sawarineko 1.1.0 → 1.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: a3f2a14c9be7b8aa29c25f511c70cb89cdb5b3a5
-  data.tar.gz: 68a1c98dca7f5c6b75b82e07358428df92b6c405
+  metadata.gz: 9557a8ea93741177f356bdc4ca74b19074ed759b
+  data.tar.gz: 1209522799b9aa7a0e00602491eb88898775bf2e
 SHA512:
-  metadata.gz: 69a20a207b87bf97cb7f86c0ce65173e673b00839b90f625acf5812dd89c2e66e90b5a8f7ed3f3f81f04d4ab6ac628c385edec15ddf04880ab504ea78ec9b593
-  data.tar.gz: d432ce8bbe7830a7f475f7ef627ca6390ccd2fa2518f7718b7f510a6893ebfbc6a29f4d3c695cd826c359c35d187b9620cae1fda6c0e2ee9f4e263961600363c
+  metadata.gz: fbd6e46d892abadc48d28f03440ec7c781ddca89768aa7336a4adc585be71b5791a482489eeab34f1e1bee9f83598ecafe5ba2ba33f2d8fa81f7e9633560b0d9
+  data.tar.gz: cdad60808e1ee4b80663882874e1d51a35fb3534b50900ba806983718a313307f57452457387e6f18b77ac89ca11d9d48507085f64022b77a19f9c7af4cb6b08

data/CHANGELOG.md

@@ -1,8 +1,13 @@
 # Changelog
 
+## 1.2.0 (2014-10-29)
+
+- [#2](https://github.com/yous/sawarineko/issues/2): Add encoding support to Converter (UTF-8, UTF-16BE, UTF-16LE, EUC-JP, Shift_JIS, EUC-KR, CP949)
+- [#2](https://github.com/yous/sawarineko/issues/2): Add `-e/--encoding` option to CLI
+
 ## 1.1.0 (2014-10-28)
 
-- Add `Sawarineko#nya`
+- Add `Sawarineko.nya`
 - Make `Converter#convert` to get the source as argument
 
 ## 1.0.0 (2014-10-27)

data/README.md

@@ -57,10 +57,11 @@ For additional command-line options:
 sawarineko -h
 ```
 
-Command flag    | Description
-----------------|--------------------
-`-h, --help`    | Print this message.
-`-v, --version` | Print version.
+Command flag              | Description
+--------------------------|-------------------------------
+`-e, --encoding ENCODING` | Specify the encoding of input.
+`-h, --help`              | Print this message.
+`-v, --version`           | Print version.
 
 ## Contributing
 

data/lib/sawarineko.rb

@@ -9,10 +9,11 @@ require 'sawarineko/cli'
 module Sawarineko
   # Make the source to Sawarineko. See Converter.
   #
-  # source - The String source to convert.
+  # source   - The String source to convert.
+  # encoding - The Encoding of the source (default: Encoding::UTF_8).
   #
   # Returns the String converted to Sawarineko.
-  def self.nya(source)
-    Converter.new.convert(source)
+  def self.nya(source, encoding = Encoding::UTF_8)
+    Converter.new(encoding).convert(source)
   end
 end

data/lib/sawarineko/cli.rb

@@ -19,9 +19,9 @@ module Sawarineko
       source = if paths.empty?
                  $stdin.read
                else
-                 IO.read(paths[0])
+                 IO.read(paths[0], encoding: @options[:encoding])
                end
-      converter = Converter.new
+      converter = Converter.new(@options[:encoding])
       puts converter.convert(source)
       0
     end

data/lib/sawarineko/converter.rb

@@ -3,16 +3,56 @@
 module Sawarineko
   # Convert plain text to Sawarineko text.
   class Converter
+    # Array of convertible character code on EUC-KR encoding.
+    CONVERTIBLE_EUC_KR =
+      %w(나 낙 난 날 남 낭).map { |ch| ch.encode(Encoding::EUC_KR).ord }.freeze
+    private_constant :CONVERTIBLE_EUC_KR
+
+    # Initialize a Converter. Get the encoding of source. Initialize Regexps for
+    # conversion.
+    #
+    # encoding - The Encoding of source (default: Encoding::UTF_8).
