Update app.py

app.py

@@ -25,9 +25,9 @@ def root():
     return jsonify({"message": "欢迎使用 GPT-OSS to OpenAI 格式代理API", "status": "ok"})
 
 # 辅助函数：创建一个标准的OpenAI流式数据块
-def create_openai_chunk(content, model="gpt-oss-120b", custom_id_prefix="chatcmpl"):
+def create_openai_chunk(content, model="gpt-oss-120b"):
     return {
-        "id": f"{custom_id_prefix}-{str(uuid.uuid4())}",
+        "id": f"chatcmpl-{str(uuid.uuid4())}",
         "object": "chat.completion.chunk",
         "created": int(time.time()),
         "model": model,
@@ -76,22 +76,21 @@ def chat_completions_proxy():
                     event_type = gpt_oss_data.get('type')
                     update_type = gpt_oss_data.get('update', {}).get('type')
                     
-                    # --- 关键改动：模拟思考过程的打字效果 ---
                     if event_type == 'thread.item_updated' and update_type == 'cot.entry_added':
                         thought = gpt_oss_data['update']['entry']['content']
-                        # 立即发送前缀
-                        yield f"data: {json.dumps(create_openai_chunk('[思考中] '))}\n\n"
-                        # 逐字流式传输思考内容
-                        for char in thought:
-                            yield f"data: {json.dumps(create_openai_chunk(char))}\n\n"
-                            time.sleep(0.02) # 加入微小延迟，以获得更好的视觉效果
+                        formatted_thought = f"[思考中] {thought}"
+                        
+                        # 为了兼容性，不逐字发送，而是一次性发送整条思考
+                        thought_chunk = create_openai_chunk(formatted_thought)
+                        yield f"data: {json.dumps(thought_chunk)}\n\n"
                         # 发送换行符以分隔
-                        yield f"data: {json.dumps(create_openai_chunk('\\n\\n'))}\n\n"
+                        newline_chunk = create_openai_chunk("\n\n")
+                        yield f"data: {json.dumps(newline_chunk)}\n\n"
 
-                    # 普通文本片段的流式传输（这部分本身就是逐字/逐词的）
                     if event_type == 'thread.item_updated' and update_type == 'assistant_message.content_part.text_delta':
                         delta_content = gpt_oss_data['update'].get('delta', '')
-                        yield f"data: {json.dumps(create_openai_chunk(delta_content))}\n\n"
+                        text_chunk = create_openai_chunk(delta_content)
+                        yield f"data: {json.dumps(text_chunk)}\n\n"
                 
                 yield "data: [DONE]\n\n"
             except IOError as e:
@@ -99,7 +98,7 @@ def chat_completions_proxy():
 
         return Response(stream_formatter(), mimetype='text/event-stream')
     
-    else: # 非流式请求逻辑保持不变
+    else:
         try:
             full_response_content = ""
             for gpt_oss_data in _internal_proxy_stream():