Customer Rating and review changes

backend/app/electronics/extract/jsonld.py

@@ -161,42 +161,67 @@ def _reviews(node: dict) -> List[Dict[str, Any]]:
     return out
 
 
+def _rating_only_nodes(nodes: List[dict]) -> Dict[Optional[str], dict]:
+    """Nameless Product nodes that only carry ratings/reviews, keyed by @id.
+
+    Review widgets (Bazaarvoice on samsung.com/in, for one) publish the
+    product's aggregateRating and reviews as a SEPARATE Product node with no
+    name, linked to the real product by the same @id. Without merging, those
+    real reviews would be skipped for lack of a name.
+    """
+    out: Dict[Optional[str], dict] = {}
+    for node in nodes:
+        if _text(node.get("name")):
+            continue
+        if not (isinstance(node.get("aggregateRating"), dict) or node.get("review") or node.get("reviews")):
+            continue
+        out.setdefault(_text(node.get("@id")), node)
+    return out
+
+
 def extract_products(html: str) -> List[Dict[str, Any]]:
     """All schema.org Product nodes on the page, flattened to plain fields."""
     products: List[Dict[str, Any]] = []
-    for block in json_ld_blocks(html):
-        for node in _walk(block):
-            if not (_types(node) & _PRODUCT_TYPES):
-                continue
-            name = _text(node.get("name"))
-            if not name:
-                continue
-            offer = _offer(node.get("offers")) if node.get("offers") else {}
-            if not offer and isinstance(node.get("hasVariant"), list):
-                offer = _offer([v.get("offers") for v in node["hasVariant"] if isinstance(v, dict) and v.get("offers")])
-            props = {}
-            for p in node.get("additionalProperty") or []:
-                if isinstance(p, dict) and p.get("name") and p.get("value") not in (None, ""):
-                    props[str(p["name"])] = str(p["value"])
-            rating = node.get("aggregateRating") if isinstance(node.get("aggregateRating"), dict) else {}
-            products.append({
-                "name": name,
-                "brand": _text(node.get("brand")),
-                "sku": _text(node.get("sku")) or _text(node.get("productID")),
-                "mpn": _text(node.get("mpn")),
-                "gtin": next((_text(node.get(k)) for k in ("gtin13", "gtin", "gtin12", "gtin14", "gtin8") if node.get(k)), None),
-                "color": _text(node.get("color")),
-                "images": _images(node.get("image")),
-                "description": _text(node.get("description")),
-                "price": offer.get("price"),
-                "currency": offer.get("currency"),
-                "availability": offer.get("availability"),
-                "in_stock": offer.get("in_stock"),
-                # Sites publish 0 for "no ratings yet"; that is not a rating.
-                "rating": (_dec(rating.get("ratingValue")) or None),
-                "review_count": int(_dec(rating.get("reviewCount") or rating.get("ratingCount")) or 0) or None,
-                "reviews": _reviews(node),
-                "properties": props,
-                "evidence": json.dumps({"name": name, "offers": offer.get("raw")}, default=str)[:1500],
-            })
+    nodes = [n for block in json_ld_blocks(html) for n in _walk(block) if _types(n) & _PRODUCT_TYPES]
+    rating_nodes = _rating_only_nodes(nodes)
+    named = [n for n in nodes if _text(n.get("name"))]
+    for node in named:
+        # Attach a ratings-only node: same @id, or - when it has no @id - the
+        # page's only named product (there is then no doubt which it is about).
+        extra = rating_nodes.get(_text(node.get("@id"))) if node.get("@id") else None
+        if extra is None and len(named) == 1:
+            extra = rating_nodes.get(None)
+        if extra is not None:
+            node = {**node,
+                    "aggregateRating": node.get("aggregateRating") or extra.get("aggregateRating"),
+                    "review": node.get("review") or extra.get("review") or extra.get("reviews")}
+        name = _text(node.get("name"))
+        offer = _offer(node.get("offers")) if node.get("offers") else {}
+        if not offer and isinstance(node.get("hasVariant"), list):
+            offer = _offer([v.get("offers") for v in node["hasVariant"] if isinstance(v, dict) and v.get("offers")])
+        props = {}
+        for p in node.get("additionalProperty") or []:
+            if isinstance(p, dict) and p.get("name") and p.get("value") not in (None, ""):
+                props[str(p["name"])] = str(p["value"])
+        rating = node.get("aggregateRating") if isinstance(node.get("aggregateRating"), dict) else {}
+        products.append({
+            "name": name,
+            "brand": _text(node.get("brand")),
+            "sku": _text(node.get("sku")) or _text(node.get("productID")),
+            "mpn": _text(node.get("mpn")),
+            "gtin": next((_text(node.get(k)) for k in ("gtin13", "gtin", "gtin12", "gtin14", "gtin8") if node.get(k)), None),
+            "color": _text(node.get("color")),
+            "images": _images(node.get("image")),
+            "description": _text(node.get("description")),
+            "price": offer.get("price"),
+            "currency": offer.get("currency"),
+            "availability": offer.get("availability"),
+            "in_stock": offer.get("in_stock"),
+            # Sites publish 0 for "no ratings yet"; that is not a rating.
+            "rating": (_dec(rating.get("ratingValue")) or None),
+            "review_count": int(_dec(rating.get("reviewCount") or rating.get("ratingCount")) or 0) or None,
+            "reviews": _reviews(node),
+            "properties": props,
+            "evidence": json.dumps({"name": name, "offers": offer.get("raw")}, default=str)[:1500],
+        })
     return products