icuSources/data/translit/ru_ru_Latn_BGN.txt

   1 # © 2016 and later: Unicode, Inc. and others.
   2 # License & terms of use: http://www.unicode.org/copyright.html#License
   3 #
   4 # File: ru_ru_Latn_BGN.txt
   5 # Generated from CLDR
   6 #
   7
   8 # BGN/PCGN 1947 System
   9 #
  10 # The BGN/PCGN system for Russian was adopted by the BGN in 1944 and
  11 # by the PCGN in 1947 for use in romanizing names written in the
  12 # Russian Cyrillic alphabet.
  13 #
  14 # The Russian Alphabet as defined by the BGN (Page 93):
  15 #     АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯ
  16 #     абвгдеёжзийклмнопрстуфхцчшщъыьэюя
  17 #
  18 # Originally prepared by Michael Everson everson@evertype.com
  19 # Fixed by Frank Yung-Fong Tang ftang@google.com
  20 #
  21 # Test Data from http://en.wikipedia.org/wiki/BGN/PCGN_romanization_of_Russian
  22 ########################################################################
  23 # MINIMAL FILTER: Russian-Latin
  24 ::[АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯабвгдеёжзийклмнопрстуфхцчшщъыьэюя];
  25 ::NFC;
  26 ########################################################################
  27 # Define All Transformation Variables
  28 ########################################################################
  29 $prime = ʹ ;
  30 $doublePrime = ʺ ;
  31 $wordBoundary = [^[:L:][:M:][:N:]] ;
  32 $upperVowels = [АЕЁЭИОУЫЮЯ] ;
  33 $lowerVowels = [аеёэиоуыюя] ;
  34 $vowels = [$upperVowels $lowerVowels] ;
  35 $upperConsonants = [[:Uppercase:]-$vowels] ;
  36 $lowerConsonants = [[:Lowercase:]-$vowels] ;
  37 $consonants = [$upperConsonants $lowerConsonants] ;
  38 $upper = [:Uppercase:];
  39 $lower = [:Lowercase:];
  40 ########################################################################
  41 # Rules moved to front to avoid masking
  42 ########################################################################
  43 $lowerVowels { ы → ·y ;
  44 $upperVowels { [Ыы] } $lower → ·y ;
  45 $upperVowels { [Ыы] } → ·Y ;
  46 [$consonants - [Йй]]{Э → ·E ;
  47 [$consonants - [Йй]]{э → ·e ;
  48 [$upperVowels [ЙЪЬ]] { Е } $upper → YE ; # CYRILLIC CAPITAL LETTER IE
  49 [$upperVowels [ЙЪЬ]] { Е → Ye ; # CYRILLIC CAPITAL LETTER IE
  50 [$upperVowels $lowerVowels [ЙйЪъЬь]] { е → ye ; # CYRILLIC SMALL LETTER IE
  51 [$upperVowels [ЙЪЬ]] { Ё } $upper → YË ; # CYRILLIC CAPITAL LETTER IO
  52 [$upperVowels [ЙЪЬ]] { Ё → Yë ; # CYRILLIC CAPITAL LETTER IO
  53 [$upperVowels $lowerVowels [ЙйЪъЬь]] { ё → yë ; # CYRILLIC SMALL LETTER IO
  54 # Since in the above rule we look at the Cyrillic context before the E/Ё/ё,
  55 # we have to transform these in a separate pass before we change the vowels.
  56 # The ::Null forces a separate pass.
  57 ::Null;
  58 ########################################################################
  59 # Start of Alphabetic Transformations
  60 ########################################################################
  61 А → A ; # CYRILLIC CAPITAL LETTER A
  62 а → a ; # CYRILLIC SMALL LETTER A
  63 Б → B ; # CYRILLIC CAPITAL LETTER BE
  64 б → b ; # CYRILLIC SMALL LETTER BE
  65 В → V ; # CYRILLIC CAPITAL LETTER VE
  66 в → v ; # CYRILLIC SMALL LETTER VE
  67 Г → G ; # CYRILLIC CAPITAL LETTER GHE
  68 г → g ; # CYRILLIC SMALL LETTER GHE
  69 Д → D ; # CYRILLIC CAPITAL LETTER DE
  70 д → d ; # CYRILLIC SMALL LETTER DE
  71 ########################################################################
  72 # BGN Page 94 Rule 1:
  73 # # The character e should be romanized ye
  74 # initially, after the vowel # characters a, e, ё, и, о, у, ы, э, ю,
  75 # and я, and after й, ъ, and ь.
  76 # In all other instances, it should
  77 # be romanized e.
