Update app.py

app.py

@@ -24,6 +24,9 @@ N_GPU_LAYERS = int(os.environ.get("N_GPU_LAYERS", "0"))
 MAX_INPUT_LENGTH = 1500
 END_TOKEN = "<|endoftext|>"
 
+# ==================== GLOBAL MODEL ====================
+llm = None
+
 # ==================== LIFECYCLE MANAGEMENT ====================
 @asynccontextmanager
 async def lifespan(app: FastAPI):
@@ -57,9 +60,7 @@ app = FastAPI(
 app.add_middleware(
     CORSMiddleware,
     allow_origins=[
-        "https://your-vercel-app.vercel.app",  # Replace with your actual Vercel URL
-        "http://localhost:3000",  # For local development
-        "http://localhost:5173",  # Vite dev server
+        "*",  # Allow all origins temporarily, update with your Vercel URL
     ],
     allow_credentials=True,
     allow_methods=["*"],
@@ -74,9 +75,6 @@ class TransformRequest(BaseModel):
 class HumanizeRequest(BaseModel):
     text: str = Field(..., min_length=1, max_length=MAX_INPUT_LENGTH)
 
-# ==================== GLOBAL MODEL ====================
-llm = None
-
 # ==================== STYLE PROMPTS ====================
 STYLE_PROMPTS = {
     "professional": """You are a professional writing assistant. Rewrite the text below in formal, corporate business language.
@@ -142,6 +140,8 @@ STYLE_TEMPERATURES = {
 # ==================== HELPER FUNCTIONS ====================
 def load_model():
     """Load the GGUF model"""
+    global MODEL_PATH  # Access the global MODEL_PATH
+    
     print(f"🔄 Loading model from: {MODEL_PATH}")
     
     try:
@@ -150,14 +150,20 @@ def load_model():
             print("📥 Downloading model from Hugging Face Hub...")
             try:
                 from huggingface_hub import hf_hub_download
-                MODEL_PATH = hf_hub_download(
+                # Use a different variable name to avoid conflict
+                downloaded_path = hf_hub_download(
                     repo_id=MODEL_REPO,
                     filename=MODEL_FILE,
                     local_dir=".",
                     token=os.environ.get("HF_TOKEN", None)
                 )
+                print(f"✅ Model downloaded to: {downloaded_path}")
             except ImportError:
                 print("⚠️  huggingface-hub not installed, using local model path")
+                # If we can't download, use fallback path
+                MODEL_PATH = os.path.join("/code", MODEL_FILE)
+        
+        print(f"📁 Model path: {MODEL_PATH}")
         
         model = Llama(
             model_path=MODEL_PATH,
@@ -174,6 +180,8 @@ def load_model():
         
     except Exception as e:
         print(f"❌ Failed to load model: {e}")
+        import traceback
+        traceback.print_exc()
         return None
 
 def clean_output(text: str) -> str:
@@ -217,9 +225,10 @@ async def transform_with_model(text: str, style: str) -> str:
     global llm
     
     if llm is None:
+        # Try to load model if not loaded
         llm = load_model()
         if llm is None:
-            raise HTTPException(status_code=503, detail="Model not available")
+            raise HTTPException(status_code=503, detail="Model not available. Please check if model file exists.")
     
     try:
         # Build prompt
@@ -259,6 +268,8 @@ async def transform_with_model(text: str, style: str) -> str:
         
     except Exception as e:
         print(f"❌ Model error: {e}")
+        import traceback
+        traceback.print_exc()
         raise HTTPException(status_code=500, detail=f"Model error: {str(e)}")
 
 # ==================== API ENDPOINTS ====================
@@ -277,10 +288,11 @@ async def root():
 async def health_check():
     """Detailed health check"""
     return {
-        "status": "healthy" if llm else "model_loading",
+        "status": "healthy" if llm else "model_loading_failed",
         "model_loaded": llm is not None,
         "threads": N_THREADS,
-        "context_size": N_CTX
+        "context_size": N_CTX,
+        "model_path": MODEL_PATH
     }
 
 @app.post("/api/humanize")
@@ -345,6 +357,8 @@ async def transform_text(request: TransformRequest):
         raise
     except Exception as e:
         print(f"❌ Transformation error: {e}")
+        import traceback
+        traceback.print_exc()
         raise HTTPException(status_code=500, detail=f"Transformation failed: {str(e)}")
 
 @app.get("/api/styles")
@@ -367,23 +381,11 @@ async def get_styles():
         "default_style": "casual"
     }
 
-# Error handler
-@app.exception_handler(Exception)
-async def general_exception_handler(request, exc):
-    return JSONResponse(
-        status_code=500,
-        content={
-            "error": "Internal server error",
-            "message": str(exc),
-            "path": request.url.path
-        }
-    )
-
 # ==================== MAIN ====================
 if __name__ == "__main__":
     import uvicorn
     
-    port = int(os.environ.get("PORT", 8000))
+    port = int(os.environ.get("PORT", 7860))
     
     uvicorn.run(
         "app:app",