Manage Azure Blob Storage with Python

This skill covers the azure-storage-blob Python SDK for Azure Blob Storage. BlobServiceClient is the account-level entry point (built with DefaultAzureCredential and the account URL), from which get_container_client() returns a ContainerClient and get_blob_client() returns a BlobClient for single-blob operations. Core operations are uploading a blob from a file, bytes, or a stream (with overwrite=True to replace an existing blob), downloading via download_blob() to a file, into memory with .readall(), or into an existing buffer with .readinto() for memory efficiency, and listing blobs in a container - all of them, filtered by name_starts_with prefix, or walked hierarchically with walk_blobs(delimiter="/") to surface virtual directories.

Use it when a Python application needs to store or serve unstructured objects - files, logs, images, backups - in Azure Blob Storage, including scenarios needing time-limited shared access via SAS tokens, large-file performance tuning, or blob metadata and content-type management. It's an object-storage client, not a database or file-system API, so hierarchical listing via walk_blobs() only simulates folders through blob-name prefixes rather than real directories.

Configuration is AZURE_STORAGE_ACCOUNT_NAME or a full AZURE_STORAGE_ACCOUNT_URL. Uploading and downloading look like this:

with open("./local-file.txt", "rb") as data:
    blob_client.upload_blob(data, overwrite=True)

with open("./downloaded.txt", "wb") as file:
    download_stream = blob_client.download_blob()
    file.write(download_stream.readall())

Large transfers can set max_block_size, max_single_put_size, and max_concurrency for parallel upload/download. A time-limited SAS URL is produced with generate_blob_sas(), taking an account name, container/blob name, an account key or user delegation key, a BlobSasPermissions object (e.g. read=True), and an expiry timestamp - the resulting token is appended as a query string to the blob's URL. get_blob_properties() returns size, content type, and last-modified time; set_blob_metadata() and set_http_headers(content_settings=...) write custom metadata and content type respectively. Deleting a blob optionally cascades to its snapshots with delete_snapshots="include". An async client under azure.storage.blob.aio mirrors every operation with await inside an async with BlobServiceClient(...) context manager, matching the async DefaultAzureCredential from azure.identity.aio.

Authenticates via Azure Identity's DefaultAzureCredential, and is distributed as the azure-storage-blob PyPI package with a fully mirrored async client under azure.storage.blob.aio.

Python developers storing or serving unstructured data in Azure Blob Storage who need uploads/downloads, hierarchical-style listing, SAS-based time-limited access, or metadata management - and who follow the SDK's guidance: prefer DefaultAzureCredential over connection strings, use context managers for async clients, set overwrite explicitly on re-upload, tune max_concurrency for large transfers, prefer readinto() over readall() for memory efficiency, and set correct content types for web-served blobs.