japanese_deinflector 0.0.1

data/.gitignore

@@ -0,0 +1,2 @@
+.DS_Store
+*.gem

data/.rspec

@@ -0,0 +1,2 @@
+--color
+--format progress

data/.travis.yml

@@ -0,0 +1,7 @@
+language: ruby
+rvm:
+  - 1.8.7
+  - 1.9.2
+  - 1.9.3
+  - jruby-18mode # JRuby in 1.8 mode
+  - jruby-19mode # JRuby in 1.9 mode

data/Gemfile

@@ -0,0 +1,4 @@
+source "http://rubygems.org"
+
+# Specify your gem's dependencies in method_decorators.gemspec
+gemspec

data/Rakefile

@@ -0,0 +1,8 @@
+#!/usr/bin/env rake
+require "bundler/gem_tasks"
+
+require 'rspec/core/rake_task'
+
+RSpec::Core::RakeTask.new(:spec)
+
+task :default => :spec

data/japanese_deinflector.gemspec

@@ -0,0 +1,21 @@
+# -*- encoding: utf-8 -*-
+$:.push File.expand_path("../lib", __FILE__)
+require 'japanese_deinflector/version'
+
+Gem::Specification.new do |s|
+  s.name = 'japanese_deinflector'
+  s.version = JapaneseDeinflector::VERSION
+  s.date = '2012-08-28'
+  s.summary = "Deinflect (unconjugate/undecline) Japanese words."
+  s.description = "Deinflect (unconjugate/undecline) Japanese words."
+  s.authors = ["Peter Graham"]
+  s.email = ["pete@gigadrill.com"]
+  s.files         = `git ls-files`.split("\n")
+  s.test_files    = `git ls-files -- {test,spec,features}/*`.split("\n")
+  s.require_paths = ["lib"]
+  s.homepage = 'http://github.com/6/japanese_deinflector'
+
+  s.add_development_dependency "json"
+  s.add_development_dependency "rake"
+  s.add_development_dependency "rspec"
+end

