langue-japanese 0.0.3 → 0.0.4

data/lib/langue/japanese/words/pronoun.rb

@@ -1,15 +1,13 @@
-require 'langue/japanese/words/noun'
+require 'langue/word'
 require 'langue/japanese/words/classifier'
 
 module Langue
   module Japanese
-    class Pronoun < Noun
-      class << self
-        include Classifier
+    class Pronoun < ::Langue::Pronoun
+      extend Classifier
 
-        def take(morphemes, index)
-          pronoun?(morphemes, index) ? 1 : 0
-        end
+      def self.take(morphemes, index)
+        pronoun?(morphemes, index) ? 1 : 0
       end
     end
   end

data/lib/langue/japanese/words/verb.rb

@@ -5,67 +5,76 @@ require 'langue/japanese/words/classifier'
 
 module Langue
   module Japanese
-    class Verb < Word
+    class Verb < ::Langue::Verb
       include Prefix
-      include Attribute
-
-      has :progressive, :passive, :aggressive, :negative, :perfective, :imperative
 
-      class << self
-        include Classifier
+      def extract_prefix_morphemes
+        size = 0
 
-        def take(morphemes, index)
-          if first_verb?(morphemes, index)
-            take_verb(morphemes, index)
-          elsif verb_prefix?(morphemes, index)
-            take_verb_with_prefix(morphemes, index)
-          elsif first_noun?(morphemes, index)
-            take_noun_conjunct_to_suru(morphemes, index)
-          elsif noun_prefix?(morphemes, index)
-            take_noun_with_prefix_conjunct_to_suru(morphemes, index)
-          else
-            0
-          end
+        if self.class.verb_prefix?(morphemes, size)
+          size += 1 while self.class.verb_prefix?(morphemes, size)
+        elsif self.class.noun_prefix?(morphemes, size)
+          size += 1 while self.class.noun_prefix?(morphemes, size)
         end
 
-        def take_verb(morphemes, index)
-          return 0 unless first_verb?(morphemes, index)
-          take_following_verb(morphemes, index)
-        end
+        morphemes[0, size]
+      end
 
-        def take_verb_with_prefix(morphemes, index)
-          size = 0
-          size += 1 while verb_prefix?(morphemes, index + size)
-          return 0 unless size > 0
-          next_size = take_verb(morphemes, index + size)
-          next_size > 0 ? size + next_size : 0
-        end
+      include Attribute
 
-        def take_noun_conjunct_to_suru(morphemes, index)
-          size = 0
-          size += 1 while following_noun?(morphemes, index + size)
-          return 0 unless size > 0
-          return 0 unless noun_conjunct_to_suru?(morphemes, index + size - 1)
-          return 0 unless suru_verb?(morphemes, index + size)
-          size + take_following_verb(morphemes, index + size)
-        end
+      has :progressive, :passive, :causative, :aggressive, :negative, :perfective, :imperative
+
+      extend Classifier
 
-        def take_noun_with_prefix_conjunct_to_suru(morphemes, index)
-          size = 0
-          size += 1 while noun_prefix?(morphemes, index + size)
-          return 0 unless size > 0
-          next_size = take_noun_conjunct_to_suru(morphemes, index + size)
-          next_size > 0 ? size + next_size : 0
+      def self.take(morphemes, index)
+        if first_verb?(morphemes, index)
+          take_verb(morphemes, index)
+        elsif verb_prefix?(morphemes, index)
+          take_verb_with_prefix(morphemes, index)
+        elsif first_noun?(morphemes, index)
+          take_noun_conjunct_to_suru(morphemes, index)
+        elsif noun_prefix?(morphemes, index)
+          take_noun_with_prefix_conjunct_to_suru(morphemes, index)
+        else
+          0
         end
+      end
 
-        private
+      def self.take_verb(morphemes, index)
+        return 0 unless first_verb?(morphemes, index)
+        take_following_verb(morphemes, index)
+      end
 
