zhongwen_tools 0.9.0 → 0.11.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: c46a1639e99601e0a9f9cb5e6961f148cf030758
-  data.tar.gz: e4dc3033e220ecd882915dadeb7e9c780a0cbe65
+  metadata.gz: 0cbc9e019d51f46d9c6b609bbc711c4f3b7dfde5
+  data.tar.gz: 06a3a6ff340d5104e205fadff0989301ddc521bd
 SHA512:
-  metadata.gz: 427f2bc4b43ea3734995aa2d4c0523244882e300457500728164eef7f297441bd400569d272e59d7b3d7218777b39f46e67ca430765f4d272b5432c186dd09d2
-  data.tar.gz: 1c6065127ee0fda328044d412b545f4b85cb53660df3070599c081f197c41f581054a69cb4d64d05377e18cf81c15808d31ec1804f9ef0951fc24350bcb84374
+  metadata.gz: 3599feaa1f681e31d8b7ed58c24c980b5317352f95d32a90a6b27f4ffdc62592c1048efcbe5dde3fbb60d1ea05bb0bb7053916833fd60ba58434599bc9fc7d18
+  data.tar.gz: 819c5c711845e7f2b9f9f39af4042c0c6f80ff64b7b86d3f3b486c7bc5418c669d9f3ee73e7988f849e5201d6ddd8b18f615927ed0511befb57dd2c21a5e9e52

data/README.md

@@ -189,7 +189,7 @@ scripts. It **does not convert Chinese characters to pinyin** (see ZhongwenTools
 
     str.to_wg
     #=> 'ni3 hao3'    #Wade-Giles
-    
+
     str.to_bpmf
     #=> 'ㄋㄧ3 ㄏㄠ3' #Zhuyin Fuhao, a.k.a. Bopomofo
 

data/Rakefile

@@ -8,46 +8,5 @@ Rake::TestTask.new do |t|
   t.test_files = FileList['test/test*.rb']
 end
 
-
-
 desc "Run tests"
 task :default => :test
-
-
-def setup_18
-  commands =  [
-    'cp Gemfile Gemfile.bak',
-    'mv .ruby-version-1.8.7 .ruby-version',
-    'cp Gemfile.1.8.7 Gemfile',
-    'bundle install --local',
-  ]
- 
-  commands.each{ |c| `#{ c }` }
-end
-
-def teardown_18
-  commands =  [
-    'mv .ruby-version .ruby-version-1.8.7 ',
-    'cp Gemfile.bak Gemfile && rm Gemfile.bak',
-    'bundle install --local'
-  ]
-
-  commands.each{ |c| 
-    `#{ c }` 
-  }
-end
-
-
-namespace :ruby_18 do
-  desc "Switch to 1.8.7"
-  task :setup do
-    setup_18 if File.exist?('.ruby-version-1.8.7')
-    puts 'Using Ruby 1.8.7'
-  end
-
-  desc "Teardown to 1.8.7"
-  task :teardown do
-    teardown_18 unless File.exist?('.ruby-version-1.8.7')
-    puts 'Using Ruby 2.1.0'
-  end
-end

data/lib/zhongwen_tools.rb

@@ -1,9 +1,7 @@
 # encoding: utf-8
-require File.expand_path("../zhongwen_tools/string", __FILE__)
-require File.expand_path("../zhongwen_tools/numbers", __FILE__)
-require File.expand_path("../zhongwen_tools/version", __FILE__)
-#require File.expand_path("../zhongwen_tools/romanization", __FILE__)
-#require File.expand_path("../zhongwen_tools/conversion", __FILE__)
+require 'zhongwen_tools/string'
+require 'zhongwen_tools/numbers'
+require 'zhongwen_tools/version'
 
 module ZhongwenTools
 end

data/lib/zhongwen_tools/conversion.rb

@@ -1,4 +1,5 @@
 # encoding: utf-8
+require 'zhongwen_tools/string'
 
 module ZhongwenTools
 
@@ -68,31 +69,33 @@ module ZhongwenTools
 
     def convert(type, str)
       load_table if ZH_CONVERSION_TABLE.length == 0
-      arr = ZH_TYPES[type] || ZH_TYPES[:zht]
+      types = ZH_TYPES[type] || ZH_TYPES[:zht]
 
       begin
-        str0 = str.dup
-        str1 = str.dup
         str_len = ZhongwenTools::String.size(str)
-        n = ( str_len < 6)? str_len : 6
-        ZH_CONVERSION_TABLE.last(n).each do |group|
-          arr.each do |t|
-            group[t].each do |key , value|
-              while !! q = str0.index( key )
-                str0[/#{key}/] = "#" * value.size
-                str1[/#{key}/] = value
-              end
-            end
-          end
-        end
-
-        str1
+        n = (str_len < 6)? str_len : 6
+        convert_zhongwen(str.dup, str.dup, types, n)
 
       rescue
         "[#{$!}]"
       end
     end
+
+    def convert_zhongwen(str0, str1, types, n)
+      ZH_CONVERSION_TABLE.last(n).each do |group|
+        types.each do |t|
+          group[t].each do |key , value|
+            until str0.index(key).nil?
+              str0[/#{key}/] = "#" * value.size
+              str1[/#{key}/] = value
+            end
+          end
+        end
+      end
+
+      str1
+    end
   end
 end
 
