Manager for test data: downloads, artifact caching, and a tmpdir context.

testing, pytest-plugins, downloads, data, manager, test, software, engineering, data-management, pytest, pytest-plugin, python, software-engineering, test-data
pip install pytest-datadir-mgr==1.3.2


datadir-mgr plugin for pytest

Logo credit Jørgen Stamp, published under a Creative Commons Attribution 2.5 Denmark License.

The datadir-mgr plugin for pytest provides the datadir_mgr fixture which allow test functions to easily download data files and cache generated data files in data directories in a manner that allows for overlaying of results. datadir-mgr is pathlib-based, so complete paths to data files are handled, not just filenames.

This plugin behaves like a limited dictionary, with datadir_mgr[item] returning a path with the most specific scope (out of global, module, [class], [function] that matches the string or path specified by item. In addition to serving data files already stored in the data directory, the fixture provides five methods useful for adding to the test data stored in the repository:

  • The download method allows downloading data files into data directories, with option MD5 checksum checks, un-gzipping, and a progressbar.
  • The savepath fixture lets an arbitrary path relative to the current working directory to be saved at a particular scope in the data directories.
  • The add_scope method lets one add directories from scopes different from the present request to be added to the search path. This lets the results of previous cached steps to be used in scopes other than global.
  • The in_tmp_dir method creates a context in a temporary directory with a list of request file paths copied in. Optionally, all output file paths can be saved at a particular scope at cleanup with an optional exclusion filter pattern (e.g., for excluding log files). Note that files in directories that begin with test_ or end with _test could be confused with scope directories and cannnot be saved. If progressbar is set to "True", then the progress of file copying will be shown, which is helpful in some long-running pytest jobs, e.g. on Travis.
  • The paths_from_scope returns a list of all paths to files from a specified scope.


Python 3.6 or greater is required. This package is tested under Linux, MacOS, and Windows using Python 3.9.