zhongwen_tools 0.12.4 → 0.15.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 466a2ff98a772d2209ee19dc8e9247f5d583a634
-  data.tar.gz: 70b02d066dd3d5fe4f77d9a44c3f7d128fd8e0e1
+  metadata.gz: f0892b76a8079a31b85cb39141d4a71d2d35f4dc
+  data.tar.gz: 75d2bab50cbde53afd7a98aa331e9d40e01b376d
 SHA512:
-  metadata.gz: ed785ce19696c432b19cb0f638f1ba600457b4ffe78712f3a0d877709003fc31752978c53bbb007709615e04bb49b4109d08169bf5709a567fec058435fe123a
-  data.tar.gz: 4c54eba7f80ab986b01eaf746d76e08730387bc16d088a55a9881fa9c1d595577a11b1a4de294564da14b6d6534884b9ca687e66a5b887f344d073b798c0da5f
+  metadata.gz: 522d46768adfbff771b16d0007e80b00aaceaa757d86d791d0811b2d495ee0cbe297ca11ade973dc2ae6dd7937c455cb27d1ad85c1e6769e0c707643f3a13cd9
+  data.tar.gz: b2e9b8739311ccdf49d637af1f633752088e002086e4a99578edd1d82d7f8509277029dbf51f8c95bbe30cd3583f818da8e981137a534748f72b04c92ccd9ee6

data/Gemfile

@@ -1,3 +1,3 @@
 source "https://rubygems.org"
-# Specify your gem's dependencies in zhongwen_tools.gemspec
+# gem dependencies are in zhongwen_tools.gemspec
 gemspec

data/README.md

@@ -1,5 +1,5 @@
 #Zhongwen Tools:
-Tools and methods for dealing with Chinese.
+Methods for dealing with Chinese.
 
 [![Build
 Status](https://travis-ci.org/stevendaniels/zhongwen_tools.png?branch=master)](https://travis-ci.org/stevendaniels/zhongwen_tools) [![Dependency Status](https://gemnasium.com/stevendaniels/zhongwen_tools.png)](https://gemnasium.com/stevendaniels/zhongwen_tools) [![Code Climate](https://codeclimate.com/github/stevendaniels/zhongwen_tools.png)](https://codeclimate.com/github/stevendaniels/zhongwen_tools) [![Coverage Status](https://coveralls.io/repos/stevendaniels/zhongwen_tools/badge.png)](https://coveralls.io/r/stevendaniels/zhongwen_tools)
@@ -12,224 +12,133 @@ Install as a gem
     $ [sudo] gem install zhongwen_tools
 
 
-Zhongwen Tools is a set of modules that makes working with Chinese
-easier. It includes the following:
-
-1. ZhongwenTools::String - deals with strings that are Chinese or pinyin.
-2. ZhongwenTools::Numbers - identifies Chinese numbers and converts numbers to and from Chinese.
-3. ZhongwenTools::Integer - converts integers into Chinese or pinyin.
-4. ZhongwenTools::Romanization - converts converting between Chinese romanization systems like Pinyin and Wade-Giles.
-5. ZhongwenTools::Conversion - converts between Chinese scripts.
-
-
-
 ## Usage
 
-Add the ZhongwenTools component you need to your classes as a module.
+You can monkey patch the String class or use refinements (Ruby 2.0+).
 
-    require 'zhongwen_tools/romanization'
+### Monkey Patch
 
-    class String
-      include ZhongwenTools::Romanization
-    end
+You can monkey patch the String class.
 
-    str = 'ni3 hao3'  #pinyin with numbers
-    str.to_pinyin
-    #=> 'nǐ hǎo'
+    require 'zhongwen_tools'
+    require 'zhongwen_tools/core_ext/string'
 
-    str.to_zhuyin_fuhao
-    #=> 'ㄋㄧ3 ㄏㄠ3'
+    # normalizes the behavior between Ruby 1.9.2, 1.9.3, 2+
+    '中国'chars #=> ['中', '国']
 
-    mzd = 'Mao Tse-tung'
-    mzd.to_pinyin
-    #=> 'Mao Zedong'
+    'zhōngguó'.capitalize #=> 'Zhōngguó'
 
-Or require the components you want.
+    'Zhōngguó'.zh_downcase #=> 'zhōngguó'
 