  78 ########################################################################
  79 # BUG(ftang)- the following two lines said BEFORE the vowels, instead of AFTER
  80 # Е}[$upperVowels [ЙЪЬ]] → YE ; # CYRILLIC CAPITAL LETTER IE
  81 # Е}[$lowerVowels [йъь]] → Ye ; # CYRILLIC CAPITAL LETTER IE
  82 $wordBoundary{Е} $upper → YE ; # CYRILLIC CAPITAL LETTER IE
  83 $wordBoundary{Е → Ye ; # CYRILLIC CAPITAL LETTER IE
  84 Е → E ; # CYRILLIC CAPITAL LETTER IE
  85 #
  86 # BUG(ftang)- the following line said BEFORE the vowels, instead of AFTER
  87 # е}[$upperVowels $lowerVowels [ЙйЪъЬь]] → ye ; # CYRILLIC SMALL LETTER IE
  88 $wordBoundary{е → ye ; # CYRILLIC SMALL LETTER IE
  89 е → e ; # CYRILLIC SMALL LETTER IE
  90 ########################################################################
  91 # End of Rule 1
  92 ########################################################################
  93 ########################################################################
  94 # BGN Page 94 Rule 2:
  95 #
  96 # The character ё is not considered a separate character of the
  97 # Russian alphabet and the dieresis is generally not shown. When the
  98 # dieresis is shown, the character should be romanized yë initially,
  99 # after the vowel characters a, e, ё, и, о, у, ы, э, ю, and я, and
 100 # after й, ъ, and ь, In all other instances, it should be romanized
 101 # ё. When the dieresis is not shown, the character may still be
 102 # romanized in the preceding manner or, alternatively, in accordance
 103 # with note 1.
 104 ########################################################################
 105 # BUG(ftang)- the following two lines said BEFORE the vowels, instead of AFTER
 106 # Ё}[$upperVowels [ЙЪЬ]] → YË ; # CYRILLIC CAPITAL LETTER IO
 107 # Ё}[$lowerVowels [йъь]] → Yë ; # CYRILLIC CAPITAL LETTER IO
 108 $wordBoundary {Ё} [·]? $upper → YË ; # CYRILLIC CAPITAL LETTER IO
 109 $wordBoundary {Ё} [·]? $lower → Yë ; # CYRILLIC CAPITAL LETTER IO
 110 Ё → Ë ; # CYRILLIC CAPITAL LETTER IO
 111 # BUG(ftang)- the following line said BEFORE the vowels, instead of AFTER
 112 # ё}[$upperVowels $lowerVowels [ЙйЪъЬь]] → yë ; # CYRILLIC SMALL LETTER IO
 113 $wordBoundary{ё → yë ; # CYRILLIC SMALL LETTER IO
 114 ё → ë ; # CYRILLIC SMALL LETTER IO
 115 ########################################################################
 116 # End of Rule 2
 117 ########################################################################
 118 Ж} $lower → Zh ; # CYRILLIC CAPITAL LETTER ZHE
 119 Ж → ZH ; # CYRILLIC CAPITAL LETTER ZHE
 120 ж → zh ; # CYRILLIC SMALL LETTER ZHE
 121 ########################################################################
 122 # BGN Page 94 Rule 3.4
 123 # э after any consonant character except
 124 # й becomes ·е
 125 ########################################################################
 126 З → Z ; # CYRILLIC CAPITAL LETTER ZE
 127 з → z ; # CYRILLIC SMALL LETTER ZE
 128 # BUG(ftang) The following two lines said those consonant becomes ·е
 129 # [$consonants - [Йй]]}Э → ·Е ;
 130 # [$consonants - [Йй]]}э → ·е ;
 131 ########################################################################
 132 # End of Rule 3.4
 133 ########################################################################
 134 И → I ; # CYRILLIC CAPITAL LETTER I
 135 и → i ; # CYRILLIC SMALL LETTER I
 136 ########################################################################
 137 # BGN Page 94 Rule 3:
 138 #
 139 # Unusual Russian character sequences occurring primarily in
 140 # non-Russian-language names may be romanized as shown below in order
 141 # to provide differentiation from regularly-occurring digraphs and
 142 # character sequences.
 143 #
 144 # BGN Page 94 Rule 3.1
 145 # й before а, у, ы, or э becomes у·
 146 ########################################################################
 147 Й}[АаУуЫыЭэ] → Y· ; # CYRILLIC CAPITAL LETTER I
 148 й}[АаУуЫыЭэ] → y· ; # CYRILLIC SMALL LETTER I
 149 Й → Y ; # CYRILLIC CAPITAL LETTER I
 150 й → y ; # CYRILLIC SMALL LETTER I
 151 ########################################################################
 152 # End Rule 3.1
 153 ########################################################################
 154 К → K ; # CYRILLIC CAPITAL LETTER KA
 155 к → k ; # CYRILLIC SMALL LETTER KA
 156 Л → L ; # CYRILLIC CAPITAL LETTER EL
 157 л → l ; # CYRILLIC SMALL LETTER EL
 158 М → M ; # CYRILLIC CAPITAL LETTER EM
 159 м → m ; # CYRILLIC SMALL LETTER EM
 160 Н → N ; # CYRILLIC CAPITAL LETTER EN
 161 н → n ; # CYRILLIC SMALL LETTER EN
 162 О → O ; # CYRILLIC CAPITAL LETTER O
 163 о → o ; # CYRILLIC SMALL LETTER O
 164 П → P ; # CYRILLIC CAPITAL LETTER PE
 165 п → p ; # CYRILLIC SMALL LETTER PE
 166 Р → R ; # CYRILLIC CAPITAL LETTER ER
 167 р → r ; # CYRILLIC SMALL LETTER ER
 168 С → S ; # CYRILLIC CAPITAL LETTER ES
 169 с → s ; # CYRILLIC SMALL LETTER ES
 170 ########################################################################
 171 # BGN Page 94 Rule 3.5
 172 # тс becomes t·s
 173 ########################################################################
 174 ТС → T·S ; # CYRILLIC CAPITAL LETTER TE
 175 Тс → T·s ; # CYRILLIC CAPITAL LETTER TE
 176 тс → t·s ; # CYRILLIC SMALL LETTER TE
 177 Т → T ; # CYRILLIC CAPITAL LETTER TE
 178 т → t ; # CYRILLIC SMALL LETTER TE
 179 ########################################################################
 180 # End Rule 3.5
 181 ########################################################################
 182 У → U ; # CYRILLIC CAPITAL LETTER U
 183 у → u ; # CYRILLIC SMALL LETTER U
 184 Ф → F ; # CYRILLIC CAPITAL LETTER EF
 185 ф → f ; # CYRILLIC SMALL LETTER EF
 186 Х} $lower → Kh ; # CYRILLIC CAPITAL LETTER HA
 187 Х → KH ; # CYRILLIC CAPITAL LETTER HA
 188 х → kh ; # CYRILLIC SMALL LETTER HA
 189 Ц} $lower → Ts ; # CYRILLIC CAPITAL LETTER TSE
 190 Ц → TS ; # CYRILLIC CAPITAL LETTER TSE
 191 ц → ts ; # CYRILLIC SMALL LETTER TSE
 192 Ч} $lower → Ch ; # CYRILLIC CAPITAL LETTER CHE
 193 Ч → CH ; # CYRILLIC CAPITAL LETTER CHE
 194 ч → ch ; # CYRILLIC SMALL LETTER CHE
 195 ########################################################################
 196 # BGN Page 94 Rule 3.6
 197 # шч becomes sh·ch
 198 ########################################################################
 199 ШЧ → SH·CH ; # CYRILLIC CAPITAL LETTER SHA
 200 Шч → Sh·ch ; # CYRILLIC CAPITAL LETTER SHA
 201 шч → sh·ch ; # CYRILLIC SMALL LETTER SHA
 202 Ш} $lower → Sh ; # CYRILLIC CAPITAL LETTER SHA
 203 Ш → SH ; # CYRILLIC CAPITAL LETTER SHA
 204 ш → sh ; # CYRILLIC SMALL LETTER SHA
 205 Щ} $lower → Shch ; # CYRILLIC CAPITAL LETTER SHCHA
 206 Щ → SHCH ; # CYRILLIC CAPITAL LETTER SHCHA
 207 щ → shch ; # CYRILLIC SMALL LETTER SHCHA
 208 ########################################################################
 209 # End Rule 3.6
 210 ########################################################################
 211 Ъ → $doublePrime ; # CYRILLIC CAPITAL LETTER HARD SIGN
 212 ъ → $doublePrime ; # CYRILLIC SMALL LETTER HARD SIGN
 213 ########################################################################
 214 # BGN Page 94 Rule 3.2
 215 # ы before а, у, ы, or э becomes у·
 216 #
 217 # BGN Page 94 Rule 3.3
 218 # ы after any vowel character becomes ·у
 219 ########################################################################
 220 #
 221 # BUG(ftang) the following line said the vowels will change
 222 # $vowels}Ы → ·Y ; # CYRILLIC CAPITAL LETTER I
 223 # $vowels}ы → ·y ; # CYRILLIC CAPITAL LETTER I
 224 Ы}[АаУуЫыЭэ] → Y· ; # CYRILLIC CAPITAL LETTER YERU
 225 ы}[ауыэ] → y· ; # CYRILLIC SMALL LETTER YERU
 226 Ы → Y ; # CYRILLIC CAPITAL LETTER YERU
 227 ы → y ; # CYRILLIC SMALL LETTER YERU
 228 ########################################################################
 229 # End Rule 3.2 and 3.3
 230 ########################################################################
 231 Ь → $prime ; # CYRILLIC CAPITAL LETTER SOFT SIGN
 232 ь → $prime ; # CYRILLIC SMALL LETTER SOFT SIGN
 233 Э → E ; # CYRILLIC CAPITAL LETTER E
 234 э → e ; # CYRILLIC SMALL LETTER E
 235 Ю} $lower → Yu ; # CYRILLIC CAPITAL LETTER YU
 236 Ю → YU ; # CYRILLIC CAPITAL LETTER YU
 237 ю → yu ; # CYRILLIC SMALL LETTER YU
 238 Я} $lower → Ya ; # CYRILLIC CAPITAL LETTER YA
 239 Я → YA ; # CYRILLIC CAPITAL LETTER YA
 240 я → ya ; # CYRILLIC SMALL LETTER YA
 241