data/lib/data/deinflect.dat

@@ -0,0 +1,336 @@
+Deinflect Rules 20081220-0509 | by Jonathan Zarate | http://www.polarcloud.com
+polite past negative
+polite negative
+polite volitional
+-chau
+-sugiru
+-nasai
+polite past
+-tara
+-tari
+causative
+potential or passive
+-sou
+-tai
+polite
+past
+negative
+passive
+-ba
+volitional
+potential
+passive or causative
+-te
+-zu
+imperative
+masu stem
+adv
+noun
+imperative negative
+くありませんでした	い	1152	0
+いませんでした	う	640	0
+きませんでした	く	640	0
+きませんでした	くる	2176	0
+ぎませんでした	ぐ	640	0
+しませんでした	す	640	0
+しませんでした	する	4224	0
+ちませんでした	つ	640	0
+にませんでした	ぬ	640	0
+びませんでした	ぶ	640	0
+みませんでした	む	640	0
+りませんでした	る	640	0
+くありません	い	1152	1
+ませんでした	る	2432	0
+いましょう	う	640	2
+きましょう	く	640	2
+きましょう	くる	2176	2
+ぎましょう	ぐ	640	2
+しましょう	す	640	2
+しましょう	する	4224	2
+ちましょう	つ	640	2
+にましょう	ぬ	640	2
+びましょう	ぶ	640	2
+みましょう	む	640	2
+りましょう	る	640	2
+いじゃう	ぐ	514	3
+いすぎる	う	513	4
+いちゃう	く	514	3
+いなさい	う	640	5
+いました	う	640	6
+いません	う	640	1
+かったら	い	1152	7
+かったり	い	1152	8
+きすぎる	く	513	4
+きすぎる	くる	2049	4
+ぎすぎる	ぐ	513	4
+きちゃう	くる	2050	3
+きなさい	く	640	5
+きなさい	くる	2176	5
+ぎなさい	ぐ	640	5
+きました	く	640	6
+きました	くる	2176	6
+ぎました	ぐ	640	6
+きません	く	640	1
+きません	くる	2176	1
+ぎません	ぐ	640	1
+こさせる	くる	2049	9
+こられる	くる	2049	10
+しすぎる	す	513	4
+しすぎる	する	4097	4
+しちゃう	す	514	3
+しちゃう	する	4098	3
+しなさい	す	640	5
+しなさい	する	4224	5
+しました	す	640	6
+しました	する	4224	6
+しません	す	640	1
+しません	する	4224	1
+ちすぎる	つ	513	4
+ちなさい	つ	640	5
+ちました	つ	640	6
+ちません	つ	640	1
+っちゃう	う	514	3
+っちゃう	く	514	3
+っちゃう	つ	514	3
+っちゃう	る	514	3
+にすぎる	ぬ	513	4
+になさい	ぬ	640	5
+にました	ぬ	640	6
+にません	ぬ	640	1
+びすぎる	ぶ	513	4
+びなさい	ぶ	640	5
+びました	ぶ	640	6
+びません	ぶ	640	1
+ましょう	る	2432	2
+みすぎる	む	513	4
+みなさい	む	640	5
+みました	む	640	6
+みません	む	640	1
+りすぎる	る	513	4
+りなさい	る	640	5
+りました	る	640	6
+りません	る	640	1
+んじゃう	ぬ	514	3
+んじゃう	ぶ	514	3
+んじゃう	む	514	3
+いそう	う	640	11
+いたい	う	516	12
+いたら	く	640	7
+いだら	ぐ	640	7
+いたり	く	640	8
+いだり	ぐ	640	8
+います	う	640	13
+かせる	く	513	9
+がせる	ぐ	513	9
+かった	い	1152	14
+かない	く	516	15
+がない	ぐ	516	15
+かれる	く	513	16
+がれる	ぐ	513	16
+きそう	く	640	11
+きそう	くる	2176	11
+ぎそう	ぐ	640	11
+きたい	く	516	12
+きたい	くる	2052	12
+ぎたい	ぐ	516	12
+きたら	くる	2176	7
+きたり	くる	2176	8
+きます	く	640	13
+きます	くる	2176	13
+ぎます	ぐ	640	13
+くない	い	1028	15
+ければ	い	1152	17
+こない	くる	2052	15
+こよう	くる	2176	18
+これる	くる	2049	19
+させる	する	4097	9
+させる	る	2305	9
+さない	す	516	15
+される	す	513	20
+される	する	4097	16
+しそう	す	640	11
+しそう	する	4224	11
+したい	す	516	12
+したい	する	4100	12
+したら	す	640	7
+したら	する	4224	7
+したり	す	640	8
+したり	する	4224	8
+しない	する	4100	15
+します	す	640	13
+します	する	4224	13
+しよう	する	4224	18
+すぎる	い	1025	4
+すぎる	る	2305	4
+たせる	つ	513	9
+たない	つ	516	15
+たれる	つ	513	16
+ちそう	つ	640	11
+ちたい	つ	516	12
+ちます	つ	640	13
+ちゃう	る	2306	3
+ったら	う	640	7
+ったら	つ	640	7
+ったら	る	640	7
+ったり	う	640	8
+ったり	つ	640	8
+ったり	る	640	8
+なさい	る	2432	5
+なせる	ぬ	513	9
+なない	ぬ	516	15
+なれる	ぬ	513	16
+にそう	ぬ	640	11
+にたい	ぬ	516	12
+にます	ぬ	640	13
+ばせる	ぶ	513	9
+ばない	ぶ	516	15
+ばれる	ぶ	513	16
+びそう	ぶ	640	11
+びたい	ぶ	516	12
+びます	ぶ	640	13
+ました	る	2432	6
+ませる	む	513	9
+ません	る	2432	1
+まない	む	516	15
+まれる	む	513	16
+みそう	む	640	11
+みたい	む	516	12
+みます	む	640	13
+らせる	る	513	9
+らない	る	516	15
+られる	る	2817	10
+りそう	る	640	11
+りたい	る	516	12
+ります	る	640	13
+わせる	う	513	9
+わない	う	516	15
+われる	う	513	16
+んだら	ぬ	640	7
+んだら	ぶ	640	7
+んだら	む	640	7
+んだり	ぬ	640	8
+んだり	ぶ	640	8
+んだり	む	640	8
+いた	く	640	14
+いだ	ぐ	640	14
+いて	く	640	21
+いで	ぐ	640	21
+えば	う	640	17
+える	う	513	19
+おう	う	640	18
+かず	く	640	22
+がず	ぐ	640	22
+きた	くる	2176	14
+きて	くる	2176	21
+くて	い	1152	21
+けば	く	640	17
+げば	ぐ	640	17
+ける	く	513	19
+げる	ぐ	513	19
+こい	くる	2176	23
+こう	く	640	18
+ごう	ぐ	640	18
+こず	くる	2176	22
+さず	す	640	22
+した	す	640	14
+した	する	4224	14
+して	す	640	21
+して	する	4224	21
+しろ	する	4224	23
+せず	する	4224	22
+せば	す	640	17
+せよ	する	4224	23
+せる	す	513	19
+そう	い	1152	11
+そう	す	640	18
+そう	る	2432	11
+たい	る	2308	12
+たず	つ	640	22
+たら	る	2432	7
+たり	る	2432	8
+った	う	640	14
+った	く	640	14
+った	つ	640	14
+った	る	640	14
+って	う	640	21
+って	く	640	21
+って	つ	640	21
+って	る	640	21
+てば	つ	640	17
+てる	つ	513	19
+とう	つ	640	18
+ない	る	2308	15
+なず	ぬ	640	22
+ねば	ぬ	640	17
+ねる	ぬ	513	19
+のう	ぬ	640	18
+ばず	ぶ	640	22
+べば	ぶ	640	17
+べる	ぶ	513	19
+ぼう	ぶ	640	18
+ます	る	2432	13
+まず	む	640	22
+めば	む	640	17
+める	む	513	19
+もう	む	640	18
+よう	る	2432	18
+らず	る	640	22
+れば	る	7040	17
+れる	る	2817	19
+ろう	る	640	18
+わず	う	640	22
+んだ	ぬ	640	14
+んだ	ぶ	640	14
+んだ	む	640	14
+んで	ぬ	640	21
+んで	ぶ	640	21
+んで	む	640	21
+い	いる	384	24
+い	う	640	24
+い	る	2176	23
+え	う	640	23
+え	える	384	24
+き	きる	384	24
+き	く	640	24
+ぎ	ぎる	384	24
+ぎ	ぐ	640	24
+く	い	1152	25
+け	く	640	23
+け	ける	384	24
+げ	ぐ	640	23
+げ	げる	384	24
+さ	い	1152	26
+し	す	640	24
+じ	じる	384	24
+ず	る	2432	22
+せ	す	640	23
+せ	せる	384	24
+ぜ	ぜる	384	24
+た	る	2432	14
+ち	ちる	384	24
+ち	つ	640	24
+て	つ	640	23
+て	てる	384	24
+て	る	2432	21
+で	でる	384	24
+な		7040	27
+に	にる	384	24
+に	ぬ	640	24
+ね	ぬ	640	23
+ね	ねる	384	24
+ひ	ひる	384	24
+び	びる	384	24
+び	ぶ	640	24
+へ	へる	384	24
+べ	ぶ	640	23
+べ	べる	384	24
+み	みる	384	24
+み	む	640	24
+め	む	640	23
+め	める	384	24
+よ	る	384	23
+り	りる	384	24
+り	る	640	24
+れ	る	640	23
+れ	れる	384	24
+ろ	る	384	23

