rudachi 1.0.1 → 1.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 0a4e15eb13cac06919c3703fc9c3efa185267b3d90cfa965f393126e145ff8f9
-  data.tar.gz: c270fc32d9a7cc3bf06179e667dd5a264d880873b6004a1118ea6f3bcb57722c
+  metadata.gz: 515a2588b078bce3a050c6e1c4cd115524daaf5edfb71b76024e44ca9cb1cf26
+  data.tar.gz: eeb369e360a30b5dc050820c0b77ce4f9e98f6f9babac3e7700009ba8d2b372f
 SHA512:
-  metadata.gz: ab74a535d435b84dc0b25affde71cdea3eb4a9decec781a60f3e40069468b6fe02e4c26d2bbca3a3b5e317b5096c4766b02721fbf11cb093722494ef47db9bd3
-  data.tar.gz: a890984b99a0fdefaa4122db1e3f13bc9ba7df100af36e0e9dc9d2290e4b6d21172e1e416dc4c8d90e40fca830530ddf84fb5042c872c3966f452c3762f3d3bb
+  metadata.gz: 989c5f7acc769e3b9304592db515f30192bdf51d1aa2ce3578b380d30a06d229cfa8f96bc818d7245adeb3b72a8fa657178796a2a08cab3addb6658236219e7d
+  data.tar.gz: 704c0e1053077802101e7218fae201a53a2adf44203f66ef2bf8093c3c8f4de208b06c7e905b2c54aeb3c2a009bb392f73f92181633c4417081a4675245f866f

data/README.md

@@ -1,23 +1,29 @@
 # Rudachi
-[Sudachi](https://github.com/WorksApplications/Sudachi) wrapper Gem for JRuby.
+JRuby wrapper for [Sudachi](https://github.com/WorksApplications/Sudachi).
 
-#### Text base
+#### Text
 ```rb
 Rudachi::TextParser.parse('東京都へ行く')
 => "東京都\t名詞,固有名詞,地名,一般,*,*\t東京都\nへ\t助詞,格助詞,*,*,*,*\tへ\n行く\t動詞,非自立可能,*,*,五段-カ行,終止形-一般\t行く\nEOS\n"
 ```
 
-#### File base
+#### File
 ```rb
-File.open('sample.txt', 'w') { |f| f << '東京都へ行く' }
-Rudachi::FileParser.parse('sample.txt')
+File.open('input.txt', 'w') { |f| f << '東京都へ行く' }
+Rudachi::FileParser.parse('input.txt')
+=> "東京都\t名詞,固有名詞,地名,一般,*,*\t東京都\nへ\t助詞,格助詞,*,*,*,*\tへ\n行く\t動詞,非自立可能,*,*,五段-カ行,終止形-一般\t行く\nEOS\n"
+```
+
+#### IO
+```rb
+Rudachi::StreamParser.parse(StringIO.new('東京都へ行く'))
 => "東京都\t名詞,固有名詞,地名,一般,*,*\t東京都\nへ\t助詞,格助詞,*,*,*,*\tへ\n行く\t動詞,非自立可能,*,*,五段-カ行,終止形-一般\t行く\nEOS\n"
 ```
 
 #### With [some options](https://github.com/WorksApplications/Sudachi#options)
 ```rb
-Rudachi::TextParser.new(o: 'result.txt', m: 'A').parse('東京都へ行く')
-File.read('result.txt')
+Rudachi::TextParser.new(o: 'output.txt', m: 'A').parse('東京都へ行く')
+File.read('output.txt')
 => "東京\t名詞,固有名詞,地名,一般,*,*\t東京\n都\t名詞,普通名詞,一般,*,*,*\t都\nへ\t助詞,格助詞,*,*,*,*\tへ\n行く\t動詞,非自立可能,*,*,五段-カ行,終止形-一般\t行く\nEOS\n"
 ```
 
