_path: FoodCollector\n  env_args: null\n  base_port: 5005\n  num_envs: 1\n  timeout_wait: 10\n  seed: -1\n  max_lifetime_restarts: 10\n  restarts_rate_limit_n: 1\n  restarts_rate_limit_period_s: 60\n</code></pre>"},{"location":"Training-ML-Agents/#engine-settings","title":"Engine settings","text":"<pre><code>engine_settings:\n  width: 84\n  height: 84\n  quality_level: 5\n  time_scale: 20\n  target_frame_rate: -1\n  capture_frame_rate: 60\n  no_graphics: false\n</code></pre>"},{"location":"Training-ML-Agents/#checkpoint-settings","title":"Checkpoint settings","text":"<pre><code>checkpoint_settings:\n  run_id: foodtorch\n  initialize_from: null\n  load_model: false\n  resume: false\n  force: true\n  train_model: false\n  inference: false\n</code></pre>"},{"location":"Training-ML-Agents/#torch-settings","title":"Torch settings:","text":"<pre><code>torch_settings:\n  device: cpu\n</code></pre>"},{"location":"Training-ML-Agents/#behavior-configurations","title":"Behavior Configurations","text":"<p>The primary section of the trainer config file is a set of configurations for each Behavior in your scene. These are defined under the sub-section <code>behaviors</code> in your trainer config file. Some of the configurations are required while others are optional. To help us get started, below is a sample file that includes all the possible settings if we're using a PPO trainer with all the possible training functionalities enabled (memory, behavioral cloning, curiosity, GAIL and self-play). You will notice that curriculum and environment parameter randomization settings are not part of the <code>behaviors</code> configuration, but in their own section called <code>environment_parameters</code>.</p> <pre><code>behaviors:\n  BehaviorPPO:\n    trainer_type: ppo\n\n    hyperparameters:\n      # Hyperparameters common to PPO and SAC\n      batch_size: 1024\n      buffer_size: 10240\n      learning_rate: 3.0e-4\n      learning_rate_schedule: linear\n\n      # PPO-specific hyperparameters\n      beta: 5.0e-3\n      beta_schedule: constant\n      epsilon: 0.2\n      epsilon_schedule: linear\n      lambd: 0.95\n      num_epoch: 3\n      shared_critic: False\n\n    # Configuration of the neural network (common to PPO/SAC)\n    network_settings:\n      vis_encode_type: simple\n      normalize: false\n      hidden_units: 128\n      num_layers: 2\n      # memory\n      memory:\n        sequence_length: 64\n        memory_size: 256\n\n    # Trainer configurations common to all trainers\n    max_steps: 5.0e5\n    time_horizon: 64\n    summary_freq: 10000\n    keep_checkpoints: 5\n    checkpoint_interval: 50000\n    threaded: false\n    init_path: null\n\n    # behavior cloning\n    behavioral_cloning:\n      demo_path: Project/Assets/ML-Agents/Examples/Pyramids/Demos/ExpertPyramid.demo\n      strength: 0.5\n      steps: 150000\n      batch_size: 512\n      num_epoch: 3\n      samples_per_update: 0\n\n    reward_signals:\n      # environment reward (default)\n      extrinsic:\n        strength: 1.0\n        gamma: 0.99\n\n      # curiosity module\n      curiosity:\n        strength: 0.02\n        gamma: 0.99\n        encoding_size: 256\n        learning_rate: 3.0e-4\n\n      # GAIL\n      gail:\n        strength: 0.01\n        gamma: 0.99\n        encoding_size: 128\n        demo_path: Project/Assets/ML-Agents/Examples/Pyramids/Demos/ExpertPyramid.demo\n        learning_rate: 3.0e-4\n        use_actions: false\n        use_vail: false\n\n    # self-play\n    self_play:\n      window: 10\n      play_against_latest_model_ratio: 0.5\n      save_steps: 50000\n      swap_steps: 2000\n      team_change: 100000\n</code></pre> <p>Here is an equivalent file if we use an SAC trainer instead. Notice that the configurations for the additional functionalities (memory, behavioral cloning, curiosity and self-play) remain unchanged.</p> <pre><code>behaviors:\n  BehaviorSAC:\n    trainer_type: sac\n\n    # Trainer configs common to PPO/SAC (excluding reward signals)\n    # same as PPO config\n\n    # SAC-specific configs (replaces the hyperparameters section above)\n    hyperparameters:\n      # Hyperparameters common to PPO and SAC\n      # Same as PPO config\n\n      # SAC-specific hyperparameters\n      # Replaces the \"PPO-specific hyperparameters\" section above\n      buffer_init_steps: 0\n      tau: 0.005\n      steps_per_update: 10.0\n      save_replay_buffer: false\n      init_entcoef: 0.5\n      reward_signal_steps_per_update: 10.0\n\n    # Configuration of the neural network (common to PPO/SAC)\n    network_settings:\n      # Same as PPO config\n\n    # Trainer configurations common to all trainers\n      # &lt;Same as PPO config&gt;\n\n    # pre-training using behavior cloning\n    behavioral_cloning:\n      # same as PPO config\n\n    reward_signals:\n      # environment reward\n      extrinsic:\n        # same as PPO config\n\n      # curiosity module\n      curiosity:\n        # same as PPO config\n\n      # GAIL\n      gail:\n        # same as PPO config\n\n    # self-play\n    self_play:\n      # same as PPO config\n</code></pre> <p>We now break apart the components of the configuration file and describe what each of these parameters mean and provide guidelines on how to set them. See Training Configuration File for a detailed description of all the configurations listed above, along with their defaults. Unless otherwise specified, omitting a configuration will revert it to its default.</p>"},{"location":"Training-ML-Agents/#default-behavior-settings","title":"Default Behavior Settings","text":"<p>In some cases, you may want to specify a set of default configurations for your Behaviors. This may be useful, for instance, if your Behavior names are generated procedurally by the environment and not known before runtime, or if you have many Behaviors with very similar settings. To specify a default configuration, insert a <code>default_settings</code> section in your YAML. This section should be formatted exactly like a configuration for a Behavior.</p> <pre><code>default_settings:\n  # &lt; Same as Behavior configuration &gt;\nbehaviors:\n  # &lt; Same as above &gt;\n</code></pre> <p>Behaviors found in the environment that aren't specified in the YAML will now use the <code>default_settings</code>, and unspecified settings in behavior configurations will default to the values in <code>default_settings</code> if specified there.</p>"},{"location":"Training-ML-Agents/#environment-parameters","title":"Environment Parameters","text":"<p>In order to control the <code>EnvironmentParameters</code> in the Unity simulation during training, you need to add a section called <code>environment_parameters</code>. For example you can set the value of an <code>EnvironmentParameter</code> called <code>my_environment_parameter</code> to <code>3.0</code> with the following code :</p> <pre><code>behaviors:\n  BehaviorY:\n    # &lt; Same as above &gt;\n\n# Add this section\nenvironment_parameters:\n  my_environment_parameter: 3.0\n</code></pre> <p>Inside the Unity simulation, you can access your Environment Parameters by doing :</p> <pre><code>Academy.Instance.EnvironmentParameters.GetWithDefault(\"my_environment_parameter\", 0.0f);\n</code></pre>"},{"location":"Training-ML-Agents/#environment-parameter-randomization","title":"Environment Parameter Randomization","text":"<p>To enable environment parameter randomization, you need to edit the <code>environment_parameters</code> section of your training configuration yaml file. Instead of providing a single float value for your environment parameter, you can specify a sampler instead. Here is an example with three environment parameters called <code>mass</code>, <code>length</code> and <code>scale</code>:</p> <pre><code>behaviors:\n  BehaviorY:\n    # &lt; Same as above &gt;\n\n# Add this section\nenvironment_parameters:\n  mass:\n    sampler_type: uniform\n    sampler_parameters:\n        min_value: 0.5\n        max_value: 10\n\n  length:\n    sampler_type: multirangeuniform\n    sampler_parameters:\n        intervals: [[7, 10], [15, 20]]\n\n  scale:\n    sampler_type: gaussian\n    sampler_parameters:\n        mean: 2\n        st_dev: .3\n</code></pre> Setting Description <code>sampler_type</code> A string identifier for the type of sampler to use for this <code>Environment Parameter</code>. <code>sampler_parameters</code> The parameters for a given <code>sampler_type</code>. Samplers of different types can have different <code>sampler_parameters</code>"},{"location":"Training-ML-Agents/#supported-sampler-types","title":"Supported Sampler Types","text":"<p>Below is a list of the <code>sampler_type</code> values supported by the toolkit.</p> <ul> <li><code>uniform</code> - Uniform sampler</li> <li>Uniformly samples a single float value from a range with a given minimum     and maximum value (inclusive).</li> <li>parameters - <code>min_value</code>, <code>max_value</code></li> <li><code>gaussian</code> - Gaussian sampler</li> <li>Samples a single float value from a normal distribution with a given mean     and standard deviation.</li> <li>parameters - <code>mean</code>, <code>st_dev</code></li> <li><code>multirange_uniform</code> - Multirange uniform sampler</li> <li>First, samples an interval from a set of intervals in proportion to relative     length of the intervals. Then, uniformly samples a single float value from the     sampled interval (inclusive). This sampler can take an arbitrary number of     intervals in a list in the following format:     [[<code>interval_1_min</code>, <code>interval_1_max</code>], [<code>interval_2_min</code>,     <code>interval_2_max</code>], ...]</li> <li>parameters - <code>intervals</code></li> </ul> <p>The implementation of the samplers can be found in the Samplers.cs file.</p>"},{"location":"Training-ML-Agents/#training-with-environment-parameter-randomization","title":"Training with Environment Parameter Randomization","text":"<p>After the sampler configuration is defined, we proceed by launching <code>mlagents-learn</code> and specify trainer configuration with  parameter randomization enabled. For example, if we wanted to train the 3D ball agent with parameter randomization, we would run</p> <pre><code>mlagents-learn config/ppo/3DBall_randomize.yaml --run-id=3D-Ball-randomize\n</code></pre> <p>We can observe progress and metrics via TensorBoard.</p>"},{"location":"Training-ML-Agents/#curriculum","title":"Curriculum","text":"<p>To enable curriculum learning, you need to add a <code>curriculum</code> sub-section to your environment parameter. Here is one example with the environment parameter <code>my_environment_parameter</code> :</p> <pre><code>behaviors:\n  BehaviorY:\n    # &lt; Same as above &gt;\n\n# Add this section\nenvironment_parameters:\n  my_environment_parameter:\n    curriculum:\n      - name: MyFirstLesson # The '-' is important as this is a list\n        completion_criteria:\n          measure: progress\n          behavior: my_behavior\n          signal_smoothing: true\n          min_lesson_length: 100\n          threshold: 0.2\n        value: 0.0\n      - name: MySecondLesson # This is the start of the second lesson\n        completion_criteria:\n          measure: progress\n          behavior: my_behavior\n          signal_smoothing: true\n          min_lesson_length: 100\n          threshold: 0.6\n          require_reset: true\n        value:\n          sampler_type: uniform\n          sampler_parameters:\n            min_value: 4.0\n            max_value: 7.0\n      - name: MyLastLesson\n        value: 8.0\n</code></pre> <p>Note that this curriculum only applies to <code>my_environment_parameter</code>. The <code>curriculum</code> section contains a list of <code>Lessons</code>. In the example, the lessons are named <code>MyFirstLesson</code>, <code>MySecondLesson</code> and <code>MyLastLesson</code>. Each <code>Lesson</code> has 3 fields :</p> <ul> <li><code>name</code> which is a user defined name for the lesson (The name of the lesson will be displayed in  the console when the lesson changes)</li> <li><code>completion_criteria</code> which determines what needs to happen in the simulation before the lesson  can be considered complete. When that condition is met, the curriculum moves on to the next  <code>Lesson</code>. Note that you do not need to specify a <code>completion_criteria</code> for the last <code>Lesson</code></li> <li><code>value</code> which is the value the environment parameter will take during the lesson. Note that this  can be a float or a sampler.