data/lib/data/deinflect.json

@@ -0,0 +1 @@
+{"reasons":["polite past negative","polite negative","polite volitional","-chau","-sugiru","-nasai","polite past","-tara","-tari","causative","potential or passive","-sou","-tai","polite","past","negative","passive","-ba","volitional","potential","passive or causative","-te","-zu","imperative","masu stem","adv","noun","imperative negative"],"rules":{"9":[{"from":"くありませんでした","to":"い","reason_id":0}],"7":[{"from":"いませんでした","to":"う","reason_id":0},{"from":"きませんでした","to":"く","reason_id":0},{"from":"きませんでした","to":"くる","reason_id":0},{"from":"ぎませんでした","to":"ぐ","reason_id":0},{"from":"しませんでした","to":"す","reason_id":0},{"from":"しませんでした","to":"する","reason_id":0},{"from":"ちませんでした","to":"つ","reason_id":0},{"from":"にませんでした","to":"ぬ","reason_id":0},{"from":"びませんでした","to":"ぶ","reason_id":0},{"from":"みませんでした","to":"む","reason_id":0},{"from":"りませんでした","to":"る","reason_id":0}],"6":[{"from":"くありません","to":"い","reason_id":1},{"from":"ませんでした","to":"る","reason_id":0}],"5":[{"from":"いましょう","to":"う","reason_id":2},{"from":"きましょう","to":"く","reason_id":2},{"from":"きましょう","to":"くる","reason_id":2},{"from":"ぎましょう","to":"ぐ","reason_id":2},{"from":"しましょう","to":"す","reason_id":2},{"from":"しましょう","to":"する","reason_id":2},{"from":"ちましょう","to":"つ","reason_id":2},{"from":"にましょう","to":"ぬ","reason_id":2},{"from":"びましょう","to":"ぶ","reason_id":2},{"from":"みましょう","to":"む","reason_id":2},{"from":"りましょう","to":"る","reason_id":2}],"4":[{"from":"いじゃう","to":"ぐ","reason_id":3},{"from":"いすぎる","to":"う","reason_id":4},{"from":"いちゃう","to":"く","reason_id":3},{"from":"いなさい","to":"う","reason_id":5},{"from":"いました","to":"う","reason_id":6},{"from":"いません","to":"う","reason_id":1},{"from":"かったら","to":"い","reason_id":7},{"from":"かったり","to":"い","reason_id":8},{"from":"きすぎる","to":"く","reason_id":4},{"from":"きすぎる","to":"くる","reason_id":4},{"from":"ぎすぎる","to":"ぐ","reason_id":4},{"from":"きちゃう","to":"くる","reason_id":3},{"from":"きなさい","to":"く","reason_id":5},{"from":"きなさい","to":"くる","reason_id":5},{"from":"ぎなさい","to":"ぐ","reason_id":5},{"from":"きました","to":"く","reason_id":6},{"from":"きました","to":"くる","reason_id":6},{"from":"ぎました","to":"ぐ","reason_id":6},{"from":"きません","to":"く","reason_id":1},{"from":"きません","to":"くる","reason_id":1},{"from":"ぎません","to":"ぐ","reason_id":1},{"from":"こさせる","to":"くる","reason_id":9},{"from":"こられる","to":"くる","reason_id":10},{"from":"しすぎる","to":"す","reason_id":4},{"from":"しすぎる","to":"する","reason_id":4},{"from":"しちゃう","to":"す","reason_id":3},{"from":"しちゃう","to":"する","reason_id":3},{"from":"しなさい","to":"す","reason_id":5},{"from":"しなさい","to":"する","reason_id":5},{"from":"しました","to":"す","reason_id":6},{"from":"しました","to":"する","reason_id":6},{"from":"しません","to":"す","reason_id":1},{"from":"しません","to":"する","reason_id":1},{"from":"ちすぎる","to":"つ","reason_id":4},{"from":"ちなさい","to":"つ","reason_id":5},{"from":"ちました","to":"つ","reason_id":6},{"from":"ちません","to":"つ","reason_id":1},{"from":"っちゃう","to":"う","reason_id":3},{"from":"っちゃう","to":"く","reason_id":3},{"from":"っちゃう","to":"つ","reason_id":3},{"from":"っちゃう","to":"る","reason_id":3},{"from":"にすぎる","to":"ぬ","reason_id":4},{"from":"になさい","to":"ぬ","reason_id":5},{"from":"にました","to":"ぬ","reason_id":6},{"from":"にません","to":"ぬ","reason_id":1},{"from":"びすぎる","to":"ぶ","reason_id":4},{"from":"びなさい","to":"ぶ","reason_id":5},{"from":"びました","to":"ぶ","reason_id":6},{"from":"びません","to":"ぶ","reason_id":1},{"from":"ましょう","to":"る","reason_id":2},{"from":"みすぎる","to":"む","reason_id":4},{"from":"みなさい","to":"む","reason_id":5},{"from":"みました","to":"む","reason_id":6},{"from":"みません","to":"む","reason_id":1},{"from":"りすぎる","to":"る","reason_id":4},{"from":"りなさい","to":"る","reason_id":5},{"from":"りました","to":"る","reason_id":6},{"from":"りません","to":"る","reason_id":1},{"from":"んじゃう","to":"ぬ","reason_id":3},{"from":"んじゃう","to":"ぶ","reason_id":3},{"from":"んじゃう","to":"む","reason_id":3}],"3":[{"from":"いそう","to":"う","reason_id":11},{"from":"いたい","to":"う","reason_id":12},{"from":"いたら","to":"く","reason_id":7},{"from":"いだら","to":"ぐ","reason_id":7},{"from":"いたり","to":"く","reason_id":8},{"from":"いだり","to":"ぐ","reason_id":8},{"from":"います","to":"う","