fix: mount gradio on fastapi directly

app.py

@@ -655,29 +655,16 @@ def verify_claim():
 init_db()      # Ensure DB exists before accepting requests
 load_models()
 
-# ─── Gradio + Flask Integration via Thread Proxy ──────────────────────────────
-# Flask runs on internal port 7861. Gradio exposes it via a FastAPI proxy route
-# at /backend/{path}, which forwards to http://127.0.0.1:7861/{path}.
-# This avoids all WSGIMiddleware path-stripping conflicts with HF's Gradio proxy.
+# ─── Gradio + Flask Integration via FastAPI mount ──────────────────────────────
+# The official way to combine Gradio and custom APIs in HF Spaces is to create
+# a FastAPI app, mount the custom API, and then mount Gradio on top.
 try:
-    import threading
-    import httpx
     import gradio as gr
     import spaces
-    from fastapi import Request as _FastAPIRequest
-    from fastapi.responses import Response as _FastAPIResponse
+    from fastapi import FastAPI
+    from fastapi.middleware.wsgi import WSGIMiddleware
 
-    # ── 1. Start Flask on internal port 7861 ─────────────────────────────────
-    _FLASK_PORT = 7861
-
-    def _run_flask_internal():
-        logger.info(f"Starting internal Flask server on port {_FLASK_PORT}...")
-        app.run(host='127.0.0.1', port=_FLASK_PORT, debug=False, use_reloader=False)
-
-    _flask_thread = threading.Thread(target=_run_flask_internal, daemon=True)
-    _flask_thread.start()
-
-    # ── 2. Gradio UI ─────────────────────────────────────────────────────────
+    # ── 1. Gradio UI ─────────────────────────────────────────────────────────
     @spaces.GPU
     def predict_gradio(statement, model_type):
         if not statement or len(statement) < 5:
@@ -715,58 +702,22 @@ try:
         ],
         outputs="text",
         title="Truth Detector API & Interactive Demo",
-        description="Backend API for Fake News Detection. REST API available at /backend/api/*"
+        description="Backend API for Fake News Detection. REST API available at /api/*"
     )
 
-    # ── 3. FastAPI middleware proxy: /backend/{path} → Flask on 127.0.0.1:7861/{path} ──
-    _SKIP_HEADERS = {'host', 'content-length', 'transfer-encoding', 'content-encoding'}
-
-    @demo.app.middleware("http")
-    async def flask_proxy_middleware(request: _FastAPIRequest, call_next):
-        if request.url.path.startswith("/backend/"):
-            target_path = request.url.path[len("/backend"):]
-            if not target_path.startswith("/"):
-                target_path = "/" + target_path
-                
-            target_url = f"http://127.0.0.1:{_FLASK_PORT}{target_path}"
-            
-            # Read the request body
-            body = await request.body()
-            
-            fwd_headers = {
-                k: v for k, v in request.headers.items()
-                if k.lower() not in _SKIP_HEADERS
-            }
-            
-            try:
-                async with httpx.AsyncClient(timeout=60) as client:
-                    resp = await client.request(
-                        method=request.method,
-                        url=target_url,
-                        headers=fwd_headers,
-                        content=body,
-                        params=dict(request.query_params),
-                        follow_redirects=True,
-                    )
-                resp_headers = {
-                    k: v for k, v in resp.headers.items()
-                    if k.lower() not in _SKIP_HEADERS
-                }
-                return _FastAPIResponse(
-                    content=resp.content,
-                    status_code=resp.status_code,
-                    headers=resp_headers,
-                )
-            except Exception as proxy_err:
-                logger.error(f"Proxy error for {request.url.path}: {proxy_err}")
-                return _FastAPIResponse(
-                    content=b'{"error":"Backend proxy error","detail":"' + str(proxy_err).encode() + b'"}',
-                    status_code=502,
-                    headers={"Content-Type": "application/json"},
-                )
-                
-        # If not /backend/*, let Gradio handle it normally
-        return await call_next(request)
+    # ── 2. Create master FastAPI app ─────────────────────────────────────────
+    # We rename the existing Flask app to flask_app, and create a new FastAPI app
+    # named `app` because Hugging Face will automatically run the variable named `app`.
+    flask_app = app
+    app = FastAPI()
+
+    # Mount Flask API routes at /api/ (This handles /api/predict, /api/auth/*, etc.)
+    # Flask routes internally start with /api (e.g. @app.route('/api/health')), 
+    # so we mount it at root "/" so that /api/health goes straight to Flask's /api/health
+    app.mount("/flask", WSGIMiddleware(flask_app))
+
+    # Mount Gradio app at root
+    app = gr.mount_gradio_app(app, demo, path="/")
 
     _GRADIO_AVAILABLE = True
 
@@ -777,7 +728,9 @@ except ImportError as _ie:
 if __name__ == '__main__':
     port = int(os.environ.get('PORT', 7860))
     if _GRADIO_AVAILABLE:
-        demo.launch(server_name="0.0.0.0", server_port=port)
+        # For local testing, run the master FastAPI app with Uvicorn
+        import uvicorn
+        uvicorn.run(app, host="0.0.0.0", port=port)
     else:
         app.run(host='0.0.0.0', port=port, debug=False)