okura 0.0.0

data/test/okura_spec.rb

@@ -0,0 +1,721 @@
+#-*- coding:utf-8
+require File.join(File.dirname(__FILE__),'spec_helper.rb')
+require File.join(File.dirname(__FILE__),'..','lib','okura')
+require File.join(File.dirname(__FILE__),'..','lib','okura','parser')
+require File.join(File.dirname(__FILE__),'..','lib','okura','serializer')
+
+require 'tmpdir'
+
+def with_dict_dir &block
+  Dir.mktmpdir {|src_dir|
+    Dir.mktmpdir {|bin_dir|
+      yield src_dir,bin_dir
+    }
+  }
+end
+
+def set_content(dir,filename,content)
+  File.open(File.join(dir,filename),'w'){|file|
+    file.write content
+  }
+end
+
+def as_io str
+  StringIO.new str
+end
+def w surface,l,r,cost
+  l=f(l) unless l.respond_to? :id
+  r=f(r) unless r.respond_to? :id
+  Okura::Word.new surface,l,r,cost
+end
+def f id,name="F#{id}"
+  Okura::Feature.new id,name
+end
+def n *args
+  Okura::Node.new *args
+end
+
+describe Okura::Parser do
+  describe 'Matrix' do
+    it 'MeCab形式のMatrixファイルを読める' do
+      parser=Okura::Parser::Matrix.new as_io(<<-EOS)
+2 3
+0 0 0
+0 1 1
+1 0 2
+1 1 3
+1 2 10
+      EOS
+      parser.rid_size.should == 2
+      parser.lid_size.should == 3
+      parser.each.to_a.should == [
+        [0,0,0],
+        [0,1,1],
+        [1,0,2],
+        [1,1,3],
+        [1,2,10]
+      ]
+    end
+  end
+  describe 'Word' do
+    it 'MeCab形式の単語ファイルを読める' do
+      parser=Okura::Parser::Word.new as_io(<<-EOS)
+あがなう,854,458,6636,動詞,自立,*,*,五段・ワ行促音便,基本形,あがなう,アガナウ,アガナウ,あがなう/購う/贖う,
+あがめる,645,546,1234,動詞,自立,*,*,一段,基本形,あがめる,アガメル,アガメル,あがめる/崇める,
+      EOS
+      parser.each.to_a.map{|x|x[0..3]}.should == [
+        ['あがなう',854,458,6636],
+        ['あがめる',645,546,1234]
+      ]
+    end
+    it 'ダブルクオートでエスケープされた単語定義も扱える'
+  end
+  describe 'Feature' do
+    it 'MeCab形式の品詞ファイルを読める' do
+      parser=Okura::Parser::Feature.new as_io(<<-EOS)
+0 BOS/EOS,*,*,*,*,*,BOS/EOS
+1 その他,間投,*,*,*,*,*
+      EOS
+      parser.each.to_a.should == [
+        [0,'BOS/EOS,*,*,*,*,*,BOS/EOS'],
+        [1,'その他,間投,*,*,*,*,*']
+      ]
+    end
+  end
+  describe 'CharType' do
+    it 'MeCab形式の文字種定義ファイルを読める' do
+      parser=Okura::Parser::CharType.new
+      h={single:[],range:[],type:[]}
+      parser.on_mapping_single {|code,type,ctypes| h[:single]<<[code,type,ctypes]}
+      parser.on_mapping_range{|from,to,type,ctypes| h[:range]<<[from,to,type,ctypes]}
+      parser.on_chartype_def{|name,invoke,group,length| h[:type]<<[name,invoke,group,length]}
+
+      parser.parse_all as_io(<<-EOS)
+DEFAULT        0 1 0  # DEFAULT is a mandatory category!
+KATAKANA       1 0 2
+
+0x000D SPACE  # CR
+0x003A..0x0040 SYMBOL
+# KANJI
+0x5146 KANJINUMERIC KANJI
+      EOS
+
+      h[:single].should == [
+        [0x000D, 'SPACE', []],
+        [0x5146, 'KANJINUMERIC', %w(KANJI)]
+      ]
+      h[:range].should == [
+        [0x003A, 0x0040, 'SYMBOL', []]
+      ]
+      h[:type].should == [
+        ['DEFAULT', false, true, 0],
+        ['KATAKANA', true, false, 2]
+      ]
+    end
+  end
+  describe 'UnkDic' do
+    it '未知語の定義を読める' do
+      parser=Okura::Parser::UnkDic.new as_io(<<-EOS)
+A,5,6,3274,記号,一般,*,*,*,*,*
+Z,9,10,5244,記号,空白,*,*,*,*,*
+      EOS
+      parser.to_a.should == [
+        ['A',5,6,3274],
+        ['Z',9,10,5244]
+      ]
+    end
+  end
+end
+
+describe 'Compile and load' do
+  describe Okura::Serializer::FormatInfo do
+    it 'シリアライズして復元できる' do
+      info=Okura::Serializer::FormatInfo.new
+      info.word_dic=:Naive
+      info.features=:Marshal
+      info.char_types=:Marshal
+      info.unk_dic=:Marshal
+      info.matrix=:Marshal
+
+      out=StringIO.new
+      info.compile(out)
+      out.rewind
+
+      loaded=Okura::Serializer::FormatInfo.load(out)
+      loaded.word_dic.should == :Naive
+      loaded.features.should == :Marshal
+      loaded.char_types.should == :Marshal
+      loaded.unk_dic.should == :Marshal
+      loaded.matrix.should == :Marshal
+    end
+    it '設定に基づいて辞書をコンパイル/ロードできる' do
+      with_dict_dir{|src_dir,bin_dir|
+        set_content(src_dir,'w1.csv',<<-EOS)
+w1,1,2,1000,
+        EOS
+        set_content(src_dir,'w2.csv',<<-EOS)
+w2,5,6,2000,
+w3,9,10,3000,
+        EOS
+        set_content(src_dir,'left-id.def',<<-EOS)
+1 F1
+5 F5
+9 F9
+        EOS
+        set_content(src_dir,'right-id.def',<<-EOS)
+2 F2
+6 F6
+10 F10
+        EOS
+        set_content(src_dir,'char.def',<<-EOS)
+A 0 0 1
+Z 1 1 3
+        EOS
+        set_content(src_dir,'unk.def',<<-EOS)
+A,5,6,3274,記号,一般,*,*,*,*,*
+Z,9,10,5244,記号,空白,*,*,*,*,*
+        EOS
+        set_content(src_dir,'matrix.def',<<-EOS)
+2 3
+0 0 10
+0 1 5
+        EOS
+
+        fi=Okura::Serializer::FormatInfo.new
+        fi.encoding='UTF-8'
+        fi.compile_dict(src_dir,bin_dir)
+
+        tagger=Okura::Serializer::FormatInfo.create_tagger(bin_dir)
+
+        tagger.dic.unk_dic.rule_size.should == 2
+        tagger.dic.word_dic.word_size.should == 3
+        tagger.mat.cost(0,1).should == 5
+
+        pending {
+          w2=tagger.dic.word_dic.possible_words('w2',0)[0]
+          w2.left.name.should == 'F5'
+          w2.right.name.should == 'F6'
+        }
+        penging {
+          u1=tagger.dic.unk_dic.word_templates_for('A')
+          u1.left.name.should == 'F5'
+          u1.right.name.should == 'F6'
+        }
+      }
+    end
+  end
+  describe Okura::Serializer::Features::Marshal do
+    it 'コンパイルして復元できる' do
+      serializer=Okura::Serializer::Features::Marshal.new
+      out=StringIO.new
+      serializer.compile(as_io(<<-EOS),out)
+0 BOS/EOS,*,*,*,*,*,BOS/EOS
+1 その他,間投,*,*,*,*,*
+      EOS
+      out.rewind
+
+      features=serializer.load(out)
+      features.from_id(0).text.should == 'BOS/EOS,*,*,*,*,*,BOS/EOS'
+      features.from_id(1).text.should == 'その他,間投,*,*,*,*,*'
+    end
+  end
+  describe Okura::Serializer::CharTypes::Marshal do
+    it 'コンパイルして復元できる' do
+      serializer=Okura::Serializer::CharTypes::Marshal.new
+      out=StringIO.new
+      serializer.compile(as_io(<<-EOS),out)
+DEFAULT 0 1 0  # DEFAULT is a mandatory category!
+TYPE1   1 0 0
+TYPE2   0 1 0
+TYPE3   0 1 3
+
+# comment
+
+0x0021 TYPE1
+0x0022 TYPE2 # comment
+0x0023..0x0040 TYPE3
+0x0099 TYPE1 TYPE2 # 互換カテゴリ
+0xABCd TYPE1 DEFAULT
+      EOS
+      out.rewind
+
+      cts=serializer.load(out)
+
+	  cts.type_for(0x21).name.should == 'TYPE1'
+	  cts.type_for(0x22).name.should == 'TYPE2'
+	  cts.type_for(0x23).name.should == 'TYPE3'
+	  cts.type_for(0x40).name.should == 'TYPE3'
+	  cts.type_for(0x41).name.should == 'DEFAULT'
+	  cts.type_for(0x99).name.should == 'TYPE1'
+
+	  t1,t2,t3=cts.named('TYPE1'), cts.named('TYPE2'), cts.named('TYPE3')
+
+	  t1.name.should == 'TYPE1'
+
+	  t1.invoke?.should be_true
+	  t2.invoke?.should be_false
+
+	  t1.group?.should be_false
+	  t2.group?.should be_true
+
+	  t2.length.should == 0
+	  t3.length.should == 3
+
+	  t1.should be_accept(0x21)
+	  t1.should_not be_accept(0x22)
+	  t2.should be_accept(0x22)
+
+	  t1.should be_accept(0x99)
+    end
+  end
+  shared_examples_for 'WordDic serializer' do
+    # subject : Serializer class
+    it 'コンパイルして復元できる' do
+      serializer=subject.new
+      features=Okura::Features.new
+      features.add 854,f(854)
+      features.add 458,f(458)
+      features.add 645,f(645)
+      features.add 546,f(546)
+      out=StringIO.new
+      src=<<-EOS
+あがなう,854,458,6636,動詞,自立,*,*,五段・ワ行促音便,基本形,あがなう,アガナウ,アガナウ,あがなう/購う/贖う,
+あがめる,645,546,1234,動詞,自立,*,*,一段,基本形,あがめる,アガメル,アガメル,あがめる/崇める,
+      EOS
+      serializer.compile(features,[as_io(src)],'UTF-8',out)
+      out.rewind
+      wd=serializer.load(out)
+
+      wd.possible_words('あがなう',0).should == [w('あがなう',f(854),f(458),6636)]
+      wd.possible_words('あがめる',0).should == [w('あがめる',f(645),f(546),1234)]
+      wd.possible_words('あがめる',1).should == []
+    end
+  end
+  describe Okura::Serializer::WordDic::Naive do
+    subject { Okura::Serializer::WordDic::Naive }
+    it_should_behave_like 'WordDic serializer'
+  end
+  describe Okura::Serializer::WordDic::DoubleArray do
+    subject { Okura::Serializer::WordDic::DoubleArray }
+    it_should_behave_like 'WordDic serializer'
+  end
+  describe Okura::Serializer::UnkDic::Marshal do
+    it 'コンパイルして復元できる' do
+      serializer=Okura::Serializer::UnkDic::Marshal.new
+      cts=Okura::CharTypes.new
+      cts.define_type 'A',true,false,10
+      cts.define_type 'Z',false,true,0
+      cts.define_map 0x0001,cts.named('A'),[]
+      cts.define_map 0x0002,cts.named('Z'),[]
+      features=Okura::Features.new
+      features.add 5,'F5'
+      features.add 6,'F6'
+      features.add 9,'F9'
+      features.add 10,'F10'
+      out=StringIO.new
+      serializer.compile(cts,features,as_io(<<-EOS),out)
+A,5,6,3274,記号,一般,*,*,*,*,*
+Z,9,10,5244,記号,空白,*,*,*,*,*
+      EOS
+      out.rewind
+
+      unk=serializer.load(out)