</li> </ul> <p>There are the different settings of the <code>completion_criteria</code> :</p> Setting Description <code>measure</code> What to measure learning progress, and advancement in lessons by. <code>reward</code> uses a measure of received reward, <code>progress</code> uses the ratio of steps/max_steps, while <code>Elo</code> is available only for self-play situations and uses Elo score as a curriculum completion measure. <code>behavior</code> Specifies which behavior is being tracked. There can be multiple behaviors with different names, each at different points of training. This setting allows the curriculum to track only one of them. <code>threshold</code> Determines at what point in value of <code>measure</code> the lesson should be increased. <code>min_lesson_length</code> The minimum number of episodes that should be completed before the lesson can change. If <code>measure</code> is set to <code>reward</code>, the average cumulative reward of the last <code>min_lesson_length</code> episodes will be used to determine if the lesson should change. Must be nonnegative.  Important: the average reward that is compared to the thresholds is different than the mean reward that is logged to the console. For example, if <code>min_lesson_length</code> is <code>100</code>, the lesson will increment after the average cumulative reward of the last <code>100</code> episodes exceeds the current threshold. The mean reward logged to the console is dictated by the <code>summary_freq</code> parameter defined above. <code>signal_smoothing</code> Whether to weight the current progress measure by previous values. <code>require_reset</code> Whether changing lesson requires the environment to reset (default: false) ##### Training with a Curriculum <p>Once we have specified our metacurriculum and curricula, we can launch <code>mlagents-learn</code> to point to the config file containing our curricula and PPO will train using Curriculum Learning. For example, to train agents in the Wall Jump environment with curriculum learning, we can run:</p> <pre><code>mlagents-learn config/ppo/WallJump_curriculum.yaml --run-id=wall-jump-curriculum\n</code></pre> <p>We can then keep track of the current lessons and progresses via TensorBoard. If you've terminated the run, you can resume it using <code>--resume</code> and lesson progress will start off where it ended.</p>"},{"location":"Training-ML-Agents/#training-using-concurrent-unity-instances","title":"Training Using Concurrent Unity Instances","text":"<p>In order to run concurrent Unity instances during training, set the number of environment instances using the command line option <code>--num-envs=&lt;n&gt;</code> when you invoke <code>mlagents-learn</code>. Optionally, you can also set the <code>--base-port</code>, which is the starting port used for the concurrent Unity instances.</p> <p>Some considerations:</p> <ul> <li>Buffer Size - If you are having trouble getting an agent to train, even   with multiple concurrent Unity instances, you could increase <code>buffer_size</code> in   the trainer config file. A common practice is to multiply   <code>buffer_size</code> by <code>num-envs</code>.</li> <li>Resource Constraints - Invoking concurrent Unity instances is constrained   by the resources on the machine. Please use discretion when setting   <code>--num-envs=&lt;n&gt;</code>.</li> <li>Result Variation Using Concurrent Unity Instances - If you keep all the   hyperparameters the same, but change <code>--num-envs=&lt;n&gt;</code>, the results and model   would likely change.</li> </ul>"},{"location":"Training-Plugins/","title":"Customizing Training via Plugins","text":"<p>ML-Agents provides support for running your own python implementations of specific interfaces during the training process. These interfaces are currently fairly limited, but will be expanded in the future.</p> <p>Note: Plugin interfaces should currently be considered \"in beta\", and they may change in future releases.</p>"},{"location":"Training-Plugins/#how-to-write-your-own-plugin","title":"How to Write Your Own Plugin","text":"<p>This video explains the basics of how to create a plugin system using setuptools, and is the same approach that ML-Agents' plugin system is based on.</p> <p>The <code>ml-agents-plugin-examples</code> directory contains a reference implementation of each plugin interface, so it's a good starting point.</p>"},{"location":"Training-Plugins/#setuppy","title":"setup.py","text":"<p>If you don't already have a <code>setup.py</code> file for your python code, you'll need to add one. <code>ml-agents-plugin-examples</code> has a minimal example of this.</p> <p>In the call to <code>setup()</code>, you'll need to add to the <code>entry_points</code> dictionary for each plugin interface that you implement. The form of this is <code>{entry point name}={plugin module}:{plugin function}</code>. For example, in  <code>ml-agents-plugin-examples</code>:</p> <pre><code>entry_points={\n    ML_AGENTS_STATS_WRITER: [\n        \"example=mlagents_plugin_examples.example_stats_writer:get_example_stats_writer\"\n    ]\n}\n</code></pre> <ul> <li><code>ML_AGENTS_STATS_WRITER</code> (which is a string constant, <code>mlagents.stats_writer</code>) is the name of the plugin interface. This must be one of the provided interfaces (see below).</li> <li><code>example</code> is the plugin implementation name. This can be anything.</li> <li><code>mlagents_plugin_examples.example_stats_writer</code> is the plugin module. This points to the module where the plugin registration function is defined.</li> <li><code>get_example_stats_writer</code> is the plugin registration function. This is called when running <code>mlagents-learn</code>. The arguments and expected return type for this are different for each plugin interface.</li> </ul>"},{"location":"Training-Plugins/#local-installation","title":"Local Installation","text":"<p>Once you've defined <code>entry_points</code> in your <code>setup.py</code>, you will need to run</p> <pre><code>pip install -e [path to your plugin code]\n</code></pre> <p>in the same python virtual environment that you have <code>mlagents</code> installed.</p>"},{"location":"Training-Plugins/#plugin-interfaces","title":"Plugin Interfaces","text":""},{"location":"Training-Plugins/#statswriter","title":"StatsWriter","text":"<p>The StatsWriter class receives various information from the training process, such as the average Agent reward in each summary period. By default, we log this information to the console and write it to TensorBoard.</p>"},{"location":"Training-Plugins/#interface","title":"Interface","text":"<p>The <code>StatsWriter.write_stats()</code> method must be implemented in any derived classes. It takes a \"category\" parameter, which typically is the behavior name of the Agents being trained, and a dictionary of <code>StatSummary</code> values with string keys. Additionally, <code>StatsWriter.on_add_stat()</code> may be extended to register a callback handler for each stat emission.</p>"},{"location":"Training-Plugins/#registration","title":"Registration","text":"<p>The <code>StatsWriter</code> registration function takes a <code>RunOptions</code> argument and returns a list of <code>StatsWriter</code>s. An example implementation is provided in <code>mlagents_plugin_examples</code></p>"},{"location":"Training-on-Amazon-Web-Service/","title":"Training on Amazon Web Service","text":"<p>:warning: Note: We no longer use this guide ourselves and so it may not work correctly. We've decided to keep it up just in case it is helpful to you.</p> <p>This page contains instructions for setting up an EC2 instance on Amazon Web Service for training ML-Agents environments.</p>"},{"location":"Training-on-Amazon-Web-Service/#pre-configured-ami","title":"Pre-configured AMI","text":"<p>We've prepared a pre-configured AMI for you with the ID: <code>ami-016ff5559334f8619</code> in the <code>us-east-1</code> region. It was created as a modification of Deep Learning AMI (Ubuntu). The AMI has been tested with p2.xlarge instance. Furthermore, if you want to train without headless mode, you need to enable X Server.</p> <p>After launching your EC2 instance using the ami and ssh into it, run the following commands to enable it:</p> <pre><code># Start the X Server, press Enter to come to the command line\n$ sudo /usr/bin/X :0 &amp;\n\n# Check if Xorg process is running\n# You will have a list of processes running on the GPU, Xorg should be in the\n# list, as shown below\n$ nvidia-smi\n\n# Thu Jun 14 20:27:26 2018\n# +-----------------------------------------------------------------------------+\n# | NVIDIA-SMI 390.67                 Driver Version: 390.67                    |\n# |-------------------------------+----------------------+----------------------+\n# | GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n# | Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n# |===============================+======================+======================|\n# |   0  Tesla K80           On   | 00000000:00:1E.0 Off |                    0 |\n# | N/A   35C    P8    31W / 149W |      9MiB / 11441MiB |      0%      Default |\n# +-------------------------------+----------------------+----------------------+\n#\n# +-----------------------------------------------------------------------------+\n# | Processes:                                                       GPU Memory |\n# |  GPU       PID   Type   Process name                             Usage      |\n# |=============================================================================|\n# |    0      2331      G   /usr/lib/xorg/Xorg                             8MiB |\n# +-----------------------------------------------------------------------------+\n\n# Make the ubuntu use X Server for display\n$ export DISPLAY=:0\n</code></pre>"},{"location":"Training-on-Amazon-Web-Service/#configuring-your-own-instance","title":"Configuring your own instance","text":"<p>You could also choose to configure your own instance. To begin with, you will need an EC2 instance which contains the latest Nvidia drivers, CUDA9, and cuDNN. In this tutorial we used the Deep Learning AMI (Ubuntu) listed under AWS Marketplace with a p2.xlarge instance.</p>"},{"location":"Training-on-Amazon-Web-Service/#installing-the-ml-agents-toolkit-on-the-instance","title":"Installing the ML-Agents Toolkit on the instance","text":"<p>After launching your EC2 instance using the ami and ssh into it:</p> <ol> <li>Activate the python3 environment</li> </ol> <p><code>sh    source activate python3</code></p> <ol> <li>Clone the ML-Agents repo and install the required Python packages</li> </ol> <p><code>sh    git clone --branch release_21 https://github.com/Unity-Technologies/ml-agents.git    cd ml-agents/ml-agents/    pip3 install -e .</code></p>"},{"location":"Training-on-Amazon-Web-Service/#setting-up-x-server-optional","title":"Setting up X Server (optional)","text":"<p>X Server setup is only necessary if you want to do training that requires visual observation input. Instructions here are adapted from this Medium post on running general Unity applications in the cloud.</p> <p>Current limitations of the Unity Engine require that a screen be available to render to when using visual observations. In order to make this possible when training on a remote server, a virtual screen is required. We can do this by installing Xorg and creating a virtual screen. Once installed and created, we can display the Unity environment in the virtual environment, and train as we would on a local machine. Ensure that <code>headless</code> mode is disabled when building linux executables which use visual observations.</p>"},{"location":"Training-on-Amazon-Web-Service/#install-and-setup-xorg","title":"Install and setup Xorg:","text":"<pre><code>```sh\n# Install Xorg\n$ sudo apt-get update\n$ sudo apt-get install -y xserver-xorg mesa-utils\n$ sudo nvidia-xconfig -a --use-display-device=None --virtual=1280x1024\n\n# Get the BusID information\n$ nvidia-xconfig --query-gpu-info\n\n# Add the BusID information to your /etc/X11/xorg.conf file\n$ sudo sed -i 's/    BoardName      \"Tesla K80\"/    BoardName      \"Tesla K80\"\\n    BusID          \"0:30:0\"/g' /etc/X11/xorg.conf\n\n# Remove the Section \"Files\" from the /etc/X11/xorg.conf file\n# And remove two lines that contain Section \"Files\" and EndSection\n$ sudo vim /etc/X11/xorg.conf\n```\n</code></pre>"},{"location":"Training-on-Amazon-Web-Service/#update-and-setup-nvidia-driver","title":"Update and setup Nvidia driver:","text":"<pre><code>```sh\n# Download and install the latest Nvidia driver for ubuntu\n# Please refer to http://download.nvidia.com/XFree86/Linux-#x86_64/latest.txt\n$ wget http://download.nvidia.com/XFree86/Linux-x86_64/390.87/NVIDIA-Linux-x86_64-390.87.run\n$ sudo /bin/bash ./NVIDIA-Linux-x86_64-390.87.run --accept-license --no-questions --ui=none\n\n# Disable Nouveau as it will clash with the Nvidia driver\n$ sudo echo 'blacklist nouveau'  | sudo tee -a /etc/modprobe.d/blacklist.conf\n$ sudo echo 'options nouveau modeset=0'  | sudo tee -a /etc/modprobe.d/blacklist.conf\n$ sudo echo options nouveau modeset=0 | sudo tee -a /etc/modprobe.d/nouveau-kms.conf\n$ sudo update-initramfs -u\n```\n</code></pre>"},{"location":"Training-on-Amazon-Web-Service/#restart-the-ec2-instance","title":"Restart the EC2 instance:","text":"<pre><code>```sh\nsudo reboot now\n```\n</code></pre>"},{"location":"Training-on-Amazon-Web-Service/#make-sure-there-are-no-xorg-processes-running","title":"Make sure there are no Xorg processes running:","text":"<pre><code># Kill any possible running Xorg processes\n# Note that you might have to run this command multiple times depending on\n# how Xorg is configured.