RuntimeError: aclose(): asynchronous generator is already running

[pors]

Describe the bug

I see this error after an async streaming call to the completion end-point:

an error occurred during closing of asynchronous generator <async_generator object aiohttp_session at 0x1029de110>
asyncgen: <async_generator object aiohttp_session at 0x1029de110>
RuntimeError: aclose(): asynchronous generator is already running

To Reproduce

Run the code below. The completion_text is printed and below that the error above is shown.

Code snippets

import asyncio
from types import AsyncGeneratorType
from dotenv import load_dotenv
import os
import openai

load_dotenv()

openai.api_key = os.getenv("OPENAI_API_KEY")

async def test_completion_async_stream():
    completion = await openai.Completion.acreate(
        model="text-davinci-003",
        prompt="Hello!",
        stream=True,
    )

    # Assert response body
    assert isinstance(completion, AsyncGeneratorType)

    collected_chunks = []
    completion_text = ''
    # iterate through the stream, if it breaks, the test failed
    async for chunk in completion:
        collected_chunks.append(chunk)
        finish_reason = chunk['choices'][0]['finish_reason'] # type: ignore
        if finish_reason is not None:
            break
        chunk_text = chunk['choices'][0]['text'] # type: ignore
        completion_text += chunk_text  # append the text
    print(completion_text)

if __name__ == "__main__":
    asyncio.run(test_completion_async_stream())

OS

macOS 13.4 (22F66)

Python version

3.11.4

Library version

0.27.8

[jamesjohnson1025]

hi, did you find any alternative way to fix the issue ?.

[RobertCraigie]

This should be fixed in our upcoming v1 beta as we've moved away from aiohttp to httpx!