image repairment in own products

scripts/repair_brand_images.py

@@ -576,14 +576,36 @@ def repair(cur, conn, brands: List[str], apply: bool, timeout: int, workers: int
             # three providers were caught serving a confidently wrong image.
             # --recheck re-opens rows whose stored URL names neither the brand
             # nor any distinctive word of the product, so they are re-searched
-            # under the corroboration rule rather than grandfathered in.
+            # under the corroboration rule rather than grandfathered in - and,
+            # for the produce bucket, any row that simply is not the reviewed
+            # image (see below).
             have = {r["image_id"] for r in needs}
             for row in rows:
                 if row["image_id"] in have:
                     continue
+                name = row.get("product_name") or ""
+
+                # A REVIEWED IMAGE OUTRANKS THE HEURISTIC.
+                #
+                # The corroboration test below asks whether the URL contains a
+                # distinctive word of the product, and it cannot tell an image OF
+                # the thing from an image merely NAMED after it. It passed, and so
+                # grandfathered in, `Bitter_Gourd_Curry.jpg` for Bitter Gourd,
+                # `Zunaid_Ahmed_Palak.jpg` - a photograph of a person - for Palak,
+                # and `BANANA_CREPE_CON_BASIL_LEAVES.jpg` for Basil Leaves. That
+                # left 57 rows untouched on the first full run.
+                #
+                # Where a person has already chosen the image, no heuristic gets a
+                # vote: if the stored URL is not the curated one, it is wrong.
+                if _is_own_products(suffix):
+                    curated = produce_reference.image_url(name)
+                    if curated and (row.get("image_url") or "") != curated:
+                        needs.append(row)
+                        continue
+
                 stored = _row_urls(row)
                 if stored and not any(
-                    _names_product(u, row.get("product_name") or "", suffix) for u in stored
+                    _names_product(u, name, suffix) for u in stored
                 ):
                     needs.append(row)
         totals["scanned"] += len(rows)