Merge branch 'main' into batch-job-spellcheck

.github/workflows/container-deploy.yml

@@ -43,8 +43,7 @@ jobs:
         # This is the OSM proxy server, that does have an ipv4 address
         echo "SSH_PROXY_HOST=45.147.209.254" >> $GITHUB_ENV
         echo "SSH_USERNAME=off" >> $GITHUB_ENV
-        # We use 'raphael' user as we don't have an off user yet on the proxy machine
-        echo "SSH_PROXY_USERNAME=raphael" >> $GITHUB_ENV
+        echo "SSH_PROXY_USERNAME=off" >> $GITHUB_ENV
         echo "SSH_PROTOCOL=tcp" >> $GITHUB_ENV
         echo "SSH_HOST=10.3.0.200" >> $GITHUB_ENV
         echo "ROBOTOFF_INSTANCE=prod" >> $GITHUB_ENV

CHANGELOG.md

@@ -1,5 +1,24 @@
 # Changelog
 
+## [1.50.4](https://github.com/openfoodfacts/robotoff/compare/v1.50.3...v1.50.4) (2024-09-03)
+
+
+### Bug Fixes
+
+* fix product weight detection bug ([fe7e758](https://github.com/openfoodfacts/robotoff/commit/fe7e75814bbbce42e6d6ee86e73102d58c5eff1b))
+* only run nutrition table detection for food type ([c019df9](https://github.com/openfoodfacts/robotoff/commit/c019df94b99c7260308b9aeb54d43b685668a57a))
+* remove Hacendado store ([e946444](https://github.com/openfoodfacts/robotoff/commit/e946444ee1dabd67b4269d18fafa7639fbc2bac9))
+* remove unused class ([22fae32](https://github.com/openfoodfacts/robotoff/commit/22fae32f798240bdfecaf04f0a7adfd08fa41d23))
+
+
+### Technical
+
+* add tmp volume for ES ([6225bcb](https://github.com/openfoodfacts/robotoff/commit/6225bcb02273aa72f40caa0d837676cd03ccd31f))
+* add volume for /tmp ([15130d2](https://github.com/openfoodfacts/robotoff/commit/15130d21747592f816360928f93f5a5e4db84569))
+* **deps:** bump sentry-sdk from 1.14.0 to 2.8.0 ([f991bf5](https://github.com/openfoodfacts/robotoff/commit/f991bf536300660613c97f43b294b50a7740f5f0))
+* improve robotoff documentation ([b9365a7](https://github.com/openfoodfacts/robotoff/commit/b9365a7cb9c91e25ca0eb2688c95b15506fac2c4))
+* make api depends on elasticsearch in docker-compose.yml ([bf99ca8](https://github.com/openfoodfacts/robotoff/commit/bf99ca8f75b1adc8b0fa2f6032cc1e68992220d9))
+
 ## [1.50.3](https://github.com/openfoodfacts/robotoff/compare/v1.50.2...v1.50.3) (2024-08-21)
 
 

data/ocr/store_regex.txt

@@ -67,7 +67,6 @@ Franprix
 Fred Meyer
 Froiz
 Giant Eagle
-Hacendado
 Haggen
 Harris Teeter
 Harrods

docker-compose.yml

@@ -83,6 +83,7 @@ services:
     depends_on:
       - postgres
       - redis
+      - elasticsearch
 
   update-listener:
     <<: *robotoff-base

pyproject.toml

@@ -38,7 +38,7 @@ ignore_missing_imports = true
 
 [tool.poetry]
 name = "robotoff"
-version = "1.50.3"
+version = "1.50.4"
 description = "Real-time and batch prediction service for Open Food Facts."
 authors = ["Open Food Facts Team"]
 license = "GNU Affero General Public License v3"
@@ -85,7 +85,7 @@ pandas = "~2.2.2"
 pyarrow = "~17.0.0"
 
 [tool.poetry.dependencies.sentry-sdk]
-version = "~1.14.0"
+version = ">=1.14,<2.9"
 extras = ["falcon"]
 
 [tool.poetry.dev-dependencies]

robotoff/insights/annotate.py

@@ -717,7 +717,3 @@ def annotate(
         auth=auth,
         is_vote=is_vote,
     )
-
-
-class InvalidInsight(Exception):
-    pass

robotoff/prediction/ocr/product_weight.py

@@ -94,7 +94,23 @@ def is_valid_weight(weight_value: str) -> bool:
     return True
 
 
-def is_extreme_weight(normalized_value: float, unit: str) -> bool:
+def is_extreme_weight(
+    normalized_value: float, unit: str, count: int | None = None
+) -> bool:
+    """Return True if the weight is extreme, i.e is likely wrongly detected.
+
+    If considered extreme, a prediction won't be generated.
+
+    :param normalized_value: the normalized weight value
+    :param unit: the normalized weight unit
+    :param count: the number of items in the pack, if any
+    :return: True if the weight is extreme, False otherwise
+    """
+    if count is not None and int(count) > 20:
+        # More than 20 items in a pack is quite unlikely for
+        # a consumer product
+        return True
+
     if unit == "g":
         # weights above 10 kg
         return normalized_value >= 10000 or normalized_value <= 10
@@ -200,7 +216,7 @@ def process_multi_packaging(match) -> Optional[dict]:
     normalized_value, normalized_unit = normalize_weight(value, unit)
 
     # Check that the weight is not extreme
-    if is_extreme_weight(normalized_value, normalized_unit):
+    if is_extreme_weight(normalized_value, normalized_unit, count):
         return None
 
     text = f"{count} x {value} {unit}"

robotoff/workers/tasks/import_image.py

@@ -148,13 +148,15 @@ def run_import_image_job(product_id: ProductIdentifier, image_url: str, ocr_url:
         image_url=image_url,
         ocr_url=ocr_url,
     )
-    enqueue_job(
-        run_nutrition_table_object_detection,
-        get_high_queue(product_id),
-        job_kwargs={"result_ttl": 0},
-        product_id=product_id,
-        image_url=image_url,
-    )
+
+    if product_id.server_type.is_food():
+        enqueue_job(
+            run_nutrition_table_object_detection,
+            get_high_queue(product_id),
+            job_kwargs={"result_ttl": 0},
+            product_id=product_id,
+            image_url=image_url,
+        )
 
     # Run UPC detection to detect if the image is dominated by a UPC (and thus
     # should not be a product selected image)