@@ -54,11 +60,13 @@ LEGAL  LICENSE-2.0.txt	system_core.dic
 gem 'rudachi'
 ```
 
-Then run `bundle install` .
+Then run `bundle install`.
 
 3. Initialize Rudachi
 
 ```rb
+require 'rudachi'
+
 Rudachi.configure do |config|
   config.jar_path = 'sudachi-0.5.3/sudachi-0.5.3.jar'
 end
@@ -71,8 +79,6 @@ end
 4. Did it !!
 
 ```rb
-require 'rudachi'
-
 Rudachi::TextParser.parse('こんにちは世界')
 => "こんにちは\t感動詞,一般,*,*,*,*\t今日は\n世界\t名詞,普通名詞,一般,*,*,*\t世界\nEOS\n"
 ```

data/lib/rudachi/config.rb

@@ -1,32 +1,8 @@
 require 'rudachi/configurable'
+require 'rudachi/option/string_option'
 
 module Rudachi
   extend Configurable
 
-  config_accessor :jar_path,  default: '/usr/java/lib/sudachi.jar'
-
-  module Option
-    extend Configurable
-
-    # @see https://github.com/WorksApplications/Sudachi#options
-    config_accessor :r,  default: nil
-    config_accessor :s,  default: nil
-    config_accessor :p,  default: '/usr/java/lib'
-    config_accessor :m,  default: 'C'
-    config_accessor :o,  default: nil
-    config_accessor :t,  default: nil
-    config_accessor :ts, default: nil
-    config_accessor :a,  default: nil
-    config_accessor :f,  default: nil
-    config_accessor :d,  default: nil
-    config_accessor :h,  default: nil
-
-    def self.cmds(opts)
-      class_variables.each_with_object([]) do |name, flags|
-        key = name.to_s.delete('@@')
-        val = opts[key] || opts[key.to_sym] || class_variable_get(name) or next
-        flags << "-#{key}" << val.to_s
-      end
-    end
-  end
+  config_accessor :jar_path, klass: Option::StringOption, default: '/usr/java/lib/sudachi.jar'
 end

data/lib/rudachi/configurable.rb

@@ -6,13 +6,12 @@ module Rudachi
 
     private
 
-    def config_accessor(name, default: nil)
-      attr_def = <<~EOS
+    def config_accessor(name, klass:, default:)
+      module_eval <<~EOS
         def self.#{name}; @@#{name}; end
-        def self.#{name}=(val); @@#{name} = val; end
+        def self.#{name}=(val); @@#{name} = #{klass}.new(val); end
       EOS
-      module_eval(attr_def)
-      class_variable_set("@@#{name}", default)
+      public_send("#{name}=", default)
     end
   end
 end

data/lib/rudachi/dependencies.rb

@@ -8,12 +8,13 @@ java_import 'java.nio.charset.StandardCharsets'
 require Rudachi.jar_path
 java_import 'com.worksap.nlp.sudachi.SudachiCommandLine'
 
-module Java
-  String                = JavaLang::String
-  System                = JavaLang::System
-  ByteArrayInputStream  = JavaIo::ByteArrayInputStream
-  ByteArrayOutputStream = JavaIo::ByteArrayOutputStream
-  PrintStream           = JavaIo::PrintStream
-  UTF_8                 = JavaNioCharset::StandardCharsets::UTF_8
-  SudachiCommandLine    = ComWorksapNlpSudachi::SudachiCommandLine
+module Rudachi
+  module Java
+    System                = ::Java::JavaLang::System
+    ByteArrayInputStream  = ::Java::JavaIo::ByteArrayInputStream
+    ByteArrayOutputStream = ::Java::JavaIo::ByteArrayOutputStream
+    PrintStream           = ::Java::JavaIo::PrintStream
+    UTF_8                 = ::Java::JavaNioCharset::StandardCharsets::UTF_8
+    SudachiCommandLine    = ::Java::ComWorksapNlpSudachi::SudachiCommandLine
+  end
 end

