zhongwen_tools 0.0.6 → 0.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: b1e19e456d7cf778c9a749a75284044981086a02
-  data.tar.gz: 103ae6d8d26029b2854bdd09e02a10bff64d5df1
+  metadata.gz: e2db83235c285238192917440b33bbb8f2f1d16f
+  data.tar.gz: abb0da06eeefc95b9a3fc1b86ef98f291430a295
 SHA512:
-  metadata.gz: dff5a94d7af2e65b6f6a63ae8a5593312eef78df4fe3b8fe9c1280bf05874db12d4230c266f6de63de655b1d07db06fa430d49584daf120d65cabc33fd9cd94a
-  data.tar.gz: 0078f0cb0ca8724c34403c04472c063ea53836b261047d968c4a78eb18eba2985356004d3dabf6314e3b930635e4a1c1058f154f45fea1f14750e903991d21b3
+  metadata.gz: a753b626668df9e3f1e229fb25769fdd5c5a0edda04bf5027c904fb8fbc644a76b8bd448d96f84c2eb431dc36114c867cfc9e39764290e6a70b2fd695707fcb0
+  data.tar.gz: d77bba5b9e8345253bb2fa5a6ac154ffca1322a9d32de374bd248d0d41b808b6a7b4e667e47215fc93108c4af74adeda7aebed83144164bd107f18e75ea603a8

data/README.md

@@ -1,6 +1,8 @@
-#Zhongwen Tools: tools and methods for dealing with Chinese.
+#Zhongwen Tools:
+Tools and 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)
+[![GitHub version](https://badge.fury.io/gh/stevendaniels%2Fzhongwen_tools.png)](http://badge.fury.io/gh/stevendaniels%2Fzhongwen_tools)
 ##INSTALLATION
 
 Install as a gem
@@ -12,7 +14,7 @@ Install as a gem
 Add the ZhongwenTools component you need to your classes as a module.
 
     class String
-      include ZhongwenToolsRomanization
+      include ZhongwenTools::Romanization
     end
 
     str = "ni3 hao3"  #pinyin with numbers
@@ -20,11 +22,12 @@ Add the ZhongwenTools component you need to your classes as a module.
     str.to_zhuyinfuhao  #=>
 
     mzd = "Mao Tse-tung"
-    mzd.to_pinyin   #=> Mao Zedong
+    mzd.to_pinyin   #=> "Mao Zedong"
 
 Or you can require the components you want
+
     require 'zhongwen_tools/numbers'
-    ZhongwenTools::Numbers.to_pinyin '一百二十' #=> 'yi1-bai2-er4-shi2'
+    ZhongwenTools::Numbers.to_pyn '一百二十' #=> 'yi1-bai2-er4-shi2'
 
 ZhongwenTools includes the following modules:
 
@@ -32,17 +35,17 @@ ZhongwenTools includes the following modules:
 2. ZhongwenTools::Numbers => functions for identifying and converting numbers.
 3. ZhongwenTools::Integer => some useful integer functions for Chinese:
    e.g. 12.to_pinyin 12.to_zht
-4. ZhongwenTools::Romanization => functions for converting between Chinese romanization systems
-5. ZhongwenTools::Conversion => functions for converting between Chinese scripts.
-6. ZhongwenTools::ToneSandhi => functions for identifying and dealing with tone sandhi. (Wiki URL)
+4. [TODO] ZhongwenTools::Romanization => functions for converting between Chinese romanization systems
+5. [TODO] ZhongwenTools::Conversion => functions for converting between Chinese scripts.
+6. [TODO] ZhongwenTools::ToneSandhi => functions for identifying and dealing with tone sandhi. (Wiki URL)
 7. [TODO] ZhongwenTools::Segmentation => functions for segmenting Chinese. Can provide different methods for converting
-8. ZhongwenTools::Tagging => functions for tagging Chinese POS, NER, etc.
-  
-  
-### ZhongwenTools::String: useful string functions for ZhongwenTools language
+8. [TODO] ZhongwenTools::Tagging => functions for tagging Chinese POS, NER, etc.
+
+
+### ZhongwenTools::String: useful string functions for Chinese.
     ZhongwenTools::String.ascii? 'hello'    #=> true #non-multibyle strings
     ZhongwenTools::String.multibyte? '中文'  #=> true #multibtye strings
-    ZhongwenTools::String.halfwidth? 
+    ZhongwenTools::String.halfwidth?
     ZhongwenTools::String.fullwidth?
     ZhongwenTools::String.to_halfwidth
     ZhongwenTools::String.uri_encode  #=> just because I'm lazy
@@ -52,12 +55,12 @@ ZhongwenTools includes the following modules:
     ZhongwenTools::String.upcase --> does pinyin uppercase
     ZhongwenTools::String.capitalize ---> does pinyin / fullwidth capitalization
 