\n$ sudo killall Xorg\n\n# Check if there is any Xorg process left\n# You will have a list of processes running on the GPU, Xorg should not be in\n# the list, as shown below.\n$ nvidia-smi\n\n# Thu Jun 14 20:21:11 2018\n# +-----------------------------------------------------------------------------+\n# | NVIDIA-SMI 390.67                 Driver Version: 390.67                    |\n# |-------------------------------+----------------------+----------------------+\n# | GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n# | Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n# |===============================+======================+======================|\n# |   0  Tesla K80           On   | 00000000:00:1E.0 Off |                    0 |\n# | N/A   37C    P8    31W / 149W |      0MiB / 11441MiB |      0%      Default |\n# +-------------------------------+----------------------+----------------------+\n#\n# +-----------------------------------------------------------------------------+\n# | Processes:                                                       GPU Memory |\n# |  GPU       PID   Type   Process name                             Usage      |\n# |=============================================================================|\n# |  No running processes found                                                 |\n# +-----------------------------------------------------------------------------+\n\n</code></pre>"},{"location":"Training-on-Amazon-Web-Service/#start-x-server-and-make-the-ubuntu-use-x-server-for-display","title":"Start X Server and make the ubuntu use X Server for display:","text":"<pre><code>```console\n# Start the X Server, press Enter to come back to the command line\n$ sudo /usr/bin/X :0 &amp;\n\n# Check if Xorg process is running\n# You will have a list of processes running on the GPU, Xorg should be in the list.\n$ nvidia-smi\n\n# Make the ubuntu use X Server for display\n$ export DISPLAY=:0\n```\n</code></pre>"},{"location":"Training-on-Amazon-Web-Service/#ensure-the-xorg-is-correctly-configured","title":"Ensure the Xorg is correctly configured:","text":"<pre><code>```sh\n# For more information on glxgears, see ftp://www.x.org/pub/X11R6.8.1/doc/glxgears.1.html.\n$ glxgears\n# If Xorg is configured correctly, you should see the following message\n\n# Running synchronized to the vertical refresh.  The framerate should be\n# approximately the same as the monitor refresh rate.\n# 137296 frames in 5.0 seconds = 27459.053 FPS\n# 141674 frames in 5.0 seconds = 28334.779 FPS\n# 141490 frames in 5.0 seconds = 28297.875 FPS\n\n```\n</code></pre>"},{"location":"Training-on-Amazon-Web-Service/#training-on-ec2-instance","title":"Training on EC2 instance","text":"<ol> <li>In the Unity Editor, load a project containing an ML-Agents environment (you    can use one of the example environments if you have not created your own).</li> <li>Open the Build Settings window (menu: File &gt; Build Settings).</li> <li>Select Linux as the Target Platform, and x86_64 as the target architecture    (the default x86 currently does not work).</li> <li>Check Headless Mode if you have not setup the X Server. (If you do not use    Headless Mode, you have to setup the X Server to enable training.)</li> <li>Click Build to build the Unity environment executable.</li> <li>Upload the executable to your EC2 instance within <code>ml-agents</code> folder.</li> <li>Change the permissions of the executable.</li> </ol> <p><code>sh    chmod +x &lt;your_env&gt;.x86_64</code></p> <ol> <li>(Without Headless Mode) Start X Server and use it for display:</li> </ol> <p>```sh    # Start the X Server, press Enter to come back to the command line    $ sudo /usr/bin/X :0 &amp;</p> <p># Check if Xorg process is running    # You will have a list of processes running on the GPU, Xorg should be in the list.    $ nvidia-smi</p> <p># Make the ubuntu use X Server for display    $ export DISPLAY=:0    ```</p> <ol> <li>Test the instance setup from Python using:</li> </ol> <p>```python    from mlagents_envs.environment import UnityEnvironment</p> <p>env = UnityEnvironment()    ``` <p>Where <code>&lt;your_env&gt;</code> corresponds to the path to your environment executable.</p> <p>You should receive a message confirming that the environment was loaded    successfully.</p> <ol> <li> <p>Train your models</p> <p><code>console mlagents-learn &lt;trainer-config-file&gt; --env=&lt;your_env&gt; --train</code></p> </li> </ol>"},{"location":"Training-on-Amazon-Web-Service/#faq","title":"FAQ","text":""},{"location":"Training-on-Amazon-Web-Service/#the-_data-folder-hasnt-been-copied-cover","title":"The _Data folder hasn't been copied cover <p>If you've built your Linux executable, but forget to copy over the corresponding _Data folder, you will see error message like the following: <pre><code>Set current directory to /home/ubuntu/ml-agents/ml-agents\nFound path: /home/ubuntu/ml-agents/ml-agents/3dball_linux.x86_64\nno boot config - using default values\n\n(Filename:  Line: 403)\n\nThere is no data folder\n</code></pre>","text":""},{"location":"Training-on-Amazon-Web-Service/#unity-environment-not-responding","title":"Unity Environment not responding <p>If you didn't setup X Server or hasn't launched it properly, or your environment somehow crashes, or you haven't <code>chmod +x</code> your Unity Environment, all of these will cause connection between Unity and Python to fail. Then you will see something like this:</p> <pre><code>Logging to /home/ubuntu/.config/unity3d/&lt;Some_Path&gt;/Player.log\nTraceback (most recent call last):\n  File \"&lt;stdin&gt;\", line 1, in &lt;module&gt;\n  File \"/home/ubuntu/ml-agents/ml-agents/mlagents_envs/environment.py\", line 63, in __init__\n    aca_params = self.send_academy_parameters(rl_init_parameters_in)\n  File \"/home/ubuntu/ml-agents/ml-agents/mlagents_envs/environment.py\", line 489, in send_academy_parameters\n    return self.communicator.initialize(inputs).rl_initialization_output\n  File \"/home/ubuntu/ml-agents/ml-agents/mlagents_envs/rpc_communicator.py\", line 60, in initialize\nmlagents_envs.exception.UnityTimeOutException: The Unity environment took too long to respond. Make sure that :\n         The environment does not need user interaction to launch\n         The environment and the Python interface have compatible versions.\n</code></pre> <p>It would be also really helpful to check your /home/ubuntu/.config/unity3d//Player.log to see what happens with your Unity environment.","text":""},{"location":"Training-on-Amazon-Web-Service/#could-not-launch-x-server","title":"Could not launch X Server <p>When you execute:</p> <pre><code>sudo /usr/bin/X :0 &amp;\n</code></pre> <p>You might see something like:</p> <pre><code>X.Org X Server 1.18.4\n...\n(==) Log file: \"/var/log/Xorg.0.log\", Time: Thu Oct 11 21:10:38 2018\n(==) Using config file: \"/etc/X11/xorg.conf\"\n(==) Using system config directory \"/usr/share/X11/xorg.conf.d\"\n(EE)\nFatal server error:\n(EE) no screens found(EE)\n(EE)\nPlease consult the X.Org Foundation support\n         at http://wiki.x.org\n for help.\n(EE) Please also check the log file at \"/var/log/Xorg.0.log\" for additional information.\n(EE)\n(EE) Server terminated with error (1). Closing log file.\n</code></pre> <p>And when you execute:</p> <pre><code>nvidia-smi\n</code></pre> <p>You might see something like:</p> <pre><code>NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.\n</code></pre> <p>This means the NVIDIA's driver needs to be updated. Refer to this section for more information.</p>","text":""},{"location":"Training-on-Microsoft-Azure/","title":"Training on Microsoft Azure (works with ML-Agents Toolkit v0.3)","text":"<p>:warning: Note: We no longer use this guide ourselves and so it may not work correctly. We've decided to keep it up just in case it is helpful to you.</p> <p>This page contains instructions for setting up training on Microsoft Azure through either Azure Container Instances or Virtual Machines. Non \"headless\" training has not yet been tested to verify support.</p>"},{"location":"Training-on-Microsoft-Azure/#pre-configured-azure-virtual-machine","title":"Pre-Configured Azure Virtual Machine","text":"<p>A pre-configured virtual machine image is available in the Azure Marketplace and is nearly completely ready for training. You can start by deploying the Data Science Virtual Machine for Linux (Ubuntu) into your Azure subscription.</p> <p>Note that, if you choose to deploy the image to an N-Series GPU optimized VM, training will, by default, run on the GPU. If you choose any other type of VM, training will run on the CPU.</p>"},{"location":"Training-on-Microsoft-Azure/#configuring-your-own-instance","title":"Configuring your own Instance","text":"<p>Setting up your own instance requires a number of package installations. Please view the documentation for doing so here.</p>"},{"location":"Training-on-Microsoft-Azure/#installing-ml-agents","title":"Installing ML-Agents","text":"<ol> <li>Move    the <code>ml-agents</code> sub-folder of this ml-agents repo to the remote Azure    instance, and set it as the working directory.</li> <li>Install the required packages:    Torch: <code>pip3 install torch==1.7.0 -f https://download.pytorch.org/whl/torch_stable.html</code> and    MLAgents: <code>python -m pip install mlagents==1.0.0</code></li> </ol>"},{"location":"Training-on-Microsoft-Azure/#testing","title":"Testing","text":"<p>To verify that all steps worked correctly:</p> <ol> <li>In the Unity Editor, load a project containing an ML-Agents environment (you    can use one of the example environments if you have not created your own).</li> <li>Open the Build Settings window (menu: File &gt; Build Settings).</li> <li>Select Linux as the Target Platform, and x86_64 as the target architecture.</li> <li>Check Headless Mode.</li> <li>Click Build to build the Unity environment executable.</li> <li>Upload the resulting files to your Azure instance.</li> <li>Test the instance setup from Python using:</li> </ol> <pre><code>from mlagents_envs.environment import UnityEnvironment\n\nenv = UnityEnvironment(file_name=\"&lt;your_env&gt;\", seed=1, side_channels=[])\n</code></pre> <p>Where <code>&lt;your_env&gt;</code> corresponds to the path to your environment executable (i.e. <code>/home/UserName/Build/yourFile</code>).</p> <p>You should receive a message confirming that the environment was loaded successfully.</p> <p>Note: When running your environment in headless mode, you must append <code>--no-graphics</code> to your mlagents-learn command, as it won't train otherwise. You can test this simply by aborting a training and check if it says \"Model Saved\" or \"Aborted\", or see if it generated the .onnx in the result folder.</p>"},{"location":"Training-on-Microsoft-Azure/#running-training-on-your-virtual-machine","title":"Running Training on your Virtual Machine","text":"<p>To run your training on the VM:</p> <ol> <li>Move    your built Unity application to your Virtual Machine.</li> <li>Set the directory where the ML-Agents Toolkit was installed to your working    directory.</li> <li>Run the following command:</li> </ol> <pre><code>mlagents-learn &lt;trainer_config&gt; --env=&lt;your_app&gt; --run-id=&lt;run_id&gt; --train\n</code></pre> <p>Where <code>&lt;your_app&gt;</code> is the path to your app (i.e. <code>~/unity-volume/3DBallHeadless</code>) and <code>&lt;run_id&gt;</code> is an identifier you would like to identify your training run with.</p> <p>If you've selected to run on a N-Series VM with GPU support, you can verify that the GPU is being used by running <code>nvidia-smi</code> from the command line.</p>"},{"location":"Training-on-Microsoft-Azure/#monitoring-your-training-run-with-tensorboard","title":"Monitoring your Training Run with TensorBoard","text":"<p>Once you have started training, you can use TensorBoard to observe the training.</p> <ol> <li> <p>Start by    opening the appropriate port for web traffic to connect to your VM.</p> </li> <li> <p>Note that you don't need to generate a new <code>Network Security Group</code> but      instead, go to the Networking tab under Settings for your VM.</p> </li> <li> <p>As an example, you could use the following settings to open the Port with      the following Inbound Rule settings:</p> <ul> <li>Source: Any</li> <li>Source Port Ranges: *</li> <li>Destination: Any</li> <li>Destination Port Ranges: 6006</li> <li>Protocol: Any</li> <li>Action: Allow</li> <li>Priority: (Leave as default)</li> </ul> </li> <li> <p>Unless you started the training as a background process, connect to your VM    from another terminal instance.</p> </li> <li>Run the following command from your terminal    <code>tensorboard --logdir results --host 0.0.0.0</code></li> <li>You should now be able to open a browser and navigate to    <code>&lt;Your_VM_IP_Address&gt;:6060</code> to view the TensorBoard report.