fix car mapping

dubizzle_scraper/core/config.py

@@ -222,7 +222,7 @@ class AlgoliaConfig:
     index_name: str = _env_str("DUBIZZLE_ALGOLIA_INDEX_NAME", "motors.com")
     category_slug: str = _env_str("DUBIZZLE_ALGOLIA_CATEGORY_SLUG", "motors/used-cars")
     base_url: str = _env_str("DUBIZZLE_ALGOLIA_BASE_URL", "https://WD0PTZ13ZS-dsn.algolia.net")
-    hits_per_page: int = _env_int("DUBIZZLE_ALGOLIA_HITS_PER_PAGE", 100)
+    hits_per_page: int = _env_int("DUBIZZLE_ALGOLIA_HITS_PER_PAGE", 20)
 
 
 # --- Конфиг Celery (лимиты задач, concurrency, beat-расписание) ---

dubizzle_scraper/discovery/algolia.py

@@ -99,11 +99,22 @@ class _AlgoliaHttpClient:
             },
             method="POST",
         )
-        with urlopen(request, timeout=30) as response:
-            body = response.read().decode("utf-8", errors="ignore")
-        parsed = json.loads(body)
-        results = parsed.get("results") or []
-        return results[0] if results and isinstance(results[0], dict) else {}
+
+        last_exc: Exception | None = None
+        for attempt in range(1, 4):
+            try:
+                with urlopen(request, timeout=15) as response:
+                    body = response.read().decode("utf-8", errors="ignore")
+                parsed = json.loads(body)
+                results = parsed.get("results") or []
+                return results[0] if results and isinstance(results[0], dict) else {}
+            except Exception as exc:
+                last_exc = exc
+                if attempt >= 3:
+                    break
+                time.sleep(0.8 * attempt)
+
+        raise last_exc or RuntimeError("Algolia request failed")
 
 
 def _build_origin_id_from_hit(hit: dict[str, Any]) -> str | None:
@@ -282,6 +293,7 @@ def _fetch_shard_hits(
     year_min: int | None,
     year_max: int | None,
     known_origin_ids: set[str] | None,
+    limit: int | None,
     threshold: float,
     max_duration_seconds: float | None,
     started_at: float,
@@ -347,6 +359,12 @@ def _fetch_shard_hits(
             hit_records[url] = raw_hit
             if origin_id:
                 origin_ids_by_url[url] = origin_id
+            if limit is not None and limit > 0 and len(urls) >= limit:
+                early_stopped = True
+                break
+
+        if limit is not None and limit > 0 and len(urls) >= limit:
+            break
 
         if known_origin_ids is not None and threshold > 0 and (page_known + page_new) > 0:
             ratio = page_known / (page_known + page_new)
@@ -438,6 +456,7 @@ def discover_vehicle_hits_from_algolia(
             year_min=year_min,
             year_max=year_max,
             known_origin_ids=known_origin_ids,
+            limit=(limit - len(collected_urls)) if limit is not None and limit > 0 else None,
             threshold=threshold,
             max_duration_seconds=max_duration_seconds,
             started_at=started_at,