Update app.py

app.py

@@ -10,32 +10,31 @@ import requests
 import nest_asyncio
 import uvicorn
 
-# ✅ Load RAG-related files
+# ✅ Load documents and FAISS index
 with open("texts.json", "r", encoding="utf-8") as f:
     texts = json.load(f)
 
 index = faiss.read_index("faiss_index.bin")
 embed_model = SentenceTransformer("all-MiniLM-L6-v2")
 
+# ✅ API keys and config
 API_KEY = os.environ.get("OPENROUTER_API_KEY")
 MODEL = "qwen/qwq-32b:free"
 
-# ✅ FastAPI app
 app = FastAPI()
 
-# ✅ Context Retriever with threshold
+# ✅ Context fetcher
 def get_context(query, top_k=5, threshold=0.3):
     query_vec = embed_model.encode([query])
     D, I = index.search(np.array(query_vec), top_k)
-
-    # Filter based on similarity distance threshold
-    results = [(texts[i], d) for i, d in zip(I[0], D[0]) if d >= threshold]
-    if not results:
+    
+    matches = [(texts[i], d) for i, d in zip(I[0], D[0]) if d >= threshold]
+    if not matches:
         return ""
     
-    return "\n".join([text for text, _ in results])
+    return "\n".join([text for text, _ in matches])
 
-# ✅ Chat Function (RAG-aware + fallback)
+# ✅ Core chat function
 def chat_fn(message, history):
     headers = {
         "Authorization": f"Bearer {API_KEY}",
@@ -45,15 +44,15 @@ def chat_fn(message, history):
     context = get_context(message)
 
     if not context.strip():
-        return "❌ Sorry, I can't answer that."
+        return "❌ I couldn’t find any relevant info to answer that. Please ask something else."
 
     messages = [
         {
             "role": "system",
             "content": (
-                "You are a concise and helpful assistant. Use only the context below to answer. "
-                "Respond clearly and briefly. Avoid unnecessary details. If the context doesn't contain the answer, reply: "
-                "'Sorry, I can't answer that.'\n\nContext:\n" + context
+                "You are a precise, concise assistant. Use ONLY the provided context to answer."
+                " Do not make up facts. Answer briefly and clearly. If unsure, respond with:"
+                " 'I couldn’t find the answer based on available context.'\n\nContext:\n" + context
             )
         }
     ]
@@ -67,39 +66,47 @@ def chat_fn(message, history):
     payload = {
         "model": MODEL,
         "messages": messages,
-        "max_tokens": 200  # Optional: limit response length
+        "max_tokens": 200,
     }
 
     try:
-        response = requests.post("https://openrouter.ai/api/v1/chat/completions", headers=headers, json=payload)
+        response = requests.post(
+            "https://openrouter.ai/api/v1/chat/completions",
+            headers=headers,
+            json=payload,
+            timeout=30  # Prevent hanging
+        )
         response.raise_for_status()
         reply = response.json()["choices"][0]["message"]["content"]
     except Exception as e:
-        reply = f"❌ Error: {e}"
+        reply = f"⚠️ API error: {str(e)[:100]}"
 
-    return reply
+    return reply.strip()
 
-# ✅ FastAPI POST endpoint
+# ✅ POST endpoint for API usage
 @app.post("/chat")
 async def chat_api(request: Request):
     body = await request.json()
-    message = body.get("message")
+    message = body.get("message", "").strip()
     history = body.get("history", [])
+    
+    if not message:
+        return JSONResponse(content={"response": "⚠️ Please enter a valid message."})
+    
     response = chat_fn(message, history)
     return JSONResponse(content={"response": response})
 
-# ✅ Gradio Interface
+# ✅ Gradio interface
 demo = gr.ChatInterface(
     fn=chat_fn,
-    title="CODEX MIRXA KAMRAN",
-    description="Chat with AI MODEL trained by Mirxa Kamran",
+    title="💬 CODEX MIRXA KAMRAN",
+    description="Ask anything related to our knowledge base. Fast, short and precise responses!",
     theme="soft"
 )
 
-# ✅ Mount Gradio at root
 app = gr.mount_gradio_app(app, demo, path="/")
 
-# ✅ For local development
+# ✅ Local run
 if __name__ == "__main__":
     nest_asyncio.apply()
     uvicorn.run(app, host="0.0.0.0", port=7860)