From 591b7e0edd792794c9521e1d77c8e550e70bd200 Mon Sep 17 00:00:00 2001
From: Yedidya Feldblum <yfeldblum@meta.com>
Date: Tue, 8 Oct 2024 03:19:33 -0700
Subject: [PATCH] let RegexMatchCache store only unique hashes of regexes

Summary: `RegexMatchCache` does not need to store full regex strings. It needs only some unique identifier of the regex, and cryptographic hashes are commonly used for this purpose.

Reviewed By: mdas7

Differential Revision: D63338698

fbshipit-source-id: af9093b1db4bff2dc30607e7bb31b743ee7510d9
---
 fb303/CallbackValuesMap-inl.h      | 2 +-
 fb303/CallbackValuesMap.h          | 5 +++--
 fb303/ServiceData.cpp              | 7 ++++---
 fb303/detail/QuantileStatMap-inl.h | 2 +-
 fb303/detail/QuantileStatMap.h     | 5 +++--
 fb303/detail/RegexUtil.cpp         | 2 +-
 fb303/detail/RegexUtil.h           | 4 ++--
 7 files changed, 15 insertions(+), 12 deletions(-)

diff --git a/fb303/CallbackValuesMap-inl.h b/fb303/CallbackValuesMap-inl.h
index 34cebd018..9981a0783 100644
--- a/fb303/CallbackValuesMap-inl.h
+++ b/fb303/CallbackValuesMap-inl.h
@@ -77,7 +77,7 @@ void CallbackValuesMap<T>::getKeys(std::vector<std::string>* keys) const {
 template <typename T>
 void CallbackValuesMap<T>::getRegexKeys(
     std::vector<std::string>& keys,
-    const std::string& regex,
+    const folly::RegexMatchCache::regex_key_and_view& regex,
     const folly::RegexMatchCache::time_point now) const {
   detail::cachedFindMatches(keys, callbackMap_, regex, now);
 }
diff --git a/fb303/CallbackValuesMap.h b/fb303/CallbackValuesMap.h
index cbf330f1d..6d3924c49 100644
--- a/fb303/CallbackValuesMap.h
+++ b/fb303/CallbackValuesMap.h
@@ -59,12 +59,13 @@ class CallbackValuesMap {
   /* Returns the keys in the map that matches regex pattern */
   void getRegexKeys(std::vector<std::string>& keys, const std::string& regex)
       const {
+    const auto key = folly::RegexMatchCache::regex_key_and_view(regex);
     const auto now = folly::RegexMatchCache::clock::now();
-    getRegexKeys(keys, regex, now);
+    getRegexKeys(keys, key, now);
   }
   void getRegexKeys(
       std::vector<std::string>& keys,
-      const std::string& regex,
+      const folly::RegexMatchCache::regex_key_and_view& regex,
       const folly::RegexMatchCache::time_point now) const;
 
   /** Returns the number of keys present in the map */
diff --git a/fb303/ServiceData.cpp b/fb303/ServiceData.cpp
index f5420047e..dba4c25d6 100644
--- a/fb303/ServiceData.cpp
+++ b/fb303/ServiceData.cpp
@@ -486,11 +486,12 @@ void ServiceData::getRegexCounters(
 void ServiceData::getRegexCountersOptimized(
     map<string, int64_t>& output,
     const string& regex) const {
+  const auto key = folly::RegexMatchCache::regex_key_and_view(regex);
   const auto now = folly::RegexMatchCache::clock::now();
   std::vector<std::string> keys;
-  detail::cachedFindMatches(keys, counters_, regex, now);
-  quantileMap_.getRegexKeys(keys, regex, now);
-  dynamicCounters_.getRegexKeys(keys, regex, now);
+  detail::cachedFindMatches(keys, counters_, key, now);
+  quantileMap_.getRegexKeys(keys, key, now);
+  dynamicCounters_.getRegexKeys(keys, key, now);
   getSelectedCounters(output, keys);
 }
 
diff --git a/fb303/detail/QuantileStatMap-inl.h b/fb303/detail/QuantileStatMap-inl.h
index 4cc87a73b..b8177ba69 100644
--- a/fb303/detail/QuantileStatMap-inl.h
+++ b/fb303/detail/QuantileStatMap-inl.h
@@ -179,7 +179,7 @@ void BasicQuantileStatMap<ClockT>::getKeys(
 template <typename ClockT>
 void BasicQuantileStatMap<ClockT>::getRegexKeys(
     std::vector<std::string>& keys,
-    const std::string& regex,
+    const folly::RegexMatchCache::regex_key_and_view& regex,
     const folly::RegexMatchCache::time_point now) const {
   detail::cachedFindMatches(keys, counters_, regex, now);
 }
diff --git a/fb303/detail/QuantileStatMap.h b/fb303/detail/QuantileStatMap.h
index af91fd154..84e6c325e 100644
--- a/fb303/detail/QuantileStatMap.h
+++ b/fb303/detail/QuantileStatMap.h
@@ -78,12 +78,13 @@ class BasicQuantileStatMap {
   /* Returns the keys in the map that matches regex pattern */
   void getRegexKeys(std::vector<std::string>& keys, const std::string& regex)
       const {
+    const auto key = folly::RegexMatchCache::regex_key_and_view(regex);
     const auto now = folly::RegexMatchCache::clock::now();
-    getRegexKeys(keys, regex, now);
+    getRegexKeys(keys, key, now);
   }
   void getRegexKeys(
       std::vector<std::string>& keys,
-      const std::string& regex,
+      const folly::RegexMatchCache::regex_key_and_view& regex,
       const folly::RegexMatchCache::time_point now) const;
 
   size_t getNumKeys() const;
diff --git a/fb303/detail/RegexUtil.cpp b/fb303/detail/RegexUtil.cpp
index 926e108a2..389c82e61 100644
--- a/fb303/detail/RegexUtil.cpp
+++ b/fb303/detail/RegexUtil.cpp
@@ -23,7 +23,7 @@ namespace facebook::fb303::detail {
 void cachedFindMatchesCopyUnderSharedLock(
     std::vector<std::string>& out,
     folly::RegexMatchCache const& cache,
-    std::string_view const regex,
+    folly::RegexMatchCacheKeyAndView const& regex,
     folly::RegexMatchCache::time_point const now) {
   auto const matches = cache.findMatchesUnsafe(regex, now);
   folly::grow_capacity_by(out, matches.size());
diff --git a/fb303/detail/RegexUtil.h b/fb303/detail/RegexUtil.h
index 5709c35cd..aa801de6b 100644
--- a/fb303/detail/RegexUtil.h
+++ b/fb303/detail/RegexUtil.h
@@ -65,14 +65,14 @@ void cachedClearStrings(Map& map) {
 void cachedFindMatchesCopyUnderSharedLock(
     std::vector<std::string>& out,
     folly::RegexMatchCache const& cache,
-    std::string_view regex,
+    folly::RegexMatchCacheKeyAndView const& regex,
     folly::RegexMatchCache::time_point now);
 
 template <typename SyncMap>
 void cachedFindMatches(
     std::vector<std::string>& out,
     SyncMap& map,
-    std::string_view const regex,
+    folly::RegexMatchCacheKeyAndView const& regex,
     folly::RegexMatchCache::time_point const now) {
   auto r = map.rlock();
   if (!r->matches.isReadyToFindMatches(regex)) {