cached-path

A file utility library that provides a unified, simple interface for accessing both local and remote files. This can be used behind other APIs that need to access files agnostic to where they are located.

Quick links

• Documentation
• PyPI Package
• Contributing
• License

Installation

cached-path requires Python 3.7 or later.

Installing with pip

cached-path is available on PyPI. Just run

pip install cached-path

Installing from source

To install cached-path from source, first clone the repository:

git clone https://github.com/allenai/cached_path.git
cd cached_path

Then run

pip install -e .

Usage

from cached_path import cached_path

Given something that might be a URL or local path, cached_path() determines which. If it's a remote resource, it downloads the file and caches it to the cache directory, and then returns the path to the cached file. If it's already a local path, it makes sure the file exists and returns the path.

For URLs, http://, https://, s3:// (AWS S3), gs:// (Google Cloud Storage), and hf:// (HuggingFace Hub) are all supported out-of-the-box. Optionally beaker:// URLs in the form of beaker://{user_name}/{dataset_name}/{file_path} are supported, which requires beaker-py to be installed.

For example, to download the PyTorch weights for the model epwalsh/bert-xsmall-dummy on HuggingFace, you could do:

cached_path("hf://epwalsh/bert-xsmall-dummy/pytorch_model.bin")

For paths or URLs that point to a tarfile or zipfile, you can also add a path to a specific file to the url_or_filename preceeded by a "!", and the archive will be automatically extracted (provided you set extract_archive to True), returning the local path to the specific file. For example:

cached_path("model.tar.gz!weights.th", extract_archive=True)

Cache directory

By default the cache directory is ~/.cache/cached_path/, however there are several ways to override this setting:

• set the environment variable CACHED_PATH_CACHE_ROOT,
• call set_cache_dir(), or
• set the cache_dir argument each time you call cached_path().

Team

cached-path is developed and maintained by the AllenNLP team, backed by the Allen Institute for Artificial Intelligence (AI2). AI2 is a non-profit institute with the mission to contribute to humanity through high-impact AI research and engineering. To learn more about who specifically contributed to this codebase, see our contributors page.

License

cached-path is licensed under Apache 2.0. A full copy of the license can be found on GitHub.