Update main.py

main.py

@@ -4,35 +4,72 @@ from sentence_transformers import SentenceTransformer
 import uvicorn
 import asyncio
 from concurrent.futures import ThreadPoolExecutor
+from typing import List  # ← إضافة جديدة
+import numpy as np  # ← إضافة جديدة
 
 app = FastAPI(title="Text Embedding API (Qwen/Qwen3-Embedding-0.6B)")
 
 class TextRequest(BaseModel):
-    text: str = Field(..., min_length=1, max_length=10000, description="Text to embed")
+    text: str = Field(..., min_length=1, description="Text to embed")  # ← تم حذف max_length=10000
 
 # Globals
 model = None
+tokenizer = None  # ← إضافة جديدة
 model_id = 'Qwen/Qwen3-Embedding-0.6B'
 executor = ThreadPoolExecutor(max_workers=4)
+MAX_TOKENS = 512  # ← إضافة جديدة
 
 @app.on_event("startup")
 async def load_model():
-    global model
+    global model, tokenizer  # ← تم إضافة tokenizer
     print(f"Loading model: {model_id}...")
     model = SentenceTransformer(model_id)
+    tokenizer = model.tokenizer  # ← إضافة جديدة
     print("Model loaded successfully")
 
 @app.get("/")
 def home():
     return {"status": "online", "model": model_id, "endpoint": "/embed/text"}
 
+# ↓↓↓ Function جديدة كاملة ↓↓↓
+def chunk_and_embed(text: str) -> List[float]:
+    """Split text into chunks if too long, then pool embeddings"""
+    tokens = tokenizer.encode(text, add_special_tokens=False)
+    
+    # If text is short, embed directly
+    if len(tokens) <= MAX_TOKENS:
+        return model.encode(text, normalize_embeddings=True).tolist()
+    
+    # Split into chunks
+    chunks = []
+    overlap = 50
+    start = 0
+    while start < len(tokens):
+        end = start + MAX_TOKENS
+        chunk_tokens = tokens[start:end]
+        chunk_text = tokenizer.decode(chunk_tokens, skip_special_tokens=True)
+        chunks.append(chunk_text)
+        
+        if end >= len(tokens):
+            break
+        start = end - overlap
+    
+    # Embed all chunks
+    chunk_embeddings = [model.encode(chunk, normalize_embeddings=True) for chunk in chunks]
+    
+    # Pool embeddings (mean)
+    final_embedding = np.mean(chunk_embeddings, axis=0).tolist()
+    
+    return final_embedding
+# ↑↑↑ نهاية Function الجديدة ↑↑↑
+
 @app.post("/embed/text")
 async def embed_text(request: TextRequest):
     try:
         loop = asyncio.get_event_loop()
         embedding = await loop.run_in_executor(
             executor,
-            lambda: model.encode(request.text, normalize_embeddings=True).tolist()
+            lambda: chunk_and_embed(request.text)  # ← تم التعديل من model.encode إلى chunk_and_embed
         )
         
         return {