-        def take_following_verb(morphemes, index)
-          size = 1
-          size += 1 while following_verb?(morphemes, index + size) || conjunctive_particle?(morphemes, index + size) && following_verb?(morphemes, index + size + 1)
-          size += 1 while auxiliary_verb?(morphemes, index + size)
-          size
-        end
+      def self.take_verb_with_prefix(morphemes, index)
+        size = 0
+        size += 1 while verb_prefix?(morphemes, index + size)
+        return 0 unless size > 0
+        next_size = take_verb(morphemes, index + size)
+        next_size > 0 ? size + next_size : 0
+      end
+
+      def self.take_noun_conjunct_to_suru(morphemes, index)
+        size = 0
+        size += 1 while following_noun?(morphemes, index + size)
+        return 0 unless size > 0
+        return 0 unless noun_conjunct_to_suru?(morphemes, index + size - 1)
+        return 0 unless suru_verb?(morphemes, index + size)
+        size + take_following_verb(morphemes, index + size)
+      end
+
+      def self.take_noun_with_prefix_conjunct_to_suru(morphemes, index)
+        size = 0
+        size += 1 while noun_prefix?(morphemes, index + size)
+        return 0 unless size > 0
+        next_size = take_noun_conjunct_to_suru(morphemes, index + size)
+        next_size > 0 ? size + next_size : 0
+      end
+
+      def self.take_following_verb(morphemes, index)
+        size = 1
+        size += 1 while following_verb?(morphemes, index + size)
+        size += 1 while final_particle?(morphemes, index + size)
+        size
       end
 
       def key_morpheme
@@ -81,20 +90,6 @@ module Langue
 
         @key_morpheme
       end
-
-      def prefix_morphemes
-        @prefix_morphemes ||= begin
-                                size = 0
-
-                                if self.class.verb_prefix?(morphemes, size)
-                                  size += 1 while self.class.verb_prefix?(morphemes, size)
-                                elsif self.class.noun_prefix?(morphemes, size)
-                                  size += 1 while self.class.noun_prefix?(morphemes, size)
-                                end
-
-                                morphemes[0, size]
-                              end
-      end
     end
   end
 end

data/spec/langue/japanese/data.yaml

@@ -3,9 +3,12 @@
   sentences:
     -
       - [今日, Noun]
+      - [は, Particle]
       - [妹, Noun]
+      - [と, Particle]
       - [一緒, Noun]
-      - - お買い物してきた
+      - [に, Particle]
+      - - お買い物してきたよ
         - Verb
         - body: 買い物してくる
           prefix: お
@@ -22,24 +25,39 @@
   sentences:
     -
       - [例, Noun]
+      - [の, Particle]
       - [女子会, Noun]
+      - [が, Particle]
       - [延期, Noun]
+      - [に, Particle]
+      - [なって, Verb]
+      - [ちょっとホッ, Adverb]
+      - [と, Particle]
+      - [している, Verb]
 
 -
   text: こんなに部屋が寒くてはインターネットもままならぬ
   sentences:
     -
+      - [こんなに, Adverb]
       - [部屋, Noun]
+      - [が, Particle]
       - - 寒く
         - Adjective
         - body: 寒い
+      - [は, Particle]
       - [インターネット, Noun]
+      - [も, Particle]
+      - [まま, Adverb]
+      - [ならぬ, Verb]
 
 -
   text: 牛乳って意外と甘い？
   sentences:
     -
       - [牛乳, Noun]
+      - [って, Particle]
+      - [意外と, Adverb]
       - - 甘い
         - Adjective
         - body: 甘い
@@ -49,7 +67,9 @@
   text: むしろ精神的疾患とは……
   sentences:
     -
+      - [むしろ, Adverb]
       - [精神的疾患, Noun]
+      - [とは, Particle]
       - [……, Period]
 
 -
@@ -58,12 +78,16 @@
     -
       - [情報系大学生, Noun]
       - [卒業, Noun]
+      - [までに, Particle]
       - [応用情報レベル, Noun]
+      - [の, Particle]
       - [知識, Noun]
+      - [を, Particle]
       - - 習っている
         - Verb
         - body: 習う
           attributes: [progressive]
+      - [と, Particle]
       - [., Period]
 
 -
@@ -71,12 +95,16 @@
   sentences:
     -
       - [プログラマー, Noun]
+      - [の, Particle]
       - [資格, Noun]
+      - [を, Particle]
       - - 持つ
         - Verb
         - body: 持つ
       - [息子, Noun]
+      - [が, Particle]
       - [面接, Noun]
+      - [に, Particle]
       - [全て, Noun]
       - - 落ちました
         - Verb
@@ -85,40 +113,52 @@
       - [。, Period]
     -
       - [理系, Noun]
+      - [でも, Particle]
       - [旧帝, Noun]
-      - [ダメ, AdjectiveNoun]
+      - [じゃ, Particle]
+      - [ダメ, AdjectivalNoun]
+      - [か, Particle]
       - [？, Period]
 
 -
   text: ちゃんと病院いこうね…。はい…。
   sentences:
     -
+      - [ちゃんと, Adverb]
       - [病院, Noun]
-      - - いこう
+      - - いこうね
         - Verb
         - body: いく
       - […。, Period]
     -
+      - [はい, Interjection]
       - […。, Period]
 
 -
   text: んー、お昼がスープパスタだけとか。。。あふんあふん。。。後で絶対お腹すくなー。でも今あんま食べる気しにゃい。。。
   sentences:
     -
+      - [ん, Particle]
       - [お昼, Noun]
+      - [が, Particle]
       - [スープパスタ, Noun]
+      - [だけとか, Particle]
       - [。。。, Period]
     -
       - [あふん, Verb]
+      - [ふん, Interjection]
       - [。。。, Period]
     -
+      - [後で, Adverb]
       - [絶対, Noun]
       - [お腹, Noun]
       - - すく
         - Adjective
         - body: すい
+      - [なー, Particle]
       - [。, Period]
     -
+      - [でも, Conjunction]
       - [今あんま, Noun]
       - - 食べる
         - Verb
@@ -130,19 +170,24 @@
   text: iPhone 4Sにただで替えられるって聞いたけど、本当に使うかは疑問なので……
   sentences:
     -
-      - [iphone4s, Noun]
+      - [iPhone4S, Noun]
+      - [に, Particle]
       - [ただ, Noun]
+      - [で, Particle]
       - - 替えられる
         - Verb
         - body: 替える
           attributes: [passive]
+      - [って, Particle]
       - - 聞いた
         - Verb
         - body: 聞く
           attributes: [perfective]
-      - - 使う
+      - [本当に, Adverb]
+      - - 使うか
         - Verb
         - body: 使う
+      - [は, Particle]
       - [疑問, Noun]
       - [……, Period]
 
@@ -151,6 +196,8 @@
   sentences:
     -
       - [俺, Pronoun]
+      - [の, Particle]
+      - [に, Particle]
       - - ご奉仕しろ
         - Verb
         - body: 奉仕する
@@ -163,6 +210,7 @@
   sentences:
     -
       - [風, Noun]
+      - [と共に, Particle]
       - - 去らぬ
         - Verb
         - body: 去る

data/spec/langue/japanese/inflector/inflection_spec.rb

@@ -0,0 +1,80 @@
+# -*- coding: utf-8 -*-
+require 'langue/japanese/inflector/inflection'
+
+describe Langue::Japanese::Inflector::Inflection, '#inflect' do
+  before do
+    suffixes = {
+      'form'       => 'form_suffix',
+      'proc'       => lambda { |options| options[:suffix] },
+      '連用タ接続' => 'ta_suffix',
+      '連用テ接続' => 'te_suffix'
+    }
+
+    @inflection = described_class.new 'suffix', suffixes
+  end
+
+  it 'inflects the word to the inflectional form' do
+    word = @inflection.inflect('body-suffix', 'form')
+    word.should == 'body-form_suffix'
+  end
+
+  it 'adds value returning from calling of its Proc as suffix to the word if the inflectional form is an instance of Proc' do
+    word = @inflection.inflect('body-suffix', 'proc', :suffix => 'proc_suffix')
+    word.should == 'body-proc_suffix'
+  end
+
+  it 'raises ArgumentError if the word does not end with the base suffix' do
+    lambda { @inflection.inflect('body-suffi', 'form') }.should raise_error(ArgumentError, 'the word does not end with "suffix"')
+  end
+
+  it 'raises ArgumentError if the inflectional form does not defined' do
+    lambda { @inflection.inflect('body-suffix', 'form1') }.should raise_error(ArgumentError, '"form1" inflectional form does not defined in the inflection')
+  end
+
+  context 'with :following option' do
+    it 'adds value of :following option to the word' do
+      word = @inflection.inflect('body-suffix', 'form', :following => '-following')
+      word.should == 'body-form_suffix-following'
+    end
+
+    it 'converts to "タ行" from "ダ行" in first character of the following word if the inflectional form is "連用タ接続" or "連用テ接続"' do
+      {
+        '連用タ接続' => 'ta_suffix',
+        '連用テ接続' => 'te_suffix'
+      }.each do |form, suffix|
+        {
+          'だ' => 'た',
+          'じ' => 'ち',
+          'ぢ' => 'ち',
+          'で' => 'て',
+          'ど' => 'と'
+        }.each do |following, converted_following|
+          word = @inflection.inflect('body-suffix', form, :following => following)
+          word.should == "body-#{suffix}#{converted_following}"
+        end
+      end
+    end
+
+    it 'converts to "ダ行" from "タ行" in first character of the following word if the inflectional form is "連用タ接続" or "連用テ接続"' do
+      inflection = described_class.new('suffix', {
+        '連用タ接続' => ['ta_suffix', true],
+        '連用テ接続' => ['te_suffix', true]
+      })
+
+      {
+        '連用タ接続' => 'ta_suffix',
+        '連用テ接続' => 'te_suffix'
+      }.each do |form, suffix|
+        {
+          'た' => 'だ',
+          'ち' => 'じ',
+          'て' => 'で',
+          'と' => 'ど'
+        }.each do |following, converted_following|
+          word = inflection.inflect('body-suffix', form, :following => following)
+          word.should == "body-#{suffix}#{converted_following}"
+        end
+      end
+    end
+  end
+end

data/spec/langue/japanese/inflector/inflections_spec.rb

@@ -0,0 +1,83 @@
+require 'langue/japanese/inflector/inflections'
+
+describe Langue::Japanese::Inflector::Inflections, '#inflection' do
+  before do
+    @inflections = described_class.new
+  end
+
+  it 'calls Langue::Japanese::Inflector::Inflection.new with the base suffix and the suffixes' do
+    Langue::Japanese::Inflector::Inflection.should_receive(:new).with('suffix', 'form' => 'suffix')
+    @inflections.inflection('classification', 'suffix', 'form' => 'suffix')
+  end
+
+  it 'defines the inflection' do
+    @inflections.inflection('classification', 'suffix', 'form' => 'suffix')
+    @inflections['classification'].should be_a(Langue::Japanese::Inflector::Inflection)
+  end
+
+  context 'in call #category' do
+    it 'does not raise ArgumentError if defined the inflectional forms in just proportion' do
+      lambda {
+        @inflections.category 'form' do
+          inflection 'classification', 'suffix', 'form' => 'suffix'
+        end
+      }.should_not raise_error(ArgumentError)
+    end
+
+    it 'raises ArgumentError if the inflectional forms is excess' do
+      lambda {
+        @inflections.category *%w(form1 form2 form3) do
+          inflection 'classification', 'suffix'
+        end
+      }.should raise_error(ArgumentError, 'form1, form2 and form3 has not been defined')
+    end
+
+    it 'raises ArgumentError if the inflectional forms is inadequate' do
+      lambda {
+        @inflections.category 'form' do
+          inflection 'classification', 'suffix', {
+            'form'  => 'suffix',
+            'form1' => 'suffix1'
+          }
+        end
+      }.should raise_error(ArgumentError, 'form1 should not be defined')
+    end
+
+    it 'raises ArgumentError if the inflectional forms is excess and inadequate' do
+      lambda {
+        @inflections.category *%w(form1 form2) do
+          inflection 'classification', 'suffix', {
+            'form3' => 'suffix3',
+            'form4' => 'suffix4'
+          }
+        end
+      }.should raise_error(ArgumentError, 'form1 and form2 has not been defined, and form3 and form4 should not be defined')
+    end
+  end
+end
+
+describe Langue::Japanese::Inflector::Inflections, '#category' do
+  before do
+    @inflections = described_class.new
+  end
+
+  it 'calls the block in scope of the instance' do
+    matcher = equal(@inflections)
+    @inflections.category { should matcher }
+  end
+end
+
+describe Langue::Japanese::Inflector::Inflections, '#categorizing?' do
+  before do
+    @inflections = described_class.new
+  end
+
+  it 'returns false after initializing' do
+    @inflections.should_not be_categorizing
+  end
+
+  it 'returns true in call #category' do
+    matcher = be_categorizing
+    @inflections.category { should matcher }
+  end
+end