Hadoop FileSystem Java Class Wrapper

Typed Python wrappers for Hadoop FileSystem class family.

Installation

You can install this package from pypi on any Hadoop or Spark runtime:

pip install hadoop-fs-wrapper

Select a version that matches hadoop version you are using:

Usage

Common use case is accessing Hadoop FileSystem from Spark session object:

from hadoop_fs_wrapper.wrappers.file_system import FileSystem

file_system = FileSystem.from_spark_session(spark=spark_session)

Then, for example, one can check if there are any files under specified path:

from hadoop_fs_wrapper.wrappers.file_system import FileSystem

def is_valid_source_path(file_system: FileSystem, path: str) -> bool:
    """
     Checks whether a regexp path refers to a valid set of paths
    :param file_system: pyHadooopWrapper FileSystem
    :param path: path e.g. (s3a|abfss|file|...)://hello@world.com/path/part*.csv
    :return: true if path resolves to existing paths, otherwise false
    """
    return len(file_system.glob_status(path)) > 0

Contribution

Currently basic filesystem operations (listing, deleting, search, iterative listing etc.) are supported. If an operation you require is not yet wrapped, please open an issue or create a PR.

All changes are tested against Spark 3.4 running in local mode.