light-mecab 0.0.4 → 0.0.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 68e3e72d4f025c81ade1963d6d0d556ceb91e2ad
-  data.tar.gz: f2d313ced0c67a085dcc1a4a33c0b500b4225f23
+  metadata.gz: 06ae6b62245e3fb9df0bdd05bf95f41b0ef489e6
+  data.tar.gz: 86ed420fb61f9774d903101c7c6e8f90c90550a0
 SHA512:
-  metadata.gz: a1abeafdabb72ba2056735ff8d7473802c04ce8be5b81b2a9efebb54605d4b5d47e59fb61e326819ee0802a692d547f12916832816b51bcd1f44aab83332a173
-  data.tar.gz: eaba3a9e21dc751204e59e3bdf16243ae333e50a94354bf3a1c04d2f165ae04c957ae1ffcaa8e84a1d1fb0503be66513d4d4f2a16ca6378c1d1bfe2119e26745
+  metadata.gz: f07bb8aa5563ec6bc1a0463efc5cd15fb6ef11a6cad34b7285ab77caf26a430d9a179c120fa5d0f6f6e3bced49856525627d239cfb0cfb2d6620da1e02735643
+  data.tar.gz: 09cdefcf4faa2c8836b39f5b97eb7cdb8e4e885b4115e310b160dcbed271f036d3e15c8baf6ec9970238013a8fce7188a0950f1ef3e2768c926e60efc9b445c4

data/README.md

@@ -4,7 +4,8 @@ LightMecabは，mecab-rubyをラッピングすることで，より簡単に形
 
 ## Environment
 
-MeCabおよびmecab-rubyが動作する環境が必要．
+MeCabおよびmecab-rubyが動作する環境が必要．  
+MeCabの辞書にはIPA辞書を用いる．
 