-    ZhongwenTools::String.has_zh? '1月'     #=> true 
+    ZhongwenTools::String.has_zh? '1月'     #=> true
     ZhongwenTools::String.is_zh? '1月'      #=> false can't be mixed.
     ZhongwenTools::String.is_zhs? '中国'    #=> true
     ZhongwenTools::String.is_zht? '中国'    #=> false
 
-#### ruby 1.8 safe methods 
+#### Ruby 1.8 safe methods
     ZhongwenTools::String.chars '中文' #=> ['中','文']
     ZhongwenTools::String.size '中文'  #=> 2
     ZhongwenTools::String.reverse '中文' #=> '文中'
@@ -67,21 +70,36 @@ ZhongwenTools includes the following modules:
 ###Numbers
 Functions for converting to and from Chinese numbers.
 
+    ZhongwenTools::Number.number_zht 12000        #=> '一萬二千'
+    ZhongwenTools::Number.number_zhs 42           #=> '四十二'
+    ZhongwenTools::Number.number_to_pyn 42        #=> 'si4-shi2-er4'
+    ZhongwenTools::Number.number_to_int '四十二'  #=> 42
+    ZhongwenTools::Number.number? '四十二'        #=> true
+
 ###Integers
+Monkey-patch your integers for Chinese.
+
+    class Integer
+      include ZhongwenTools::Integer
+    end
+
+    12.to_pinyin #=> 'shi2-er4'
+    12.to_zht    #=> '十二'
+
 
 ### Romanization
 ZhongwenTools::Chinese has tools for converting between Chinese language romanization systems and
 scripts.
 
     class String
-      include ZhongwenToolsRomanization
+      include ZhongwenTools::Romanization
     end
 
 
     str = "ni3 hao3"
     romanization_system = "pyn" #pyn|wg|yale|bpmf|zhyfh|wade-giles|bopomofo
 
-    str.to_pinyin romanization_system   
+    str.to_pinyin romanization_system
     #=> "nǐ hǎo"
 
     str.to_py romanization_system
@@ -106,13 +124,13 @@ Functions for converting between scripts (e.g. traditional Chinese to
 simplified Chinese) and between chinese and romanization systems (e.g.
 Chinese to pinyin).
 
-ZhongwenTools::Conversion.to_zhs
-ZhongwenTools::Conversion.to_zht
-ZhongwenTools::Conversion.to_zhtw
-ZhongwenTools::Conversion.to_zhhk
-ZhongwenTools::Conversion.to_zhmc
-ZhongwenTools::Conversion.to_zhsg
-ZhongwenTools::Conversion.to_zhprc
+    ZhongwenTools::Conversion.to_zhs ‘華語'
+    ZhongwenTools::Conversion.to_zht '华语'
+    ZhongwenTools::Conversion.to_zhtw '方便面'
+    ZhongwenTools::Conversion.to_zhhk '方便面'
+    ZhongwenTools::Conversion.to_zhmc
+    ZhongwenTools::Conversion.to_zhsg
+    ZhongwenTools::Conversion.to_zhprc '马铃薯'
 