</li> </ol>"},{"location":"Training-on-Microsoft-Azure/#running-on-azure-container-instances","title":"Running on Azure Container Instances","text":"<p>Azure Container Instances allow you to spin up a container, on demand, that will run your training and then be shut down. This ensures you aren't leaving a billable VM running when it isn't needed. Using ACI enables you to offload training of your models without needing to install Python and TensorFlow on your own computer.</p>"},{"location":"Training-on-Microsoft-Azure/#custom-instances","title":"Custom Instances","text":"<p>This page contains instructions for setting up a custom Virtual Machine on Microsoft Azure so you can running ML-Agents training in the cloud.</p> <ol> <li>Start by    deploying an Azure VM    with Ubuntu Linux (tests were done with 16.04 LTS). To use GPU support, use a    N-Series VM.</li> <li>SSH into your VM.</li> <li>Start with the following commands to install the Nvidia driver:</li> </ol> <p>```sh    wget http://us.download.nvidia.com/tesla/375.66/nvidia-diag-driver-local-repo-ubuntu1604_375.66-1_amd64.deb</p> <p>sudo dpkg -i nvidia-diag-driver-local-repo-ubuntu1604_375.66-1_amd64.deb</p> <p>sudo apt-get update</p> <p>sudo apt-get install cuda-drivers</p> <p>sudo reboot    ```</p> <ol> <li>After a minute you should be able to reconnect to your VM and install the    CUDA toolkit:</li> </ol> <p>```sh    wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/cuda-repo-ubuntu1604_8.0.61-1_amd64.deb</p> <p>sudo dpkg -i cuda-repo-ubuntu1604_8.0.61-1_amd64.deb</p> <p>sudo apt-get update</p> <p>sudo apt-get install cuda-8-0    ```</p> <ol> <li> <p>You'll next need to download cuDNN from the Nvidia developer site. This    requires a registered account.</p> </li> <li> <p>Navigate to http://developer.nvidia.com and    create an account and verify it.</p> </li> <li> <p>Download (to your own computer) cuDNN from    this url.</p> </li> <li> <p>Copy the deb package to your VM:</p> </li> </ol> <p><code>sh    scp libcudnn6_6.0.21-1+cuda8.0_amd64.deb &lt;VMUserName&gt;@&lt;VMIPAddress&gt;:libcudnn6_6.0.21-1+cuda8.0_amd64.deb</code></p> <ol> <li>SSH back to your VM and execute the following:</li> </ol> <p>```console    sudo dpkg -i libcudnn6_6.0.21-1+cuda8.0_amd64.deb</p> <p>export LD_LIBRARY_PATH=/usr/local/cuda/lib64/:/usr/lib/x86_64-linux-gnu/:$LD_LIBRARY_PATH    . ~/.profile</p> <p>sudo reboot    ```</p> <ol> <li> <p>After a minute, you should be able to SSH back into your VM. After doing so,     run the following:</p> <p><code>sh sudo apt install python-pip sudo apt install python3-pip</code></p> </li> <li> <p>At this point, you need to install TensorFlow. The version you install     should be tied to if you are using GPU to train:</p> <p><code>sh pip3 install tensorflow-gpu==1.4.0 keras==2.0.6</code></p> <p>Or CPU to train:</p> <p><code>sh pip3 install tensorflow==1.4.0 keras==2.0.6</code></p> </li> <li> <p>You'll then need to install additional dependencies:</p> <p><code>sh pip3 install pillow pip3 install numpy</code></p> </li> </ol>"},{"location":"Tutorial-Custom-Trainer-Plugin/","title":"Custom Trainer Plugin","text":""},{"location":"Tutorial-Custom-Trainer-Plugin/#how-to-write-a-custom-trainer-plugin","title":"How to write a custom trainer plugin","text":""},{"location":"Tutorial-Custom-Trainer-Plugin/#step-1-write-your-custom-trainer-class","title":"Step 1: Write your custom trainer class","text":"<p>Before you start writing your code, make sure to use your favorite environment management tool(e.g. <code>venv</code> or <code>conda</code>) to create and activate a Python virtual environment. The following command uses <code>conda</code>, but other tools work similarly:</p> <pre><code>conda create -n trainer-env python=3.10.12\nconda activate trainer-env\n</code></pre> <p>Users of the plug-in system are responsible for implementing the trainer class subject to the API standard. Let us follow an example by implementing a custom trainer named \"YourCustomTrainer\". You can either extend <code>OnPolicyTrainer</code> or <code>OffPolicyTrainer</code> classes depending on the training strategies you choose.</p> <p>Please refer to the internal PPO implementation for a complete code example. We will not provide a workable code in the document. The purpose of the tutorial is to introduce you to the core components and interfaces of our plugin framework. We use code snippets and patterns to demonstrate the control and data flow.</p> <p>Your custom trainers are responsible for collecting experiences and training the models. Your custom trainer class acts like a coordinator to the policy and optimizer. To start implementing methods in the class, create a policy class objects from method <code>create_policy</code>:</p> <pre><code>def create_policy(\n    self, parsed_behavior_id: BehaviorIdentifiers, behavior_spec: BehaviorSpec\n) -&gt; TorchPolicy:\n\n    actor_cls: Union[Type[SimpleActor], Type[SharedActorCritic]] = SimpleActor\n    actor_kwargs: Dict[str, Any] = {\n        \"conditional_sigma\": False,\n        \"tanh_squash\": False,\n    }\n    if self.shared_critic:\n        reward_signal_configs = self.trainer_settings.reward_signals\n        reward_signal_names = [\n            key.value for key, _ in reward_signal_configs.items()\n        ]\n        actor_cls = SharedActorCritic\n        actor_kwargs.update({\"stream_names\": reward_signal_names})\n\n    policy = TorchPolicy(\n        self.seed,\n        behavior_spec,\n        self.trainer_settings.network_settings,\n        actor_cls,\n        actor_kwargs,\n    )\n    return policy\n\n</code></pre> <p>Depending on whether you use shared or separate network architecture for your policy, we provide <code>SimpleActor</code> and <code>SharedActorCritic</code> from <code>mlagents.trainers.torch_entities.networks</code> that you can choose from. In our example above, we use a <code>SimpleActor</code>.</p> <p>Next, create an optimizer class object from <code>create_optimizer</code> method and connect it to the policy object you created above:</p> <pre><code>def create_optimizer(self) -&gt; TorchOptimizer:\n    return TorchPPOOptimizer(  # type: ignore\n        cast(TorchPolicy, self.policy), self.trainer_settings  # type: ignore\n    )  # type: ignore\n\n</code></pre> <p>There are a couple of abstract methods(<code>_process_trajectory</code> and <code>_update_policy</code>) inherited from <code>RLTrainer</code> that you need to implement in your custom trainer class. <code>_process_trajectory</code> takes a trajectory and processes it, putting it into the update buffer. Processing involves calculating value and advantage targets for the model updating step. Given input <code>trajectory: Trajectory</code>, users are responsible for processing the data in the trajectory and append <code>agent_buffer_trajectory</code> to the back of the update buffer by calling <code>self._append_to_update_buffer(agent_buffer_trajectory)</code>, whose output will be used in updating the model in <code>optimizer</code> class.</p> <p>A typical <code>_process_trajectory</code> function(incomplete) will convert a trajectory object to an agent buffer then get all value estimates from the trajectory by calling <code>self.optimizer.get_trajectory_value_estimates</code>. From the returned dictionary of value estimates we extract reward signals keyed by their names:</p> <pre><code>def _process_trajectory(self, trajectory: Trajectory) -&gt; None:\n    super()._process_trajectory(trajectory)\n    agent_id = trajectory.agent_id  # All the agents should have the same ID\n\n    agent_buffer_trajectory = trajectory.to_agentbuffer()\n\n    # Get all value estimates\n    (\n        value_estimates,\n        value_next,\n        value_memories,\n    ) =  self.optimizer.get_trajectory_value_estimates(\n        agent_buffer_trajectory,\n        trajectory.next_obs,\n        trajectory.done_reached and not trajectory.interrupted,\n    )\n\n    for name, v in value_estimates.items():\n        agent_buffer_trajectory[RewardSignalUtil.value_estimates_key(name)].extend(\n            v\n        )\n        self._stats_reporter.add_stat(\n            f\"Policy/{self.optimizer.reward_signals[name].name.capitalize()} Value Estimate\",\n            np.mean(v),\n        )\n\n    # Evaluate all reward functions\n    self.collected_rewards[\"environment\"][agent_id] += np.sum(\n        agent_buffer_trajectory[BufferKey.ENVIRONMENT_REWARDS]\n    )\n    for name, reward_signal in self.optimizer.reward_signals.items():\n        evaluate_result = (\n            reward_signal.evaluate(agent_buffer_trajectory) * reward_signal.strength\n        )\n        agent_buffer_trajectory[RewardSignalUtil.rewards_key(name)].extend(\n            evaluate_result\n        )\n        # Report the reward signals\n        self.collected_rewards[name][agent_id] += np.sum(evaluate_result)\n\n    self._append_to_update_buffer(agent_buffer_trajectory)\n\n</code></pre> <p>A trajectory will be a list of dictionaries of strings mapped to <code>Anything</code>. When calling <code>forward</code> on a policy, the argument will include an \u201cexperience\u201d dictionary from the last step. The <code>forward</code> method will generate an action and the next \u201cexperience\u201d dictionary. Examples of fields in the \u201cexperience\u201d dictionary include observation, action, reward, done status, group_reward, LSTM memory state, etc.</p>"},{"location":"Tutorial-Custom-Trainer-Plugin/#step-2-implement-your-custom-optimizer-for-the-trainer","title":"Step 2: implement your custom optimizer for the trainer.","text":"<p>We will show you an example we implemented - <code>class TorchPPOOptimizer(TorchOptimizer)</code>, which takes a Policy and a Dict of trainer parameters and creates an Optimizer that connects to the policy. Your optimizer should include a value estimator and a loss function in the <code>update</code> method.</p> <p>Before writing your optimizer class, first define setting class <code>class PPOSettings(OnPolicyHyperparamSettings)</code> for your custom optimizer:</p> <pre><code>class PPOSettings(OnPolicyHyperparamSettings):\n    beta: float = 5.0e-3\n    epsilon: float = 0.2\n    lambd: float = 0.95\n    num_epoch: int = 3\n    shared_critic: bool = False\n    learning_rate_schedule: ScheduleType = ScheduleType.LINEAR\n    beta_schedule: ScheduleType = ScheduleType.LINEAR\n    epsilon_schedule: ScheduleType = ScheduleType.LINEAR\n\n</code></pre> <p>You should implement <code>update</code> function following interface:</p> <pre><code>def update(self, batch: AgentBuffer, num_sequences: int) -&gt; Dict[str, float]:\n\n</code></pre> <p>In which losses and other metrics are calculated from an <code>AgentBuffer</code> that is generated from your trainer class, depending on which model you choose to implement the loss functions will be different. In our case we calculate value loss from critic and trust region policy loss. A typical pattern(incomplete) of the calculations will look like the following:</p> <pre><code>run_out = self.policy.actor.get_stats(\n    current_obs,\n    actions,\n    masks=act_masks,\n    memories=memories,\n    sequence_length=self.policy.sequence_length,\n)\n\nlog_probs = run_out[\"log_probs\"]\nentropy = run_out[\"entropy\"]\n\nvalues, _ = self.critic.critic_pass(\n    current_obs,\n    memories=value_memories,\n    sequence_length=self.policy.sequence_length,\n)\npolicy_loss = ModelUtils.trust_region_policy_loss(\n    ModelUtils.list_to_tensor(batch[BufferKey.ADVANTAGES]),\n    log_probs,\n    old_log_probs,\n    loss_masks,\n    decay_eps,\n)\nloss = (\n    policy_loss\n    + 0.5 * value_loss\n    - decay_bet * ModelUtils.masked_mean(entropy, loss_masks)\n)\n\n</code></pre> <p>Finally update the model and return the a dictionary including calculated losses and updated decay learning rate:</p> <pre><code>ModelUtils.update_learning_rate(self.optimizer, decay_lr)\nself.optimizer.zero_grad()\nloss.backward()\n\nself.optimizer.step()\nupdate_stats = {\n    \"Losses/Policy Loss\": torch.abs(policy_loss).item(),\n    \"Losses/Value Loss\": value_loss.item(),\n    \"Policy/Learning Rate\": decay_lr,\n    \"Policy/Epsilon\": decay_eps,\n    \"Policy/Beta\": decay_bet,\n}\n\n</code></pre>"},{"location":"Tutorial-Custom-Trainer-Plugin/#step-3-integrate-your-custom-trainer-into-the-plugin-system","title":"Step 3: Integrate your custom trainer into the plugin system","text":"<p>By integrating a custom trainer into the plugin system, a user can use their published packages which have their implementations. To do that, you need to add a setup.py file. In the call to setup(), you'll need to add to the entry_points dictionary for each plugin interface that you implement. The form of this is {entry point name}={plugin module}:{plugin function}. For example:</p> <pre><code>entry_points={\n        ML_AGENTS_TRAINER_TYPE: [\n            \"your_trainer_type=your_package.your_custom_trainer:get_type_and_setting\"\n        ]\n    },\n</code></pre> <p>Some key elements in the code:</p> <pre><code>ML_AGENTS_TRAINER_TYPE: a string constant for trainer type\nyour_trainer_type: name your trainer type, used in configuration file\nyour_package: your pip installable package containing custom trainer implementation\n</code></pre> <p>Also define <code>get_type_and_setting</code> method in <code>YourCustomTrainer</code> class:</p> <pre><code>def get_type_and_setting():\n    return {YourCustomTrainer.get_trainer_name(): YourCustomTrainer}, {\n        YourCustomTrainer.get_trainer_name(): YourCustomSetting\n    }\n\n</code></pre> <p>Finally, specify trainer type in the config file:</p> <pre><code>behaviors:\n  3DBall:\n    trainer_type: your_trainer_type\n...