data/lib/rudachi/file_parser.rb

@@ -1,4 +1,4 @@
-require 'rudachi/config'
+require 'rudachi/option/config'
 require 'rudachi/loader'
 
 module Rudachi
@@ -9,30 +9,33 @@ module Rudachi
 
     def initialize(**opts)
       Rudachi.load!
-
-      @output = Java::ByteArrayOutputStream.new
-      @opts   = opts
+      @opts = Option.new(opts)
     end
 
     def parse(path)
-      take_stdout do
-        Java::SudachiCommandLine.main(
-          Option.cmds(@opts).push(Java::String.new(path))
-        )
+      output_stream do |output|
+        take_stdout(output) do
+          Java::SudachiCommandLine.main(
+            Option.cmds(@opts).push(path)
+          )
+        end
       end
-      @output.toString
     end
 
     private
 
-    def take_stdout
+    def take_stdout(output)
       stdout = Java::System.out
-      stream = Java::PrintStream.new(@output)
-      Java::System.setOut(stream)
 
+      Java::System.setOut(output)
       yield
-
       Java::System.setOut(stdout)
     end
+
+    def output_stream
+      Java::ByteArrayOutputStream.new.tap do |output|
+        yield Java::PrintStream.new(output)
+      end.toString
+    end
   end
 end

data/lib/rudachi/loader.rb

@@ -1,5 +1,16 @@
 module Rudachi
-  def self.load!
-    require 'rudachi/dependencies'
+  class UnavailableError < StandardError; end;
+
+  class << self
+    def load!
+      raise UnavailableError, 'jruby_required' unless jruby?
+      require 'rudachi/dependencies'
+    end
+
+    private
+
+    def jruby?
+      RUBY_PLATFORM == 'java'
+    end
   end
 end

data/lib/rudachi/option/boolean_option.rb

@@ -0,0 +1,14 @@
+module Rudachi
+  class Option
+    class BooleanOption < Delegator
+      def initialize(bool)
+        raise ArgumentError, 'must be `false` or `true`' unless bool.is_a?(FalseClass) || bool.is_a?(TrueClass)
+        @value = bool
+      end
+
+      def __getobj__; @value; end
+      def enable?; @value; end
+      def with_arg?; false; end
+    end
+  end
+end

data/lib/rudachi/option/config.rb

@@ -0,0 +1,47 @@
+require 'rudachi/configurable'
+require 'rudachi/option/boolean_option'
+require 'rudachi/option/string_option'
+
+module Rudachi
+  class Option
+    extend Configurable
+
+    # @see https://github.com/WorksApplications/Sudachi#options
+    config_accessor :r,  klass: StringOption,  default: nil
+    config_accessor :s,  klass: StringOption,  default: nil
+    config_accessor :p,  klass: StringOption,  default: nil
+    config_accessor :m,  klass: StringOption,  default: nil
+    config_accessor :o,  klass: StringOption,  default: nil
+    config_accessor :a,  klass: BooleanOption, default: false
+    config_accessor :d,  klass: BooleanOption, default: false
+    config_accessor :t,  klass: BooleanOption, default: false
+    config_accessor :ts, klass: BooleanOption, default: false
+    config_accessor :f,  klass: BooleanOption, default: false
+    config_accessor :h,  klass: BooleanOption, default: false
+
+    def self.cmds(opts=Option.new)
+      class_variables.each_with_object([]) do |name, flags|
+        key = name[2..-1].to_sym
+        opt = opts.get(key) { class_variable_get(name) }
+        next unless opt&.enable?
+        flags << "-#{key}"
+        flags << opt.to_s if opt.with_arg?
+      end
+    end
+
+    def initialize(**hash)
+      @opts = hash.each_with_object({}) do |(key, val), _hash|
+        raise ArgumentError, %{unknown option "#{key}"} unless self.class.class_variable_defined?("@@#{key}")
+        begin
+          _hash[key.to_sym] = self.class.class_variable_get("@@#{key}").class.new(val)
+        rescue ArgumentError => e
+          raise ArgumentError, %{"#{key}" #{e.message}}
+        end
+      end
+    end
+
+    def get(key, &block)
+      @opts.key?(key) ? @opts[key] : block&.call
+    end
+  end
+end