 ## Installation
 
@@ -23,20 +24,33 @@ Gemfileに以下を記述する．
     $ gem install light-mecab
 
 ## Usage
-例えば「太郎はこの本を二郎を見た女性に渡した。」という文に含まれる名詞は，
-
 ```ruby
 require 'light-mecab'
+
 sentence = '太郎はこの本を二郎を見た女性に渡した。'
-LightMecab::Morpheme.new(sentence).noun
-```
-によって取得できる（返り値はString型の配列）．
+m = LightMecab::Morpheme.new(sentence)
 
-その他の品詞（形態素）を取得するメソッド名は，以下を参照．
+# 品詞数の取得
+m.num
+# => 15
 
-    lib/locale/morpheme.yml
+# 品詞分解
+m.parse
+# => [{"太郎"=>"名詞"}, {"は"=>"助詞"}, …, {"。"=>"記号"}]
 
+# わかち書き
+m.parse.map{|p| p.keys}.flatten.join(' ')
+# => "太郎 は この 本 を 二 郎 を 見 た 女性 に 渡し た 。"
 
+# 名詞一覧
+m.noun
+# => ["太郎", "本", "二", "郎", "女性"]
+# その他の品詞を取得するメソッド名は locale/morpheme.yml を参照
+
+# MeCab::Nodeオブジェクトを取得したい場合
+LightMecab::Morpheme.analyze(sentence)
+# => MeCab::Nodeの配列が返ってきます
+```
 
 ## Contributing
 

data/lib/light-mecab.rb

@@ -1,63 +1,4 @@
-require 'light-mecab/version'
 require 'MeCab'
 require 'yaml'
-
-module LightMecab
-  class Morpheme
-    @@tagger = ::MeCab::Tagger.new
-    @@i18n = ::YAML.load_file(File.expand_path(File.join(__FILE__, '..', '..', 'locale', 'morpheme.yml')))
-
-    class << self
-      # @param text [String]
-      # @return [Array <MeCab::Node>] 
-      def analyze(text)
-        nodes = Array.new
-        node = @@tagger.parseToNode(text)
-        while node
-          nodes << node
-          node = node.next
-        end
-        nodes.shift
-        nodes.pop
-        nodes
-      end
-
-      # @return [Hash]
-      def i18n
-        @@i18n
-      end
-    end
-
-    # @param text [String]
-    def initialize(text)
-      @nodes = self.class.analyze(text)
-    end
-
-    # @return [Integer]
-    def count
-      @nodes.size
-    end
-
-    # @param method_name [Symbol]
-    def method_missing(method_name)
-      if !self.class.i18n[method_name.to_s]
-        raise NoMethodError
-      end
-      extract(self.class.i18n[method_name.to_s])
-    end
-
-    private
-
-    # @param name [String]
-    # @return [Array <String>]
-    def extract(name)
-      morpheme = Array.new
-      @nodes.each do |node|
-        if name == node.feature.force_encoding('UTF-8').split(',').first
-          morpheme << node.surface.force_encoding('UTF-8')
-        end
-      end
-      morpheme
-    end
-  end
-end
+require 'light-mecab/morpheme'
+require 'light-mecab/version'

data/lib/light-mecab/morpheme.rb

@@ -0,0 +1,69 @@
+module LightMecab
+  class Morpheme
+    @@tagger = ::MeCab::Tagger.new
+    @@i18n = ::YAML.load_file(File.expand_path(File.join(__FILE__, '..', '..', '..', 'locale', 'morpheme.yml')))
+
+    class << self
+      # @param text [String]
+      # @return [Array <MeCab::Node>] 
+      def analyze(text)
+        nodes = Array.new
+        node = @@tagger.parseToNode(text)
+        while node
+          nodes << node
+          node = node.next
+        end
+        nodes.shift
+        nodes.pop
+        nodes
+      end
+
+      # @param node [MeCab::Node]
+      # @return [String]
+      def pos(node)
+        node.surface.force_encoding('UTF-8')
+      end
+
+      # @param node [MeCab::Node]
+      # @return [String]
+      def pos_name(node)
+        node.feature.force_encoding('UTF-8').split(',').first
+      end
+
+      # @return [Hash]
+      def i18n
+        @@i18n
+      end
+    end
+
+    # @param text [String]
+    def initialize(text)
+      @nodes = self.class.analyze(text)
+    end
+
+    # @return [Integer]
+    def num
+      @nodes.size
+    end
+
+    # return [Array <Hash>]
+    def parse
+      @nodes.map {|node| {self.class.pos(node) => self.class.pos_name(node)}}
+    end
+
+    # @param method_name [Symbol]
+    # @raise [NoMethodError]
+    def method_missing(method_name)
+      raise NoMethodError unless self.class.i18n[method_name.to_s]
+      extract(self.class.i18n[method_name.to_s])
+    end
+
+    private
+
+    # @param name [String]
+    # @return [Array <String>]
+    def extract(name)
+      @nodes.map {|node| self.class.pos(node) if name == self.class.pos_name(node)}.compact
+    end
+  end
+end

data/lib/light-mecab/version.rb

@@ -1,3 +1,3 @@
 module LightMecab
-  VERSION = '0.0.4'
+  VERSION = '0.0.5'
 end

data/locale/morpheme.yml

@@ -1,10 +1,14 @@
+ad_adjective: '連体詞'
+prefix:       '接頭詞'
 noun:         '名詞'
 verb:         '動詞'
-aux_verb:     '助動詞'
 adjective:    '形容詞'
 adverb:       '副詞'
 conjunction:  '接続詞'
-interjection: '感動詞'
 particle:     '助詞'
-ad_adjective: '連体詞'
+aux_verb:     '助動詞'
+interjection: '感動詞'
 symbol:       '記号'
+filler:       'フィラー'
+other:        'その他'
+unknown:      '未知語'

data/test/test_morpheme.rb

@@ -10,8 +10,39 @@ class TC_LightMecab_Morpheme < Test::Unit::TestCase
     @morpheme = LightMecab::Morpheme.new(@sentence)
   end
 
-  def test_count
-    assert_equal(17, @morpheme.count)
+  def test_num
+    assert_equal(17, @morpheme.num)
+  end
+
+  def test_parse()
+    parse_result = [
+      {"そして"=>"接続詞"},
+      {"太郎"=>"名詞"},
+      {"は"=>"助詞"},
+      {"この"=>"連体詞"},
+      {"本"=>"名詞"},
+      {"を"=>"助詞"},
+      {"二"=>"名詞"},
+      {"郎"=>"名詞"},
+      {"を"=>"助詞"},
+      {"見"=>"動詞"},
+      {"た"=>"助動詞"},
+      {"美しい"=>"形容詞"},
+      {"女性"=>"名詞"},
+      {"に"=>"助詞"},
+      {"渡し"=>"動詞"},
+      {"た"=>"助動詞"},
+      {"。"=>"記号"}
+    ]
+    assert_equal(parse_result, @morpheme.parse)
+  end
+
+  def test_method_missing_ad_adjective
+    assert_equal(['この'], @morpheme.ad_adjective)
+  end
+
+  def test_method_missing_prefix
+    assert_equal([], @morpheme.prefix)
   end
 
   def test_method_missing_noun
@@ -22,10 +53,6 @@ class TC_LightMecab_Morpheme < Test::Unit::TestCase
     assert_equal(['見', '渡し'], @morpheme.verb)
   end
 
-  def test_method_missing_aux_verb
-    assert_equal(['た', 'た'], @morpheme.aux_verb)
-  end
-
   def test_method_missing_adjective
     assert_equal(['美しい'], @morpheme.adjective)
   end
@@ -38,23 +65,37 @@ class TC_LightMecab_Morpheme < Test::Unit::TestCase
     assert_equal(['そして'], @morpheme.conjunction)
   end
 
-  def test_method_missing_interjection
-    assert_equal([], @morpheme.interjection)
-  end
-
   def test_method_missing_particle
     assert_equal(['は', 'を', 'を', 'に'], @morpheme.particle)
   end
 
-  def test_method_missing_ad_adjective
-    assert_equal(['この'], @morpheme.ad_adjective)
+  def test_method_missing_aux_verb
+    assert_equal(['た', 'た'], @morpheme.aux_verb)
+  end
+
+  def test_method_missing_interjection
+    assert_equal([], @morpheme.interjection)
   end
 
   def test_method_missing_symbol
     assert_equal(['。'], @morpheme.symbol)
   end
 
+  def test_method_missing_filler
+    assert_equal([], @morpheme.filler)
+  end
+
+  def test_method_missing_other
+    assert_equal([], @morpheme.other)
+  end
+
+  def test_method_missing_unknown
+    assert_equal([], @morpheme.unknown)
+  end
+
   def test_method_missing_hoge
-    assert_raise{@morpheme.hoge}
+    assert_raise(NoMethodError) do
+      @morpheme.hogei
+    end
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: light-mecab
 version: !ruby/object:Gem::Version
-  version: 0.0.4
+  version: 0.0.5
 platform: ruby
 authors:
 - Kei Tsuchiya
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-10-24 00:00:00.000000000 Z
+date: 2013-11-21 00:00:00.000000000 Z
 dependencies: []
 description: Execute morphological analysis by MeCab
 email:
@@ -23,6 +23,7 @@ files:
 - README.md
 - Rakefile
 - lib/light-mecab.rb
+- lib/light-mecab/morpheme.rb
 - lib/light-mecab/version.rb
 - light-mecab.gemspec
 - locale/morpheme.yml