+    def initialize(encoding = Encoding::UTF_8)
+      @encoding = Encoding.find(encoding)
+      @hiragana_regex = Regexp.new('な'.encode(@encoding)).freeze
+      @katakana_regex = Regexp.new('ナ'.encode(@encoding)).freeze
+      @hangul_regex = case @encoding
+                      when Encoding::UTF_8, Encoding::UTF_16BE,
+                           Encoding::UTF_16LE, Encoding::EUC_KR
+                        Regexp.new('[나-낳]'.encode(@encoding)).freeze
+                      when Encoding::CP949
+                        Regexp.new('[나-낳낛-낤낥-낲]'.encode(@encoding)).freeze
+                      end
+    end
+
     # Convert the source.
     #
     # source - The String source to convert.
     #
     # Returns the String converted to Sawarineko.
     def convert(source)
-      source
-        .gsub(/な/, 'にゃ')
-        .gsub(/ナ/, 'ニャ')
-        .gsub(/[나-낳]/) { |ch| (ch.ord + 56).chr(Encoding::UTF_8) }
+      new_source = source.gsub(@hiragana_regex, 'にゃ'.encode(@encoding).freeze)
+                         .gsub(@katakana_regex, 'ニャ'.encode(@encoding).freeze)
+      if @hangul_regex
+        new_source.gsub(@hangul_regex) { |ch| convert_hangul(ch) }
+      else
+        new_source
+      end
+    end
+
+    private
+
+    # Convert a hangul character to Sawarineko.
+    #
+    # ch - A hangul character to convert.
+    #
+    # Returns the converted character.
+    def convert_hangul(ch)
+      if @encoding == Encoding::EUC_KR && !CONVERTIBLE_EUC_KR.include?(ch.ord)
+        return ch
+      end
+      (ch.encode(Encoding::UTF_8).ord + 56).chr(Encoding::UTF_8)
+                                           .encode(@encoding)
     end
   end
 end

data/lib/sawarineko/option.rb

@@ -7,7 +7,7 @@ module Sawarineko
   class Option
     # Initialize a Option.
     def initialize
-      @options = {}
+      @options = { encoding: Encoding::UTF_8 }
     end
 
     # Parse the passed arguments to a Hash.
@@ -21,24 +21,37 @@ module Sawarineko
         opts.banner = 'Usage: sawarineko [options] [source]'
 
         add_options(opts)
+        add_options_on_tail(opts)
       end.parse!(args)
       [@options, args]
     end
 
     private
 
-    # Add command line options to OptionParser.
+    # Add command line options.
     #
     # opts - An OptionParser object to add options.
     #
     # Returns nothing.
     def add_options(opts)
-      opts.on('-h', '--help', 'Print this message.') do
+      opts.on('-e', '--encoding ENCODING',
+              'Specify the encoding of input.') do |encoding|
+        @options[:encoding] = encoding
+      end
+    end
+
+    # Add command line options printed at tail.
+    #
+    # opts - An OptionParser object to add options.
+    #
+    # Returns nothing.
+    def add_options_on_tail(opts)
+      opts.on_tail('-h', '--help', 'Print this message.') do
         puts opts
         exit 0
       end
 
-      opts.on('-v', '--version', 'Print version.') do
+      opts.on_tail('-v', '--version', 'Print version.') do
         puts Version::STRING
         exit 0
       end

data/lib/sawarineko/version.rb

@@ -3,6 +3,6 @@
 module Sawarineko
   # Hold Sawarineko's version information.
   module Version
-    STRING = '1.1.0'
+    STRING = '1.2.0'
   end
 end

data/spec/sawarineko/cli_spec.rb

@@ -36,4 +36,20 @@ RSpec.describe Sawarineko::CLI do
                                  'にゃにゃはんにゃにゃだいにゃんにゃく' \
                                  "にゃらべてにゃがにゃがめ\n")
   end
+
+  describe '-e/--encoding' do
+    it 'converts text file with given encoding' do
+      create_file('sawarineko.txt',
+                  ['ななめななじゅうななどの' \
+                   'ならびでなくなくいななく' \
+                   'ななはんななだいなんなく' \
+                   'ならべてながながめ'.encode(Encoding::CP949)])
+      expect(cli.run(['--encoding', 'cp949',
+                      'sawarineko.txt'])).to be(0)
+      expect($stdout.string).to eq('にゃにゃめにゃにゃじゅうにゃにゃどの' \
+                                   'にゃらびでにゃくにゃくいにゃにゃく' \
+                                   'にゃにゃはんにゃにゃだいにゃんにゃく' \
+                                   "にゃらべてにゃがにゃがめ\n")
+    end
+  end
 end

data/spec/sawarineko/converter_spec.rb

@@ -3,44 +3,108 @@
 require 'spec_helper'
 
 RSpec.describe Sawarineko::Converter do
-  subject(:converter) { described_class.new }
-
-  it 'converts hiragana な to にゃ' do
-    expect(converter.convert('な')).to eq('にゃ')
-    expect(converter.convert('ななめななじゅうななどのならびで' \
-                             'なくなくいななくななはんななだい' \
-                             'なんなくならべてながながめ'))
-      .to eq('にゃにゃめにゃにゃじゅうにゃにゃどのにゃらびで' \
-             'にゃくにゃくいにゃにゃくにゃにゃはんにゃにゃだい' \
-             'にゃんにゃくにゃらべてにゃがにゃがめ')
-  end
-
-  it 'converts katakana ナ to ニャ' do
-    expect(converter.convert('ナ')).to eq('ニャ')
-    expect(converter.convert('ナナメナナジュウナナドノナラビデ' \
-                             'ナクナクイナナクナナハンナナダイ' \
-                             'ナンナクナラベテナガナガメ'))
-      .to eq('ニャニャメニャニャジュウニャニャドノニャラビデ' \
-             'ニャクニャクイニャニャクニャニャハンニャニャダイ' \
-             'ニャンニャクニャラベテニャガニャガメ')
-  end
-
-  it 'converts hangul 나-낳 to 냐-냫' do
-    ('나'..'낳').zip('냐'..'냫').each do |na, nya|
-      expect(converter.convert(na)).to eq(nya)
+  include ConverterHelper
+
+  describe '#initialize' do
+    subject(:converter) { described_class.new }
+
+    it 'stores default encoding' do
+      encoding = converter.instance_variable_get(:@encoding)
+      expect(encoding).to be(Encoding::UTF_8)
+    end
+
+    context 'with Encoding object' do
+      subject(:converter) { described_class.new(encoding) }
+      let(:encoding) { Encoding::UTF_16LE }
+
+      it 'stores given encoding' do
+        encoding = converter.instance_variable_get(:@encoding)
+        expect(encoding).to be(Encoding::UTF_16LE)
+      end
+    end
+
+    context 'with string encoding name' do
+      subject(:converter) { described_class.new(encoding) }
+      let(:encoding) { 'utf-16le' }
+
+      it 'converts given name to Encoding object and stores' do
+        encoding = converter.instance_variable_get(:@encoding)
+        expect(encoding).to be(Encoding::UTF_16LE)
+      end
     end
-    expect(converter.convert('나랑 너랑 봄나들이 배낭 매고 봄나들이 ' \
-                             '버드나무 낭창낭창 남실바람 남실남실 ' \
-                             '개나리 꽃에 나비가 하나 ' \
-                             '배낭 속에 바나나가 하나'))
-      .to eq('냐랑 너랑 봄냐들이 배냥 매고 봄냐들이 ' \
-             '버드냐무 냥창냥창 냠실바람 냠실냠실 ' \
-             '개냐리 꽃에 냐비가 하냐 ' \
-             '배냥 속에 바냐냐가 하냐')
-  end
-
-  it 'converts all languages at once' do
-    expect(converter.convert(%w(な ナ 나).join("\n")))
-      .to eq(%w(にゃ ニャ 냐).join("\n"))
+
+    context 'with string encoding alias name' do
+      subject(:converter) { described_class.new(encoding) }
+      let(:encoding) { 'eucKR' }
+
+      it 'converts given name to Encoding object and stores' do
+        encoding = converter.instance_variable_get(:@encoding)
+        expect(encoding).to be(Encoding::EUC_KR)
+      end
+    end
+  end
+
+  context 'with UTF-8 encoding' do
+    subject(:converter) { described_class.new(encoding) }
+    let(:encoding) { Encoding::UTF_8 }
+    it_behaves_like 'a converter supports japanese', Encoding::UTF_8
+    it_behaves_like 'a converter supports hangul', Encoding::UTF_8
+
+    it 'converts all languages at once' do
+      new_source = convert_source(converter, encoding, %w(な ナ 나).join("\n"))
+      expect(new_source).to eq(%w(にゃ ニャ 냐).join("\n").encode(encoding))
+    end
+  end
+
+  context 'with UTF-16BE encoding' do
+    it_behaves_like 'a converter supports japanese', Encoding::UTF_16BE
+    it_behaves_like 'a converter supports hangul', Encoding::UTF_16BE
+  end
+
+  context 'with UTF-16LE encoding' do
+    it_behaves_like 'a converter supports japanese', Encoding::UTF_16LE
+    it_behaves_like 'a converter supports hangul', Encoding::UTF_16LE
+  end
+
+  context 'with EUC-JP encoding' do
+    it_behaves_like 'a converter supports japanese', Encoding::EUC_JP
+  end
+
+  context 'with Shift_JIS encoding' do
+    it_behaves_like 'a converter supports japanese', Encoding::Shift_JIS
+  end
+
+  context 'with EUC-KR encoding' do
+    subject(:converter) { described_class.new(encoding) }
+    let(:encoding) { Encoding::EUC_KR }
+    it_behaves_like 'a converter supports japanese', Encoding::EUC_KR
+
+    it 'converts hangul 나, 낙, 난, 날, 남, 낭 to 냐, 냑, 냔, 냘, 냠, 냥' do
+      %w(나 낙 난 날 남 낭).zip(%w(냐 냑 냔 냘 냠 냥)).each do |na, nya|
+        new_source = convert_source(converter, encoding, na)
+        expect(new_source).to eq(nya.encode(encoding))
+      end
+      new_source = convert_source(converter, encoding,
+                                  '나랑 너랑 봄나들이 배낭 매고 봄나들이 ' \
+                                  '버드나무 낭창낭창 남실바람 남실남실 ' \
+                                  '개나리 꽃에 나비가 하나 ' \
+                                  '배낭 속에 바나나가 하나')
+      expect(new_source).to eq('냐랑 너랑 봄냐들이 배냥 매고 봄냐들이 ' \
+                               '버드냐무 냥창냥창 냠실바람 냠실냠실 ' \
+                               '개냐리 꽃에 냐비가 하냐 ' \
+                               '배냥 속에 바냐냐가 하냐'.encode(encoding))
+    end
+
+    it 'preserves other hangul in range 나-낳' do
+      %w(낚 낟 낡 낢 납 낫 났 낮 낯 낱 낳).each do |ch|
+        new_source = convert_source(converter, encoding, ch)
+        expect(new_source).to eq(ch.encode(encoding))
+      end
+    end
+  end
+
+  context 'with CP949 encoding' do
+    it_behaves_like 'a converter supports japanese', Encoding::CP949
+    it_behaves_like 'a converter supports hangul', Encoding::CP949
   end
 end

data/spec/sawarineko/option_spec.rb

@@ -25,6 +25,7 @@ RSpec.describe Sawarineko::Option do
 
         expected_help = <<-END
 Usage: sawarineko [options] [source]
+    -e, --encoding ENCODING          Specify the encoding of input.
     -h, --help                       Print this message.
     -v, --version                    Print version.
         END

data/spec/sawarineko_spec.rb

@@ -5,7 +5,7 @@ require 'spec_helper'
 RSpec.describe Sawarineko do
   subject(:sawarineko) { described_class }
 
-  describe '#nya' do
+  describe '.nya' do
     it { is_expected.to respond_to(:nya) }
 
     it 'converts string passed as an argument' do

data/spec/support/converter_helper.rb

@@ -0,0 +1,15 @@
+# encoding: utf-8
+
+# Helper methods for testing Converter.
+module ConverterHelper
+  # Convert the source with given encoding.
+  #
+  # converter - The Converter instance to process.
+  # encoding  - The Encoding used in converter.
+  # source    - The String source with UTF-8 encoding.
+  #
+  # Returns the converted source String.
+  def convert_source(converter, encoding, source)
+    converter.convert(source.encode(encoding))
+  end
+end

data/spec/support/shared_converter_supports_hangul.rb

@@ -0,0 +1,27 @@
+# encoding: utf-8
+
+require 'spec_helper'
+
+shared_examples 'a converter supports hangul' do |encoding|
+  include ConverterHelper
+
+  subject(:converter) { described_class.new(encoding) }
+
+  describe '#convert' do
+    it 'converts hangul 나-낳 to 냐-냫' do
+      ('나'..'낳').zip('냐'..'냫').each do |na, nya|
+        new_source = convert_source(converter, encoding, na)
+        expect(new_source).to eq(nya.encode(encoding))
+      end
+      new_source = convert_source(converter, encoding,
+                                  '나랑 너랑 봄나들이 배낭 매고 봄나들이 ' \
+                                  '버드나무 낭창낭창 남실바람 남실남실 ' \
+                                  '개나리 꽃에 나비가 하나 ' \
+                                  '배낭 속에 바나나가 하나')
+      expect(new_source).to eq('냐랑 너랑 봄냐들이 배냥 매고 봄냐들이 ' \
+                               '버드냐무 냥창냥창 냠실바람 냠실냠실 ' \
+                               '개냐리 꽃에 냐비가 하냐 ' \
+                               '배냥 속에 바냐냐가 하냐'.encode(encoding))
+    end
+  end
+end

data/spec/support/shared_converter_supports_japanese.rb

@@ -0,0 +1,39 @@
+# encoding: utf-8
+
+require 'spec_helper'
+
+shared_examples 'a converter supports japanese' do |encoding|
+  include ConverterHelper
+
+  subject(:converter) { described_class.new(encoding) }
+
+  describe '#convert' do
+    it 'converts hiragana な to にゃ' do
+      new_source = convert_source(converter, encoding, 'な')
+      expect(new_source).to eq('にゃ'.encode(encoding))
+      new_source = convert_source(converter, encoding,
+                                  'ななめななじゅうななどの' \
+                                  'ならびでなくなくいななく' \
+                                  'ななはんななだいなんなく' \
+                                  'ならべてながながめ')
+      expect(new_source).to eq('にゃにゃめにゃにゃじゅうにゃにゃどの' \
+                               'にゃらびでにゃくにゃくいにゃにゃく' \
+                               'にゃにゃはんにゃにゃだいにゃんにゃく' \
+                               'にゃらべてにゃがにゃがめ'.encode(encoding))
+    end
+
+    it 'converts katakana ナ to ニャ' do
+      new_source = convert_source(converter, encoding, 'ナ')
+      expect(new_source).to eq('ニャ'.encode(encoding))
+      new_source = convert_source(converter, encoding,
+                                  'ナナメナナジュウナナドノ' \
+                                  'ナラビデナクナクイナナク' \
+                                  'ナナハンナナダイナンナク' \
+                                  'ナラベテナガナガメ')
+      expect(new_source).to eq('ニャニャメニャニャジュウニャニャドノ' \
+                               'ニャラビデニャクニャクイニャニャク' \
+                               'ニャニャハンニャニャダイニャンニャク' \
+                               'ニャラベテニャガニャガメ'.encode(encoding))
+    end
+  end
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: sawarineko
 version: !ruby/object:Gem::Version
-  version: 1.1.0
+  version: 1.2.0
 platform: ruby
 authors:
 - ChaYoung You
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-10-28 00:00:00.000000000 Z
+date: 2014-10-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -111,9 +111,12 @@ files:
 - spec/sawarineko/version_spec.rb
 - spec/sawarineko_spec.rb
 - spec/spec_helper.rb
+- spec/support/converter_helper.rb
 - spec/support/exit_code_matchers.rb
 - spec/support/file_helper.rb
 - spec/support/isolated_environment.rb
+- spec/support/shared_converter_supports_hangul.rb
+- spec/support/shared_converter_supports_japanese.rb
 homepage: ''
 licenses:
 - MIT
@@ -145,6 +148,9 @@ test_files:
 - spec/sawarineko/version_spec.rb
 - spec/sawarineko_spec.rb
 - spec/spec_helper.rb
+- spec/support/converter_helper.rb
 - spec/support/exit_code_matchers.rb
 - spec/support/file_helper.rb
 - spec/support/isolated_environment.rb
+- spec/support/shared_converter_supports_hangul.rb
+- spec/support/shared_converter_supports_japanese.rb