Update app.py

app.py

@@ -72,7 +72,6 @@ def chat_completions():
     def generate():
         nonlocal messages
         full_response = ""
-        final_finish_reason = None
         while True:
             conversation = "\n".join([f"{msg['role']}: {msg['content']}" for msg in messages])
             conversation += "\n请关注并回复user最近的消息并避免总结对话历史的回答"
@@ -98,19 +97,20 @@ def chat_completions():
                 elif '"final":true' in event.data:
                     final_data = json.loads(event.data)
                     finish_reason = final_data.get('responseMessage', {}).get('finish_reason', 'stop')
-                    if finish_reason != 'stop':
+                    if finish_reason == 'length':
                         messages.append({"role": "assistant", "content": full_response})
                         messages.append({"role": "user", "content": "请继续你的输出，不要重复之前的内容"})
-                        break
+                        break  # 跳出当前循环，继续下一次请求
                     else:
-                        final_finish_reason = finish_reason
                         yield f"data: {json.dumps(format_openai_response('', finish_reason))}\n\n"
                         yield "data: [DONE]\n\n"
-                        return
+                        return  # 完全结束生成
+            
+            # 如果因为 length 而跳出内部循环，则继续外部循环
 
-        if final_finish_reason is None:
-            yield f"data: {json.dumps(format_openai_response('', 'stop'))}\n\n"
-            yield "data: [DONE]\n\n"
+        # 如果所有请求都因 length 而结束，发送一个最终的停止信号
+        yield f"data: {json.dumps(format_openai_response('', 'stop'))}\n\n"
+        yield "data: [DONE]\n\n"
 
     if stream:
         return Response(stream_with_context(generate()), content_type='text/event-stream')