reason_id":13},{"from":"かせる","to":"く","reason_id":9},{"from":"がせる","to":"ぐ","reason_id":9},{"from":"かった","to":"い","reason_id":14},{"from":"かない","to":"く","reason_id":15},{"from":"がない","to":"ぐ","reason_id":15},{"from":"かれる","to":"く","reason_id":16},{"from":"がれる","to":"ぐ","reason_id":16},{"from":"きそう","to":"く","reason_id":11},{"from":"きそう","to":"くる","reason_id":11},{"from":"ぎそう","to":"ぐ","reason_id":11},{"from":"きたい","to":"く","reason_id":12},{"from":"きたい","to":"くる","reason_id":12},{"from":"ぎたい","to":"ぐ","reason_id":12},{"from":"きたら","to":"くる","reason_id":7},{"from":"きたり","to":"くる","reason_id":8},{"from":"きます","to":"く","reason_id":13},{"from":"きます","to":"くる","reason_id":13},{"from":"ぎます","to":"ぐ","reason_id":13},{"from":"くない","to":"い","reason_id":15},{"from":"ければ","to":"い","reason_id":17},{"from":"こない","to":"くる","reason_id":15},{"from":"こよう","to":"くる","reason_id":18},{"from":"これる","to":"くる","reason_id":19},{"from":"させる","to":"する","reason_id":9},{"from":"させる","to":"る","reason_id":9},{"from":"さない","to":"す","reason_id":15},{"from":"される","to":"す","reason_id":20},{"from":"される","to":"する","reason_id":16},{"from":"しそう","to":"す","reason_id":11},{"from":"しそう","to":"する","reason_id":11},{"from":"したい","to":"す","reason_id":12},{"from":"したい","to":"する","reason_id":12},{"from":"したら","to":"す","reason_id":7},{"from":"したら","to":"する","reason_id":7},{"from":"したり","to":"す","reason_id":8},{"from":"したり","to":"する","reason_id":8},{"from":"しない","to":"する","reason_id":15},{"from":"します","to":"す","reason_id":13},{"from":"します","to":"する","reason_id":13},{"from":"しよう","to":"する","reason_id":18},{"from":"すぎる","to":"い","reason_id":4},{"from":"すぎる","to":"る","reason_id":4},{"from":"たせる","to":"つ","reason_id":9},{"from":"たない","to":"つ","reason_id":15},{"from":"たれる","to":"つ","reason_id":16},{"from":"ちそう","to":"つ","reason_id":11},{"from":"ちたい","to":"つ","reason_id":12},{"from":"ちます","to":"つ","reason_id":13},{"from":"ちゃう","to":"る","reason_id":3},{"from":"ったら","to":"う","reason_id":7},{"from":"ったら","to":"つ","reason_id":7},{"from":"ったら","to":"る","reason_id":7},{"from":"ったり","to":"う","reason_id":8},{"from":"ったり","to":"つ","reason_id":8},{"from":"ったり","to":"る","reason_id":8},{"from":"なさい","to":"る","reason_id":5},{"from":"なせる","to":"ぬ","reason_id":9},{"from":"なない","to":"ぬ","reason_id":15},{"from":"なれる","to":"ぬ","reason_id":16},{"from":"にそう","to":"ぬ","reason_id":11},{"from":"にたい","to":"ぬ","reason_id":12},{"from":"にます","to":"ぬ","reason_id":13},{"from":"ばせる","to":"ぶ","reason_id":9},{"from":"ばない","to":"ぶ","reason_id":15},{"from":"ばれる","to":"ぶ","reason_id":16},{"from":"びそう","to":"ぶ","reason_id":11},{"from":"びたい","to":"ぶ","reason_id":12},{"from":"びます","to":"ぶ","reason_id":13},{"from":"ました","to":"る","reason_id":6},{"from":"ませる","to":"む","reason_id":9},{"from":"ません","to":"る","reason_id":1},{"from":"まない","to":"む","reason_id":15},{"from":"まれる","to":"む","reason_id":16},{"from":"みそう","to":"む","reason_id":11},{"from":"みたい","to":"む","reason_id":12},{"from":"みます","to":"む","reason_id":13},{"from":"らせる","to":"る","reason_id":9},{"from":"らない","to":"る","reason_id":15},{"from":"られる","to":"る","reason_id":10},{"from":"りそう","to":"る","reason_id":11},{"from":"りたい","to":"る","reason_id":12},{"from":"ります","to":"る","reason_id":13},{"from":"わせる","to":"う","reason_id":9},{"from":"わない","to":"う","reason_id":15},{"from":"われる","to":"う","reason_id":16},{"from":"んだら","to":"ぬ","reason_id":7},{"from":"んだら","to":"ぶ","reason_id":7},{"from":"んだら","to":"む","reason_id":7},{"from":"んだり","to":"ぬ","reason_id":8},{"from":"んだり","to":"ぶ","reason_id":8},{"from":"んだり","to":"む","reason_id":8}],"2":[{"from":"いた","to":"く","reason_id":14},{"from":"いだ","to":"ぐ","reason_id":14},{"from":"いて","to":"く","reason_id":21},{"from":"いで","to":"ぐ","reason_id":21},{"from":"えば","to":"う","reason_id":17},{"from":"える","to":"う","reason_id":19},{"from":"おう","to":"う","reason_id":18},{"from":"かず","to":"く","reason_id":22},{"from":"がず","to":"ぐ","reason_id":22},{"from":"きた","to":"くる","reason_id":14},{"from":"きて","to":"くる","reason_id":21},{"from":"くて","to":"い","reason_id":21},{"from":"けば","to":"く","reason_id":17},{"from":"げば","to":"ぐ","reason_id":17},{"from":"ける","to":"く