-require File.expand_path("../conversion/string", __FILE__)
+require 'zhongwen_tools/conversion/string'

data/lib/zhongwen_tools/integer.rb

@@ -1,5 +1,5 @@
 # encoding: utf-8
-require File.expand_path("../numbers", __FILE__)
+require 'zhongwen_tools/numbers'
 
 module ZhongwenTools
   module Integer

data/lib/zhongwen_tools/numbers.rb

@@ -1,5 +1,6 @@
 # encoding: utf-8
-require File.expand_path("../regex", __FILE__)
+require 'zhongwen_tools/regex'
+
 # TODO: more testing
 module ZhongwenTools
   module Numbers

data/lib/zhongwen_tools/regex.rb

@@ -4,13 +4,13 @@ module ZhongwenTools
     extend self
 
     def pyn
-      # NOTE: might not need / want the space on the end.
-      /(#{pyn_regexes.values.join('|')})([1-5])?([\s\-]+)?/
+      /(#{pyn_regexes.values.join('|')}|r)([1-5])?([\s\-]+)?/
     end
 
     def py
-      # NOTE: might not need / want the space on the end.
-      /(#{pyn_regexes.map{|k,v| v.to_s[7..-2].gsub_with_hash(/[aeiouv]/,py_tones)}.join('|')}(\s\-))/
+      # FIXME: need to detect Ālābó
+      # ([ĀÁǍÀA][io]?|[io]?|[][āáǎàaēéěèeūúǔùu]?o?|[ĒÉĚÈE]i?|[]i?|[ŌÓǑÒO]u?|[]u?|u[āáǎàaēoēéěèe]?i?|[]e?)(n?g?r?)){1,}
+      /(#{pyn_regexes.map{|k,v| v.to_s[7..-2].gsub_with_hash(/[aeiouv]/,py_tones)}.join('|')}([\s\-])?)/
     end
 
     def fullwidth
@@ -85,7 +85,6 @@ module ZhongwenTools
         'o' => '[ōóǒòo]',
         'u' => '[ūúǔùu]',
         'v' => '[ǖǘǚǜü]'
-        #([ĀÁǍÀA][io]?|[io]?|[][āáǎàaēéěèeūúǔùu]?o?|[ĒÉĚÈE]i?|[]i?|[ŌÓǑÒO]u?|[]u?|u[āáǎàaēoēéěèe]?i?|[]e?)(n?g?r?)){1,}
       }
     end
   end

data/lib/zhongwen_tools/romanization.rb

@@ -1,58 +1,52 @@
 # encoding: utf-8
-require File.expand_path("../romanization/conversion_table", __FILE__)
-require File.expand_path("../romanization/detect", __FILE__)
-require File.expand_path("../romanization/string", __FILE__)
-require File.expand_path("../romanization/pyn_to_py", __FILE__)
-
+require 'zhongwen_tools/string'
+require 'zhongwen_tools/romanization/conversion_table'
+require 'zhongwen_tools/romanization/string'
+require 'zhongwen_tools/romanization/pyn_to_py'
+
+# TODO: follow tone conventions for different systems.
+#       IPA	mä˥˥	mä˧˥	mä˨˩˦	mä˥˩	mä
+#       Pinyin	mā	má	mǎ	mà	ma
+#       Tongyong Pinyin	ma	má	mǎ	mà	må # this will be difficult.
+#       Wade–Giles	ma¹	ma²	ma³	ma⁴	ma⁰
+#       Zhuyin	ㄇㄚ	ㄇㄚˊ	ㄇㄚˇ	ㄇㄚˋ	•ㄇㄚ
 module ZhongwenTools
   module Romanization
     extend self
 
-    def to_pinyin(*args)
-      str, from = _romanization_options(args)
-
-      _convert_romanization str, :py, from
-    end
-
-    def to_bopomofo *args
-      str, from = _romanization_options(args)
-
-      _convert_romanization str, :zyfh, from
-    end
-
-    def to_yale(*args)
-      str, from = _romanization_options(args)
-      _convert_romanization str, :yale, from
-    end
-
-    def to_wade_giles(*args)
-      str, from = _romanization_options(args)
-      _convert_romanization str, :wg, from
-    end
-
-    def to_typy(*args)
-      str, from = _romanization_options(args)
-      _convert_romanization str, :typy, from
-    end
-
-    def to_pyn(*args)
-      # needs to guess what the romanization type is.
-      str, from = _romanization_options(args)
-      _convert_romanization str, :pyn, from
+    %w(pinyin py pyn bopomofo bpmf zhuyin zyfh zhyfh zhuyin_fuhao yale wade_giles wg typy tongyong mps2).each do |type|
+      define_method("to_#{type}") do |*args|
+        str, from = _romanization_options(args)
+        _convert_romanization str, _set_type(type.to_sym), _set_type(from)
+      end
     end
 
     private
 
+    # Private: Provides romanization options for romanization methods. If no :from argument is given, then
+    #          the method will try to guess the romanization. This can sometimes provide sub-optimal
+    #          romanization suggestions. See lib/zhongwen_tools/romanization/detect.rb#romanization? for details.
+    #
+    # args - an Array of arguments. If the Object is a String, then the first argument should be the :from option.
+    #        Otherwise, the first argument is a String and the second argument is the :from option.
+    #
+    # Examples:
+    #
+    #
+    #   _romanization_options('hao3', :pyn) #=> 'hao3' :pyn
+    #   _romanization_options('hao3') #=> 'hao3', :pyn
+    #
+    # Returns an Array. The first item is a String to be converted. The second item is a Symbol for the :from option.
     def _romanization_options(args)
       if self.class.to_s != 'String'
         str = args[0]
-        from = (args[1] || :pyn).to_sym
+        from = args[1] || str.romanization? || :pyn
       else
         str = self
-        from = (args[0] || :pyn).to_sym
+        from = args[0] || str.romanization? || :pyn
       end
 
-      [str, from]
+      [str, from.to_sym]
     end
 
     #  Private: Replaces numbered pinyin with actual pinyin. Pinyin separated with hyphens are combined as one word.
@@ -70,8 +64,12 @@ module ZhongwenTools
       # TODO: move regex to ZhongwenTools::Regex
       regex = /(([BPMFDTNLGKHZCSRJQXWYbpmfdtnlgkhzcsrjqxwy]?[h]?)(A[io]?|a[io]?|i[aeu]?o?|Ei?|ei?|Ou?|ou?|u[aoe]?i?|ve?)?(n?g?)(r?)([1-5])(\-+)?)/
 
-      # doing the substitution in a block is ~8x faster than using scan and each.
-      # Explanation: if it's pinyin without vowels, e.g. m, ng, then convert, otherwise, check if it needs an apostrophe (http://www.pinyin.info/romanization/hanyu/apostrophes.html). If it does, add it and then convert. Otherwise, just convert. Oh, and if double hyphens are used, replace them with one hyphen. And finally, correct those apostrophes at the very end.
+      # Using gsub is ~8x faster than using scan and each.
+      # Explanation: if it's pinyin without vowels, e.g. m, ng, then convert,
+      #              otherwise, check if it needs an apostrophe (http://www.pinyin.info/romanization/hanyu/apostrophes.html).
+      #              If it does, add it and then convert. Otherwise, just convert.
+      #              Oh, and if double hyphens are used, replace them with one hyphen.
+      #              And finally, correct those apostrophes at the very end.
       str.gsub(regex) do
         ($3.nil? ? "#{PYN_PY[$1]}" : ($2 == '' && ['a','e','o'].include?($3[0,1]))? "'#{PYN_PY["#{$3}#{$6}"]}#{$4}#{$5}" : "#{$2}#{PYN_PY["#{$3}#{$6}"]}#{$4}#{$5}") + (($7.to_s.length > 1) ? '-' : '')
       end.gsub("-'","-").sub(/^'/,'')
@@ -88,10 +86,10 @@ module ZhongwenTools
       tokens.collect do |t|
         search = t.gsub(/[1-5].*/,'')
 
-        if from.nil? || convert_from.nil?
-          replace = (_replacement(t) || {}).fetch(convert_to){search}
+        if from.nil?
+          replace = (_replacement(t) || {}).fetch(to){search}
         else
-          replace = (_replacement(t, convert_from) || {}).fetch(convert_to){search}
+          replace = (_replacement(t, from) || {}).fetch(to){search}
         end
 
         replace = _fix_capitalization(str, t, replace)
@@ -121,32 +119,28 @@ module ZhongwenTools
     def _convert_romanization str, to, from
       return str if to == from
 
