Skip to content

Releases: morfologik/morfologik-stemming

Release 2.0.1

10 Sep 21:23
@dweiss dweiss
2.0.1
9dd7d8c
Compare
Choose a tag to compare
Release 2.0.1

Bug Fixes

  • GH-65: Dictionary.read(URL) ends in NPE when reading from a JAR resource
    (Dawid Weiss)
Loading

Release 2.0.0

10 Sep 07:53
@dweiss dweiss
2.0.0
a3b0b3f
Compare
Choose a tag to compare
Release 2.0.0

This release comes with a cleanup of the API for Java 1.7. There are
several aspects of the code that have been dropped (or added):

  • NIO is used extensively, mostly for better error reporting.
  • There is a simplified lookup of resources, no class-relative loading
    of dictionaries for example. The caller is in charge of looking
    up either an URL to the dictionary or providing an InputStream to it.
  • Removed internal caching of dictionaries from Dictionary. The
    Polish stemmer is initialized lazily and reuses its dictionary
    internally.
  • Numerous minor tweaks of parameters. JavaDocs.
  • A complete rewrite of the tools to compile (and decompile) FSA automata
    and complete stemming dictionaries. The tools now assert the validity
    of input data files and ensure no corrupt dictionaries can be produced.

Changes in backwards compatibility policy

  • GH-64: Removed OSGi support because of Maven issues (forks build
    phases, tests, etc.).
  • GH-62: Recompress Polish dictionary to use ';' as the separator.
    (Dawid Weiss)
  • GH-59: Moved Dictionary.convertText utility to
    DictionaryLookup.applyReplacements and fixed current reliance on map
    ordering. (Dawid Weiss)
  • GH-55: Removed the "distribution" module entirely. The tools module
    should be self-organizing. Complete overhaul of all the tools.
    Examples. Simplified syntax, options and assumptions.
    Input sanity checks and validation. (Dawid Weiss)
  • GH-57: Restructured the project into FSA traversal/ reading (only)
    and FSA Builders (construction). This cleans up dependency
    structure as well (HPPC is not required for FSA traversals).
    (Dawid Weiss)
  • GH-54: Make Java 1.7 the minimum required version. Certain methods
    that relied on File as arguments have been removed or changed to
    accept Path. (Dawid Weiss)

New Features

  • GH-53: Review library dependencies and bring them up to date.
    (Dawid Weiss)
  • Added OSGi support (Michal Hlavac)
  • GH-51: Remove and fail on deprecated metadata (fsa.dict.uses-*).
    (Dawid Weiss)

Optimizations

  • GH-61: Refactored the code to use one encoding/ decoding routine
    and ByteBuffers. Removed dependency on Guava.

Bug Fixes

  • GH-32: make replaceRunOnWords return "a lot" for "alot", etc.
    (Daniel Naber)
  • GH-34: ArrayIndexOutOfBoundsException with replacement-pairs.
    (Jaume Ortolà, Daniel Naber)

Check out the changelog.

Loading