natto 0.0.1 → 0.0.2

data/lib/natto.rb

@@ -1,39 +1,98 @@
+# -*- coding: UTF-8 -*-
 require 'rubygems' if RUBY_VERSION.to_f < 1.9
-require 'ffi'
-require 'rbconfig'
 
 module Natto
-  extend FFI::Library
+  require 'ffi'
 
-  def self.included(base)
-    base.extend(ClassMethods)
-  end
+  class MeCab
+    SUPPORTED_OPTS = [ :rcfile, :dicdir, :userdic, :output_format_type, :lattice_level, 
+                       :node_format, :unk_format, :bos_format, :eos_format, :eon_format,
+                       :unk_feature, :nbest, :theta, :cost_factor ]
+                       # :all_morphs, :partial, :allocate_sentence ]
+    attr_reader :ptr
+
+    def initialize(options={})
+      opt_str = self.class.build_options_str(options)
+      #@ptr = FFI::MemoryPointer.new :pointer
+      @ptr = Natto::Binding.mecab_new2(opt_str)
+      puts @ptr.inspect
+      raise MeCabError.new("MeCab initialiation error with '#{opt_str}'") if @ptr.address == 0
+      #@dict = Natto::DictionaryInfo.new(Natto::Binding.mecab_dictionary_info(@ptr))
+      ObjectSpace.define_finalizer(self, self.class.create_free_proc(@ptr))
+    end
 
-  def self.find_library
-    host_os = RbConfig::CONFIG['host_os']
+    def parse(s)
+      Natto::Binding.mecab_sparse_tostr(@ptr, s) || 
+        raise(MeCabError.new(Natto::Binding.mecab_strerror(@ptr)))
+    end
+    
+    def self.create_free_proc(ptr)
+      Proc.new do
+        #puts "mecab_destroy #{ptr}"
+        Natto::Binding.mecab_destroy(ptr)
+      end
+    end
 
-    if host_os =~ /mswin|mingw/
-      raise LoadError, "Please set MECAB_PATH to full path to libmecab.dll"
-    elsif host_os =~ /cygwin/
-      'cygmecab-1'
-    else
-      'mecab'
+    def self.build_options_str(options={})
+      opt = []
+      SUPPORTED_OPTS.each do |k|
+        if options.has_key? k
+          key = k.to_s.gsub('_', '-')  
+          opt << "--#{key}=#{options[k]}"
+        end
+      end
+      opt.join(" ")
     end
   end
+ 
+  class MeCabError < RuntimeError; end
+
+  class DictionaryInfo < FFI::Struct
+    layout  :filename, :string,
+            :charset,  :string,
+            :size,     :uint,
+            :type,     :int,
+            :lsize,    :uint,
+            :rsize,    :uint,
+            :version,  :ushort,
+            :next,     :pointer 
+  end
 
-  ffi_lib(ENV['MECAB_PATH'] || find_library)
+  module Binding
+    require 'rbconfig'
+    extend FFI::Library
 
-  attach_function :mecab_new2, [:string], :pointer
-  attach_function :mecab_version, [], :string
-  attach_function :mecab_sparse_tostr, [:pointer, :string], :string
-  attach_function :mecab_strerror, [:pointer],:string
-  attach_function :mecab_destroy, [:pointer], :void
+    MECAB_PATH = 'MECAB_PATH'
 
-  module ClassMethods
-    def mecab_version
-      Natto.mecab_version
+    def self.included(base)
+      base.extend(ClassMethods)
+    end
+
+    def self.find_library
+      host_os = RbConfig::CONFIG['host_os']
+
+      if host_os =~ /mswin|mingw/i
+        raise LoadError, "Please set #{MECAB_PATH} to full path to libmecab.dll"
+      elsif host_os =~ /cygwin/i
+        'cygmecab-1'
+      else
+        'mecab'
+      end
     end
-  end
-end
 
+    ffi_lib(ENV[MECAB_PATH] || find_library)
 
+    attach_function :mecab_version, [], :string
+    attach_function :mecab_new2, [:string], :pointer
+    attach_function :mecab_destroy, [:pointer], :void
+    attach_function :mecab_sparse_tostr, [:pointer, :string], :string
+    attach_function :mecab_strerror, [:pointer],:string
+    attach_function :mecab_dictionary_info, [:pointer], :pointer
+
+    module ClassMethods
+      def mecab_version
+        Natto::Binding.mecab_version
+      end
+    end
+  end
+end

data/test/test_natto.rb

@@ -1,20 +1,46 @@
+# -*- coding: UTF-8 -*-
 $:.unshift('lib')
-
 require 'rubygems' if RUBY_VERSION.to_f < 1.9
 require 'test/unit'
 require 'natto'
 
 class TestNatto < Test::Unit::TestCase
   def setup
+    @klass = Class.new do
+      include Natto::Binding
+    end
   end
 
   def teardown
   end
 
   def test_classmethods_include
-    klass = Class.new do
-      include Natto
+    assert_equal('0.98', @klass.mecab_version)
+  end
+
+  def test_build_options_str
+    res = Natto::MeCab.build_options_str
+    assert_equal('', res)
+
+    res = Natto::MeCab.build_options_str(:unknown=>"ignore")
+    assert_equal('', res)
+
+    res = Natto::MeCab.build_options_str(:dicdir=>"a")
+    assert_equal('--dicdir=a', res)
+
+    res = Natto::MeCab.build_options_str(:userdic=>"b")
+    assert_equal('--userdic=b', res)
+
+    res = Natto::MeCab.build_options_str(:output_format_type=>"c")
+    assert_equal('--output-format-type=c', res)
+
+    res = Natto::MeCab.build_options_str(:output_format_type=>"c", :userdic=>"b", :dicdir=>"a")
+    assert_equal('--dicdir=a --userdic=b --output-format-type=c', res)
+  end
+
+  def test_initialize
+    assert_raise Natto::MeCabError do
+      Natto::MeCab.new(:output_format_type=>'UNDEFINED')
     end
-    assert_equal('0.98', klass.mecab_version)
   end
 end

metadata

@@ -5,16 +5,16 @@ version: !ruby/object:Gem::Version
   segments: 
   - 0
   - 0
-  - 1
-  version: 0.0.1
+  - 2
+  version: 0.0.2
 platform: ruby
 authors: 
 - Brooke M. Fujita
-autorequire: natto
+autorequire: 
 bindir: bin
 cert_chain: []
 
-date: 2010-12-14 00:00:00 +09:00
+date: 2010-12-20 00:00:00 +09:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -72,7 +72,6 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       - 0
       version: "0"
 requirements: 
-- none
 - MeCab, 0.98 or greater
 rubyforge_project: 
 rubygems_version: 1.3.7