使用OpenAI Assistant流式响应报错：'AsyncStream'对象不可迭代

按照OpenAI Assistant官方文档实现流式响应时，始终无法正常运行，持续抛出以下错误：

File "/home/vscode/.local/lib/python3.12/site-packages/openai/lib/streaming/_assistants.py", line 404, in __stream__
    for event in stream:
TypeError: 'AsyncStream' object is not iterable

我的FastAPI后端API端点代码如下：

@router.post("/new")
async def post_new(request: Request):
    """Start a new chat."""
    ...
        thread = await client.beta.threads.create()
        await client.beta.threads.messages.create(
            thread_id=thread.id,
            content=content_start,
            role="user",
            metadata={"type": "hidden"},
        )
        # 创建流式线程运行实例
        async with client.beta.threads.runs.stream(
            thread_id=thread.id,
            assistant_id=request.state.assistant_id,
            event_handler=EventHandler(),
        ) as stream:
            print("\n\nStream started")  # 该行能正常打印
            
            stream.until_done() # 调用此方法时触发报错

如果不在with语句前添加async，会出现以下警告：

/workspace/backend/routers/chat.py:54: RuntimeWarning: coroutine 'AsyncAPIClient.post' was never awaited
  with client.beta.threads.runs.stream(

环境信息：

• Python 3.12
• OpenAI SDK版本1.25.1（也尝试过低版本）
• 基于FastAPI开发后端

问题出在事件处理器的实现上，需改用异步版本的事件处理器，而非官方示例中的同步实现。代码示例如下：

from openai import AsyncAssistantEventHandler
from typing_extensions import override


class EventHandler(AsyncAssistantEventHandler):
    def __init__(self, response_stream) -> None:
        super().__init__()
        self.response_stream = response_stream

    @override
    async def on_text_delta(self, delta, snapshot):
        print("\n on_text_delta", delta.value, end="", flush=True)
        return delta.value
...

内容的提问来源于stack exchange，提问作者cookieDope