when_exe 0.4.2 → 0.5.0

data/lib/when_exe/locales/akt.rb

@@ -147,6 +147,35 @@ module When
     # @private
     IAST_K_keys = transliteration_keys(IAST_K)
 
+    # @private
+    NumAlter1 = [['〇','零'], [/[壱壹]/,'一'], [/[弐貳]/,'二'], [/[弎参參]/,'三'], ['肆','四'],
+                 ['伍','五'], ['陸','六'], ['柒','七'], ['捌','八'], ['玖','九'], ['拾','十'],
+                 ['廿', '二十'], [/[卅丗]/, '三十'], ['卌', '四十'],
+                 ['佰', '百'], ['仟', '千'], ['萬', '万'], ['秭', '𥝱']]
+
+    # @private
+    NumAlter2 = [['零〇','０'], ['一壱壹','１'], ['二弐貳','２'], ['三弎参參','３'], ['四肆','４'],
+                 ['五伍','５'], ['六陸','６'], ['七柒','７'], ['八捌','８'], ['九玖','９']]
+
+    # @private
+    Numbers  = %w(零 一 二 三 四 五 六 七 八 九 十 百 千 万 億 兆 京 垓 𥝱
+                  穣 溝 澗 正 載 極 恒河沙 阿僧祇 那由他 不可思議 無量大数)
+
+    # @private
+    NumRExp1 = /#{Numbers[10..12].reverse.map {|num| "(?:(.*)#{num})"}.join('?')}?(.*)/
+
+    # public
+    NumRExp3 = /([#{Numbers[0..12].join('')}廿卅丗卌]+)/
+
+    # @private
+    NumRExp4 = /#{Numbers[13..-1].reverse.map {|num| "(?:(.*)#{num})"}.join('?')}?(.*)/
+
+    # @private
+    NumMap   = Hash[*(Numbers[0..9].zip((0..9).to_a)).flatten]
+
+    # @private
+    DigitMap = [1,1,1,0]
+
     #
     # Convert AKT string to katakana scripts
     #
@@ -173,5 +202,39 @@ module When
         gsub(/([rm])y/, '\1uy'),
         locale)
     end
+ 
+    #
+    # Convert kanji scripts to numeric
+    #
+    def self.k2a_digits(figures, force_alter=false)
+      if force_alter
+        figures = figures.dup
+        NumAlter1.each {|alter| figures.gsub!(*alter)}
+      end
+      sum = 0
+      NumRExp4 =~ figures
+      $~.to_a[1..-1].each do |wide_match|
+        if NumRExp1 =~ wide_match
+          $~.to_a[1..-1].each_with_index do |match,digit|
+            sum *= 10
+            sum += (NumMap[match] || DigitMap[digit]) if match
+          end
+        else
+          sum *= 10000
+        end
+      end
+      sum
+    end
+
+    #
+    # Convert zenkaku figures to numeric
+    #
+    def self.z2h_digits(figures, force_alter=false)
+      if force_alter
+        figures = figures.dup
+        NumAlter2.each {|alter| figures.tr!(*alter)}
+      end
+      figures.tr('０１２３４５６７８９', '0123456789').to_i
+    end
   end
 end

data/lib/when_exe/locales/encoding_conversion.rb

@@ -1,134 +1,134 @@
-# -*- coding: utf-8 -*-
-=begin
-  Copyright (C) 2014 Takashi SUGA
-
-  You may use and/or modify this file according to the license described in the LICENSE.txt file included in this archive.
-=end
-
-module When
-
-  #
-  # エンコーディングの変換
-  #
-  module EncodingConversion
-
-    class << self
-      if String.method_defined?(:encode)
-        #
-        # 内部エンコーディング文字列化
-        #
-        # @param [Object] source もとにする Object
-        #
-        # @return [Object] 内部エンコーディング化した Object
-        #
-        # @private
-        def to_internal_encoding(source)
-          default_internal = Encoding.default_internal||'UTF-8'
-          case source
-          when Locale ;  source.to_internal_encoding
-          when String ; case source.encoding
-                        when default_internal     ; source
-                        when Encoding::ASCII_8BIT ; source.dup.force_encoding(default_internal)
-                        else                      ; source.encode(default_internal)
-                        end
-          when Regexp ; case source.encoding
-                        when default_internal     ; source
-                        when Encoding::ASCII_8BIT ; Regexp.new(source.dup.force_encoding(default_internal))
-                        else                      ; Regexp.new(source.to_s.encode(default_internal))
-                        end
-          when Array  ;  source.map {|e| to_internal_encoding(e)}
-          when Hash
-            hash = {}
-            source.each_pair do |key, value|
-              hash[to_internal_encoding(key)] = to_internal_encoding(value)
-            end
-            hash
-          else        ;  source.respond_to?(:to_internal_encoding) ? source.to_internal_encoding : source
-          end
-        end
-
-        #
-        # 外部エンコーディング文字列化
-        #
-        # @param [Object] source もとにする Object
-        #
-        # @return [Object] 外部エンコーディング化した Object
-        #
-        # @private
-        def to_external_encoding(source)
-          default_external = Encoding.default_external
-          case source
-          when Locale ;  source.to_external_encoding
-          when String ; case source.encoding
-                        when default_external     ; source
-                        when Encoding::ASCII_8BIT ; source.dup.force_encoding(default_external)
-                        else                      ; source.encode(default_external)
-                        end
-          when Regexp ; case source.encoding
-                        when default_external     ; source
-                        when Encoding::ASCII_8BIT ; Regexp.new(source.dup.force_encoding(default_external))
-                        else                      ; Regexp.new(source.to_s.encode(default_external))
-                        end
-          when Array  ;  source.map {|e| to_external_encoding(e)}
-          when Hash
-            hash = {}
-            source.each_pair do |key, value|
-              hash[to_external_encoding(key)] = to_external_encoding(value)
-            end
-            hash
-          else        ;  source.respond_to?(:to_external_encoding) ? source.to_external_encoding : source
-          end
-        end
-
-      else
-        # 内部エンコーディング文字列化(ダミー)
-        #
-        # @param [String] string もとにする String または M17n
-        #
-        # @return [String] そのまま返す
-        #
-        def to_internal_encoding(string)
-          string
-        end
-        #
-        # 外部エンコーディング文字列化(ダミー)
-        #
-        # @param [String] string もとにする String または M17n
-        #
-        # @return [String] そのまま返す
-        #
-        def to_external_encoding(string)
-          string
-        end
-      end
-    end
-
-    # 
-    # 内部エンコーディング文字列化
-    #
-    def to_internal_encoding
-      EncodingConversion.to_internal_encoding(self)
-    end
-
-    # 
-    # 外部エンコーディング文字列化
-    #
-    def to_external_encoding
-      EncodingConversion.to_external_encoding(self)
-    end
-
-    # 
-    # 内部エンコーディング文字列化(単項)
-    #
-    def +@
-      to_internal_encoding
-    end
-
-    # 
-    # 外部エンコーディング文字列化(単項)
-    #
-    def -@
-      to_external_encoding
-    end
-  end
-end
+# -*- coding: utf-8 -*-
+=begin
+  Copyright (C) 2014 Takashi SUGA
+
+  You may use and/or modify this file according to the license described in the LICENSE.txt file included in this archive.
+=end
+
+module When
+
+  #
+  # エンコーディングの変換
+  #
+  module EncodingConversion
+
+    class << self
+      if String.method_defined?(:encode)
+        #
+        # 内部エンコーディング文字列化
+        #
+        # @param [Object] source もとにする Object
+        #
+        # @return [Object] 内部エンコーディング化した Object
+        #
+        # @private
+        def to_internal_encoding(source)
+          default_internal = Encoding.default_internal||'UTF-8'
+          case source
+          when Locale ;  source.to_internal_encoding
+          when String ; case source.encoding
+                        when default_internal     ; source
+                        when Encoding::ASCII_8BIT ; source.dup.force_encoding(default_internal)
+                        else                      ; source.encode(default_internal)
+                        end
+          when Regexp ; case source.encoding
+                        when default_internal     ; source
+                        when Encoding::ASCII_8BIT ; Regexp.new(source.dup.force_encoding(default_internal))
+                        else                      ; Regexp.new(source.to_s.encode(default_internal))
+                        end
+          when Array  ;  source.map {|e| to_internal_encoding(e)}
+          when Hash
+            hash = {}
+            source.each_pair do |key, value|
+              hash[to_internal_encoding(key)] = to_internal_encoding(value)
+            end
+            hash
+          else        ;  source.respond_to?(:to_internal_encoding) ? source.to_internal_encoding : source
+          end
+        end
+
+        #
+        # 外部エンコーディング文字列化
+        #
+        # @param [Object] source もとにする Object
+        #
+        # @return [Object] 外部エンコーディング化した Object
+        #
+        # @private
+        def to_external_encoding(source)
+          default_external = Encoding.default_external
+          case source
+          when Locale ;  source.to_external_encoding
+          when String ; case source.encoding
+                        when default_external     ; source
+                        when Encoding::ASCII_8BIT ; source.dup.force_encoding(default_external)
+                        else                      ; source.encode(default_external)
+                        end
+          when Regexp ; case source.encoding
+                        when default_external     ; source
+                        when Encoding::ASCII_8BIT ; Regexp.new(source.dup.force_encoding(default_external))
+                        else                      ; Regexp.new(source.to_s.encode(default_external))
+                        end
+          when Array  ;  source.map {|e| to_external_encoding(e)}
+          when Hash
+            hash = {}
+            source.each_pair do |key, value|
+              hash[to_external_encoding(key)] = to_external_encoding(value)
+            end
+            hash
+          else        ;  source.respond_to?(:to_external_encoding) ? source.to_external_encoding : source
+          end
+        end
+
+      else
+        # 内部エンコーディング文字列化(ダミー)
+        #
+        # @param [String] string もとにする String または M17n
+        #
+        # @return [String] そのまま返す
+        #
+        def to_internal_encoding(string)
+          string
+        end
+        #
+        # 外部エンコーディング文字列化(ダミー)
+        #
+        # @param [String] string もとにする String または M17n
+        #
+        # @return [String] そのまま返す
+        #
+        def to_external_encoding(string)
+          string
+        end
+      end
+    end
+
+    # 
+    # 内部エンコーディング文字列化
+    #
+    def to_internal_encoding
+      EncodingConversion.to_internal_encoding(self)
+    end
+
+    # 
+    # 外部エンコーディング文字列化
+    #
+    def to_external_encoding
+      EncodingConversion.to_external_encoding(self)
+    end
+
+    # 
+    # 内部エンコーディング文字列化(単項)
+    #
+    def +@
+      to_internal_encoding
+    end
+
+    # 
+    # 外部エンコーディング文字列化(単項)
+    #
+    def -@
+      to_external_encoding
+    end
+  end
+end

data/lib/when_exe/locales/iast.rb

@@ -1,90 +1,90 @@
-# -*- coding: utf-8 -*-
-=begin
-  Copyright (C) 2014 Takashi SUGA
-
-  You may use and/or modify this file according to the license described in the LICENSE.txt file included in this archive.
-=end
-
-require 'when_exe/locales/transliteration_table'
-
-module When
-  module Locale
-
-    _IAST_Part_I = {
-        "A"  => "अ",      "a"  => "अ",
-        "Ā"  => "आ",     "ā"  => "आ",
-        "I"  => "इ",      "i"  => "इ",
-        "Ī"  => "ई",      "ī"  => "ई",
-        "U"  => "उ",      "u"  => "उ",
-        "Ū"  => "ऊ",      "ū"  => "ऊ",
-        "Ṛ"  => "ऋ",     "ṛ"  => "ऋ",
-        "Ṝ"  => "ॠ",     "ṝ"  => "ॠ",
-        "Ḷ"  => "ऌ",      "ḷ"  => "ऌ",
-        "Ḹ"  => "ॡ",      "ḹ"  => "ॡ",
-        "E"  => "ए",      "e"  => "ए",
-        "Ai" => "ऐ",      "ai" => "ऐ",
-        "O"  => "ओ",     "o"  => "ओ",
-        "Au" => "औ",     "au" => "औ",
-        "Ṃ"  => "अं",     "ṃ"  => "अं",
-        "Ḥ"  => "अः",    "ḥ"  => "अः",
-        "'"  => "अऽ",
-
-        "K"  => "क",      "k"  => "क",
-        "C"  => "च",      "c"  => "च",
-        "Ṭ"  => "ट",      "ṭ"  => "ट",
-        "T"  => "त",      "t"  => "त",
-        "P"  => "प",      "p"  => "प",
-        "Kh" => "ख",     "kh" => "ख",
-        "Ch" => "छ",      "ch" => "छ",
-        "Ṭh" => "ठ",      "ṭh" => "ठ",
-        "Th" => "थ",      "th" => "थ",
-        "Ph" => "फ",      "ph" => "फ",
-        "G"  => "ग",      "g"  => "ग",
-        "J"  => "ज",      "j"  => "ज",
-        "Ḍ"  => "ड",      "ḍ"  => "ड",
-        "D"  => "द",      "d"  => "द",
-        "B"  => "ब",      "b"  => "ब",
-        "Gh" => "घ",      "gh" => "घ",
-        "Jh" => "झ",      "jh" => "झ",
-        "Ḍh" => "ढ",      "ḍh" => "ढ",
-        "Dh" => "ध",      "dh" => "ध",
-        "Bh" => "भ",      "bh" => "भ",
-        "Ṅ"  => "ङ",      "ṅ"  => "ङ",
-        "Ñ"  => "ञ",      "ñ"  => "ञ",
-        "Ṇ"  => "ण",      "ṇ"  => "ण",
-        "N"  => "न",      "n"  => "न",
-        "M"  => "म",      "m"  => "म",
-        "H"  => "ह",      "h"  => "ह",
-        "Y"  => "य",      "y"  => "य",
-        "R"  => "र",      "r"  => "र",
-        "L"  => "ल",      "l"  => "ल",
-        "V"  => "व",      "v"  => "व",
-        "Ś"  => "श",      "ś"  => "श",
-        "Ṣ"  => "ष",      "ṣ"  => "ष",
-        "S"  => "स",      "s"  => "स"
-    }
-
-    _IAST_Part_II = {
-        "R̥"  => "ऋ",     "r̥"  => "ऋ",
-        "R̥̄"  => "ॠ",     "r̥̄"  => "ॠ",
-        "Ē"  => "ए",      "ē"  => "ए",
-        "Ō"  => "ओ",     "ō"  => "ओ",
-        "Ṁ"  => "अं",     "ṁ"  => "अं"
-    }
-
-    #
-    # IAST: International Alphabet of Sanskrit Transliteration
-    #
-    IAST = {
-      'hi' => _IAST_Part_I.merge(_IAST_Part_II)
-    }
-
-    #
-    # IASTR: International Alphabet of Sanskrit Transliteration - Reverse
-    #
-    IASTR = {
-      'hi' => _IAST_Part_I.invert
-    }
-
-  end
-end
+# -*- coding: utf-8 -*-
+=begin
+  Copyright (C) 2014 Takashi SUGA
+
+  You may use and/or modify this file according to the license described in the LICENSE.txt file included in this archive.
+=end
+
+require 'when_exe/locales/transliteration_table'
+
+module When
+  module Locale
+
+    _IAST_Part_I = {
+        "A"  => "अ",      "a"  => "अ",
+        "Ā"  => "आ",     "ā"  => "आ",
+        "I"  => "इ",      "i"  => "इ",
+        "Ī"  => "ई",      "ī"  => "ई",
+        "U"  => "उ",      "u"  => "उ",
+        "Ū"  => "ऊ",      "ū"  => "ऊ",
+        "Ṛ"  => "ऋ",     "ṛ"  => "ऋ",
+        "Ṝ"  => "ॠ",     "ṝ"  => "ॠ",
+        "Ḷ"  => "ऌ",      "ḷ"  => "ऌ",
+        "Ḹ"  => "ॡ",      "ḹ"  => "ॡ",
+        "E"  => "ए",      "e"  => "ए",
+        "Ai" => "ऐ",      "ai" => "ऐ",
+        "O"  => "ओ",     "o"  => "ओ",
+        "Au" => "औ",     "au" => "औ",
+        "Ṃ"  => "अं",     "ṃ"  => "अं",
+        "Ḥ"  => "अः",    "ḥ"  => "अः",
+        "'"  => "अऽ",
+
+        "K"  => "क",      "k"  => "क",
+        "C"  => "च",      "c"  => "च",
+        "Ṭ"  => "ट",      "ṭ"  => "ट",
+        "T"  => "त",      "t"  => "त",
+        "P"  => "प",      "p"  => "प",
+        "Kh" => "ख",     "kh" => "ख",
+        "Ch" => "छ",      "ch" => "छ",
+        "Ṭh" => "ठ",      "ṭh" => "ठ",
+        "Th" => "थ",      "th" => "थ",
+        "Ph" => "फ",      "ph" => "फ",
+        "G"  => "ग",      "g"  => "ग",
+        "J"  => "ज",      "j"  => "ज",
+        "Ḍ"  => "ड",      "ḍ"  => "ड",
+        "D"  => "द",      "d"  => "द",
+        "B"  => "ब",      "b"  => "ब",
+        "Gh" => "घ",      "gh" => "घ",
+        "Jh" => "झ",      "jh" => "झ",
+        "Ḍh" => "ढ",      "ḍh" => "ढ",
+        "Dh" => "ध",      "dh" => "ध",
+        "Bh" => "भ",      "bh" => "भ",
+        "Ṅ"  => "ङ",      "ṅ"  => "ङ",
+        "Ñ"  => "ञ",      "ñ"  => "ञ",
+        "Ṇ"  => "ण",      "ṇ"  => "ण",
+        "N"  => "न",      "n"  => "न",
+        "M"  => "म",      "m"  => "म",
+        "H"  => "ह",      "h"  => "ह",
+        "Y"  => "य",      "y"  => "य",
+        "R"  => "र",      "r"  => "र",
+        "L"  => "ल",      "l"  => "ल",
+        "V"  => "व",      "v"  => "व",
+        "Ś"  => "श",      "ś"  => "श",
+        "Ṣ"  => "ष",      "ṣ"  => "ष",
+        "S"  => "स",      "s"  => "स"
+    }
+
+    _IAST_Part_II = {
+        "R̥"  => "ऋ",     "r̥"  => "ऋ",
+        "R̥̄"  => "ॠ",     "r̥̄"  => "ॠ",
+        "Ē"  => "ए",      "ē"  => "ए",
+        "Ō"  => "ओ",     "ō"  => "ओ",
+        "Ṁ"  => "अं",     "ṁ"  => "अं"
+    }
+
+    #
+    # IAST: International Alphabet of Sanskrit Transliteration
+    #
+    IAST = {
+      'hi' => _IAST_Part_I.merge(_IAST_Part_II)
+    }
+
+    #
+    # IASTR: International Alphabet of Sanskrit Transliteration - Reverse
+    #
+    IASTR = {
+      'hi' => _IAST_Part_I.invert
+    }
+
+  end
+end