+      unk.word_templates_for('A').first.cost.should == 3274
+      unk.word_templates_for('Z').first.cost.should == 5244
+    end
+  end
+  describe Okura::Serializer::Matrix::Marshal do
+    it 'コンパイルして復元できる' do
+      serializer=Okura::Serializer::Matrix::Marshal.new
+      out=StringIO.new
+      serializer.compile(as_io(<<-EOS),out)
+2 3
+0 0 0
+0 1 1
+1 0 2
+1 1 3
+1 2 10
+      EOS
+      out.rewind
+
+      mat=serializer.load(out)
+      mat.cost(0,0).should == 0
+      mat.cost(1,2).should == 10
+    end
+  end
+end
+
+describe Okura::Matrix do
+  describe '#cost' do
+    it '渡された二つのFeature idを元にコストを返せる' do
+      m=Okura::Matrix.new 2,2
+      m.set(0,0,0)
+      m.set(0,1,1)
+      m.set(1,0,2)
+      m.set(1,1,3)
+
+      m.cost(1,1).should == 3
+    end
+  end
+end
+
+shared_examples_for 'WordDic' do
+  # subject = dict builder
+  def w surface
+    Okura::Word.new surface,f(1),f(1),1
+  end
+
+  describe '#possible_words' do
+    it '登録された単語のサイズを取得できる' do
+      subject.build.word_size.should == 0
+      subject.define w('aaa')
+      subject.define w('bbb')
+      subject.build.word_size.should == 2
+    end
+    it '同じ表記の単語を複数登録できる' do
+      w1=Okura::Word.new 'w',f(1),f(2),100
+      w2=Okura::Word.new 'w',f(10),f(20),200
+      subject.define w1
+      subject.define w1
+      subject.define w2
+
+      wd=subject.build
+
+      wd.possible_words('w',0).should == [w1,w1,w2]
+    end
+    it '文字列と位置から､辞書に登録された単語を返せる' do
+      subject.define w('aaa')
+      subject.define w('bbb')
+      subject.define w('aa')
+      subject.define w('aaaa')
+      subject.define w('aaaaa')
+
+      wd=subject.build
+
+      wd.possible_words('bbbaaa',0).should == [w('bbb')]
+      wd.possible_words('bbbaaa',1).should == []
+      wd.possible_words('bbbaaa',3).should == [w('aa'),w('aaa')]
+    end
+    it 'マルチバイト文字にも対応している' do
+      subject.define w('ニワトリ')
+      wd=subject.build
+
+      wd.possible_words('ニワトリ',0).should == [w('ニワトリ')]
+      wd.possible_words('ニワトリ',1).should == []
+    end
+    def matches words,str,dest
+      words.each{|word| subject.define w(word) }
+      dic=subject.build
+      dic.possible_words(str,0).should == dest.map{|d|w(d)}
+    end
+    it { matches %w()        , ''        , %w() }
+    it { matches %w()        , 'aaa'     , %w() }
+    it { matches %w(a)       , ''        , %w() }
+    it { matches %w(a)       , 'a'       , %w(a) }
+    it { matches %w(a)       , 'aa'      , %w(a) }
+    it { matches %w(a)       , 'b'       , %w() }
+    it { matches %w(aa)      , 'a'       , %w() }
+    it { matches %w(aa)      , 'aa'      , %w(aa) }
+    it { matches %w(aa)      , 'aaa'     , %w(aa) }
+    it { matches %w(aa)      , 'ab'      , %w() }
+    it { matches %w(a aa)    , 'a'       , %w(a) }
+    it { matches %w(a aa)    , 'aa'      , %w(a aa) }
+    it { matches %w(a aa)    , 'aaa'     , %w(a aa) }
+    it { matches %w(a aa)    , 'aab'     , %w(a aa) }
+    it { matches %w(a aa ab) , 'aab'     , %w(a aa) }
+    it { matches %w(a aa ab) , 'ab'      , %w(a ab) }
+    it { matches %w(a aa ab) , 'aa'      , %w(a aa) }
+    it { matches %w(a b)     , 'ba'      , %w(b) }
+    it { matches %w(アイウ)  , 'アイウ'  , %w(アイウ) }
+    it { matches %w(ア アイ) , 'アイウ'  , %w(ア アイ) }
+    it { matches %w(ア アイ) , 'aアイウ' , %w() }
+  end
+end
+
+describe Okura::WordDic::Naive do
+  class NaiveBuilder
+    def initialize
+      @wd=Okura::WordDic::Naive.new
+    end
+    def define *args
+      @wd.define *args
+    end
+    def build
+      @wd
+    end
+  end
+  subject { NaiveBuilder.new }
+  it_should_behave_like 'WordDic'
+end
+
+describe Okura::WordDic::DoubleArray do
+  subject { Okura::WordDic::DoubleArray::Builder.new }
+  def base(dic)
+    dic.instance_eval{@base}
+  end
+  def check(dic)
+    dic.instance_eval{@check}
+  end
+  def words(dic)
+    dic.instance_eval{@words}
+  end
+  it_should_behave_like 'WordDic'
+end
+
+describe Okura::Features do
+end
+
+describe Okura::CharTypes do
+  describe '#type_for' do
+	describe '文字に対するCharTypeが定義されていない場合' do
+	  describe '文字種DEFAULTが定義されている場合' do
+		subject {
+		  cts=Okura::CharTypes.new
+		  cts.define_type 'DEFAULT',false,false,0
+		  cts
+		}
+		it 'CharType#default_typeが返る' do
+		  subject.type_for('a'.ord).name.should == subject.default_type.name
+		end
+	  end
+	  describe '文字種DEFAULTが定義されてない場合' do
+		subject { cts=Okura::CharTypes.new }
+		it 'エラーになる' do
+		  expect { subject.type_for('a'.ord) }.to raise_error
+		end
+	  end
+	end
+  end
+  describe '#define_map' do
+	describe '互換カテゴリが指定された場合' do
+	  subject {
+		cts=Okura::CharTypes.new
+		cts.define_type 'A',true,true,10
+		cts.define_type 'B',true,true,10
+		cts.define_map 1,cts.named('A'),[cts.named('B')]
+		cts
+	  }
+	  it '互換カテゴリが正しく認識される' do
+		subject.named('A').accept?(1).should be_true
+		subject.named('B').accept?(1).should be_true
+	  end
+	end
+  end
+end
+
+describe Okura::UnkDic do
+  describe '#possible_words' do
+	describe '互換カテゴリ' do
+	  subject {
+		cts=Okura::CharTypes.new
+		cts.define_type 'KATAKANA',false,true,0
+		cts.define_type 'HIRAGANA',false,true,0
+		cts.define_map 'ア'.ord,cts.named('KATAKANA'),[]
+		cts.define_map 'ー'.ord,cts.named('HIRAGANA'),[cts.named('KATAKANA')]
+		ud=Okura::UnkDic.new cts
+		ud.define 'KATAKANA',f(10),f(20),1000
+		ud.define 'HIRAGANA',f(1),f(2),1000
+		ud
+	  }
+	  it '互換カテゴリを正しく解釈する' do
+		subject.possible_words('アーー',0,false).should == [w('アーー',10,20,1000)]
+	  end
+	end
+	describe '未知語定義' do
+	  describe '同一文字種に複数の未知語定義があった場合' do
+		subject do
+		  cts=Okura::CharTypes.new
+		  cts.define_type 'A',true,true,0
+		  cts.define_map 'A'.ord,cts.named('A'),[]
+		  ud=Okura::UnkDic.new cts
+		  ud.define 'A',f(10),f(20),1000
+		  ud.define 'A',f(11),f(21),1111
+		  ud
+		end
+		it 'すべての定義から未知語を抽出する' do
+		  subject.possible_words('A',0,false).should == [
+			w('A',10,20,1000),
+			w('A',11,21,1111)
+		  ]
+		end
+	  end
+	end
+  end
+  describe '#possible_words: 文字コードによる挙動:' do
+	subject do
+	  cts=Okura::CharTypes.new
+	  cts.define_type 'A',true,true,0
+	  cts.define_map 'あ'.ord,cts.named('A'),[]
+	  ud=Okura::UnkDic.new cts
+	  ud.define 'A',f(10),f(20),1000
+	  ud
+	end
+	describe 'UTF8文字列が来たとき' do
+	  it '正しく解析できる' do
+		subject.possible_words('あいう'.encode('UTF-8'),0,false).map(&:surface).should == %w(あ)
+	  end
+	end
+	describe 'UTF8じゃない文字列が来たとき' do
+	  it 'エラーになる' do
+		expect { subject.possible_words('あいう'.encode('SHIFT_JIS'),0,false) }.to raise_error
+	  end
+	end
+  end
+  describe '#possible_words: 先頭文字のカテゴリによる挙動:' do
+	def create_chartypes typename_under_test
+	  cts=Okura::CharTypes.new
+	  cts.define_type 'T000',false,false,0
+	  cts.define_type 'T012',false,true,2
+	  cts.define_type 'T100',true,false,0
+	  cts.define_type 'T102',true,false,2
+	  cts.define_type 'T110',true,true,0
+	  cts.define_type 'T112',true,true,2
+	  cts.define_type 'ZZZZ',true,true,2
+
+	  cts.define_map 'A'.ord,cts.named(typename_under_test),[]
+	  cts.define_map 'Z'.ord,cts.named('ZZZZ'),[]
+
+	  cts
+	end
+	def create_subject typename_under_test
+	  udic=Okura::UnkDic.new create_chartypes(typename_under_test)
+	  udic.define typename_under_test,f(10),(20),1000
+	  udic
+	end
+	describe 'invoke=0のとき' do
+	  subject { create_subject 'T012' }
+	  describe '辞書に単語がある場合' do
+		it '未知語を抽出しない' do
+		  subject.possible_words('AAA',0,true).should be_empty
+		end
+	  end
+	end
+	describe 'invoke=1のとき' do
+	  describe '辞書に単語がある場合' do
+		subject { create_subject 'T102' }
+		it 'も､未知語を抽出する' do
+		  subject.possible_words('AAAZ',0,true).should_not be_empty
+		end
+	  end
+	  describe '先頭文字のカテゴリに対応する未知語定義がなかった場合' do
+		subject { create_subject 'T112' }
+		it '未知語を抽出しない' do
+		  subject.possible_words('ZZ',0,false).should be_empty
+		end
+	  end
+	  describe '辞書に単語がない場合' do
+		describe 'group=0のとき' do
+		  describe 'length=0のとき' do
+			subject { create_subject 'T100' }
+			it '未知語を抽出しない' do
+			  subject.possible_words('AAAZ',0,false).should be_empty
+			end
+		  end
+		  describe 'length=2のとき' do
+			subject { create_subject 'T102' }
+			it '2文字までの同種文字列を未知語とする' do
+			  subject.possible_words('AAAZ',0,false).map(&:surface).should == %w(A AA)
+			end
+		  end
+		end
+		describe 'group=1のとき' do
+		  describe 'length=0のとき' do
+			subject { create_subject 'T110' }
+			it '同種の文字列を長さ制限なしでまとめて未知語とする' do
+			  subject.possible_words('AAAAAZ',0,false).map(&:surface).should == %w(AAAAA)
+			end
+			it '連続が一文字の場合も未知語として取れる' do
+			  subject.possible_words('AZZZ',0,false).map(&:surface).should == %w(A)
+			end
+			it '1文字しかなくても正しく扱える' do
+			  subject.possible_words('A',0,false).map(&:surface).should == %w(A)
+			end
+		  end
+		  describe 'length=2のとき' do
+			subject { create_subject 'T112' }
+			it 'length=0の結果に加え､2文字までの同種文字列を未知語とする' do
+			  subject.possible_words('AAAAAZ',0,false).map(&:surface).should == %w(A AA AAAAA)
+			end
+			it '1文字しかなくても正しく扱える' do
+			  subject.possible_words('A',0,false).map(&:surface).should == %w(A)
+			end
+			it '2文字しかなくても正しく扱える' do
+			  subject.possible_words('AA',0,false).map(&:surface).should == %w(A AA)
+			end
+			it '3文字しかなくても正しく扱える' do
+			  subject.possible_words('AAA',0,false).map(&:surface).should == %w(A AA AAA)
+			end
+		  end
+		end
+	  end
+	end
+  end
+end
+
+describe Okura::Tagger do
+  describe '#parse' do
+    it '文字列を解析してNodesを返せる' do
+      dic=Okura::WordDic::Naive.new
+      dic.define w('a',1,1,0)
+      dic.define w('aa',1,1,10)
+      dic.define w('b',2,2,3)
+      tagger=Okura::Tagger.new dic,nil
+
+      nodes=tagger.parse('aab')
+
+      nodes[0][0].word.should == w('BOS/EOS',0,0,0)
+      nodes[4][0].word.should == w('BOS/EOS',0,0,0)
+      nodes[1].size.should == 2
+      nodes[3][0].word.should == w('b',2,2,3)
+    end
+  end
+end
+
+describe Okura::Node do
+  describe '#make_bos_eos' do
+    describe '#length' do
+      it 'returns 1' do
+        Okura::Node.mk_bos_eos.length.should == 1
+      end
+    end
+  end
+end
+
+describe Okura::Nodes do
+  describe '#mincost_path' do
+    it '最小コストのパスを返せる' do
+      mat=Okura::Matrix.new 2,2
+      mat.set(0,1,10)
+      mat.set(1,0,10)
+      nodes=Okura::Nodes.new 3,mat
+      nodes.add(0,Okura::Node.mk_bos_eos)
+      nodes.add(1,n(w('a',1,1,10)))
+      nodes.add(1,n(w('b',1,1,0)))
+      nodes.add(2,Okura::Node.mk_bos_eos)
+
+      mcp=nodes.mincost_path
+      mcp.length.should == 3
+      mcp[0].word.surface.should == 'BOS/EOS'
+      mcp[1].word.surface.should == 'b'
+      mcp[2].word.surface.should == 'BOS/EOS'
+    end
+    it '単語長が1を超えても動く' do
+      mat=Okura::Matrix.new 2,2
+      mat.set(0,1,10)
+      mat.set(1,0,10)
+      mat.set(1,1,10)
+      nodes=Okura::Nodes.new 4,mat
+      nodes.add(0,Okura::Node.mk_bos_eos)
+      nodes.add(1,n(w('a',1,1,10)))
+      nodes.add(1,n(w('bb',1,1,0)))
+      nodes.add(2,n(w('a',1,1,10)))
+      nodes.add(3,Okura::Node.mk_bos_eos)
+
+      mcp=nodes.mincost_path
+      mcp.length.should == 3
+      mcp[0].word.surface.should == 'BOS/EOS'
+      mcp[1].word.surface.should == 'bb'
+      mcp[2].word.surface.should == 'BOS/EOS'
+    end
+  end
+end