Update app.py

app.py

@@ -1,11 +1,16 @@
-from fastapi import FastAPI
+from fastapi import FastAPI, HTTPException
 from fastapi.middleware.cors import CORSMiddleware
+from fastapi.responses import StreamingResponse
 from pydantic import BaseModel
 import requests
 import os
+import json
+import asyncio
+from typing import List, Dict, Any
 
 DEEPSEEK_API_KEY = os.getenv("DEEPSEEK_API_KEY")
 DEEPSEEK_URL = "https://api.deepseek.com/chat/completions"
+DEEPSEEK_STREAM_URL = "https://api.deepseek.com/chat/completions"
 
 app = FastAPI()
 
@@ -17,15 +22,45 @@ app.add_middleware(
     allow_headers=["*"],
 )
 
+class Message(BaseModel):
+    role: str
+    content: str
+
 class AskRequest(BaseModel):
-    messages: list
+    messages: List[Message]
+
+class StreamRequest(BaseModel):
+    messages: List[Message]
 
 @app.get("/")
 def health():
-    return {"status": "HF DeepSeek backend running"}
+    return {"status": "HF DeepSeek backend running with streaming support"}
 
 @app.post("/ask")
 def ask(req: AskRequest):
+    """Non-streaming endpoint for backward compatibility"""
+    headers = {
+        "Content-Type": "application/json",
+        "Authorization": f"Bearer {DEEPSEEK_API_KEY}"
+    }
+
+    payload = {
+        "model": "deepseek-chat",
+        "messages": [msg.dict() for msg in req.messages],
+        "temperature": 0.2,
+        "stream": False
+    }
+
+    try:
+        r = requests.post(DEEPSEEK_URL, headers=headers, json=payload, timeout=60)
+        r.raise_for_status()
+        return r.json()
+    except requests.exceptions.RequestException as e:
+        raise HTTPException(status_code=500, detail=f"DeepSeek API error: {str(e)}")
+
+@app.post("/stream")
+async def stream(req: StreamRequest):
+    """Streaming endpoint for real-time responses"""
     headers = {
         "Content-Type": "application/json",
         "Authorization": f"Bearer {DEEPSEEK_API_KEY}"
@@ -33,13 +68,90 @@ def ask(req: AskRequest):
 
     payload = {
         "model": "deepseek-chat",
-        "messages": req.messages,
-        "temperature": 0.2
+        "messages": [msg.dict() for msg in req.messages],
+        "temperature": 0.2,
+        "stream": True
     }
 
-    r = requests.post(DEEPSEEK_URL, headers=headers, json=payload, timeout=60)
+    async def generate():
+        try:
+            # Make streaming request to DeepSeek API
+            response = requests.post(
+                DEEPSEEK_STREAM_URL, 
+                headers=headers, 
+                json=payload, 
+                stream=True,
+                timeout=60
+            )
+            
+            response.raise_for_status()
+            
+            # Stream the response from DeepSeek to the client
+            for line in response.iter_lines():
+                if line:
+                    decoded_line = line.decode('utf-8').strip()
+                    if decoded_line.startswith('data: '):
+                        data = decoded_line[6:]  # Remove "data: " prefix
+                        
+                        if data == '[DONE]':
+                            yield f"data: {json.dumps({'choices': [{'delta': {'content': ''}}]})}\n\n"
+                            yield "data: [DONE]\n\n"
+                            break
+                        
+                        try:
+                            parsed_data = json.loads(data)
+                            
+                            # Extract content from DeepSeek response
+                            if 'choices' in parsed_data and len(parsed_data['choices']) > 0:
+                                choice = parsed_data['choices'][0]
+                                if 'delta' in choice and 'content' in choice['delta']:
+                                    content = choice['delta']['content']
+                                    if content:
+                                        # Forward the SSE event to client
+                                        yield f"data: {json.dumps({'choices': [{'delta': {'content': content}}]})}\n\n"
+                                        
+                        except json.JSONDecodeError:
+                            continue
+                        
+        except requests.exceptions.RequestException as e:
+            error_message = f"DeepSeek API streaming error: {str(e)}"
+            yield f"data: {json.dumps({'error': error_message})}\n\n"
+            yield "data: [DONE]\n\n"
+        except Exception as e:
+            error_message = f"Unexpected error: {str(e)}"
+            yield f"data: {json.dumps({'error': error_message})}\n\n"
+            yield "data: [DONE]\n\n"
 
-    if r.status_code != 200:
-        return {"error": "DeepSeek API failed", "detail": r.text}
+    return StreamingResponse(
+        generate(),
+        media_type="text/event-stream",
+        headers={
+            "Cache-Control": "no-cache",
+            "Connection": "keep-alive",
+            "X-Accel-Buffering": "no"  # Disable buffering for nginx
+        }
+    )
 
-    return r.json()
+@app.get("/test-stream")
+async def test_stream():
+    """Test endpoint to verify streaming is working"""
+    async def generate_test():
+        test_messages = [
+            "Hello! ",
+            "I'm your AI teacher. ",
+            "How can I help you today? ",
+            "I can explain mathematics, ",
+            "science, and other subjects. ",
+            "What would you like to learn?",
+        ]
+        
+        for msg in test_messages:
+            yield f"data: {json.dumps({'choices': [{'delta': {'content': msg}}]})}\n\n"
+            await asyncio.sleep(0.5)
+        
+        yield "data: [DONE]\n\n"
+    
+    return StreamingResponse(
+        generate_test(),
+        media_type="text/event-stream"
+    )