Upload 21 files

app.py

@@ -86,11 +86,22 @@ MATCHING_RESULT_CACHE_LOCK = threading.Lock()
 MATCHING_RESULT_CACHE_MAX = int(os.getenv("MATCHING_RESULT_CACHE_MAX", "500"))
 MATCHING_RESULT_CACHE_TTL_SECONDS = int(os.getenv("MATCHING_RESULT_CACHE_TTL_SECONDS", "86400"))
 
+SCRAPER_QUERY_CACHE: dict[str, tuple[dict[str, Any], float]] = {}
+SCRAPER_QUERY_CACHE_LOCK = threading.Lock()
+SCRAPER_QUERY_CACHE_TTL_SECONDS = int(os.getenv("SCRAPER_QUERY_CACHE_TTL_SECONDS", "1296000"))  # 15 days
+
 
 def _matching_cache_storage_key(key: str) -> str:
     return f"matching:{key}"
 
 
+def _scraper_cache_key(user_prompt: str, store: str, gender: str, target_category: str) -> str:
+    """Create deterministic cache key for scraper queries"""
+    import hashlib
+    key_str = f"{user_prompt.lower().strip()}|{store}|{gender}|{target_category}"
+    return hashlib.md5(key_str.encode()).hexdigest()
+
+
 def _normalize_cache_category(value: Any) -> str:
     category = _norm(value)
     if category in {"topwear", "bottomwear", "others"}:
@@ -3186,8 +3197,24 @@ def scraper_recommend(payload: dict[str, Any] = Body(default_factory=dict)) -> d
     if isinstance(max_products, int) and max_products < 1:
         raise HTTPException(status_code=400, detail="max_products must be at least 1")
 
+    # Check cache first for faster repeat queries
+    cache_key = _scraper_cache_key(user_prompt, store, gender, target_category)
+    with SCRAPER_QUERY_CACHE_LOCK:
+        if cache_key in SCRAPER_QUERY_CACHE:
+            cached_result, cached_timestamp = SCRAPER_QUERY_CACHE[cache_key]
+            if time.time() - cached_timestamp < SCRAPER_QUERY_CACHE_TTL_SECONDS:
+                print(f"[CACHE HIT] Returning cached scraper results for: {user_prompt[:50]}...")
+                return cached_result
+        # Clean up expired cache entries
+        expired_keys = [
+            k for k, (_, ts) in SCRAPER_QUERY_CACHE.items()
+            if time.time() - ts >= SCRAPER_QUERY_CACHE_TTL_SECONDS
+        ]
+        for k in expired_keys:
+            del SCRAPER_QUERY_CACHE[k]
+
     try:
-        return _generate_scraper_plan_with_kimi(
+        result = _generate_scraper_plan_with_kimi(
             occasion=occasion,
             gender=gender,
             preferences=preferences,
@@ -3196,8 +3223,12 @@ def scraper_recommend(payload: dict[str, Any] = Body(default_factory=dict)) -> d
             filters=filters,
             max_products=max_products,
             store=store,
-            strict_kimi=True,
+            strict_kimi=False,
         )
+        # Cache the result
+        with SCRAPER_QUERY_CACHE_LOCK:
+            SCRAPER_QUERY_CACHE[cache_key] = (result, time.time())
+        return result
     except NvidiaGatewayError as exc:
         raise HTTPException(status_code=502, detail=str(exc)) from exc
     except NvidiaPayloadError as exc: