Update api.py

api.py

@@ -38,8 +38,8 @@ app = FastAPI(title="Hadith FAISS API", version="1.0")
 
 app.add_middleware(
     CORSMiddleware,
-    allow_origins=["*"],   # لاحقاً: استبدلها بدومين موقعك للأمان
-    allow_credentials=False,
+    allow_origins=["*"],  # لو تبي تقفلها على دومين موقعك فقط قل لي
+    allow_credentials=True,
     allow_methods=["*"],
     allow_headers=["*"],
 )
@@ -92,15 +92,11 @@ def load_items():
     with open(JSON_PATH, "r", encoding="utf-8") as f:
         _items = json.load(f)
 
-    if not isinstance(_items, list):
-        raise RuntimeError("Dataset JSON root must be a list")
-
     _item_by_id = {}
     for it in _items:
         cid = it.get("corpusID")
-        if cid is None:
-            continue
-        _item_by_id[int(cid)] = it
+        if cid is not None:
+            _item_by_id[int(cid)] = it
 
 
 def get_model() -> SentenceTransformer:
@@ -110,14 +106,19 @@ def get_model() -> SentenceTransformer:
     return _model
 
 
-def save_artifacts(index: faiss.Index, emb: np.ndarray, id_by_pos: List[int], pos_by_id: Dict[int, int]):
+def save_artifacts(
+    index: faiss.Index,
+    emb: np.ndarray,
+    id_by_pos: List[int],
+    pos_by_id: Dict[int, int],
+):
     ensure_dirs()
 
     faiss.write_index(index, INDEX_PATH)
-    np.save(EMB_PATH, emb)  # creates EMB_PATH + ".npy"
+    np.save(EMB_PATH, emb)  # will create embeddings.npy
 
     with open(ID_BY_POS_PATH, "w", encoding="utf-8") as f:
-        json.dump(id_by_pos, f, ensure_ascii=False)
+        json.dump([int(x) for x in id_by_pos], f, ensure_ascii=False)
 
     pos_by_id_str = {str(k): int(v) for k, v in pos_by_id.items()}
     with open(POS_BY_ID_PATH, "w", encoding="utf-8") as f:
@@ -128,7 +129,7 @@ def load_artifacts():
     global _index, _emb, _id_by_pos, _pos_by_id, _DIM
 
     _index = faiss.read_index(INDEX_PATH)
-    _emb = np.load(EMB_PATH + ".npy").astype("float32", copy=False)
+    _emb = np.load(EMB_PATH + ".npy").astype("float32")
 
     with open(ID_BY_POS_PATH, "r", encoding="utf-8") as f:
         _id_by_pos = [int(x) for x in json.load(f)]
@@ -143,6 +144,7 @@ def load_artifacts():
 def build_all():
     """
     Build embeddings + FAISS then save.
+    This should run only if artifacts are missing.
     """
     global _index, _emb, _id_by_pos, _pos_by_id, _DIM
 
@@ -160,17 +162,11 @@ def build_all():
     )
     emb = np.asarray(emb, dtype="float32")
 
-    dim = emb.shape[1]
+    dim = int(emb.shape[1])
     index = faiss.IndexFlatIP(dim)  # cosine via IP since normalized
     index.add(emb)
 
-    # Build ID mappings
-    id_by_pos = []
-    for x in _items:
-        if "corpusID" not in x:
-            raise RuntimeError("Each item must have corpusID")
-        id_by_pos.append(int(x["corpusID"]))
-
+    id_by_pos = [int(x["corpusID"]) for x in _items]
     pos_by_id = {cid: i for i, cid in enumerate(id_by_pos)}
 
     save_artifacts(index, emb, id_by_pos, pos_by_id)
@@ -179,14 +175,14 @@ def build_all():
     _emb = emb
     _id_by_pos = id_by_pos
     _pos_by_id = pos_by_id
-    _DIM = int(dim)
+    _DIM = dim
 
     dt = time.time() - t0
     print(f"[build_all] Built + saved artifacts in {dt:.2f}s. dim={_DIM}, n={len(_id_by_pos)}")
 
 
 def require_ready():
-    if (not _READY) or (_index is None) or (_emb is None):
+    if not _READY or _index is None or _emb is None:
         raise HTTPException(status_code=503, detail="API is not ready yet. Try again in a moment.")
 
 
@@ -241,11 +237,17 @@ def on_startup():
     except Exception as e:
         _READY = False
         print("[startup] FAILED ❌", str(e))
+        # keep app up but not ready
 
 
 # -----------------------------
 # Routes
 # -----------------------------
+@app.get("/")
+def root():
+    return {"name": "Hadith FAISS API", "ready": _READY}
+
+
 @app.get("/health")
 def health():
     return {
@@ -253,6 +255,7 @@ def health():
         "items": len(_items),
         "dim": _DIM,
         "has_artifacts": artifacts_exist(),
+        "model": MODEL_NAME,
     }
 
 
@@ -263,7 +266,7 @@ def stats():
         "items": len(_items),
         "dim": _DIM,
         "index_type": type(_index).__name__,
-        "model": MODEL_NAME,
+        "topk_max": TOPK_MAX,
     }
 
 
@@ -279,7 +282,6 @@ def get_item(corpus_id: int):
 @app.get("/similar/{corpus_id}")
 def similar(corpus_id: int, topk: int = 10):
     require_ready()
-
     cid = int(corpus_id)
     if cid not in _pos_by_id:
         raise HTTPException(status_code=404, detail="corpusID not found in index")
@@ -317,7 +319,6 @@ def similar(corpus_id: int, topk: int = 10):
 @app.post("/search")
 def search(req: SearchRequest):
     require_ready()
-
     q = (req.query or "").strip()
     if not q:
         raise HTTPException(status_code=400, detail="query is empty")