import httpx
from typing import Callable, Awaitable, Optional


# 流式请求LLm的方法
async def stream_post_request(
        url,
        headers=None,
        json=None,
        chunk_handler: Optional[Callable[[bytes], Awaitable[bytes]]] = None
):
    async with httpx.AsyncClient(http2=True) as client:
        async with client.stream(
                method="POST", url=url, headers=headers, json=json
        ) as response:
            async for chunk in response.aiter_bytes():
                if chunk_handler:
                    # 支持异步处理
                    chunk = await chunk_handler(chunk)
                yield chunk