v0.20

Python support

• Support Python 3.11 and 3.12.
• Drop support for Python 3.7.

Added

• Add a py.typed file so mypy can lint listparser use in dependent applications.
• Support lxml v5.0.0 while maintaining support for v4.6.2.

Changed

• Support malformed XML documents by using lxml's HTML parsers.

Fixed

• Fix the feed URL to the blog entries about listparser.

Documentation

• Add the OPML 1.0 and OPML 2.0 specifications to the listparser documentation for posterity.

Development

• Randomize the order of listparser's unit tests on each run
  to help ensure there are no testing interdependencies.
• Add black, flake8, and isort as pre-commit hooks.
• Add mypy as a tox test environment.
• Add pyupgrade as a pre-commit hook and enforce Python 3.8+ syntax.
• Python warnings encountered during testing are now escalated to errors.
• Move tool dependencies into separate requirements.txt files.
• Add a tox label, update, so it's easy to update tool dependencies.

v0.19

"Spring cleaning"

Added

• Add requests as an optional dependency.
• Add lxml as an optional dependency.
  This is currently the only way to parse malformed XML files.
• Add support for Python 3.7 through Python 3.10.

Removed

• Drop support for CPython 3.6 and lower.

• Drop support for IronPython.

• Drop support for Jython.

• Remove some ancient Jython compatibility code.

• HTML entities are no longer automatically injected into the DTD.

  If lxml is installed and undeclared HTML entities are encountered,
  they will be ignored.

  If lxml is not installed and undeclared HTML entities are encountered,
  Python's xml.sax parser will fail to parse the document.

Fixed

• Fix some coverage messages that were output during testing.
• Fix the combined coverage HTML report's paths.

Changed

• Migrate to a src/-based directory structure.
• Migrate to Read the Docs for documentation hosting.
• parse() no longer accepts etag, modified, or agent arguments.
• HTTP response information is no longer available in the return dictionary.
• The unit tests no longer launch an HTTP server.
• Coverage data is now erased before the unit tests begin.
• bozo is now True or False, not 1 or 0.
• bozo_exception is guaranteed to exist in the return dict.
• The RFC 822 parser now returns timezone-aware datetime objects
  instead of converting to UTC and losing timezone info.
• Move and split lptest.py into the tests/ subdirectory
  and into specific test_* files.
• Increased test coverage reporting to include test runners.
• Decreased the number of tox environments where listparser must be built and installed.
• Internally rely on dict key-based lookups instead of SuperDict attribute-based lookups.
• Cache XML parsing method lookup results to avoid excessive string formatting.
• Change the name of the ListError to ListparserError.
• Split the codebase into multiple files.
• Use tox to automate testing across multiple interpreters and versions.
• Migrate to pytest for unit testing.
• Remove dependence on the six package.
• Add type annotations.
• Remove compatibility code.
• Migrate to Poetry and pyproject.toml for project configuration.
• Change the license from LGPLv3 to MIT.
• Use scriv to manage the CHANGELOG.