","reason_id":19},{"from":"げる","to":"ぐ","reason_id":19},{"from":"こい","to":"くる","reason_id":23},{"from":"こう","to":"く","reason_id":18},{"from":"ごう","to":"ぐ","reason_id":18},{"from":"こず","to":"くる","reason_id":22},{"from":"さず","to":"す","reason_id":22},{"from":"した","to":"す","reason_id":14},{"from":"した","to":"する","reason_id":14},{"from":"して","to":"す","reason_id":21},{"from":"して","to":"する","reason_id":21},{"from":"しろ","to":"する","reason_id":23},{"from":"せず","to":"する","reason_id":22},{"from":"せば","to":"す","reason_id":17},{"from":"せよ","to":"する","reason_id":23},{"from":"せる","to":"す","reason_id":19},{"from":"そう","to":"い","reason_id":11},{"from":"そう","to":"す","reason_id":18},{"from":"そう","to":"る","reason_id":11},{"from":"たい","to":"る","reason_id":12},{"from":"たず","to":"つ","reason_id":22},{"from":"たら","to":"る","reason_id":7},{"from":"たり","to":"る","reason_id":8},{"from":"った","to":"う","reason_id":14},{"from":"った","to":"く","reason_id":14},{"from":"った","to":"つ","reason_id":14},{"from":"った","to":"る","reason_id":14},{"from":"って","to":"う","reason_id":21},{"from":"って","to":"く","reason_id":21},{"from":"って","to":"つ","reason_id":21},{"from":"って","to":"る","reason_id":21},{"from":"てば","to":"つ","reason_id":17},{"from":"てる","to":"つ","reason_id":19},{"from":"とう","to":"つ","reason_id":18},{"from":"ない","to":"る","reason_id":15},{"from":"なず","to":"ぬ","reason_id":22},{"from":"ねば","to":"ぬ","reason_id":17},{"from":"ねる","to":"ぬ","reason_id":19},{"from":"のう","to":"ぬ","reason_id":18},{"from":"ばず","to":"ぶ","reason_id":22},{"from":"べば","to":"ぶ","reason_id":17},{"from":"べる","to":"ぶ","reason_id":19},{"from":"ぼう","to":"ぶ","reason_id":18},{"from":"ます","to":"る","reason_id":13},{"from":"まず","to":"む","reason_id":22},{"from":"めば","to":"む","reason_id":17},{"from":"める","to":"む","reason_id":19},{"from":"もう","to":"む","reason_id":18},{"from":"よう","to":"る","reason_id":18},{"from":"らず","to":"る","reason_id":22},{"from":"れば","to":"る","reason_id":17},{"from":"れる","to":"る","reason_id":19},{"from":"ろう","to":"る","reason_id":18},{"from":"わず","to":"う","reason_id":22},{"from":"んだ","to":"ぬ","reason_id":14},{"from":"んだ","to":"ぶ","reason_id":14},{"from":"んだ","to":"む","reason_id":14},{"from":"んで","to":"ぬ","reason_id":21},{"from":"んで","to":"ぶ","reason_id":21},{"from":"んで","to":"む","reason_id":21}],"1":[{"from":"い","to":"いる","reason_id":24},{"from":"い","to":"う","reason_id":24},{"from":"い","to":"る","reason_id":23},{"from":"え","to":"う","reason_id":23},{"from":"え","to":"える","reason_id":24},{"from":"き","to":"きる","reason_id":24},{"from":"き","to":"く","reason_id":24},{"from":"ぎ","to":"ぎる","reason_id":24},{"from":"ぎ","to":"ぐ","reason_id":24},{"from":"く","to":"い","reason_id":25},{"from":"け","to":"く","reason_id":23},{"from":"け","to":"ける","reason_id":24},{"from":"げ","to":"ぐ","reason_id":23},{"from":"げ","to":"げる","reason_id":24},{"from":"さ","to":"い","reason_id":26},{"from":"し","to":"す","reason_id":24},{"from":"じ","to":"じる","reason_id":24},{"from":"ず","to":"る","reason_id":22},{"from":"せ","to":"す","reason_id":23},{"from":"せ","to":"せる","reason_id":24},{"from":"ぜ","to":"ぜる","reason_id":24},{"from":"た","to":"る","reason_id":14},{"from":"ち","to":"ちる","reason_id":24},{"from":"ち","to":"つ","reason_id":24},{"from":"て","to":"つ","reason_id":23},{"from":"て","to":"てる","reason_id":24},{"from":"て","to":"る","reason_id":21},{"from":"で","to":"でる","reason_id":24},{"from":"な","to":"","reason_id":27},{"from":"に","to":"にる","reason_id":24},{"from":"に","to":"ぬ","reason_id":24},{"from":"ね","to":"ぬ","reason_id":23},{"from":"ね","to":"ねる","reason_id":24},{"from":"ひ","to":"ひる","reason_id":24},{"from":"び","to":"びる","reason_id":24},{"from":"び","to":"ぶ","reason_id":24},{"from":"へ","to":"へる","reason_id":24},{"from":"べ","to":"ぶ","reason_id":23},{"from":"べ","to":"べる","reason_id":24},{"from":"み","to":"みる","reason_id":24},{"from":"み","to":"む","reason_id":24},{"from":"め","to":"む","reason_id":23},{"from":"め","to":"める","reason_id":24},{"from":"よ","to":"る","reason_id":23},{"from":"り","to":"りる","reason_id":24},{"from":"り","to":"る","reason_id":24},{"from":"れ","to":"る","reason_id":23},{"from":"れ","to":"れる","reason_id":24},{"from":"ろ","to":"る","reason_id":23}]}}