\n</code></pre>"},{"location":"Tutorial-Custom-Trainer-Plugin/#step-4-install-your-custom-trainer-and-run-training","title":"Step 4: Install your custom trainer and run training:","text":"<p>Before installing your custom trainer package, make sure you have <code>ml-agents-env</code> and <code>ml-agents</code> installed</p> <pre><code>pip3 install -e ./ml-agents-envs &amp;&amp; pip3 install -e ./ml-agents\n</code></pre> <p>Install your custom trainer package(if your package is pip installable):</p> <pre><code>pip3 install your_custom_package\n</code></pre> <p>Or follow our internal implementations:</p> <pre><code>pip3 install -e ./ml-agents-trainer-plugin\n</code></pre> <p>Following the previous installations your package is added as an entrypoint and you can use a config file with new trainers:</p> <pre><code>mlagents-learn ml-agents-trainer-plugin/mlagents_trainer_plugin/a2c/a2c_3DBall.yaml --run-id &lt;run-id-name&gt;\n--env &lt;env-executable&gt;\n</code></pre>"},{"location":"Tutorial-Custom-Trainer-Plugin/#validate-your-implementations","title":"Validate your implementations:","text":"<p>Create a clean Python environment with Python 3.10.12 and activate it before you start, if you haven't done so already:</p> <pre><code>conda create -n trainer-env python=3.10.12\nconda activate trainer-env\n</code></pre> <p>Make sure you follow previous steps and install all required packages. We are testing internal implementations in this tutorial, but ML-Agents users can run similar validations once they have their own implementations installed:</p> <pre><code>pip3 install -e ./ml-agents-envs &amp;&amp; pip3 install -e ./ml-agents\npip3 install -e ./ml-agents-trainer-plugin\n</code></pre> <p>Once your package is added as an <code>entrypoint</code>, you can add to the config file the new trainer type. Check if trainer type is specified in the config file <code>a2c_3DBall.yaml</code>:</p> <pre><code>trainer_type: a2c\n</code></pre> <p>Test if custom trainer package is installed by running:</p> <pre><code>mlagents-learn ml-agents-trainer-plugin/mlagents_trainer_plugin/a2c/a2c_3DBall.yaml --run-id test-trainer\n</code></pre> <p>You can also list all trainers installed in the registry. Type <code>python</code> in your shell to open a REPL session. Run the python code below, you should be able to see all trainer types currently installed:</p> <pre><code>&gt;&gt;&gt; import pkg_resources\n&gt;&gt;&gt; for entry in pkg_resources.iter_entry_points('mlagents.trainer_type'):\n...     print(entry)\n...\ndefault = mlagents.plugins.trainer_type:get_default_trainer_types\na2c = mlagents_trainer_plugin.a2c.a2c_trainer:get_type_and_setting\ndqn = mlagents_trainer_plugin.dqn.dqn_trainer:get_type_and_setting\n</code></pre> <p>If it is properly installed, you will see Unity logo and message indicating training will start:</p> <pre><code>[INFO] Listening on port 5004. Start training by pressing the Play button in the Unity Editor.\n</code></pre> <p>If you see the following error message, it could be due to trainer type is wrong or the trainer type specified is not installed:</p> <pre><code>mlagents.trainers.exception.TrainerConfigError: Invalid trainer type a2c was found\n</code></pre>"},{"location":"Unity-Environment-Registry/","title":"Unity Environment Registry [Experimental]","text":"<p>The Unity Environment Registry is a database of pre-built Unity environments that can be easily used without having to install the Unity Editor. It is a great way to get started with our UnityEnvironment API.</p>"},{"location":"Unity-Environment-Registry/#loading-an-environment-from-the-registry","title":"Loading an Environment from the Registry","text":"<p>To get started, you can access the default registry we provide with our Example Environments. The Unity Environment Registry implements a Mapping, therefore, you can access an entry with its identifier with the square brackets <code>[ ]</code>. Use the following code to list all of the environment identifiers present in the default registry:</p> <pre><code>from mlagents_envs.registry import default_registry\n\nenvironment_names = list(default_registry.keys())\nfor name in environment_names:\n   print(name)\n</code></pre> <p>The <code>make()</code> method on a registry value will return a <code>UnityEnvironment</code> ready to be used. All arguments passed to the make method will be passed to the constructor of the <code>UnityEnvironment</code> as well. Refer to the documentation on the Python-API for more information about the arguments of the <code>UnityEnvironment</code> constructor. For example, the following code will create the environment under the identifier <code>\"my-env\"</code>, reset it, perform a few steps and finally close it:</p> <pre><code>from mlagents_envs.registry import default_registry\n\nenv = default_registry[\"my-env\"].make()\nenv.reset()\nfor _ in range(10):\n  env.step()\nenv.close()\n</code></pre>"},{"location":"Unity-Environment-Registry/#create-and-share-your-own-registry","title":"Create and share your own registry","text":"<p>In order to share the <code>UnityEnvironment</code> you created, you must:</p> <ul> <li>Create a Unity executable of your environment for each platform (Linux, OSX and/or Windows)</li> <li>Place each executable in a <code>zip</code> compressed folder</li> <li>Upload each zip file online to your preferred hosting platform</li> <li>Create a <code>yaml</code> file that will contain the description and path to your environment</li> <li>Upload the <code>yaml</code> file online The <code>yaml</code> file must have the following format :</li> </ul> <pre><code>environments:\n  - &lt;environment-identifier&gt;:\n     expected_reward: &lt;expected-reward-float&gt;\n     description: &lt;description-of-the-environment&gt;\n     linux_url: &lt;url-to-the-linux-zip-folder&gt;\n     darwin_url: &lt;url-to-the-osx-zip-folder&gt;\n     win_url: &lt;url-to-the-windows-zip-folder&gt;\n     additional_args:\n      - &lt;an-optional-list-of-command-line-arguments-for-the-executable&gt;\n      - ...\n</code></pre> <p>Your users can now use your environment with the following code :</p> <pre><code>from mlagents_envs.registry import UnityEnvRegistry\n\nregistry = UnityEnvRegistry()\nregistry.register_from_yaml(\"url-or-path-to-your-yaml-file\")\n</code></pre> <p>Note: The <code>\"url-or-path-to-your-yaml-file\"</code> can be either a url or a local path.</p>"},{"location":"Using-Docker/","title":"Using Docker For ML-Agents (Deprecated)","text":"<p>:warning: Note: We no longer use this guide ourselves and so it may not work correctly. We've decided to keep it up just in case it is helpful to you.</p> <p>We currently offer a solution for Windows and Mac users who would like to do training or inference using Docker. This option may be appealing to those who would like to avoid installing Python and TensorFlow themselves. The current setup forces both TensorFlow and Unity to only rely on the CPU for computations. Consequently, our Docker simulation does not use a GPU and uses <code>Xvfb</code> to do visual rendering. <code>Xvfb</code> is a utility that enables <code>ML-Agents</code> (or any other application) to do rendering virtually i.e. it does not assume that the machine running <code>ML-Agents</code> has a GPU or a display attached to it. This means that rich environments which involve agents using camera-based visual observations might be slower.</p>"},{"location":"Using-Docker/#requirements","title":"Requirements","text":"<ul> <li>Docker</li> <li>Unity Linux Build Support Component. Make sure to select the Linux Build   Support component when installing Unity.</li> </ul>"},{"location":"Using-Docker/#setup","title":"Setup","text":"<ul> <li> <p>Download the Unity Installer and add   the Linux Build Support Component</p> </li> <li> <p>Download and install   Docker if you don't have it setup on your machine.</p> </li> <li> <p>Since Docker runs a container in an environment that is isolated from the host   machine, a mounted directory in your host machine is used to share data, e.g.   the trainer configuration file, Unity executable and   TensorFlow graph. For convenience, we created an empty <code>unity-volume</code>   directory at the root of the repository for this purpose, but feel free to use   any other directory. The remainder of this guide assumes that the   <code>unity-volume</code> directory is the one used.</p> </li> </ul>"},{"location":"Using-Docker/#usage","title":"Usage","text":"<p>Using Docker for ML-Agents involves three steps: building the Unity environment with specific flags, building a Docker container and, finally, running the container. If you are not familiar with building a Unity environment for ML-Agents, please read through our Getting Started with the 3D Balance Ball Example guide first.</p>"},{"location":"Using-Docker/#build-the-environment-optional","title":"Build the Environment (Optional)","text":"<p>If you want to used the Editor to perform training, you can skip this step.</p> <p>Since Docker typically runs a container sharing a (linux) kernel with the host machine, the Unity environment has to be built for the linux platform. When building a Unity environment, please select the following options from the the Build Settings window:</p> <ul> <li>Set the Target Platform to <code>Linux</code></li> <li>Set the Architecture to <code>x86_64</code></li> </ul> <p>Then click <code>Build</code>, pick an environment name (e.g. <code>3DBall</code>) and set the output directory to <code>unity-volume</code>. After building, ensure that the file <code>&lt;environment-name&gt;.x86_64</code> and subdirectory <code>&lt;environment-name&gt;_Data/</code> are created under <code>unity-volume</code>.</p> <p></p>"},{"location":"Using-Docker/#build-the-docker-container","title":"Build the Docker Container","text":"<p>First, make sure the Docker engine is running on your machine. Then build the Docker container by calling the following command at the top-level of the repository:</p> <pre><code>docker build -t &lt;image-name&gt; .\n</code></pre> <p>Replace <code>&lt;image-name&gt;</code> with a name for the Docker image, e.g. <code>balance.ball.v0.1</code>.</p>"},{"location":"Using-Docker/#run-the-docker-container","title":"Run the Docker Container","text":"<p>Run the Docker container by calling the following command at the top-level of the repository:</p> <pre><code>docker run -it --name &lt;container-name&gt; \\\n           --mount type=bind,source=\"$(pwd)\"/unity-volume,target=/unity-volume \\\n           -p 5005:5005 \\\n           -p 6006:6006 \\\n           &lt;image-name&gt;:latest \\\n           &lt;trainer-config-file&gt; \\\n           --env=&lt;environment-name&gt; \\\n           --train \\\n           --run-id=&lt;run-id&gt;\n</code></pre> <p>Notes on argument values:</p> <ul> <li><code>&lt;container-name&gt;</code> is used to identify the container (in case you want to   interrupt and terminate it). This is optional and Docker will generate a   random name if this is not set. Note that this must be unique for every run   of a Docker image.</li> <li><code>&lt;image-name&gt;</code> references the image name used when building the container.</li> <li><code>&lt;environment-name&gt;</code> (Optional): If you are training with a linux   executable, this is the name of the executable. If you are training in the   Editor, do not pass a <code>&lt;environment-name&gt;</code> argument and press the Play   button in Unity when the message \"Start training by pressing the Play button   in the Unity Editor\" is displayed on the screen.</li> <li><code>source</code>: Reference to the path in your host OS where you will store the Unity   executable.</li> <li><code>target</code>: Tells Docker to mount the <code>source</code> path as a disk with this name.</li> <li><code>trainer-config-file</code>, <code>train</code>, <code>run-id</code>: ML-Agents arguments passed to   <code>mlagents-learn</code>. <code>trainer-config-file</code> is the filename of the trainer config   file, <code>train</code> trains the algorithm, and <code>run-id</code> is used to tag each   experiment with a unique identifier. We recommend placing the trainer-config   file inside <code>unity-volume</code> so that the container has access to the file.