-    require 'zhongwen_tools/numbers'
+    'ālābó'.zh_upcase #=> 'ĀLĀBÓ'
 
-    ZhongwenTools::Numbers.to_pyn '一百二十'
-    #=> 'yi1-bai2-er4-shi2'
+    '你好Kailin！'.has_zh? # => true
 
-### Using ZhongwenTools::String
-Zhongwen Tools has string methods for detecting different string formats
-and for converting to and from halfwidth, fullwidth, and utf-8 codepoints.
+    '你好Kailin！'.has_zh_punctuation? # => true
 
-    require 'zhongwen_tools/string'
+    '你好'.zh? #=> true
 
-    ZhongwenTools::String.ascii? 'hello'
-    #=> true #non-multibyle strings
+    '你好'.uri_encode
 
-    ZhongwenTools::String.multibyte? '中文'
-    #=> true #multibtye strings
+    '你好'.uri_escape
 
-    ZhongwenTools::String.halfwidth? 'hello'
-    #=> true
+    '你好！'.strip_zh_punctuation #=> '你好'
 
-    ZhongwenTools::String.fullwidth? 'ｈｅｌｌｏ'
-    #=> true
+    'hello'.ascii?
 
-    ZhongwenTools::String.to_halfwidth 'ｈｅｌｌｏ'
-    #=> 'hello'
+    '你好'.multibyte?
 
-    ZhongwenTools::String.uri_encode '我太懒'
-    #=> '%E6%88%91%E5%A4%AA%E6%87%92'
+    'hello'.halfwidth?
 
-    ZhongwenTools::String.to_codepoint '中文'
-    #=> '\u4e2d\u6587'
+    'ｈeｌｌｏ'.fullwidth?
+    
+    'ｈeｌｌｏ'.to_halfwidth?
 
-    ZhongwenTools::String.from_codepoint '\u4e2d\u6587'
-    #=> '中文' #converts string from a utf-8 codepoint.
+    '\u4e2d\u6587'.from_codepoint #=> '中文'
 
+    '中文'.to_codepoint #=>  '\u4e2d\u6587'
 
-#### Detecting Chinese or Chinese Punctuation.
-Zhongwen Tools can also detect if a string is or has Chinese or Chinese
-punctuation.
+    'Zhong1guo2'.to_pinyin #=> 'Zhōngguó'
 
-	require 'zhongwen_tools/string'
-    ZhongwenTools::String.has_zh? '1月'
-    #=> true
+    'Zhong1guo2'.to_py #=> 'Zhōngguó'
 
-    ZhongwenTools::String.zh? '1月'
-    #=> false #(The string can't be mixed.)
+    'Zhōngguó'.to_py #=> 'Zhong1guo2'
 
-    ZhongwenTools::String.has_zh_punctuation? '你在哪里？'
-    #=> true
-    ZhongwenTools::String.strip_zh_punctuation? '你在哪里？'
-    #=> '你在哪里'
+    'nǐ hǎo'.to_bpmf #=> 'ㄋㄧ3 ㄏㄠ3'
 
-#### Converting between Traditional and Simplified Chinese
-By requiring conversion module, ZhongwenTools::String gets some
-convenience methods for converting to and from traditional and
-simplified Chinese.
+    'nǐ hǎo'.to_zyfh #=> 'ㄋㄧ3 ㄏㄠ3'
 
-    require 'zhongwen_tools/conversion'
+    'nǐ hǎo'.to_zhyfh #=> 'ㄋㄧ3 ㄏㄠ3'
 
-    ZhongwenTools::String.zhs? '中国'
-    #=> true
-    ZhongwenTools::String.zht? '中国'
-    #=> false
+    'nǐ hǎo'.to_bopomofo #=> 'ㄋㄧ3 ㄏㄠ3'
 
-#### Romanization
-By requiring the romanization module ZhongwenTools::String gets some
-convenience methods for dealing with romanization.
+    'nǐ hǎo'.to_yale #=> 'ni3 hau3',
 
-    require 'zhongwen_tools/romanziation'
+    'nǐ hǎo'.to_typy #=> 'ni3 hao3',
 
-    ZhongwenTools::String.to_pinyin 'ni3 hao3'
-    #=> 'nǐ hǎo'
+    'nǐ hǎo'.to_tongyong #=> 'ni3 hao3',
 
+    'nǐ hǎo'.to_tongyong_pinyin #=> 'ni3 hao3',
 
-#### Pinyin-safe String Methods
-The following capitalization methods work for pinyin.
+    'nǐ hǎo'.to_wg #=> 'ni3 hao3',
 
-    require 'zhongwen_tools/string'
+    'nǐ hǎo'.to_wade_giles #=> 'ni3 hao3',
 
-    ZhongwenTools::String.downcase 'Àomén'
-    #=> 'àomén' does pinyin/ lowercase
-    ZhongwenTools::String.upcase 'àomén'
-    #=> 'ÀOMÉN'
-    ZhongwenTools::String.capitalize 'àomén'
-    #=> 'Àomén'
+    'nǐ hǎo'.to_mps2 #=> 'ni3 hau3'
 
+    '你们好'.zhs? #=> true
 
-### Numbers
-Functions for converting to and from Chinese numbers.
+    '你们好'.zht? #=> false
 
-    require 'zhongwen_tools/numbers'
+    '你們好'.to_zhs #=> '你们好'
 
-    ZhongwenTools::Numbers.number_to_zht :num, 12000
-    #=> '一萬二千'
-    ZhongwenTools::Numbers.number_to_zhs :num, 42
-    #=> '四十二'
-    ZhongwenTools::Numbers.number_to_pyn :num, 42
-    #=> 'si4-shi2-er4'
-    ZhongwenTools::Numbers.zh_number_to_number '四十二'
-    #=> 42
-    ZhongwenTools::Numbers.number? '四十二'
-    #=> true
+    '你们好'.to_zht #=> '你們好'
 
-### Integers
-Monkey-patch your integers for Chinese.
+    '金枪鱼'.to_zhtw #=> '鮪魚'
 
-    require 'zhongwen_tools/ingteger'
+    '鮪魚'.to_zhcn #=> '金枪鱼'
 
-    class Integer
-      include ZhongwenTools::Integer
-    end
+    '金枪鱼'.to_zhhk #=> '吞拿魚'
 
-    12.to_pinyin
-    #=> 'shi2-er4'
-    12.to_zht
-    #=> '十二'
 
+#### Integer Extensions
 
-### Romanization
-ZhongwenTools::Romanization has tools for converting between Chinese language romanization systems and
-scripts. It **does not convert Chinese characters to pinyin** (see ZhongwenTools::Conversion). Romanization methods must be required explicitly.
+You can also monkey patch the Integer class!
+    require 'zhongwen_tools'
+    require 'zhongwen_tools/core_ext/integer'
 
-    gem 'zhongwen_tools'
-    require 'zhongwen_tools/romanization'
+    1.to_pyn #=> 'yi1'
+    10_000.to_zht #=> '一萬'
+    10_000.to_zhs #=> '一万'
 
-    class String
-      include ZhongwenTools::Romanization
-    end
+### Refinements (Ruby > 2.0 only)
 
+You can also use refinements, to refine the string or integer class.
 
-    str = 'ni3 hao3'
-    py = 'nǐ hǎo'
+    # some_class.rb
+    require 'zhongwen_tools'
+    using ZhongwenTools
 
-    str.to_pinyin
-    #=> 'nǐ hǎo'
-    str.to_py
-    #=> 'nǐ hǎo'
+    class SomeClass
 
-    py.to_pyn
-    #=> 'ni3 hao3'
-
-    str.to_wg
-    #=> 'ni3 hao3'    #Wade-Giles
-
-    str.to_bpmf
-    #=> 'ㄋㄧ3 ㄏㄠ3' #Zhuyin Fuhao, a.k.a. Bopomofo
-
-    str.to_yale
-    #=> 'ni3 hau3'
-
-    str.to_typy
-    #=> 'ni3 hao3'
-
-    str.pyn?
-    #=> true
-    str.wg?
-    #=> true #(There can be overlap between Wade-Giles and Pinyin)
-    str.to_py.py? 
-    #=> true
+      def hi
+        '你们好'.to_zht
+      end
+    end
 
-    #split pinyin with numbers accurately.
-    'dong1xi1'.split_pyn    # => ['dong1', 'xi1']
-    'dongxi'.split_pyn      # => ['dong', 'xi']
+    SomeClass.new.hi #=> 你們好
 
-### Conversion
-Functions for converting between scripts (e.g. traditional Chinese to
-simplified Chinese) and [TODO] between Chinese and romanization systems (e.g.
-Chinese to pinyin).
-Conversion methods must be required explicitly.
+### ZhongwenTools::Core
 
-    gem 'zhongwen_tools'
-    require 'zhongwen_tools/conversion'
+The core functionality of ZhongwenTools excludes converting between
+simplified and traditional Chinese. You can use it by requiring
+'zhongwen_tools/core' instead of 'zhongwen_tools'
 
-    ZhongwenTools::Conversion.to_zhs '華語'
-    #=> '华语'
-    ZhongwenTools::Conversion.to_zht '华语'
-    #=> '華語'
-    ZhongwenTools::Conversion.to_zhtw '方便面'
-    #=> '泡麵'
-    ZhongwenTools::Conversion.to_zhhk '方便面'
-    #=> '即食麵'
-    ZhongwenTools::Conversion.to_zhcn '即食麵'
-    #=> '方便面'
+    require 'zhongwen_tools/core' 
+    require 'zhongwen_tools/core_ext/string'
 
+    'ni3 hao3'.to_pinyin #=> 'nǐ hǎo'
+    '你們好'.to_zhs #=> NoMethodError
 
-## TODO
-1. A character -> pinyin converter
+##TODO:
+ 1. create a generic ZhongwenTools::Romanization.split method for convenience

data/Rakefile

@@ -1,4 +1,3 @@
-require 'rubygems' if RUBY_VERSION < '1.9'
 require 'bundler/gem_tasks'
 require 'rake/testtask'
 Bundler.require :test

data/lib/zhongwen_tools/{string/caps.rb → caps.rb}

@@ -1,7 +1,24 @@
 # encoding: utf-8
 
 module ZhongwenTools
-  UNICODE_CAPS = {
+  module Caps
+
+  def self.downcase(str)
+    regex = /(#{ZhongwenTools::Caps::CAPS.keys.join('|')})/
+    str.gsub(regex, ZhongwenTools::Caps::CAPS).downcase
+  end
+
+  def self.upcase(str)
+    str.gsub(/(#{ZhongwenTools::Caps::CAPS.values.join('|')})/){
+      ZhongwenTools::Caps::CAPS.find{|k,v| v == $1}[0]
+    }.upcase
+  end
+
+  def self.capitalize(str)
+    str.sub(str[0], ZhongwenTools::Caps.upcase(str[0]))
+  end
+
+  CAPS = {
     'Ā' => 'ā',
     'Á' => 'á',
     'Ǎ' => 'ǎ',
@@ -53,4 +70,5 @@ module ZhongwenTools
     "Ｙ" => "ｙ",
     "Ｚ" => "ｚ"
   }
+  end
 end

data/lib/zhongwen_tools/core.rb

@@ -0,0 +1,19 @@
+# encoding: utf-8
+require 'zhongwen_tools/ruby_19' if RUBY_VERSION < '2.0.0'
+require 'zhongwen_tools/regex'
+require 'zhongwen_tools/fullwidth'
+require 'zhongwen_tools/unicode'
+require 'zhongwen_tools/uri'
+require 'zhongwen_tools/zhongwen'
+require 'zhongwen_tools/number'
+require 'zhongwen_tools/romanization'
+require 'zhongwen_tools/string_extension'
+
+module ZhongwenTools
+  # Allow for the use Refinements.
+  if RUBY_VERSION >= '2.0.0'
+    refine String do
+      include ::ZhongwenTools::StringExtension
+    end
+  end
+end

data/lib/zhongwen_tools/core_ext/integer.rb

@@ -0,0 +1,8 @@
+require 'zhongwen_tools/integer_extension'
+ensure_module_defined = ->(base, module_name){
+  base.const_set(module_name, Module.new) unless base.const_defined?(module_name)
+}
+ensure_module_defined[ZhongwenTools, :Number]
+ensure_module_defined[ZhongwenTools, :IntegerExtension]
+
+Integer.send(:include, ZhongwenTools::IntegerExtension)