 
 ###Tone Sandhi
@@ -125,4 +143,3 @@ that requires an external dependency is packaged as a separate gem.
 ## TODO
 1. A trad/simp script converter
 2. A character -> pinyin converter
-3. A language detector

data/lib/zhongwen_tools/integer.rb

@@ -0,0 +1,41 @@
+#encoding: utf-8
+require './lib/zhongwen_tools/numbers'
+
+module ZhongwenTools
+  module Integer
+    include ZhongwenTools::Numbers
+
+    def to_zh(type = nil)
+      type == :zht ? self.to_zht? : self.to_zhs
+    end
+
+    def to_zhs(int = nil)
+      int ||= self
+      number_to_zhs :num, int.to_s
+    end
+
+    def to_zht(int = nil)
+      int ||= self
+      number_to_zht :num, int.to_s
+    end
+
+    def to_pyn(int = nil)
+      int ||= self
+      number_to_pyn int.to_s, :num
+    end
+
+    class Basement
+      include ZhongwenTools::Integer
+    end
+
+    def self.to_zhs(*args)
+      Basement.new.to_zhs(*args)
+    end
+    def self.to_zht(*args)
+      Basement.new.to_zht(*args)
+    end
+    def self.to_pyn(*args)
+      Basement.new.to_pyn(*args)
+    end
+  end
+end

data/lib/zhongwen_tools/numbers.rb

@@ -5,46 +5,69 @@ module ZhongwenTools
     NUMBER_MULTIPLES = '拾十百佰千仟仟万萬亿億'
 
     NUMBERS_TABLE = [
-      { :zh_s => '零', :zh_t => '零', :num => 0, :pyn => 'ling2'},
-      { :zh_s => '〇', :zh_t => '〇', :num => 0, :pyn => 'ling2'},
-      { :zh_s => '一', :zh_t => '一', :num => 1, :pyn => 'yi1'},
-      { :zh_s => '壹', :zh_t => '壹', :num => 1, :pyn => 'yi1'},
-      { :zh_s => '幺', :zh_t => '幺', :num => 1, :pyn => 'yao1'},
-      { :zh_s => '二', :zh_t => '二', :num => 2, :pyn => 'er4'},
-      { :zh_s => '两', :zh_t => '兩', :num => 2, :pyn => 'liang3'},
-      { :zh_s => '贰', :zh_t => '貳', :num => 2, :pyn => 'er4'},
-      { :zh_s => '三', :zh_t => '三', :num => 3, :pyn => 'san1'},
-      { :zh_s => '弎', :zh_t => '弎', :num => 3, :pyn => 'san1'},
-      { :zh_s => '叁', :zh_t => '參', :num => 3, :pyn => 'san1'},
-      { :zh_s => '四', :zh_t => '四', :num => 4, :pyn => 'si4'},
-      { :zh_s => '䦉', :zh_t => '䦉', :num => 4, :pyn => 'si4'},
-      { :zh_s => '肆', :zh_t => '肆', :num => 4, :pyn => 'si4'},
-      { :zh_s => '五', :zh_t => '五', :num => 5, :pyn => 'wu3'},
-      { :zh_s => '伍', :zh_t => '伍', :num => 5, :pyn => 'wu3'},
-      { :zh_s => '六', :zh_t => '六', :num => 6, :pyn => 'liu4'},
-      { :zh_s => '陆', :zh_t => '陸', :num => 6, :pyn => 'liu4'},
-      { :zh_s => '七', :zh_t => '七', :num => 7, :pyn => 'qi1'},
-      { :zh_s => '柒', :zh_t => '柒', :num => 7, :pyn => 'qi1'},
-      { :zh_s => '八', :zh_t => '八', :num => 8, :pyn => 'ba1'},
-      { :zh_s => '捌', :zh_t => '捌', :num => 8, :pyn => 'ba1'},
-      { :zh_s => '九', :zh_t => '九', :num => 9, :pyn => 'jiu3'},
-      { :zh_s => '玖', :zh_t => '玖', :num => 9, :pyn => 'jiu3'},
-      { :zh_s => '十', :zh_t => '十', :num => 10, :pyn => 'shi2'},
-      { :zh_s => '拾', :zh_t => '拾', :num => 10, :pyn => 'shi2'},
-      { :zh_s => '廿', :zh_t => '廿', :num => 20, :pyn => ' nian4'},
-      { :zh_s => '百', :zh_t => '百', :num => 100, :pyn => 'bai2'},
-      { :zh_s => '佰', :zh_t => '佰', :num => 100, :pyn => 'bai2'},
-      { :zh_s => '千', :zh_t => '千', :num => 1000, :pyn => 'qian2'},
-      { :zh_s => '仟', :zh_t => '仟', :num => 1000, :pyn => 'qian2'},
-      { :zh_s => '万', :zh_t => '萬', :num => 10000, :pyn => 'wan4'},
-      { :zh_s => '亿', :zh_t => '億', :num => 100000000, :pyn => 'yi4'},
+      { :zhs => '零', :zht => '零', :num => 0, :pyn => 'ling2'},
+      { :zhs => '〇', :zht => '〇', :num => 0, :pyn => 'ling2'},
+      { :zhs => '一', :zht => '一', :num => 1, :pyn => 'yi1'},
+      { :zhs => '壹', :zht => '壹', :num => 1, :pyn => 'yi1'},
+      { :zhs => '幺', :zht => '幺', :num => 1, :pyn => 'yao1'},
+      { :zhs => '二', :zht => '二', :num => 2, :pyn => 'er4'},
+      { :zhs => '两', :zht => '兩', :num => 2, :pyn => 'liang3'},
+      { :zhs => '贰', :zht => '貳', :num => 2, :pyn => 'er4'},
+      { :zhs => '三', :zht => '三', :num => 3, :pyn => 'san1'},
+      { :zhs => '弎', :zht => '弎', :num => 3, :pyn => 'san1'},
+      { :zhs => '叁', :zht => '參', :num => 3, :pyn => 'san1'},
+      { :zhs => '四', :zht => '四', :num => 4, :pyn => 'si4'},
+      { :zhs => '䦉', :zht => '䦉', :num => 4, :pyn => 'si4'},
+      { :zhs => '肆', :zht => '肆', :num => 4, :pyn => 'si4'},
+      { :zhs => '五', :zht => '五', :num => 5, :pyn => 'wu3'},
+      { :zhs => '伍', :zht => '伍', :num => 5, :pyn => 'wu3'},
+      { :zhs => '六', :zht => '六', :num => 6, :pyn => 'liu4'},
+      { :zhs => '陆', :zht => '陸', :num => 6, :pyn => 'liu4'},
+      { :zhs => '七', :zht => '七', :num => 7, :pyn => 'qi1'},
+      { :zhs => '柒', :zht => '柒', :num => 7, :pyn => 'qi1'},
+      { :zhs => '八', :zht => '八', :num => 8, :pyn => 'ba1'},
+      { :zhs => '捌', :zht => '捌', :num => 8, :pyn => 'ba1'},
+      { :zhs => '九', :zht => '九', :num => 9, :pyn => 'jiu3'},
+      { :zhs => '玖', :zht => '玖', :num => 9, :pyn => 'jiu3'},
+      { :zhs => '十', :zht => '十', :num => 10, :pyn => 'shi2'},
+      { :zhs => '拾', :zht => '拾', :num => 10, :pyn => 'shi2'},
+      { :zhs => '廿', :zht => '廿', :num => 20, :pyn => ' nian4'},
+      { :zhs => '百', :zht => '百', :num => 100, :pyn => 'bai2'},
+      { :zhs => '佰', :zht => '佰', :num => 100, :pyn => 'bai2'},
+      { :zhs => '千', :zht => '千', :num => 1000, :pyn => 'qian2'},
+      { :zhs => '仟', :zht => '仟', :num => 1000, :pyn => 'qian2'},
+      { :zhs => '万', :zht => '萬', :num => 10000, :pyn => 'wan4'},
+      { :zhs => '亿', :zht => '億', :num => 100000000, :pyn => 'yi4'},
     ]
 
-    def is_number? word
+    def number? word
       #垓	秭	穰	溝	澗	正	載 --> beyond 100,000,000!
       "#{word}".gsub(/([\d]|[一二三四五六七八九十百千萬万億亿]){2,}/,'') == ''
     end
 