</li> </ul> <p>To train with a <code>3DBall</code> environment executable, the command would be:</p> <pre><code>docker run -it --name 3DBallContainer.first.trial \\\n           --mount type=bind,source=\"$(pwd)\"/unity-volume,target=/unity-volume \\\n           -p 5005:5005 \\\n           -p 6006:6006 \\\n           balance.ball.v0.1:latest 3DBall \\\n           /unity-volume/trainer_config.yaml \\\n           --env=/unity-volume/3DBall \\\n           --train \\\n           --run-id=3dball_first_trial\n</code></pre> <p>For more detail on Docker mounts, check out these docs from Docker.</p> <p>NOTE If you are training using docker for environments that use visual observations, you may need to increase the default memory that Docker allocates for the container. For example, see here for instructions for Docker for Mac.</p>"},{"location":"Using-Docker/#running-tensorboard","title":"Running Tensorboard","text":"<p>You can run Tensorboard to monitor your training instance on http://localhost:6006:</p> <pre><code>docker exec -it &lt;container-name&gt; tensorboard --logdir /unity-volume/results --host 0.0.0.0\n</code></pre> <p>With our previous 3DBall example, this command would look like this:</p> <pre><code>docker exec -it 3DBallContainer.first.trial tensorboard --logdir /unity-volume/results --host 0.0.0.0\n</code></pre> <p>For more details on Tensorboard, check out the documentation about Using Tensorboard.</p>"},{"location":"Using-Docker/#stopping-container-and-saving-state","title":"Stopping Container and Saving State","text":"<p>If you are satisfied with the training progress, you can stop the Docker container while saving state by either using <code>Ctrl+C</code> or <code>\u2318+C</code> (Mac) or by using the following command:</p> <pre><code>docker kill --signal=SIGINT &lt;container-name&gt;\n</code></pre> <p><code>&lt;container-name&gt;</code> is the name of the container specified in the earlier <code>docker run</code> command. If you didn't specify one, you can find the randomly generated identifier by running <code>docker container ls</code>.</p>"},{"location":"Using-Tensorboard/","title":"Using TensorBoard to Observe Training","text":"<p>The ML-Agents Toolkit saves statistics during learning session that you can view with a TensorFlow utility named, TensorBoard.</p> <p>The <code>mlagents-learn</code> command saves training statistics to a folder named <code>results</code>, organized by the <code>run-id</code> value you assign to a training session.</p> <p>In order to observe the training process, either during training or afterward, start TensorBoard:</p> <ol> <li>Open a terminal or console window:</li> <li>Navigate to the directory where the ML-Agents Toolkit is installed.</li> <li>From the command line run: <code>tensorboard --logdir results --port 6006</code></li> <li>Open a browser window and navigate to    localhost:6006.</li> </ol> <p>Note: The default port TensorBoard uses is 6006. If there is an existing session running on port 6006 a new session can be launched on an open port using the --port option.</p> <p>Note: If you don't assign a <code>run-id</code> identifier, <code>mlagents-learn</code> uses the default string, \"ppo\". You can delete the folders under the <code>results</code> directory to clear out old statistics.</p> <p>On the left side of the TensorBoard window, you can select which of the training runs you want to display. You can select multiple run-ids to compare statistics. The TensorBoard window also provides options for how to display and smooth graphs.</p>"},{"location":"Using-Tensorboard/#the-ml-agents-toolkit-training-statistics","title":"The ML-Agents Toolkit training statistics","text":"<p>The ML-Agents training program saves the following statistics:</p> <p></p>"},{"location":"Using-Tensorboard/#environment-statistics","title":"Environment Statistics","text":"<ul> <li> <p><code>Environment/Lesson</code> - Plots the progress from lesson to lesson. Only   interesting when performing curriculum training.</p> </li> <li> <p><code>Environment/Cumulative Reward</code> - The mean cumulative episode reward over all   agents. Should increase during a successful training session.</p> </li> <li> <p><code>Environment/Episode Length</code> - The mean length of each episode in the   environment for all agents.</p> </li> </ul>"},{"location":"Using-Tensorboard/#is-training","title":"Is Training","text":"<ul> <li><code>Is Training</code> - A boolean indicating if the agent is updating its model.</li> </ul>"},{"location":"Using-Tensorboard/#policy-statistics","title":"Policy Statistics","text":"<ul> <li> <p><code>Policy/Entropy</code> (PPO; SAC) - How random the decisions of the model are.   Should slowly decrease during a successful training process. If it decreases   too quickly, the <code>beta</code> hyperparameter should be increased.</p> </li> <li> <p><code>Policy/Learning Rate</code> (PPO; SAC) - How large a step the training algorithm   takes as it searches for the optimal policy. Should decrease over time.</p> </li> <li> <p><code>Policy/Entropy Coefficient</code> (SAC) - Determines the relative importance of the   entropy term. This value is adjusted automatically so that the agent retains   some amount of randomness during training.</p> </li> <li> <p><code>Policy/Extrinsic Reward</code> (PPO; SAC) - This corresponds to the mean cumulative   reward received from the environment per-episode.</p> </li> <li> <p><code>Policy/Value Estimate</code> (PPO; SAC) - The mean value estimate for all states   visited by the agent. Should increase during a successful training session.</p> </li> <li> <p><code>Policy/Curiosity Reward</code> (PPO/SAC+Curiosity) - This corresponds to the mean   cumulative intrinsic reward generated per-episode.</p> </li> <li> <p><code>Policy/Curiosity Value Estimate</code> (PPO/SAC+Curiosity) - The agent's value   estimate for the curiosity reward.</p> </li> <li> <p><code>Policy/GAIL Reward</code> (PPO/SAC+GAIL) - This corresponds to the mean cumulative   discriminator-based reward generated per-episode.</p> </li> <li> <p><code>Policy/GAIL Value Estimate</code> (PPO/SAC+GAIL) - The agent's value estimate for   the GAIL reward.</p> </li> <li> <p><code>Policy/GAIL Policy Estimate</code> (PPO/SAC+GAIL) - The discriminator's estimate   for states and actions generated by the policy.</p> </li> <li> <p><code>Policy/GAIL Expert Estimate</code> (PPO/SAC+GAIL) - The discriminator's estimate   for states and actions drawn from expert demonstrations.</p> </li> </ul>"},{"location":"Using-Tensorboard/#learning-loss-functions","title":"Learning Loss Functions","text":"<ul> <li> <p><code>Losses/Policy Loss</code> (PPO; SAC) - The mean magnitude of policy loss function.   Correlates to how much the policy (process for deciding actions) is changing.   The magnitude of this should decrease during a successful training session.</p> </li> <li> <p><code>Losses/Value Loss</code> (PPO; SAC) - The mean loss of the value function update.   Correlates to how well the model is able to predict the value of each state.   This should increase while the agent is learning, and then decrease once the   reward stabilizes.</p> </li> <li> <p><code>Losses/Forward Loss</code> (PPO/SAC+Curiosity) - The mean magnitude of the forward   model loss function. Corresponds to how well the model is able to predict the   new observation encoding.</p> </li> <li> <p><code>Losses/Inverse Loss</code> (PPO/SAC+Curiosity) - The mean magnitude of the inverse   model loss function. Corresponds to how well the model is able to predict the   action taken between two observations.</p> </li> <li> <p><code>Losses/Pretraining Loss</code> (BC) - The mean magnitude of the behavioral cloning   loss. Corresponds to how well the model imitates the demonstration data.</p> </li> <li> <p><code>Losses/GAIL Loss</code> (GAIL) - The mean magnitude of the GAIL discriminator loss.   Corresponds to how well the model imitates the demonstration data.</p> </li> </ul>"},{"location":"Using-Tensorboard/#self-play","title":"Self-Play","text":"<ul> <li><code>Self-Play/ELO</code> (Self-Play) -   ELO measures the relative   skill level between two players. In a proper training run, the ELO of the   agent should steadily increase.</li> </ul>"},{"location":"Using-Tensorboard/#exporting-data-from-tensorboard","title":"Exporting Data from TensorBoard","text":"<p>To export timeseries data in CSV or JSON format, check the \"Show data download links\" in the upper left. This will enable download links below each chart.</p> <p></p>"},{"location":"Using-Tensorboard/#custom-metrics-from-unity","title":"Custom Metrics from Unity","text":"<p>To get custom metrics from a C# environment into TensorBoard, you can use the <code>StatsRecorder</code>:</p> <pre><code>var statsRecorder = Academy.Instance.StatsRecorder;\nstatsRecorder.Add(\"MyMetric\", 1.0);\n</code></pre>"},{"location":"Using-Virtual-Environment/","title":"Using Virtual Environment","text":""},{"location":"Using-Virtual-Environment/#what-is-a-virtual-environment","title":"What is a Virtual Environment?","text":"<p>A Virtual Environment is a self contained directory tree that contains a Python installation for a particular version of Python, plus a number of additional packages. To learn more about Virtual Environments see here.</p>"},{"location":"Using-Virtual-Environment/#why-should-i-use-a-virtual-environment","title":"Why should I use a Virtual Environment?","text":"<p>A Virtual Environment keeps all dependencies for the Python project separate from dependencies of other projects. This has a few advantages:</p> <ol> <li>It makes dependency management for the project easy.</li> <li>It enables using and testing of different library versions by quickly    spinning up a new environment and verifying the compatibility of the code    with the different version.</li> </ol>"},{"location":"Using-Virtual-Environment/#python-version-requirement-required","title":"Python Version Requirement (Required)","text":"<p>This guide has been tested with Python 3.10.12. Newer versions might not have support for the dependent libraries, so are not recommended.</p>"},{"location":"Using-Virtual-Environment/#use-conda-or-mamba","title":"Use Conda (or Mamba)","text":"<p>While there are many options for setting up virtual environments for python, by far the most common and simpler approach is by using Anaconda (aka Conda). You can read the documentation on how to get started with Conda here.</p>"},{"location":"Using-Virtual-Environment/#installing-pip-required","title":"Installing Pip (Required)","text":"<ol> <li>Download the <code>get-pip.py</code> file using the command    <code>curl https://bootstrap.pypa.io/get-pip.py -o get-pip.py</code></li> <li>Run the following <code>python3 get-pip.py</code></li> <li>Check pip version using <code>pip3 -V</code></li> </ol> <p>Note (for Ubuntu users): If the <code>ModuleNotFoundError: No module named 'distutils.util'</code> error is encountered, then python3-distutils needs to be installed. Install python3-distutils using <code>sudo apt-get install python3-distutils</code></p>"},{"location":"Using-Virtual-Environment/#mac-os-x-setup","title":"Mac OS X Setup","text":"<ol> <li>Create a folder where the virtual environments will reside    <code>$ mkdir ~/python-envs</code></li> <li>To create a new environment named <code>sample-env</code> execute    <code>$ python3 -m venv ~/python-envs/sample-env</code></li> <li>To activate the environment execute    <code>$ source ~/python-envs/sample-env/bin/activate</code></li> <li>Upgrade to the latest pip version using <code>$ pip3 install --upgrade pip</code></li> <li>Upgrade to the latest setuptools version using    <code>$ pip3 install --upgrade setuptools</code></li> <li>To deactivate the environment execute <code>$ deactivate</code> (you can reactivate the    environment using the same <code>activate</code> command listed above)</li> </ol>"},{"location":"Using-Virtual-Environment/#ubuntu-setup","title":"Ubuntu Setup","text":"<ol> <li>Install the python3-venv package using <code>$ sudo apt-get install python3-venv</code></li> <li>Follow the steps in the Mac OS X installation.</li> </ol>"},{"location":"Using-Virtual-Environment/#windows-setup","title":"Windows Setup","text":"<ol> <li>Create a folder where the virtual environments will reside <code>md python-envs</code></li> <li>To create a new environment named <code>sample-env</code> execute    <code>python -m venv python-envs\\sample-env</code></li> <li>To activate the environment execute <code>python-envs\\sample-env\\Scripts\\activate</code></li> <li>Upgrade to the latest pip version using <code>pip install --upgrade pip</code></li> <li>To deactivate the environment execute <code>deactivate</code> (you can reactivate the    environment using the same <code>activate</code> command listed above)</li> </ol> <p>Note: - Verify that you are using Python version 3.10.12. Launch a   command prompt using <code>cmd</code> and execute <code>python --version</code> to verify the version. - Python3 installation may require admin privileges on Windows. - This guide is for Windows 10 using a 64-bit architecture only.