data/lib/deinflect_to_json.rb

@@ -0,0 +1,32 @@
+#encoding: utf-8
+require 'json'
+
+def parse(fpath)
+  reasons = []
+  rules_hash = {}
+  File.open(fpath).each_with_index do |line, i|
+    next  if i == 0 # Skip header
+    parts = line.strip.split(/\t/)
+    # Reasons are listed at the top of the file and are not tab-separated
+    if parts.size == 1
+      reasons << parts[0]
+    # Rules are tab-separated in the following format:
+    # <from>\t<to>\t<type>\t<reason_index>
+    else
+      from = parts.first
+      rules_hash[from.size] ||= []
+      rules_hash[from.size] << {
+        :from => from,
+        :to => parts[1],
+        :reason_id => parts[3].to_i
+      }
+    end
+  end
+
+  {:reasons => reasons, :rules => rules_hash}
+end
+
+root = File.expand_path(File.dirname(__FILE__))
+File.open(File.join(root, 'data/deinflect.json'), 'w') do |f|
+  f.write(parse(File.join(root, 'data/deinflect.dat')).to_json)
+end

data/lib/japanese_deinflector/version.rb

@@ -0,0 +1,3 @@
+class JapaneseDeinflector
+  VERSION = "0.0.1"
+end

data/lib/japanese_deinflector.rb

@@ -0,0 +1,41 @@
+#encoding: utf-8
+require 'json'
+require "japanese_deinflector/version"
+
+class JapaneseDeinflector
+  def initialize
+    File.open(File.join(File.expand_path(File.dirname(__FILE__)), 'data/deinflect.json')) do |f|
+      rules_and_reasons = JSON.parse(f.read, :symbolize_names => true)
+      @reasons = rules_and_reasons[:reasons]
+      # Convert hash keys to integers
+      @rules = {}
+      # Convert hash keys from something like :"9" -> 9
+      rules_and_reasons[:rules].each do |size, rules|
+        @rules[size.to_s.to_i] = rules
+      end
+    end
+  end
+
+  def deinflect(word)
+    possibilities = []
+    rules_less_than_size(word.size).each do |size, rules|
+      ending = word[-size..-1]
+      rules.each do |rule|
+        next  unless ending == rule[:from]
+        deinflected_word = "#{word[0..-size-1]}#{rule[:to]}"
+        next  if possibilities.include?(deinflected_word)
+        # Weight is between 0 and 1, 1 being a higher chance of actual deinflection
+        weight = (Float(size) / word.size).round(3)
+        reason = @reasons[rule[:reason_id]]
+        possibilities << {:weight => weight, :word => deinflected_word, :reason => reason}
+      end
+    end
+    possibilities
+  end
+
+  private
+
+  def rules_less_than_size(max_size)
+    @rules.clone.keep_if{|size, rules| size < max_size}
+  end
+end

