Customer Rating and review changes

backend/app/electronics/cli.py

@@ -143,58 +143,22 @@ def rematch(category: str = typer.Option(..., help="mobiles | laptops"),
 
 @app.command()
 def reviews(category: Optional[str] = typer.Option(None, help="mobiles | laptops (default both)"),
-            limit: int = typer.Option(200, help="Max product pages to re-read"),
+            limit: int = typer.Option(500, help="Max product pages to re-read"),
+            budget: int = typer.Option(60, help="Max web searches for finding brand-store pages"),
+            no_discover: bool = typer.Option(False, "--no-discover", help="Skip finding brand-store pages"),
             verbose: bool = False) -> None:
-    """Re-read ratings and customer reviews from the product pages already on file.
+    """Refresh real customer ratings and reviews from free, robots-allowed sources.
 
-    Only pages the collector itself reads (scraped / brand official listings)
-    are fetched, politely (robots.txt, per-site pacing, circuit breaker). A
-    rating or review is stored only when the page's own schema.org data states
-    it; nothing is generated.
+    Finds each verified product's page on brand stores that publish reviews
+    (brands.yaml `reviews_site`), then re-reads every readable product page for
+    its rating, star breakdown and reviews, plus Vijay Sales' public review
+    feed. A rating or review is stored only when its source states it.
     """
     _setup_logging(verbose)
-    from rapidfuzz import fuzz
+    from app.electronics.review_refresh import refresh_reviews
 
-    from app.electronics.db import repository as repo
-    from app.electronics.extract.jsonld import extract_products
-    from app.electronics.net.polite_client import PoliteClient
-
-    rows = repo.listings_for_review_backfill(category)[:limit]
-    run_id = repo.start_run("reviews", {"category": category, "pages": len(rows)})
-    client = PoliteClient(on_fetch=lambda r, host: repo.log_fetch(run_id, r.url, host, r.status, r.bytes,
-                                                                    r.outcome, r.robots_allowed))
-    stats = {"pages": 0, "pages_ok": 0, "rated": 0, "reviews_stored": 0, "no_matching_product": 0}
-    status, error = "done", None
-    try:
-        for row in rows:
-            stats["pages"] += 1
-            res = client.get(row["source_url"])
-            if not res.ok:
-                continue
-            stats["pages_ok"] += 1
-            products = extract_products(res.text)
-            # The same product the listing was stored from: its SKU, else its name.
-            match = next((p for p in products if p.get("sku") and p["sku"] == row["source_sku"]), None)
-            if match is None:
-                title = (row["title"] or "").lower()
-                scored = [(fuzz.token_set_ratio(p["name"].lower(), title), p) for p in products]
-                scored = [sp for sp in scored if sp[0] >= 85]
-                match = max(scored, key=lambda sp: sp[0])[1] if scored else None
-            if match is None:
-                stats["no_matching_product"] += 1
-                continue
-            if match.get("rating") is not None and Decimal(0) < match["rating"] <= Decimal(5):
-                repo.update_listing_rating(row["listing_id"], match["rating"], match.get("review_count"))
-                stats["rated"] += 1
-            if match.get("reviews"):
-                stats["reviews_stored"] += repo.save_reviews(row["listing_id"], match["reviews"])
-            typer.echo(f"  {row['domain']:22} rating={match.get('rating')} reviews={len(match.get('reviews') or [])}")
-    except Exception as exc:  # noqa: BLE001
-        status, error = "failed", repr(exc)
-        raise
-    finally:
-        client.close()
-        repo.finish_run(run_id, status, stats, error)
+    stats = refresh_reviews(category=category, limit=limit, budget=budget, discover=not no_discover,
+                            progress=typer.echo)
     typer.echo(json.dumps(stats, indent=2))