</p>"},{"location":"Versioning/","title":"ML-Agents Versioning","text":""},{"location":"Versioning/#context","title":"Context","text":"<p>As the ML-Agents project evolves into a more mature product, we want to communicate the process we use to version our packages and the data that flows into, through, and out of them clearly. Our project now has four packages (1 Unity, 3 Python) along with artifacts that are produced as well as consumed.  This document covers the versioning for these packages and artifacts.</p>"},{"location":"Versioning/#github-releases","title":"GitHub Releases","text":"<p>Up until now, all packages were in lockstep in-terms of versioning. As a result, the GitHub releases were tagged with the version of all those packages (e.g. v0.15.0, v0.15.1) and labeled accordingly. With the decoupling of package versions, we now need to revisit our GitHub release tagging. The proposal is that we move towards an integer release numbering for our repo and each such release will call out specific version upgrades of each package. For instance, with the April 30th release, we will have: - GitHub Release 1 (branch name: release_1_branch)   - com.unity.ml-agents release 1.0.0   - ml-agents release 0.16.0   - ml-agents-envs release 0.16.0   - gym-unity release 0.16.0</p> <p>Our release cadence will not be affected by these versioning changes.  We will keep having monthly releases to fix bugs and release new features.</p>"},{"location":"Versioning/#packages","title":"Packages","text":"<p>All of the software packages, and their generated artifacts will be versioned.  Any automation tools will not be versioned.</p>"},{"location":"Versioning/#unity-package","title":"Unity package","text":"<p>Package name: com.unity.ml-agents - Versioned following Semantic Versioning Guidelines - This package consumes an artifact of the training process: the <code>.nn</code> file.  These files     are integer versioned and currently at version 2. The com.unity.ml-agents package     will need to support the version of <code>.nn</code> files which existed at its 1.0.0 release.     For example, consider that com.unity.ml-agents is at version 1.0.0 and the NN files     are at version 2.  If the NN files change to version 3, the next release of     com.unity.ml-agents at version 1.1.0 guarantees it will be able to read both of these     formats.  If the NN files were to change to version 4 and com.unity.ml-agents to     version 2.0.0, support for NN versions 2 and 3 could be dropped for com.unity.ml-agents     version 2.0.0. - This package produces one artifact, the <code>.demo</code> files.  These files will have integer     versioning. This means their version will increment by 1 at each change.  The     com.unity.ml-agents package must be backward compatible with version changes     that occur between minor versions. - To summarize, the artifacts produced and consumed by com.unity.ml-agents are guaranteed     to be supported for 1.x.x versions of com.unity.ml-agents.  We intend to provide stability     for our users by moving to a 1.0.0 release of com.unity.ml-agents.</p>"},{"location":"Versioning/#python-packages","title":"Python Packages","text":"<p>Package names: ml-agents / ml-agents-envs / gym-unity - The python packages remain in \"Beta.\"  This means that breaking changes to the public     API of the python packages can change without having to have a major version bump.     Historically, the python and C# packages were in version lockstep.  This is no longer     the case.  The python packages will remain in lockstep with each other for now, while the     C# package will follow its own versioning as is appropriate.  However, the python package     versions may diverge in the future. - While the python packages will remain in Beta for now, we acknowledge that the most     heavily used portion of our python interface is the <code>mlagents-learn</code> CLI and strive     to make this part of our API backward compatible. We are actively working on this and     expect to have a stable CLI in the next few weeks.</p>"},{"location":"Versioning/#communicator","title":"Communicator","text":"<p>Packages which communicate: com.unity.ml-agents / ml-agents-envs</p> <p>Another entity of the ML-Agents Toolkit that requires versioning is the communication layer between C# and Python, which will follow also semantic versioning.  This guarantees a level of backward compatibility between different versions of C# and Python packages which communicate. Any Communicator version 1.x.x of the Unity package should be compatible with any 1.x.x Communicator Version in Python.</p> <p>An RLCapabilities struct keeps track of which features exist. This struct is passed from C# to Python, and another from Python to C#.  With this feature level granularity, we can notify users more specifically about feature limitations based on what's available in both C# and Python. These notifications will be logged to the python terminal, or to the Unity Editor Console.</p>"},{"location":"Versioning/#side-channels","title":"Side Channels","text":"<p>The communicator is what manages data transfer between Unity and Python for the core training loop. Side Channels are another means of data transfer between Unity and Python. Side Channels are not versioned, but have been designed to support backward compatibility for what they are. As of today, we provide 4 side channels: - FloatProperties: shared float data between Unity - Python (bidirectional) - RawBytes: raw data that can be sent Unity - Python (bidirectional) - EngineConfig: a set of numeric fields in a pre-defined order sent from Python to Unity - Stats: (name, value, agg) messages sent from Unity to Python</p> <p>Aside from the specific implementations of side channels we provide (and use ourselves), the Side Channel interface is made available for users to create their own custom side channels. As such, we guarantee that the built in SideChannel interface between Unity and Python is backward compatible in packages that share the same major version.</p>"},{"location":"com.unity.ml-agents.extensions/","title":"About ML-Agents Extensions package (<code>com.unity.ml-agents.extensions</code>)","text":"<p>The Unity ML-Agents Extensions package contains optional add-ons to the C# SDK for the Unity ML-Agents Toolkit.</p> <p>These extensions are all considered experimental, and their API or behavior may change between versions.</p>"},{"location":"com.unity.ml-agents.extensions/#package-contents","title":"Package contents","text":"<p>The following table describes the package folder structure:</p> Location Description Documentation~ Contains the documentation for the Unity package. Editor Contains utilities for Editor windows and drawers. Runtime Contains core C# APIs for integrating ML-Agents into your Unity scene. Tests Contains the unit tests for the package. <p>The Runtime directory currently contains these features:  * Physics-based sensors  * Input System Package Integration  * Custom Grid-based Sensors</p>"},{"location":"com.unity.ml-agents.extensions/#installation","title":"Installation","text":"<p>The ML-Agents Extensions package is not currently available in the Package Manager. There are two recommended ways to install the package:</p>"},{"location":"com.unity.ml-agents.extensions/#local-installation","title":"Local Installation","text":"<p>Clone the repository and follow the Local Installation for Development directions (substituting <code>com.unity.ml-agents.extensions</code> for the package name).</p>"},{"location":"com.unity.ml-agents.extensions/#github-via-package-manager","title":"Github via Package Manager","text":"<p>In Unity 2019.4 or later, open the Package Manager, hit the \"+\" button, and select \"Add package from git URL\".</p> <p></p> <p>In the dialog that appears, enter  ``` git+https://github.com/Unity-Technologies/ml-agents.git?path=com.unity.ml-agents.extensions#release_21</p> <pre><code>\nYou can also edit your project's `manifest.json` directly and add the following line to the `dependencies`\nsection:\n</code></pre> <p>\"com.unity.ml-agents.extensions\": \"git+https://github.com/Unity-Technologies/ml-agents.git?path=com.unity.ml-agents.extensions#release_21\", ``` See Git dependencies for more information. Note that this may take several minutes to resolve the packages the first time that you add it.</p>"},{"location":"com.unity.ml-agents.extensions/#requirements","title":"Requirements","text":"<p>This version of the Unity ML-Agents package is compatible with the following versions of the Unity Editor:</p> <ul> <li>2019.4 and later</li> </ul> <p>If using the <code>InputActuatorComponent</code> - install the <code>com.unity.inputsystem</code> package version <code>1.1.0-preview.3</code> or later.</p>"},{"location":"com.unity.ml-agents.extensions/#known-limitations","title":"Known Limitations","text":"<ul> <li>For the <code>InputActuatorComponent</code><ul> <li>Limited implementation of <code>InputControls</code></li> <li>No way to customize the action space of the <code>InputActuatorComponent</code></li> </ul> </li> </ul>"},{"location":"com.unity.ml-agents.extensions/#need-help","title":"Need Help?","text":"<p>The main README contains links for contacting the team or getting support.</p>"},{"location":"com.unity.ml-agents/","title":"About ML-Agents package (<code>com.unity.ml-agents</code>)","text":"<p>The ML-Agents package contains the primary C# SDK for the Unity ML-Agents Toolkit.</p> <p>The package allows you to convert any Unity scene into a learning environment and train character behaviors using a variety of machine learning algorithms. Additionally, it allows you to embed these trained behaviors back into Unity scenes to control your characters. More specifically, the package provides the following core functionalities:</p> <ul> <li>Define Agents: entities, or characters, whose behavior will be learned. Agents   are entities that generate observations (through sensors), take actions, and   receive rewards from the environment.</li> <li>Define Behaviors: entities that specify how an agent should act. Multiple   agents can share the same Behavior and a scene may have multiple Behaviors.</li> <li>Record demonstrations of an agent within the Editor. You can use   demonstrations to help train a behavior for that agent.</li> <li>Embedding a trained behavior into the scene via the Unity Inference Engine.   Embedded behaviors allow you to switch an Agent between learning and   inference.</li> </ul> <p>Note that the ML-Agents package does not contain the machine learning algorithms for training behaviors. The ML-Agents package only supports instrumenting a Unity scene, setting it up for training, and then embedding the trained model back into your Unity scene. The machine learning algorithms that orchestrate training are part of the companion Python package.</p> <p>Note that we also provide an ML-Agents Extensions package (<code>com.unity.ml-agents.extensions</code>) that contains early/experimental features that you may find useful. This package is only available from the ML-Agents GitHub repo.</p>"},{"location":"com.unity.ml-agents/#package-contents","title":"Package contents","text":"<p>The following table describes the package folder structure:</p> Location Description Documentation~ Contains the documentation for the Unity package. Editor Contains utilities for Editor windows and drawers. Plugins Contains third-party DLLs. Runtime Contains core C# APIs for integrating ML-Agents into your Unity scene. Runtime/Integrations Contains utilities for integrating ML-Agents into specific game genres. Tests Contains the unit tests for the package. <p></p>"},{"location":"com.unity.ml-agents/#installation","title":"Installation","text":"<p>To install this ML-Agents package, follow the instructions in the Package Manager documentation.</p> <p>To install the companion Python package to enable training behaviors, follow the installation instructions on our GitHub repository.</p>"},{"location":"com.unity.ml-agents/#advanced-installation","title":"Advanced Installation","text":"<p>With the changes to Unity Package Manager in 2021, experimental packages will not show up in the package list and have to be installed manually. There are two recommended ways to install the package manually:</p>"},{"location":"com.unity.ml-agents/#github-via-package-manager","title":"Github via Package Manager","text":"<p>In Unity 2019.4 or later, open the Package Manager, hit the \"+\" button, and select \"Add package from git URL\".</p> <p></p> <p>In the dialog that appears, enter</p> <pre><code>git+https://github.com/Unity-Technologies/ml-agents.git?path=com.unity.ml-agents#release_21\n</code></pre> <p>You can also edit your project's <code>manifest.json</code> directly and add the following line to the <code>dependencies</code> section:</p> <pre><code>\"com.unity.ml-agents\": \"git+https://github.com/Unity-Technologies/ml-agents.git?path=com.unity.ml-agents#release_21\",\n</code></pre> <p>See Git dependencies for more information. Note that this may take several minutes to resolve the packages the first time that you add it.</p>"},{"location":"com.unity.ml-agents/#local-installation-for-development","title":"Local Installation for Development","text":"<p>Clone the repository and follow the Local Installation for Development directions.</p>"},{"location":"com.unity.ml-agents/#requirements","title":"Requirements","text":"<p>This version of the Unity ML-Agents package is compatible with the following versions of the Unity Editor:</p> <ul> <li>2019.4 and later</li> </ul>"},{"location":"com.unity.ml-agents/#known-limitations","title":"Known Limitations","text":""},{"location":"com.unity.ml-agents/#training","title":"Training","text":"<p>Training is limited to the Unity Editor and Standalone builds on Windows, MacOS, and Linux with the Mono scripting backend. Currently, training does not work with the IL2CPP scripting backend. Your environment will default to inference mode if training is not supported or is not currently running.</p>"},{"location":"com.unity.ml-agents/#inference","title":"Inference","text":"<p>Inference is executed via the Unity Inference Engine.</p> <p>CPU</p> <p>All platforms supported.</p> <p>GPU</p> <p>All platforms supported except:</p> <ul> <li>WebGL and GLES 3/2 on Android / iPhone</li> </ul> <p>NOTE: Mobile platform support includes:</p> <ul> <li>Vulkan for Android</li> <li>Metal for iOS.</li> </ul>"},{"location":"com.unity.ml-agents/#headless-mode","title":"Headless Mode","text":"<p>If you enable Headless mode, you will not be able to collect visual observations from your agents.</p>"},{"location":"com.unity.ml-agents/#rendering-speed-and-synchronization","title":"Rendering Speed and Synchronization","text":"<p>Currently the speed of the game physics can only be increased to 100x real-time. The Academy also moves in time with FixedUpdate() rather than Update(), so game behavior implemented in Update() may be out of sync with the agent decision making. See Execution Order of Event Functions for more information.</p> <p>You can control the frequency of Academy stepping by calling <code>Academy.Instance.DisableAutomaticStepping()</code>, and then calling <code>Academy.Instance.EnvironmentStep()</code></p>"},{"location":"com.unity.ml-agents/#unity-inference-engine-models","title":"Unity Inference Engine Models","text":"<p>Currently, only models created with our trainers are supported for running ML-Agents with a neural network behavior.</p>"},{"location":"com.unity.ml-agents/#helpful-links","title":"Helpful links","text":"<p>If you are new to the Unity ML-Agents package, or have a question after reading the documentation, you can checkout our GitHub Repository, which also includes a number of ways to connect with us including our ML-Agents Forum.</p> <p>In order to improve the developer experience for Unity ML-Agents Toolkit, we have added in-editor analytics. Please refer to \"Information that is passively collected by Unity\" in the Unity Privacy Policy.</p>"},{"location":"ml-agents-envs/","title":"Unity ML-Agents Python Interface","text":"<p>The <code>mlagents_envs</code> Python package is part of the ML-Agents Toolkit. <code>mlagents_envs</code> provides three Python APIs that allows direct interaction with the Unity game engine: - A single agent API (Gym API) - A gym-like multi-agent API (PettingZoo API) - A low-level API (LLAPI)</p> <p>The LLAPI is used by the trainer implementation in <code>mlagents</code>. <code>mlagents_envs</code> can be used independently of <code>mlagents</code> for Python communication.</p>"},{"location":"ml-agents-envs/#installation","title":"Installation","text":"<p>Install the <code>mlagents_envs</code> package with:</p> <pre><code>python -m pip install mlagents_envs==1.0.0\n</code></pre>"},{"location":"ml-agents-envs/#usage-more-information","title":"Usage &amp; More Information","text":"<p>See - Gym API Guide - PettingZoo API Guide - Python API Guide</p> <p>for more information on how to use the API to interact with a Unity environment.</p> <p>For more information on the ML-Agents Toolkit and how to instrument a Unity scene with the ML-Agents SDK, check out the main ML-Agents Toolkit documentation.</p>"},{"location":"ml-agents-envs/#limitations","title":"Limitations","text":"<ul> <li><code>mlagents_envs</code> uses localhost ports to exchange data between Unity and   Python. As such, multiple instances can have their ports collide, leading to   errors. Make sure to use a different port if you are using multiple instances   of <code>UnityEnvironment</code>.</li> <li>Communication between Unity and the Python <code>UnityEnvironment</code> is not secure.</li> <li>On Linux, ports are not released immediately after the communication closes.   As such, you cannot reuse ports right after closing a <code>UnityEnvironment</code>.</li> </ul>"},{"location":"ml-agents/","title":"Unity ML-Agents Trainers","text":"<p>The <code>mlagents</code> Python package is part of the ML-Agents Toolkit. <code>mlagents</code> provides a set of reinforcement and imitation learning algorithms designed to be used with Unity environments. The algorithms interface with the Python API provided by the <code>mlagents_envs</code> package. See here for more information on <code>mlagents_envs</code>.</p> <p>The algorithms can be accessed using the: <code>mlagents-learn</code> access point. See here for more information on using this package.</p>"},{"location":"ml-agents/#installation","title":"Installation","text":"<p>Install the <code>mlagents</code> package with:</p> <pre><code>python -m pip install mlagents==1.0.0\n</code></pre>"},{"location":"ml-agents/#usage-more-information","title":"Usage &amp; More Information","text":"<p>For more information on the ML-Agents Toolkit and how to instrument a Unity scene with the ML-Agents SDK, check out the main ML-Agents Toolkit documentation.</p>"},{"location":"ml-agents/#limitations","title":"Limitations","text":"<ul> <li>Resuming self-play from a checkpoint resets the reported ELO to the default   value.</li> </ul>"},{"location":"doxygen/Readme/","title":"Doxygen files","text":"<p>To generate the API reference as HTML files, run:</p> <pre><code>doxygen dox-ml-agents.conf\n</code></pre>"}]}
\ No newline at end of file
diff --git a/sitemap.xml b/sitemap.xml
index 9c9f413880..08941aafdd 100644
--- a/sitemap.xml
+++ b/sitemap.xml
@@ -2,292 +2,292 @@
 <urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/API-Reference/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Background-Machine-Learning/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Background-PyTorch/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Background-Unity/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/CODE_OF_CONDUCT/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/CONTRIBUTING/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Custom-SideChannels/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/ELO-Rating-System/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/FAQ/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Getting-Started/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Glossary/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Hugging-Face-Integration/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Installation-Anaconda-Windows/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Installation/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Integrations-Match3/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Integrations/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/LICENSE/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Learning-Environment-Create-New/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Learning-Environment-Design-Agents/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Learning-Environment-Design/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Learning-Environment-Examples/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Learning-Environment-Executable/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Limitations/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/ML-Agents-Envs-README/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/ML-Agents-Overview/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/ML-Agents-README/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/ML-Agents-Toolkit-Documentation/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Migrating/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Package-Settings/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Profiling-Python/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Python-Custom-Trainer-Plugin/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Python-Gym-API-Documentation/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Python-Gym-API/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Python-LLAPI-Documentation/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Python-LLAPI/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Python-On-Off-Policy-Trainer-Documentation/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Python-Optimizer-Documentation/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Python-PettingZoo-API-Documentation/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Python-PettingZoo-API/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Readme/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Sentis/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Training-Configuration-File/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Training-ML-Agents/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Training-Plugins/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Training-on-Amazon-Web-Service/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Training-on-Microsoft-Azure/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Tutorial-Custom-Trainer-Plugin/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Unity-Environment-Registry/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Using-Docker/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Using-Tensorboard/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Using-Virtual-Environment/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/Versioning/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/com.unity.ml-agents.extensions/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/com.unity.ml-agents/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/ml-agents-envs/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/ml-agents/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
     <url>
          <loc>https://unity-technologies.github.io/ml-agents/doxygen/Readme/</loc>
-         <lastmod>2023-11-26</lastmod>
+         <lastmod>2024-04-02</lastmod>
          <changefreq>daily</changefreq>
     </url>
 </urlset>
\ No newline at end of file
diff --git a/sitemap.xml.gz b/sitemap.xml.gz
index f50c70ce7b3328f25b00296af332cc3ec41d1c4c..fada96f7bd7dc3f2e63b7fa68c52df039942118b 100644
GIT binary patch
literal 800
zcmV+*1K<1~iwFpR)C*<;|8r?{Wo=<_E_iKh0M(k`Z`v>r$KU%`Al{vH-5$1yQc(yU
z5<fBst+bcQI7edP_?Ybg>0du5L5xb%o~B8MB4RAZe7@#<xjWK(I2EdM0AnTU`|gL}
zw%Y+MFqJyH@2-<E``mr#-}FkOEVzyztnGa_cW(bR43EcSV5NhC?`?o43$4ig2OFL{
z!`t9Xw|~>|uX=*E)x3HNMZccK`e{lnUBw)TT%*EFf(<h1@^TYM3=7402HM%M7uFYD
z&%v!LFzw&n-hE=X{<!l)&V5(IM9y^vTLVA)DVM71g;zZ-TzH}T2b_a=!j`ZF<JS?$
zP(O09%M6w}Wiu{vsew&_8}03YqK3Fi(1@HO`ntY?4ES&{ie4Av*WqG5S`QOK-1B5P
zd0Z!x`4b)J(mE{IN~SRMU!+0Naf{J(!Is>4f5KLk_4fxd<m172GOSPF+9h_}I7sPm
zE3}pyI)3k4nPv4tkGTLg(GD`>jT<!?0TXQ<SE_+#1I-2Ml(QG9Q#@LdDhNSe%*Fa>
z%ZdCWNp>>O)5$QJuOd3U=g6By`XCMJ!uPVF0q%g!;YeG@2&~M!%b}Ik{Y46G4bh1g
zdj*!d3J|5^8$~DMp2|WxVvck+J@4Ck@hw|MgV8J^NWVCMIY>gD^fx-$HBS)LPI{+_
zqOA8Ut(a{lo2Z9SH7w?0=bLQRri65yOl)u~Rqf&9%H@P?q{U_DDotXz)L>YwN}nlG
z6Y*3Ptgd9!+^z3LP{W>1$*+7Pfg1F}AKNX9QAtr<_Alw*VX=2U%zUFRBW~|uKM?6V
zrT1FM&@X|f1)c0_<=-Wpn5S_zb;vVBy_H$nR@m5B5=v+<8ZI$EJ@htMBM4p>@eBUu
zPcN|XF~3i51(1|6lft0Itz(0?(h$;6Iy90i;_-j;_`$o{+7dEp9iji`*0we!A0+I+
z5Q9!YTQnQwCM8DxER8F@^<P;I83oVn^KX=@I|7TK85&&r0ysHuKBNlDe~>|j{nBEf
e4+OEv5T|&meCxlZAkE2w>;D1@-6X^wD*yl^9E_L%

literal 801
zcmV++1K#`}iwFqfFk@u`|8r?{Wo=<_E_iKh0M(jpZ`v>r$KUr;Al^IagEVaurJ@iz
zBpxydt+Y3lagM~o@pZNXq@R9Hf*6&iy_zOngov>m^ZT3g<?cxD;aI579*mW!@4Gj_
zb+-drU@CQX-(4qT_PP7ezv`7nS#TXcSljz<?%dngFgzR%ft3yl{$>L-S!hM>KiKfp
z8D0lpy8Ww;Kk5nER&(?eihezd_0yDEx{5gvxkiPV1RG@1<>e-j7#520479UhFRU-R
zoP%3eVA{XEzP)2NH|+M4A9Cuu7$$PAGuRsV*-yDtRWH2gY2m^%-9O+Q#1poJEf~L!
zK!*B}i(O{0)G3>BkxLD13fyRK2NX5LRf0z36w%l96=cANi&6Bt7{3k|^U->k5aOOE
z%gN(9narQ)K$q5G!B#Sbq5mQcijG^1rVF;@&ifO#s;s|1kRcxrzLQ~n0@p6F<HkWs
zhg+ew+|coR-^whj7kbPEu!(k%8E@RE$q1Ne>$p-4G#h9xP^X-|NS)%rl2kzm`fM)N
zM_W$hcO==#Ku;&bXugW*@SY-X7U{h-s0-iAh6cC;HirXk9V4(Z^Dc*0R`(Yvv^7LW
zUc4!=)K!2e9p5NA8TV8c(h+l{v*~Hy){Af1G8&9#5kdOJ9?V`6@}$4f$*y^VsCLpj
zO%!FlXKBT3GucEvgsNdN7dzi%t2QO1<78rkTd8UfA6G6XWFsvuJ5^~C!=(npVpaM~
zk(!97s$g{`o91r)D1sXHbV`2Z8wu2)7yh-~vKW;V)p`Gt{v8%?&WD-r)Mdo&J?sY}
zeW&zM3mN()@U)<lU9J4Pq!aTr&ZZ7|hN!nPE87Yi8%sh7?ODS)=BJ0=25SVt>mq)^
z-~H(YHa_O}>8${gGG<a3w77L_@Ln218cK&oaz#A;ZyrDRP+MC<Caoj%-`v{PrsRW!
z9T;NJ322LEgWROV$e*QgrMLbI%ORuSxqbePQgugQ5i~=Cb6)^Q=go&yLHQ3d$grPV
f4D_BLHW}g+kCkuzw-lr~S#bSdbhlDh9xDI<XXl9{