fix: use distinct cache for test assets

.gitignore

@@ -8,6 +8,7 @@ dist
 .idea
 robotoff.egg-info
 ml/data/*
+cache
 datasets
 *.joblib
 .ipynb_checkpoints
@@ -41,5 +42,4 @@ coverage.xml
 site/
 gh_pages/
 doc/README.md
-doc/references/cli.md 
-data/diskcache
+doc/references/cli.md

docker-compose.yml

@@ -3,6 +3,7 @@ version: "3.9"
 x-robotoff-base-volumes:
   &robotoff-base-volumes
   - ./data:/opt/robotoff/data
+  - ./cache:/opt/robotoff/cache
   - ./datasets:/opt/robotoff/datasets
   - ./tf_models:/opt/robotoff/tf_models
   - ./models:/opt/robotoff/models

docker/dev.yml

@@ -23,6 +23,7 @@ x-robotoff-dev: &robotoff-dev
       - ./i18n:/opt/robotoff/i18n
       # make data available
       - ./data:/opt/robotoff/data
+      - ./cache:/opt/robotoff/cache
       # make migration files available
       - ./migrations:/opt/robotoff/migrations
       # make doc generation available

robotoff/settings.py

@@ -123,6 +123,7 @@ def event_api() -> str:
 
 PROJECT_DIR = Path(__file__).parent.parent
 DATA_DIR = PROJECT_DIR / "data"
+CACHE_DIR = PROJECT_DIR / "cache"
 DATASET_DIR = PROJECT_DIR / "datasets"
 DATASET_DIR.mkdir(exist_ok=True)
 I18N_DIR = PROJECT_DIR / "i18n"
@@ -338,5 +339,8 @@ def get_package_version() -> str:
 MIGRATE_DIR = PROJECT_DIR / "migrations"
 
 
-# Path of the local disk cache, see robotoff.cache for more information
-DISKCACHE_DIR = DATA_DIR / "diskcache"
+# Path of the main local disk cache, see robotoff.cache for more information
+DISKCACHE_DIR = CACHE_DIR / "diskcache"
+
+# Path of the local disk cache used for tests
+TESTS_DISKCACHE_DIR = CACHE_DIR / "diskcache_tests_assets"

tests/unit/pytest_utils.py

@@ -1,9 +1,13 @@
 import json
 
+from diskcache import Cache
 from openfoodfacts import OCRResult
 
+from robotoff import settings
 from robotoff.utils.download import cache_asset_from_url
 
+test_cache = Cache(settings.TESTS_DISKCACHE_DIR)
+
 
 def get_asset(asset_path: str) -> bytes | None:
     """Get an asset from the test-data repository.
@@ -15,7 +19,9 @@ def get_asset(asset_path: str) -> bytes | None:
         fetched
     """
     asset_url = f"https://raw.githubusercontent.com/openfoodfacts/test-data{asset_path}"
-    return cache_asset_from_url(key=asset_url, tag="test", asset_url=asset_url)
+    return cache_asset_from_url(
+        key=asset_url, cache=test_cache, tag="test", asset_url=asset_url
+    )
 
 
 def get_ocr_result_asset(asset_path: str) -> OCRResult | None: