Python異步錯誤：僅在異步函式中允許等待

我正在嘗試異步讀取檔案，但我遇到了一個錯誤，說“ await only allowed in async function。” 據我了解，當await關鍵字在未標記為的函式中使用時會發生此錯誤async。但是，如下面的代碼所示，我將read_blob函式標記為async

我在 .ipynb Jupiter notebook 中運行它沒有任何問題（我在其中按單元格順序運行代碼），但是當我嘗試在 .py 檔案中的 VSCode 中運行它時，它會引發錯誤。

這是我的代碼：

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))

uj5u.com熱心網友回復：

異步程式的入口點應該是asyncio.run，您應該將代碼包裝在異步方法中，然后呼叫它

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())

標籤：Python 异步 异步等待 蟒蛇异步

上一篇：如何讓JS堆疊真正“等待”異步函式的結束？

下一篇：如何在ReduxAddCase完成的情況下分配陣列