data/lib/rudachi/option/string_option.rb

@@ -0,0 +1,14 @@
+module Rudachi
+  class Option
+    class StringOption < Delegator
+      def initialize(str)
+        raise ArgumentError, 'must be `nil` or `String`' unless str.nil? || str.is_a?(String)
+        @value = str
+      end
+
+      def __getobj__; @value; end
+      def enable?; !!@value; end
+      def with_arg?; true; end
+    end
+  end
+end

data/lib/rudachi/stream_parser.rb

@@ -0,0 +1,11 @@
+require 'rudachi/text_parser'
+
+module Rudachi
+  class StreamParser < TextParser
+    private
+
+    def input_stream(stream)
+      stream.to_inputstream
+    end
+  end
+end

data/lib/rudachi/text_parser.rb

@@ -3,25 +3,29 @@ require 'rudachi/file_parser'
 module Rudachi
   class TextParser < FileParser
     def parse(text)
-      @input = Java::String.new(text)
-      take_stdin do
-        take_stdout do
-          Java::SudachiCommandLine.main(Option.cmds(@opts))
+      output_stream do |output|
+        take_stdin(input_stream(text)) do
+          take_stdout(output) do
+            Java::SudachiCommandLine.main(Option.cmds(@opts))
+          end
         end
       end
-      @output.toString
     end
 
     private
 
-    def take_stdin
+    def take_stdin(input)
       stdin = Java::System.in
-      stream = Java::ByteArrayInputStream.new(@input.getBytes(Java::UTF_8))
-      Java::System.setIn(stream)
 
+      Java::System.setIn(input)
       yield
-
       Java::System.setIn(stdin)
     end
+
+    def input_stream(text)
+      Java::ByteArrayInputStream.new(
+        text.to_java.getBytes(Java::UTF_8)
+      )
+    end
   end
 end

data/lib/rudachi/version.rb

@@ -1,3 +1,3 @@
 module Rudachi
-  VERSION = '1.0.1'
+  VERSION = '1.2.0'
 end

data/lib/rudachi.rb

@@ -1,3 +1,5 @@
 require 'rudachi/config'
+require 'rudachi/option/config'
 require 'rudachi/file_parser'
 require 'rudachi/text_parser'
+require 'rudachi/stream_parser'

metadata

@@ -1,16 +1,16 @@
 --- !ruby/object:Gem::Specification
 name: rudachi
 version: !ruby/object:Gem::Version
-  version: 1.0.1
+  version: 1.2.0
 platform: ruby
 authors:
 - SongCastle
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2022-04-08 00:00:00.000000000 Z
+date: 2022-04-12 00:00:00.000000000 Z
 dependencies: []
-description: Sudachi wrapper for JRuby.
+description: A JRuby wrapper for Sudachi.
 email: "-"
 executables: []
 extensions: []
@@ -23,6 +23,10 @@ files:
 - lib/rudachi/dependencies.rb
 - lib/rudachi/file_parser.rb
 - lib/rudachi/loader.rb
+- lib/rudachi/option/boolean_option.rb
+- lib/rudachi/option/config.rb
+- lib/rudachi/option/string_option.rb
+- lib/rudachi/stream_parser.rb
 - lib/rudachi/text_parser.rb
 - lib/rudachi/version.rb
 homepage: https://github.com/SongCastle/rudachi
@@ -47,5 +51,5 @@ requirements: []
 rubygems_version: 3.0.3
 signing_key: 
 specification_version: 4
-summary: Sudachi wrapper for JRuby
+summary: A JRuby wrapper for Sudachi
 test_files: []