feat: add ingredient origin extraction

Adapted from PR by Pykrom:
#890

robotoff/app/api.py

@@ -368,6 +368,41 @@ def on_get(self, req: falcon.Request, resp: falcon.Response):
         resp.media = {"nutrients": [p.to_dict() for p in predictions]}
 
 
+class OriginPredictorResource:
+    def on_get(self, req: falcon.Request, resp: falcon.Response):
+        ocr_url = req.get_param("ocr_url", required=True)
+
+        if not ocr_url.endswith(".json"):
+            raise falcon.HTTPBadRequest("a JSON file is expected")
+
+        barcode = get_barcode_from_url(ocr_url)
+
+        if barcode is None:
+            raise falcon.HTTPBadRequest(f"invalid OCR URL: {ocr_url}")
+
+        try:
+            predictions = extract_ocr_predictions(
+                barcode, ocr_url, [PredictionType.origin]
+            )
+
+        except requests.exceptions.RequestException:
+            resp.media = {
+                "error": "download_error",
+                "error_description": "an error occurred during OCR JSON download",
+            }
+            return
+
+        except OCRParsingException as e:
+            logger.error(e)
+            resp.media = {
+                "error": "invalid_ocr",
+                "error_description": "an error occurred during OCR parsing",
+            }
+            return
+
+        resp.media = {"origin": [p.to_dict() for p in predictions]}
+
+
 class OCRInsightsPredictorResource:
     def on_get(self, req: falcon.Request, resp: falcon.Response):
         ocr_url = req.get_param("ocr_url", required=True)

robotoff/prediction/ocr/core.py

@@ -21,6 +21,7 @@
 from .label import find_labels
 from .location import find_locations
 from .nutrient import find_nutrient_mentions, find_nutrient_values
+from .origin import origin_parser
 from .packager_code import find_packager_codes
 from .packaging import find_packaging
 from .product_weight import find_product_weight
@@ -43,6 +44,7 @@
     PredictionType.trace: find_traces,
     PredictionType.nutrient: find_nutrient_values,
     PredictionType.nutrient_mention: find_nutrient_mentions,
+    PredictionType.origin: origin_parser.find_origin,
     PredictionType.brand: find_brands,
     PredictionType.store: find_stores,
     PredictionType.packaging: find_packaging,