Merge pull request #35 from axiom-data-science/caching_optional

made caching optional with config flag

docs/whats_new.md

@@ -1,5 +1,17 @@
 # What's New
 
+## v0.9.6 (November 15, 2024)
+
+* made caching directory creation and saving to cache optional with input option `use_cache`.
+
+## v0.9.5 (November 14, 2024)
+
+* fixed error in output file
+
+## v0.9.4 (November 14, 2024)
+
+* Updated locations for local model output.
+
 ## v0.9.3 (November 13, 2024)
 
 * Moved `output_format` parameter to manager config from model config

particle_tracking_manager/models/opendrift/opendrift.py

@@ -11,8 +11,6 @@
 from pathlib import Path
 from typing import Optional, Union
 
-import appdirs
-
 # using my own version of ROMS reader
 # from .reader_ROMS_native import Reader
 import pandas as pd
@@ -858,22 +856,11 @@ def run_add_reader(
                     "start_time and end_time must be set to narrow model output to simulation time"
                 )
 
-            # save interpolators to save time
-            cache_dir = Path(
-                appdirs.user_cache_dir(
-                    appname="particle-tracking-manager", appauthor="axiom-data-science"
-                )
-            )
-            cache_dir.mkdir(parents=True, exist_ok=True)
-            self.cache_dir = cache_dir
-            self.interpolator_filename = cache_dir / Path(
-                f"{self.ocean_model}_interpolator"
-            )
             reader = reader_ROMS_native.Reader(
                 filename=ds,
                 name=self.ocean_model,
                 standard_name_mapping=standard_name_mapping,
-                save_interpolator=True,
+                save_interpolator=self.save_interpolator,
                 interpolator_filename=self.interpolator_filename,
             )
 

particle_tracking_manager/the_manager.py

@@ -1,14 +1,15 @@
 """Contains logic for configuring particle tracking simulations."""
 
 
-# from docstring_inheritance import NumpyDocstringInheritanceMeta
 import datetime
 import json
 import logging
 import pathlib
 
 from typing import Optional, Union
 
+# from docstring_inheritance import NumpyDocstringInheritanceMeta
+import appdirs
 import pandas as pd
 
 from .cli import is_None
@@ -139,6 +140,8 @@ class ParticleTrackingManager:
         Name of output file to save, by default None. If None, default is set in the model. Without any suffix.
     output_format : str, default "netcdf"
         Name of input/output module type to use for writing Lagrangian model output. Default is "netcdf".
+    use_cache : bool
+        Set to True to use cache for saving interpolators, by default True.
 
     Notes
     -----
@@ -165,6 +168,7 @@ class ParticleTrackingManager:
     output_file: str
     output_format: str
     output_file_initial: Optional[str]
+    interpolator_filename: Optional[pathlib.Path]
 
     def __init__(
         self,
@@ -193,6 +197,7 @@ def __init__(
         use_static_masks: bool = config_ptm["use_static_masks"]["default"],
         output_file: Optional[str] = config_ptm["output_file"]["default"],
         output_format: str = config_ptm["output_format"]["default"],
+        use_cache: bool = config_ptm["use_cache"]["default"],
         **kw,
     ) -> None:
         """Inputs necessary for any particle tracking."""
@@ -466,6 +471,29 @@ def __setattr__(self, name: str, value) -> None:
                 )
                 self.seed_seafloor = False
 
+            # save interpolators to save time
+            if name == "use_cache":
+                if value:
+                    cache_dir = pathlib.Path(
+                        appdirs.user_cache_dir(
+                            appname="particle-tracking-manager",
+                            appauthor="axiom-data-science",
+                        )
+                    )
+                    cache_dir.mkdir(parents=True, exist_ok=True)
+                    self.cache_dir = cache_dir
+                    self.interpolator_filename = cache_dir / pathlib.Path(
+                        f"{self.ocean_model}_interpolator"
+                    )
+                    self.save_interpolator = True
+                    self.logger.info(
+                        f"Interpolators will be saved to {self.interpolator_filename}."
+                    )
+                else:
+                    self.save_interpolator = False
+                    self.interpolator_filename = None
+                    self.logger.info("Interpolators will not be saved.")
+
             # if reader, lon, and lat set, check inputs
             if (
                 name == "has_added_reader"

particle_tracking_manager/the_manager_config.json

@@ -187,5 +187,11 @@
         "default": "netcdf",
         "description": "Output file format. Options are \"netcdf\" or \"parquet\".",
         "ptm_level": 2
+    },
+    "use_cache": {
+        "type": "bool",
+        "default": true,
+        "description": "Set to True to use cache for storing interpolators. This saves time on repeat simulations, may be used for other items in the future.",
+        "ptm_level": 3
     }
 }

tests/test_manager.py

@@ -240,6 +240,13 @@ def test_input_too_many_end_of_simulation():
         )
 
 
+def test_no_cache():
+    """Test having no cache"""
+
+    m = ptm.OpenDriftModel(use_cache=False)
+    assert m.interpolator_filename is None
+
+
 def test_changing_end_of_simulation():
     """change end_time, steps, and duration