-      if to == :py
-        if from == :pyn
+      result =
+        if to == :py
+          raise NotImplementedError, 'method not implemented' if from != :pyn
+          # convert to pyn first.
+          # TODO: test :zyfh -> py
+          # str = _to_romanization str, to, :pyn if from != :pyn
           _to_pinyin str
+
+        elsif to == :pyn
+          if from == :py
+            _convert_pinyin_to_pyn(str)
+          else
+             _to_romanization str, :pyn, from
+          end
         else
-          raise NotImplementedError, 'method not implemented'
-          #convert to pyn first.
-        end
-      elsif to == :zyfh
-        if from == :py
-          #need to convert pinyin to pyn
-          raise NotImplementedError, 'method not implemented'
-        end
-        _to_romanization(str, to, from).gsub('-','')
-      elsif to == :pyn
-        if from == :py
-         _convert_pinyin_to_pyn(str)
-        else
-          raise NotImplementedError, 'method not implemented'
-        end
-      else
-        if from == :pyn
+           str = _to_romanization str, to, :pyn if from != :pyn
           _to_romanization str, to, from
-        else
-          raise NotImplementedError, 'method not implemented'
         end
-      end
+
+      # TODO: check to see if wade giles, yale etc. can have hyphens.
+      result = result.gsub('-','') if to == :zyfh
+      result
     end
 
     def _convert_pinyin_to_pyn(pinyin)
@@ -155,75 +149,51 @@ module ZhongwenTools
       words =  pinyin.split(' ')
 
       pyn = words.map do |word|
-        pys = word.split(/['\-]/).flatten.map{|x| x.scan(PY_REGEX).map{|x| (x - [nil])[0]}}.flatten
-        current_pyn = word
+        pys = word.split(/['\-]/).flatten.map{|x| x.scan(Regex.py).map{|x| (x - [nil])[0]}}.flatten
+        _current_pyn(word, pys)
+      end
+
+      pyn.join(' ')
+    end
 
-        pys.each do |py|
-          #take the longest pinyin match.
-          match = ZhongwenTools::Romanization::PYN_PY.values.select do |x|
-            py.include? x
-          end.sort{|x,y| x.length <=> y.length}[-1]
+    def _current_pyn(pyn, pinyin_arr)
+      pinyin_arr.each do |pinyin|
+        pyn = pyn.sub(pinyin, pinyin_replacement(pinyin))
+      end
 
-          # Edge case.. en/eng pyn -> py conversion is one way only.
-          match = match[/(ē|é|ě|è)n?g?/].nil? ? match : match.chars[0]
+      pyn.gsub("'",'')
+    end
 
-          replace = ZhongwenTools::Romanization::PYN_PY.find{|k,v| k if v == match}[0]
-          p = py.gsub(match, replace).gsub(/([^\d ]*)(\d)([^\d ]*)/){$1 + $3 + $2}
+    def pinyin_replacement(py)
+      #take the longest pinyin match.
+      match = PYN_PY.values.select do |x|
+        py.include? x
+      end.sort{|x,y| x.length <=> y.length}[-1]
 
-          current_pyn = current_pyn.sub(py, p)
-        end
+      # Edge case.. en/eng pyn -> py conversion is one way only.
+      match = match[/(ē|é|ě|è)n?g?/].nil? ? match : match.chars[0]
 
-        current_pyn.gsub("'",'')
-      end
+      replace = PYN_PY.find{|k,v| k if v == match}[0]
 
-      pyn.join(' ')
+      py.gsub(match, replace).gsub(/([^\d ]*)(\d)([^\d ]*)/){$1 + $3 + $2}
     end
 
 
     def _set_type(type)
       type = type.to_s.downcase.to_sym
-      case type
-      when :zhuyinfuhao
-        :zyfh
-      when :bopomofo
-        :zyfh
-      when :bpmf
-        type = :zyfh
-      when :zhyfh
-        type = :zyfh
-      when :zyfh
+      return type if [:zyfh, :wg, :typy, :py, :mps2, :yale, :pyn].include? type
+
+      if [:zhuyinfuhao, :zhuyin, :zhuyin_fuhao, :bopomofo, :bpmf, :zhyfh].include? type
         :zyfh
-      when 'wade-giles'.to_sym
-        type = :wg
-      when :yale
-        :yale
-      when :tongyong
-        type = :typy
-      when :wg
-        type = :wg
-      when :typy
+      elsif [:wade_giles, 'wade-giles'.to_sym].include? type
+        :wg
+      elsif [:tongyong, :typy, :ty].include? type
         :typy
-      when :ty
-        type = :typy
-      when :pyn
+      elsif type == :pinyin
         :py
-      when :pinyin
-        type = :py
-      when :py
-        type = :py
-      when :msp2
-        :msp2
-      else
-        nil
       end
     end
-
-    alias_method :to_py, :to_pinyin
-    alias_method :to_zhyfh, :to_bopomofo
-    alias_method :to_zhuyin, :to_bopomofo
-    alias_method :to_zhuyin_fuhao, :to_bopomofo
-    alias_method :to_bpmf, :to_bopomofo
-    alias_method :to_wg, :to_wade_giles
-    alias_method :to_tongyong, :to_typy
   end
 end
+
+require 'zhongwen_tools/romanization/detect'