Python从Azure Blob读取joblib文件报UnicodeDecodeError如何解决？

joblib.load() 入参要求为文件路径字符串或可读二进制类文件对象，你直接传入downloaded_blob.readall()返回的二进制字节串时，joblib会误将该字节串识别为文件路径，尝试按UTF-8编码解析路径字符串，由此触发解码错误。

有两种可直接落地的实现方式，和pickle.loads的读取效果完全一致：

• 方案1：将二进制内容包装为io.BytesIO类文件对象，传入joblib.load读取

from azure.storage.blob import BlobClient
import joblib
from io import BytesIO

blob_client = BlobClient.from_connection_string('connection_string', 'myContainer', 'myBlob.joblib')
downloaded_blob = blob_client.download_blob()
model = joblib.load(BytesIO(downloaded_blob.readall()))

• 方案2：直接调用joblib原生兼容pickle的loads接口，直接接收二进制字节串

from azure.storage.blob import BlobClient
import joblib

blob_client = BlobClient.from_connection_string('connection_string', 'myContainer', 'myBlob.joblib')
downloaded_blob = blob_client.download_blob()
# 用法和pickle.loads完全一致
model = joblib.loads(downloaded_blob.readall())

内容的提问来源于stack exchange，提问作者Christian