rudachi 1.0.0

checksums.yaml

@@ -0,0 +1,7 @@
+---
+SHA256:
+  metadata.gz: f43021bc769b0005c5d46c3abde244756712ce1419b82ae884729508aab64780
+  data.tar.gz: 5f3b2f74e771d2b3feb8d6de4abf7d59864da6108904437c28a498f2ae6d1e98
+SHA512:
+  metadata.gz: 38535bc9cee7b34ee2ab3219294484d30758c82b4c2a85f53cb36519700eb52d2e44f75160ba2aee24a626a7dc95c5a02774bfd10c86da18067a7d6afcd33996
+  data.tar.gz: 289e76174d4f7053e71c2c276635ab6b2283b48c220cb498615c88605ffec6a41f3888c89a80ad728383170c5052c3dd76be81a02bef91dc38f4049b18db1c21

data/README.md

@@ -0,0 +1,78 @@
+# Rudachi
+[Sudachi](https://github.com/WorksApplications/Sudachi) wrapper Gem for JRuby.
+
+- Text base
+```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
+```rb
+File.open('sample.txt', 'w') { |f| f << '東京都へ行く' }
+Rudachi::FileParser.parse('sample.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"
+```
+
+- With some options
+```rb
+Rudachi::TextParser.new(o: 'result.txt', m: 'A').parse('東京都へ行く')
+File.read('result.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"
+```
+
+## Requirements
+
+- [JRuby](https://github.com/jruby/jruby) 9.1.3.0 or later
+- [Sudachi](https://github.com/WorksApplications/Sudachi)
+
+## Installation
+
+1. Install JAR and dictionary of Sudachi ([details](https://github.com/WorksApplications/Sudachi/blob/develop/docs/tutorial.md#linux-%E3%81%AE%E5%A0%B4%E5%90%88))
+
+- Install the Sudachi JAR file
+```sh
+$ wget https://github.com/WorksApplications/Sudachi/releases/download/v0.5.3/sudachi-0.5.3-executable.zip
+$ unzip sudachi-0.5.3-executable.zip
+$ ls sudachi-0.5.3
+LICENSE-2.0.txt  README.md  javax.json-1.1.jar	jdartsclone-1.2.0.jar  licenses  sudachi-0.5.3.jar  sudachi.json  sudachi_fulldict.json
+```
+
+- Install the Sudachi dictionary
+```sh
+$ wget http://sudachi.s3-website-ap-northeast-1.amazonaws.com/sudachidict/sudachi-dictionary-latest-full.zip
+$ unzip -j -d sudachi-dictionary-latest-full sudachi-dictionary-latest-full.zip
+$ mv sudachi-dictionary-latest-full/system_full.dic sudachi-dictionary-latest-full/system_core.dic
+$ ls sudachi-dictionary-latest-full
+LEGAL  LICENSE-2.0.txt	system_core.dic
+```
+
+2. Install Rudachi
+
+```rb
+# Gemfile
+gem 'rudachi', git: 'https://github.com/SongCastle/rudachi.git', tag: 'v1.0.0'
+```
+
+Then run `bundle install` .
+
+3. Initialize Rudachi
+
+```rb
+Rudachi.configure do |config|
+  config.jar_path = 'sudachi-0.5.3/sudachi-0.5.3.jar'
+end
+
+Rudachi::Option.configure do |config|
+  config.p = 'sudachi-dictionary-latest-full'
+end
+```
+
+4. Dit 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

@@ -0,0 +1,32 @@
+require 'rudachi/configurable'
+
+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
+end

data/lib/rudachi/configurable.rb

@@ -0,0 +1,18 @@
+module Rudachi
+  module Configurable
+    def configure
+      yield self
+    end
+
+    private
+
+    def config_accessor(name, default: nil)
+      attr_def = <<~EOS
+        def self.#{name}; @@#{name}; end
+        def self.#{name}=(val); @@#{name} = val; end
+      EOS
+      module_eval(attr_def)
+      class_variable_set("@@#{name}", default)
+    end
+  end
+end

data/lib/rudachi/dependencies.rb

@@ -0,0 +1,19 @@
+require 'java'
+java_import 'java.lang.System'
+java_import 'java.io.PrintStream'
+java_import 'java.io.ByteArrayInputStream'
+java_import 'java.io.ByteArrayOutputStream'
+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
+end

data/lib/rudachi/file_parser.rb

@@ -0,0 +1,38 @@
+require 'rudachi/config'
+require 'rudachi/loader'
+
+module Rudachi
+  class FileParser
+    def self.parse(path)
+      new.parse(path)
+    end
+
+    def initialize(**opts)
+      Rudachi.load!
+
+      @output = Java::ByteArrayOutputStream.new
+      @opts   = opts
+    end
+
+    def parse(path)
+      take_stdout do
+        Java::SudachiCommandLine.main(
+          Option.cmds(@opts).push(Java::String.new(path))
+        )
+      end
+      @output.toString
+    end
+
+    private
+
+    def take_stdout
+      stdout = Java::System.out
+      stream = Java::PrintStream.new(@output)
+      Java::System.setOut(stream)
+
+      yield
+
+      Java::System.setOut(stdout)
+    end
+  end
+end

data/lib/rudachi/loader.rb

@@ -0,0 +1,5 @@
+module Rudachi
+  def self.load!
+    require 'rudachi/dependencies'
+  end
+end

data/lib/rudachi/text_parser.rb

@@ -0,0 +1,27 @@
+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))
+        end
+      end
+      @output.toString
+    end
+
+    private
+
+    def take_stdin
+      stdin = Java::System.in
+      stream = Java::ByteArrayInputStream.new(@input.getBytes(Java::UTF_8))
+      Java::System.setIn(stream)
+
+      yield
+
+      Java::System.setIn(stdin)
+    end
+  end
+end

data/lib/rudachi/version.rb

@@ -0,0 +1,3 @@
+module Rudachi
+  VERSION = '1.0.0'
+end

data/lib/rudachi.rb

@@ -0,0 +1,3 @@
+require 'rudachi/config'
+require 'rudachi/file_parser'
+require 'rudachi/text_parser'

metadata

@@ -0,0 +1,51 @@
+--- !ruby/object:Gem::Specification
+name: rudachi
+version: !ruby/object:Gem::Version
+  version: 1.0.0
+platform: ruby
+authors:
+- SongCastle
+autorequire: 
+bindir: bin
+cert_chain: []
+date: 2022-04-06 00:00:00.000000000 Z
+dependencies: []
+description: Sudachi wrapper for JRuby.
+email: "-"
+executables: []
+extensions: []
+extra_rdoc_files: []
+files:
+- README.md
+- lib/rudachi.rb
+- lib/rudachi/config.rb
+- lib/rudachi/configurable.rb
+- lib/rudachi/dependencies.rb
+- lib/rudachi/file_parser.rb
+- lib/rudachi/loader.rb
+- lib/rudachi/text_parser.rb
+- lib/rudachi/version.rb
+homepage: https://github.com/SongCastle/rudachi
+licenses:
+- MIT
+metadata: {}
+post_install_message: 
+rdoc_options: []
+require_paths:
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement
+  requirements:
+  - - ">="
+    - !ruby/object:Gem::Version
+      version: '2.3'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  requirements:
+  - - ">="
+    - !ruby/object:Gem::Version
+      version: '0'
+requirements: []
+rubygems_version: 3.0.3
+signing_key: 
+specification_version: 4
+summary: Sudachi wrapper for JRuby
+test_files: []