python async错误：等待仅在异步函数中允许

Question

我正在尝试异步读取文件，但是我遇到了一个错误，说“ 等待异步函数允许。”，从我的理解中，当等待关键字是在未标记为async的函数内部使用。但是，如下所示，我将read_blob函数标记为async

我在.ipynb jupiter Notebook中没有任何问题（我在其中运行的是按cell依次通过单元格代码），但是当我尝试在.py文件中的vscode中运行它时，它会引发错误。

这是我的代码：

from azure.storage.blob.aio import ContainerClient
import asyncio
from azure.core.exceptions import ResourceNotFoundError
from io import StringIO, BytesIO


class AsyncContainerClient(ContainerClient):

    async def read_blob(self,
                  blob_name: str,
                  add_blob_name_col=False,
                  add_blob_date_col=False,
                  preprocessing_func=None,
                  zip_regex=r'.+\.gz$',
                  csv_regex='.+\.csv(\.gz)?$',
                  parquet_regex='.+\.parquet$',
                  regex_string=None,
                  **kwargs):

        assert isinstance(blob_name, str), f'{blob_name} is not a string'

        try:
            blob = (await self.download_blob(blob_name))

            with BytesIO() as byte_stream:
                await blob.readinto(byte_stream)
                byte_stream.seek(0)
                return pd.read_parquet(byte_stream, engine='pyarrow')
        
        except ResourceNotFoundError:
            return 0


blob_sas_url = "https://proan.blob"
acc = AsyncContainerClient.from_container_url(blob_sas_url)

test_dirs = ["models1/model.parquet", "models2/model.parquet", "models3/model.parquet"]
res = await asyncio.gather(*(acc.read_blob(f) for f in test_dirs))

Answer 1

异步程序的切入点应为asyncio.run，您应该用异步方法包装代码，然后调用它

from azure.storage.blob.aio import ContainerClient
import asyncio
from azure.core.exceptions import ResourceNotFoundError
from io import StringIO, BytesIO


class AsyncContainerClient(ContainerClient):

    async def read_blob(self,
                  blob_name: str,
                  add_blob_name_col=False,
                  add_blob_date_col=False,
                  preprocessing_func=None,
                  zip_regex=r'.+\.gz
,
                  csv_regex='.+\.csv(\.gz)?
,
                  parquet_regex='.+\.parquet
,
                  regex_string=None,
                  **kwargs):

        assert isinstance(blob_name, str), f'{blob_name} is not a string'

        try:
            blob = (await self.download_blob(blob_name))

            with BytesIO() as byte_stream:
                await blob.readinto(byte_stream)
                byte_stream.seek(0)
                return pd.read_parquet(byte_stream, engine='pyarrow')
        
        except ResourceNotFoundError:
            return 0


async def main():
    blob_sas_url = "https://proan.blob"
    acc = AsyncContainerClient.from_container_url(blob_sas_url)

    test_dirs = ["models1/model.parquet", "models2/model.parquet", 
    "models3/model.parquet"]
    return await asyncio.gather(*(acc.read_blob(f) for f in test_dirs))

asyncio.run(main())