Changelog

0.2.0 (2026-08-21)

Breaking changes

  • Rename import_data() to read(); rename export_data() to write().

  • Remove load_data(), save_data(), assign_and_save_data(), and load_remote_data(). Build the path with pathlib.Path and call read/ write directly, or use pipette.cache_url() for a remote file.

  • Remove load_data_as_name(). It has no Python equivalent: it exists in R because load() dumps saved object names into the calling environment, so R needs a hook to bind a different name. Python already binds whatever name you write, e.g. new = pipette.read(path).

  • Rename encode()/decode() and remove factorize()/unfactorize(), merging all four into categorize()/uncategorize(). Their j= parameter is now columns=.

  • Rename atomize() to drop_nested_columns().

  • Rename paste_url() to join_url(). Its protocol="none" sentinel is now protocol=None.

  • Rename get_url_dir_list() to list_remote_dir().

  • Remove metadata2(). Use df.attrs directly.

  • Remove droplevels(). Use df[col] = df[col].cat.remove_unused_categories() directly.

  • Remove remove_na(). Use df.dropna() directly.

  • Remove md5() and sha256(). Use hashlib.file_digest() (stdlib since Python 3.11) directly.

  • Remove init_dir(). Use os.makedirs(path, exist_ok=True) or Path(path).mkdir(parents=True, exist_ok=True) directly.

  • Remove get_json(). Use read(url, format="json").

  • Remove match_rowname_column() from the public API; it is now private.

  • Remove the na_strings alias; use NA_STRINGS.

  • Remove the dead PIPETTE_TESTS_URL constant.

  • read() and write() now take str | os.PathLike, so a pathlib.Path works everywhere a path is accepted.

  • read()’s rownames= parameter is renamed to index=; same for write().

  • write() now returns the written path (a Path locally, or the str URI for an s3:// target) instead of a bare str.

  • write() gains a format= parameter, matching read().

  • Every parameter after the first positional argument on every public function is now keyword-only.

0.1.0 (2026-06-19)

Changes

  • Switch license to Apache-2.0.

  • Publish to python.acidgenomics.com (private PEP 503 index).

  • Update installation instructions in README.

0.0.1 (initial)

Initial release.