adlfs 0.0.5a0

Access Azure Datalake Gen1 with fsspec and dask

Dask interface to Azure-Datalake Gen1 and Gen2 Storage

Warning: this code is experimental and untested.

Quickstart

This package is on PyPi and can be installed using:

pip install adlfs

In your code, call:

from fsspec.registry import known_implementations

To use the Gen1 filesystem:

known_implementations[‘adl’] = {‘class’: ‘adlfs.AzureDatalakeFileSystem’}

To use the Gen2 filesystem:

known_implementations[‘abfs’] = {‘class’: ‘adlfs.AzureBlobFileSystem’}

This allows operations such as: import dask.dataframe as dd storage_options={ ‘tenant_id’: TENANT_ID, ‘client_id’: CLIENT_ID, ‘client_secret’: CLIENT_SECRET, ‘storage_account’: STORAGE_ACCOUNT, ‘filesystem’: FILESYSTEM, } dd.read_csv(‘abfs://folder/file.csv’, storage_options=STORAGE_OPTIONS}

Details

The package includes pythonic filesystem implementations for both Azure Datalake Gen1 and Azure Datalake Gen2, that facilitate interactions with both Azure Datalake implementations with Dask, using the intake/filesystem_spec base class.

Operations against both Gen1 and Gen2 datalakes currently require an Azure ServicePrincipal with suitable credentials to perform operations on the resources of choice.

Operations on the Azure Gen1 Datalake are implemented by leveraging multiple inheritance from both the fsspec.AbstractFileSystem and the Azure Python Gen1 Filesystem library, while operations against the Azure Gen2 Datalake are implemented by using subclassing the fsspec.AbstractFileSystem and leveraging the Azure Datalake Gen2 API. Note that the Azure Datalake Gen2 API allows calls to using either the ‘http://’ or ‘https://’ protocols, designated by an ‘abfs[s]://’ protocol. Under the hood in adlfs, this will always happen using ‘https://’ using the requests library.

An Azure Datalake Gen2 url takes the following form, which is replicated in the adlfs library, for the sake of consistency: ‘abfs[s]://{storage_account}/{filesystem}/{folder}/{file}’

Currently, when using either the ‘adl://’ or ‘abfs://’ protocols in a dask operation, it is required to explicitly declare the storage_options, as described in the Dask documentation. The intent is to eliminate this requirement for (at at minimum) Gen2 operations, by having the adlfs library parse the filesystem name