+    def zh_number_to_number(zh_number)
+      zh_number = zh_number.to_s
+      numbers = convert_date(zh_number)
+
+      #if it's a year, or an oddly formatted number
+      return numbers.join('').to_i if zh_number[/[#{NUMBER_MULTIPLES}]/u].nil?
+
+      convert_numbers numbers
+    end
+
+    #these should also be able to convert numbers to chinese numbers
+    def number_to_zhs type, number
+      convert_number_to :zhs, type.to_sym, number
+    end
+    def number_to_zht type, number
+      convert_number_to :zht, type.to_sym, number
+    end
+
+    def number_to_pyn number, type = 'zh_s'
+      convert_number_to :pyn, type.to_sym, number, '-'
+    end
+
+    private
     def convert_date(zh)
       #if it's a year, or an oddly formatted number
       zh_numbers = ZhongwenTools::String.chars zh
@@ -64,7 +87,7 @@ module ZhongwenTools
     end
 
     def convert(number)
-      NUMBERS_TABLE.find{|x|  x[:zh_s] == number || x[:zh_t] == number  || x[:num].to_s == number}
+      NUMBERS_TABLE.find{|x|  x[:zhs] == number || x[:zht] == number  || x[:num].to_s == number}
     end
 
     def convert_numbers(numbers)
@@ -99,36 +122,18 @@ module ZhongwenTools
       is_number_multiplier?(curr_num) ? number * curr_num : number + curr_num
     end
 
-    def convert_chinese_numbers_to_numbers(zh_number)
-      zh_number = zh_number.to_s
-      numbers = convert_date(zh_number)
-
-      #if it's a year, or an oddly formatted number
-      return numbers.join('').to_i if zh_number[/[#{NUMBER_MULTIPLES}]/u].nil?
-
-      convert_numbers numbers
-    end
 
     def is_number_multiplier?(number)
       [10,100,1000,10000,100000000].include? number
     end
 
-    #these should also be able to convert numbers to chinese numbers
-    def convert_number_to_simplified type, number
-      convert_number_to :zh_s, type.to_sym, number
-    end
-    def convert_number_to_traditional type, number
-      convert_number_to :zh_t, type.to_sym, number
-    end
-
-    def convert_number_to_pyn number, type = 'zh_s'
-      convert_number_to :pyn, type.to_sym, number, '-'
-    end
 
 
     def check_wan(wan, i)
       wan ||= 0
       wan += 1 if (i + 1) % 5 == 0
+
+      wan
     end
 
     def convert_from_zh number, to
@@ -170,7 +175,7 @@ module ZhongwenTools
     end
 
     def convert_number_to(to, from, number, separator = '')
-      return number unless [:zh_t, :zh_s, :num, :pyn].include? to
+      return number unless [:zht, :zhs, :num, :pyn].include? to
 
       if from == :num
         converted_number = convert_from_num(number, to)

data/test/test_helper.rb

@@ -12,3 +12,11 @@ rescue LoadError
 end
 
 require 'test/unit'
+
+if RUBY_VERSION < '1.9'
+  class Test::Unit::TestCase
+    def refute(statement, message = '')
+      assert !statement, message
+    end
+  end
+end

data/test/test_integer.rb

@@ -0,0 +1,31 @@
+#encoding: utf-8
+$:.unshift File.join(File.dirname(__FILE__),'..','lib')
+
+require './test/test_helper'
+require 'zhongwen_tools/integer'
+
+class Integer
+  include ZhongwenTools::Integer
+end
+
+class TestInteger < Test::Unit::TestCase
+  def test_zh
+    assert_equal 122.to_zh, '一百二十二'
+    assert_equal 12.to_zh, '十二'
+    assert_equal 12000.to_zht, '一萬二千'
+    assert_equal 12000.to_zhs, '一万二千'
+    refute 12000.to_zh == 12000.to_zht
+
+
+    assert_equal '十二', ZhongwenTools::Integer.to_zhs(12)
+    assert_equal '一萬二千', ZhongwenTools::Integer.to_zht(12000)
+    assert_equal '一万二千', ZhongwenTools::Integer.to_zhs(12000)
+    refute  ZhongwenTools::Integer.to_zhs(12000) == ZhongwenTools::Integer.to_zht(12000)
+  end
+
+  def test_pinyin
+    assert_equal 12.to_pyn, 'shi2-er4'
+    assert_equal 'shi2-er4', ZhongwenTools::Integer.to_pyn(12)
+  end
+end
+

data/test/test_numbers.rb

@@ -5,37 +5,35 @@ require './test/test_helper'
 require 'zhongwen_tools/string'
 require 'zhongwen_tools/numbers'
 
