0.2.0

Added

• Convert membership tests against eligible literal sets into OR patterns when
  the hashable-subjects assumption is enabled.

Fixed

• Apply --no-types patterns to qualified isinstance targets such as
  typing.Mapping, including targets contained in type tuples.

0.1.0

Added

• Added --check mode. It reports files that would be converted without
  modifying them and exits with status 1 when changes are needed or processing
  errors occur.
• Added matchify and matchify-check pre-commit hooks for automatic rewriting
  and check-only CI workflows.
• Added conversion of eligible inline dictionary lookups and single-use local
  lookup-table variables into match statements. Literal and nested tuple keys,
  arbitrary value expressions, comments, single evaluation of the lookup key,
  and the original KeyError behavior are preserved where supported. This
  conversion requires the new lookup-equality assumption.
• Added the identity-equality assumption, allowing qualified identity checks
  such as value is Enum.MEMBER to become value patterns when identity and
  equality are known to be equivalent.
• Added separate list-sequence-pattern and tuple-sequence-pattern
  assumptions for converting explicit sequence type checks into Python sequence
  patterns. Checks against both types require both assumptions.
• Exported Assumptions from the top-level matchify package.
• Added an MIT license.

Changed

• Composite match subjects selected under pure-subjects now use paths shared
  by a majority of branches. This enables more eligible mixed-subject chains to
  be converted while keeping less common checks in guards.
• Sequence transformations are more conservative: an isinstance check for
  list or tuple is no longer treated as implied by a sequence pattern unless
  the corresponding assumption is enabled. Qualified class names remain
  guarded and are not covered by these assumptions.
• CLI summaries distinguish check results with would convert, and normal as
  well as check mode now return a failure status for processing errors.
• Release automation now uses dedicated release pull requests, tags the tested
  merge commit, publishes it to PyPI, and creates GitHub release notes from
  Scriv fragments.

Fixed

• Preserved comments attached to dictionary lookup statements during
  conversion.
• Restricted dictionary lookup conversion to safe, unambiguous candidates.
  Chained subscriptions, reused local tables, multiple or unsupported uses,
  duplicate or unhashable keys, dictionary unpacking, and other unsafe forms
  remain unchanged.
• Avoided capture-name collisions in generated missing-key cases.

Development

• Expanded behavior and public-API tests, including generated runtime
  round-trips, and enforced 100% branch coverage.
• Added locked dependencies, lowest-direct-dependency CI coverage, PyPy 3.11
  testing, and updated repository template and maintenance automation.