Releases: morfologik/morfologik-stemming
release/2.2.0-rc3
What's Changed
- PR #121: fix bug in replacements: s>ss, ss>s (Jaume Ortolà).
Full Changelog: 2.2.0-rc2...2.2.0-rc3
release/2.2.0-rc2
Bug Fixes
-
PR #118: fix HMatrix not being reset between calls to
Speller.findReplacementCandidates(), causing incorrect candidates to be
returned on repeated calls (Jaume Ortolà). -
GH-38: support ^ (start) and $ (end) anchors and _ (space) in
replacement-pairs, following hunspell REP conventions. -
GH-75: Fix incorrect and incomplete CharsetDecoder usage in Speller.findRepl():
missing charBuffer.clear() before decode and missing decoder.flush() after
decode, which could produce wrong candidates for stateful encodings.
Other Changes
- Update Maven build plugins to current versions.
Release 2.1.9
Release 2.1.7
GH-102: Update vulnerable jcommander dependency
Release 2.1.6
Other Changes
- PR #101: fix replaceRunOnWords() not working for words that are uppercase at
sentence start (Daniel Naber).
Release 2.1.5
Bug Fixes
-
PR #96: incorrect logic in runOnWords (Jaume Ortolà).
-
PR #97: micro performance optimization (Daniel Naber).
Other Changes
- GH-95: Speller: findReplacementCandidates returns full CandidateData. This
commit also refactors the Speller to use a stateless returned array
list rather than reuse an internal field. Should not make a
practical difference. (Dawid Weiss)
Release 2.1.4
Release 2.1.3
Bug Fixes:
- GH-86: Speller: words containing the dictionary separator are not handled properly (Jaume Ortolà via Dawid Weiss).
Release 2.1.2
Bug Fixes:
- GH-85: Encoded sequences can clash with separator byte and cause assertion errors. (Daniel Naber, Dawid Weiss).