Changelog

Changelog#

v1.2.0 - 17.11.2025#

Version v1.2.0 ignores key=None in all find* methods and bumps the required dependencies.

  • Passing key=None is now ignored in all find* methods, and is now equivalent to passing key="*" or not passing the key. Previously this was interpreted as "None" (#192).

  • Drop support for python 3.9 (#195).

  • Bump minimum supported versions of the dependencies (#197):

    Package

    Old

    New

    numpy

    1.24

    1.26

    pandas

    2.0

    2.2

    parse

    1.19

    1.20

  • Ensure warnings are raised for user code and not internally (#199).

  • Consolidate package metadata and configuration in pyproject.toml (#174).

  • The repr now indicates when test_paths were passed (#202).

  • Enable “bugbear” linting rules (#198, and #200).

v1.1.0 - 19.03.2025#

Version v1.1.0 fixes a regression from v1.0.0 and adds FileContainer.search_single.

  • Restore behavior of FileContainer.search(key=None) - passing None to search is ignored again (#143).

  • Added FileContainer.search_single to search for exactly one path in a FileContainer (#158).

  • Fixed issues of non-unique test_path (#139).

v1.0.1 - 16.01.205#

This version patches a bug from v1.0.0. that broke FileContainer.concat().

  • compare keys of two FileContainers with pandas.Index.equals() instead of is in FileContainer.concat() (#145) - this fixes a bug where no two FileContainers that are not the same object could be concatenated together, see #143.

v1.0.0 - 07.01.2025#

Version 1.0.0 renames the package to filefisher (from filefinder) and makes the first release to pypi. It includes some modifications of the FileContainer class to be more user friendly, allows concatenating two FileContainers and adds methods to find exactly one path or file. Defines and tests minimum supported versions of the dependencies and adds documentation on readthedocs.

  • Renamed from filefinder to filefisher (#131) and deprecated filefinder (#132)

  • Added documentation for readthedocs (#134) and extended the documentation with usage (#135) and installation instructions (#136), as well as extension of the api documentation (#141)

  • Added method to concatenate two FileContainers (#126)

  • Added two methods to find exactly one file or path (and raise an error otherwise): FileFinder.find_single_file and FileFinder.find_single_path (#101).

  • Raise an error if an unnamed placeholder (e.g., "{}") is passed (#110)

  • The FileFinder.find_files arguments on_parse_error and _allow_empty can no longer be passed by position (#99).

  • FileFinder now raises an error if an invalid "{placeholder}" is used (#99).

  • Define and test the minimum supported versions of the dependencies (#125).

    Package

    Old

    New

    numpy

    undefined

    1.24

    pandas

    undefined

    2.0

    parse

    undefined

    1.19

  • Changes to FileContainer:

    • An empty FileContainer is returned instead of an empty list when no files/ paths are found (#114)

    • Renamed the "filename" column to "path" and made it a pd.Index, thus removing this column from the underlying DataFrame (#113).

    • Added meta and paths properties to FileContainer which allow to iterate over them (#121).

    • Added items() method to FileContainer, which iterates over path, meta (#128).

    • Deprecated iterating over FileContainer, use .paths, .meta or items() instead (#128).

    • Deprecated combine_by_key, create a pd.MultiIndex instead (#115).

    • Added the number of paths to the repr (#116).

    • Added capability to concat two FileContainers (#126).

    • Refactor FileContainer.search using isin (#117).

  • Explicitly test on python 3.13 (#103).

  • Drop support for python 3.9 (#102).

v0.3.0 - 27.03.2024#

New release that adds handling for parsing errors. It also drops python 3.7 and 3.8 support.

  • Change on_missing option in the priority_filter from “error” to “raise”. (#79)

  • Drop support for python 3.7 and 3.8 (#80)

  • Allow passing scalar numbers to find_paths and find_files (#58).

  • Show duplicates for non-unique queries (#73)

  • Add options on how to handle parsing errors (#75)

v0.2.0 - 23.05.2023#

New release that allows to specify a format spec which allows parsing more complex file name structures. It also drops python 3.6 support and modernizes the build system.

  • Allow passing format spec to the captured names to allow more precise name matching (#57).

  • Add tests for the cmip functionality and fix issue with filefinder.cmip.ensure_unique_grid (#35).

  • Removed support for python 3.6.

  • Explicitly test python 3.11.

v0.1.0 - 05.08.2022#

  • First version released based on the code developed for my IPCC AR6 analyses and including some additions (e.g. priority_filter, preferring kwargs over keys passed via a dictionary, more complete tests).

Contents