fix: fix previously introduced issues

robotoff/prediction/ocr/location.py

@@ -31,7 +31,7 @@ class City:
     coordinates: Optional[tuple[float, float]]
 
 
-@cache()
+@cache
 def load_cities_fr(source: Union[Path, BinaryIO, None] = None) -> set[City]:
     """Load French cities dataset.
 

robotoff/prediction/ocr/packager_code.py

@@ -57,7 +57,7 @@ def process_USDA_match_to_flashtext(match) -> Optional[str]:
     return USDA_code
 
 
-@cache()
+@cache
 def generate_USDA_code_keyword_processor() -> KeywordProcessor:
     """Builds the KeyWordProcessor for USDA codes."""
 
@@ -184,7 +184,7 @@ def find_packager_codes_regex(content: Union[OCRResult, str]) -> list[Prediction
     return results
 
 
-@cache()
+@cache
 def generate_fishing_code_keyword_processor() -> KeywordProcessor:
     codes = text_file_iter(settings.OCR_FISHING_FLASHTEXT_DATA_PATH)
     return generate_keyword_processor(("{}||{}".format(c.upper(), c) for c in codes))

robotoff/prediction/ocr/trace.py

@@ -13,6 +13,7 @@
 from robotoff import settings
 from robotoff.types import Prediction, PredictionType
 from robotoff.utils import text_file_iter
+from robotoff.utils.text.flashtext import KeywordProcessor
 
 from .utils import generate_keyword_processor
 
@@ -21,8 +22,10 @@
 PREDICTOR_VERSION = "1"
 
 
-@cache()
-def generate_trace_keyword_processor(labels: Optional[list[str]] = None):
+@cache
+def generate_trace_keyword_processor(
+    labels: Optional[list[str]] = None,
+) -> KeywordProcessor:
     if labels is None:
         labels = list(text_file_iter(settings.OCR_TRACE_ALLERGEN_DATA_PATH))
 

robotoff/utils/image.py

@@ -72,6 +72,9 @@ def get_image_from_url(
             error_raise=error_raise,
             session=session,
         )
+
+        if content_bytes is None:
+            return None
     else:
         r = _get_image_from_url(image_url, error_raise, session)
         if r is None:

tests/unit/prediction/ocr/test_location.py

@@ -28,7 +28,8 @@ def test_load_cities_fr(mocker):
         ],
     )
 
-    res = load_cities_fr()
+    # Bypass the cache
+    res = load_cities_fr.__wrapped__()
 
     m_gzip_open.assert_called_once_with(settings.OCR_CITIES_FR_PATH, "rb")
     m_json_load.assert_called_once_with(m_gzip_open.return_value.__enter__.return_value)
@@ -46,7 +47,7 @@ def test_load_cities_fr(mocker):
     with pytest.raises(
         ValueError, match="'123', invalid FR postal code for city 'yolo'"
     ):
-        load_cities_fr()
+        load_cities_fr.__wrapped__()
 
     m_gzip_open.assert_called_once_with(settings.OCR_CITIES_FR_PATH, "rb")
     m_json_load.assert_called_once_with(m_gzip_open.return_value.__enter__.return_value)
@@ -60,14 +61,14 @@ def test_load_cities_fr(mocker):
     with pytest.raises(
         ValueError, match="'12A42', invalid FR postal code for city 'yolo'"
     ):
-        load_cities_fr()
+        load_cities_fr.__wrapped__()
 
     m_gzip_open.assert_called_once_with(settings.OCR_CITIES_FR_PATH, "rb")
     m_json_load.assert_called_once_with(m_gzip_open.return_value.__enter__.return_value)
 
 
 def test_cities_fr_dataset():
-    cities_fr = load_cities_fr()
+    cities_fr = load_cities_fr.__wrapped__()
 
     assert all(isinstance(item, City) for item in cities_fr)
     assert len(set(cities_fr)) == len(cities_fr)