fix(chatCompletion): handle errors in stream processing gracefully

Ensure that the [DONE] message is always sent even if an error occurs during stream processing. This prevents the client from waiting indefinitely and improves the robustness of the chat completion endpoint.

routes/chatCompletion.py

@@ -7,9 +7,19 @@ import json
 router = APIRouter()
 
 def generate_stream(response):
-    for chunk in response:
-        yield f"data: {json.dumps(chunk.__dict__, separators=(',', ':'))}\n\n"
-    yield "data: [DONE]\n\n"
+    try:
+        for chunk in response:
+            try:
+                # Attempt to process and yield the chunk
+                yield f"data: {json.dumps(chunk.__dict__, separators=(',', ':'))}\n\n"
+            except Exception as e:
+                # Optional: Log the error for debugging
+                print(f"Error during stream processing: {e}")
+                # Stop sending chunks if an error occurs
+                break
+    finally:
+        # Ensure the [DONE] message is always sent, even if an error occurred
+        yield "data: [DONE]\n\n"
 
 @router.post("/v1/chat/completions", tags=["Chat Completion"])
 async def chat_completion(body: ChatRequest):