data/lib/zhongwen_tools/core_ext/string.rb

@@ -0,0 +1,10 @@
+require 'zhongwen_tools/string_extension'
+ensure_module_defined = ->(base, module_name){
+  base.const_set(module_name, Module.new) unless base.const_defined?(module_name)
+}
+
+ensure_module_defined[ZhongwenTools, :Romanization]
+ensure_module_defined[ZhongwenTools, :StringExtension]
+ensure_module_defined[ZhongwenTools, :Script]
+
+String.send(:include, ZhongwenTools::StringExtension)

data/lib/zhongwen_tools/fullwidth.rb

@@ -0,0 +1,102 @@
+# encoding: utf-8
+require 'zhongwen_tools/regex'
+
+module ZhongwenTools
+  module Fullwidth
+    # TODO: type checking.
+    def self.halfwidth?(str)
+      str[ZhongwenTools::Regex.fullwidth].nil?
+    end
+
+    def self.fullwidth?(str)
+      !halfwidth?(str) && to_halfwidth(str) != str
+    end
+
+    def self.to_halfwidth(str)
+      str.gsub(/(#{ZhongwenTools::Regex.fullwidth})/, ZhongwenTools::Fullwidth::FW_HW)
+    end
+
+    FW_HW ={
+      "０" => "0",
+      "１" => "1",
+      "２" => "2",
+      "３" => "3",
+      "４" => "4",
+      "５" => "5",
+      "６" => "6",
+      "７" => "7",
+      "８" => "8",
+      "９" => "9",
+      "Ａ" => "A",
+      "Ｂ" => "B",
+      "Ｃ" => "C",
+      "Ｄ" => "D",
+      "Ｅ" => "E",
+      "Ｆ" => "F",
+      "Ｇ" => "G",
+      "Ｈ" => "H",
+      "Ｉ" => "I",
+      "Ｊ" => "J",
+      "Ｋ" => "K",
+      "Ｌ" => "L",
+      "Ｍ" => "M",
+      "Ｎ" => "N",
+      "Ｏ" => "O",
+      "Ｐ" => "P",
+      "Ｑ" => "Q",
+      "Ｒ" => "R",
+      "Ｓ" => "S",
+      "Ｔ" => "T",
+      "Ｕ" => "U",
+      "Ｖ" => "V",
+      "Ｗ" => "W",
+      "Ｘ" => "X",
+      "Ｙ" => "Y",
+      "Ｚ" => "Z",
+      "ａ" => "a",
+      "ｂ" => "b",
+      "ｃ" => "c",
+      "ｄ" => "d",
+      "ｅ" => "e",
+      "ｆ" => "f",
+      "ｇ" => "g",
+      "ｈ" => "h",
+      "ｉ" => "i",
+      "ｊ" => "j",
+      "ｋ" => "k",
+      "ｌ" => "l",
+      "ｍ" => "m",
+      "ｎ" => "n",
+      "ｏ" => "o",
+      "ｐ" => "p",
+      "ｑ" => "q",
+      "ｒ" => "r",
+      "ｓ" => "s",
+      "ｔ" => "t",
+      "ｕ" => "u",
+      "ｖ" => "v",
+      "ｗ" => "w",
+      "ｘ" => "x",
+      "ｙ" => "y",
+      "ｚ" => "z",
+      "％" => '%',
+      "．" => '.',
+      '：' => ':',
+      "＃" => '#',
+      "＄" => "$",
+      "＆" => "&",
+      "＋" => "+",
+      "－" => "-",
+      "／" => "/",
+      "＼" => '\\',
+      '＝' => '=',
+      "；" => ";",
+      "＜" => "<",
+      "＞" => ">",
+      "？" => "?",
+      "。" => ".",
+      "！" => "!",
+      '，' => ','
+    }
+  end
+end