data/spec/japanese_deinflector_spec.rb

@@ -0,0 +1,70 @@
+#encoding: utf-8
+require 'spec_helper'
+
+describe JapaneseDeinflector do
+  subject{ JapaneseDeinflector.new }
+
+  it "deinflects plain positive verbs" do
+    # progressive tense
+    #subject.deinflect("見ている").first[:word].should == "見る"
+    #subject.deinflect("歌っている").first[:word].should == "歌う"
+
+    # past tense
+    subject.deinflect("見た").first[:word].should == "見る"
+    subject.deinflect("歌った").first[:word].should == "歌う"
+  end
+
+  it "deinflects polite positive verbs" do
+    # present tense
+    subject.deinflect("見ます").first[:word].should == "見る"
+    subject.deinflect("歌います").first[:word].should == "歌う"
+
+    # progressive tense
+    #subject.deinflect("見ています").first[:word].should == "見る"
+    #subject.deinflect("歌っています").first[:word].should == "歌う"
+
+    # past tense
+    subject.deinflect("見ました").first[:word].should == "見る"
+    subject.deinflect("歌いました").first[:word].should == "歌う"
+  end
+
+  it "deinflects plain negative verbs" do
+    # present tense
+    subject.deinflect("見ない").first[:word].should == "見る"
+    subject.deinflect("歌わない").first[:word].should == "歌う"
+
+    # progressive tense
+    #subject.deinflect("見ていない").first[:word].should == "見る"
+    #subject.deinflect("歌っていない").first[:word].should == "歌う"
+
+    # past tense
+    #subject.deinflect("見なかった").first[:word].should == "見る"
+    #subject.deinflect("歌わなかった").first[:word].should == "歌う"
+  end
+
+  it "deinflects polite negative formal verbs" do
+    # present tense
+    subject.deinflect("見ません").first[:word].should == "見る"
+    subject.deinflect("歌いません").first[:word].should == "歌う"
+
+    # progressive tense
+    #subject.deinflect("見ていません").first[:word].should == "見る"
+    #subject.deinflect("歌っていません").first[:word].should == "歌う"
+
+    # past tense
+    subject.deinflect("見ませんでした").first[:word].should == "見る"
+    subject.deinflect("歌いませんでした").first[:word].should == "歌う"
+  end
+
+  it "deinflects polite negative formal adjectives" do
+    # present tense
+    subject.deinflect("嬉しくありません").first[:word].should == "嬉しい"
+
+    # past tense
+    subject.deinflect("嬉しくありませんでした").first[:word].should == "嬉しい"
+  end
+
+  it "deinflects imperative verbs" do
+    subject.deinflect("歌って").first[:word].should == "歌う"
+  end
+end

data/spec/spec_helper.rb

@@ -0,0 +1,13 @@
+require 'japanese_deinflector'
+
+# This file was generated by the `rspec --init` command. Conventionally, all
+# specs live under a `spec` directory, which RSpec adds to the `$LOAD_PATH`.
+# Require this file using `require "spec_helper.rb"` to ensure that it is only
+# loaded once.
+#
+# See http://rubydoc.info/gems/rspec-core/RSpec/Core/Configuration
+RSpec.configure do |config|
+  config.treat_symbols_as_metadata_keys_with_true_values = true
+  config.run_all_when_everything_filtered = true
+  config.filter_run :focus
+end

metadata

@@ -0,0 +1,108 @@
+--- !ruby/object:Gem::Specification
+name: japanese_deinflector
+version: !ruby/object:Gem::Version
+  version: 0.0.1
+  prerelease: 
+platform: ruby
+authors:
+- Peter Graham
+autorequire: 
+bindir: bin
+cert_chain: []
+date: 2012-08-28 00:00:00.000000000 Z
+dependencies:
+- !ruby/object:Gem::Dependency
+  name: json
+  requirement: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+- !ruby/object:Gem::Dependency
+  name: rake
+  requirement: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+- !ruby/object:Gem::Dependency
+  name: rspec
+  requirement: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+description: Deinflect (unconjugate/undecline) Japanese words.
+email:
+- pete@gigadrill.com
+executables: []
+extensions: []
+extra_rdoc_files: []
+files:
+- .gitignore
+- .rspec
+- .travis.yml
+- Gemfile
+- Rakefile
+- japanese_deinflector.gemspec
+- lib/data/deinflect.dat
+- lib/data/deinflect.json
+- lib/deinflect_to_json.rb
+- lib/japanese_deinflector.rb
+- lib/japanese_deinflector/version.rb
+- spec/japanese_deinflector_spec.rb
+- spec/spec_helper.rb
+homepage: http://github.com/6/japanese_deinflector
+licenses: []
+post_install_message: 
+rdoc_options: []
+require_paths:
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+requirements: []
+rubyforge_project: 
+rubygems_version: 1.8.21
+signing_key: 
+specification_version: 3
+summary: Deinflect (unconjugate/undecline) Japanese words.
+test_files:
+- spec/japanese_deinflector_spec.rb
+- spec/spec_helper.rb