-class TestCJKTools < Test::Unit::TestCase
+class TestNumbers < Test::Unit::TestCase
   include ZhongwenTools::Numbers
   def test_convert_to_numbers
-    #skip 
+    #skip
     #your function sucks dick man
-      @numbers.each do |num|
-        number = convert_chinese_numbers_to_numbers num[:zh]
-        binding.pry if num[:en] != number
-        assert_equal num[:en], number
-      end
+    @numbers.each do |num|
+      number = zh_number_to_number num[:zh]
+      binding.pry if num[:en] != number
+      assert_equal num[:en], number
+    end
   end
 
   def test_convert_to_traditional_number
     zhs = @numbers[0][:zh]
-    zht = convert_number_to_traditional :zh_s, zhs
+    zht = number_to_zht :zht, zhs
 
     assert_equal '一萬兩千七', zht
   end
 
   def test_convert_to_simplified_from_number
-    #skip
-   num = @numbers[0][:en]
-   zht = convert_number_to_traditional :num, num
+    num = @numbers[0][:en]
+    zhs = number_to_zhs :num, num
 
-#adds garbage!!
-    assert_equal '一萬二千七', zht
+    assert_equal '一万二千七', zhs
   end
 
   def test_convert_number_to_pyn
     num = '一百三十六'
-    pyn = self.convert_number_to_pyn num
+    pyn = self.number_to_pyn num
 
     assert_equal 'yi1-bai2-san1-shi2-liu4', pyn
   end

data/test/test_string.rb

@@ -7,14 +7,6 @@ class String
   include ZhongwenTools::String
 end
 
-if RUBY_VERSION < '1.9'
-  class Test::Unit::TestCase
-    def refute(statement, message = '')
-      assert !statement, message
-    end
-  end
-end
-
 class TestString < Test::Unit::TestCase
 
   def test_size

data/zhongwen_tools.gemspec

@@ -4,7 +4,7 @@ $:.push File.expand_path("../lib", __FILE__)
 Gem::Specification.new do |s|
   s.name        = "zhongwen_tools"
   s.license     = "MIT"
-  s.version     = "0.0.6"
+  s.version     = "0.1.0"
   s.authors     = ["Steven Daniels"]
   s.email       = ["steven@tastymantou.com"]
   s.homepage    = "https://github.com/stevendaniels/zhongwen_tools"

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: zhongwen_tools
 version: !ruby/object:Gem::Version
-  version: 0.0.6
+  version: 0.1.0
 platform: ruby
 authors:
 - Steven Daniels
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-01-18 00:00:00.000000000 Z
+date: 2014-01-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake
@@ -80,6 +80,7 @@ files:
 - README.md
 - Rakefile
 - lib/zhongwen_tools.rb
+- lib/zhongwen_tools/integer.rb
 - lib/zhongwen_tools/numbers.rb
 - lib/zhongwen_tools/string.rb
 - lib/zhongwen_tools/string/fullwidth.rb
@@ -87,6 +88,7 @@ files:
 - lib/zhongwen_tools/string/ruby19.rb
 - test/test_conversion.rb
 - test/test_helper.rb
+- test/test_integer.rb
 - test/test_numbers.rb
 - test/test_romanization.rb
 - test/test_string.rb
@@ -118,6 +120,7 @@ summary: Zhongwen Tools provide romanization conversions and helper methods for
 test_files:
 - test/test_conversion.rb
 - test/test_helper.rb
+- test/test_integer.rb
 - test/test_numbers.rb
 - test/test_romanization.rb
 - test/test_string.rb