data/lib/zhongwen_tools/integer_extension.rb

@@ -0,0 +1,31 @@
+# encoding: utf-8
+require 'zhongwen_tools/number'
+
+module ZhongwenTools
+  module IntegerExtension
+
+    def to_zh(type = nil)
+      #type == :zht ? self.to_zht? : self.to_zhs
+         ZhongwenTools::Number.convert(self, type, :i)
+    end
+
+    %w(zhs zht).each do |type|
+      define_method("to_#{type}") do
+         ZhongwenTools::Number.convert(self, type.to_sym, :i)
+      end
+    end
+
+    def to_pyn
+      ZhongwenTools::Number.convert self, :pyn, :i
+    end
+
+    # TODO: add to_pyn to_bmpf, to_wg, to_mps2, to_yale, etc.
+  end
+
+  # Allow for the use Refinements.
+  if RUBY_VERSION >= '2.0.0'
+    refine Integer do
+      include ::ZhongwenTools::IntegerExtension
+    end
+  end
+end

data/lib/zhongwen_tools/number/number_table.rb

@@ -0,0 +1,44 @@
+# encoding: utf-8
+
+module ZhongwenTools
+  module Number
+    # TODO: Add huge numbers.
+    # 垓	秭	穰	溝	澗	正	載 --> beyond 100,000,000!
+    # NOTE: financial numbers i == 0 ? NT.select{ |x| x[:i] == i }.last[:zhs] : NT.find{ |x| x[:i] = i }
+    NUMBERS_TABLE = [
+      { :zhs => '零', :zht => '零', :i => 0, :pyn => 'ling2'},
+      { :zhs => '〇', :zht => '〇', :i => 0, :pyn => 'ling2'},
+      { :zhs => '一', :zht => '一', :i => 1, :pyn => 'yi1'},
+      { :zhs => '壹', :zht => '壹', :i => 1, :pyn => 'yi1'},
+      { :zhs => '幺', :zht => '幺', :i => 1, :pyn => 'yao1'},
+      { :zhs => '二', :zht => '二', :i => 2, :pyn => 'er4'},
+      { :zhs => '两', :zht => '兩', :i => 2, :pyn => 'liang3'},
+      { :zhs => '贰', :zht => '貳', :i => 2, :pyn => 'er4'},
+      { :zhs => '三', :zht => '三', :i => 3, :pyn => 'san1'},
+      { :zhs => '弎', :zht => '弎', :i => 3, :pyn => 'san1'},
+      { :zhs => '叁', :zht => '參', :i => 3, :pyn => 'san1'},
+      { :zhs => '四', :zht => '四', :i => 4, :pyn => 'si4'},
+      { :zhs => '䦉', :zht => '䦉', :i => 4, :pyn => 'si4'},
+      { :zhs => '肆', :zht => '肆', :i => 4, :pyn => 'si4'},
+      { :zhs => '五', :zht => '五', :i => 5, :pyn => 'wu3'},
+      { :zhs => '伍', :zht => '伍', :i => 5, :pyn => 'wu3'},
+      { :zhs => '六', :zht => '六', :i => 6, :pyn => 'liu4'},
+      { :zhs => '陆', :zht => '陸', :i => 6, :pyn => 'liu4'},
+      { :zhs => '七', :zht => '七', :i => 7, :pyn => 'qi1'},
+      { :zhs => '柒', :zht => '柒', :i => 7, :pyn => 'qi1'},
+      { :zhs => '八', :zht => '八', :i => 8, :pyn => 'ba1'},
+      { :zhs => '捌', :zht => '捌', :i => 8, :pyn => 'ba1'},
+      { :zhs => '九', :zht => '九', :i => 9, :pyn => 'jiu3'},
+      { :zhs => '玖', :zht => '玖', :i => 9, :pyn => 'jiu3'},
+      { :zhs => '十', :zht => '十', :i => 10, :pyn => 'shi2'},
+      { :zhs => '拾', :zht => '拾', :i => 10, :pyn => 'shi2'},
+      { :zhs => '廿', :zht => '廿', :i => 20, :pyn => ' nian4'},
+      { :zhs => '百', :zht => '百', :i => 100, :pyn => 'bai2'},
+      { :zhs => '佰', :zht => '佰', :i => 100, :pyn => 'bai2'},
+      { :zhs => '千', :zht => '千', :i => 1_000, :pyn => 'qian1'},
+      { :zhs => '仟', :zht => '仟', :i => 1_000, :pyn => 'qian1'},
+      { :zhs => '万', :zht => '萬', :i => 10_000, :pyn => 'wan4'},
+      { :zhs => '亿', :zht => '億', :i => 100_000_000, :pyn => 'yi4'},
+    ]
+  end
+end