icuSources/data/translit/ru_ru_Latn_BGN.txt

   1 # ***************************************************************************
   2 # *
   3 # *  Copyright (C) 2004-2016, International Business Machines
   4 # *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
   5 # *
   6 # ***************************************************************************
   7 # File: ru_ru_Latn_BGN.txt
   8 # Generated from CLDR
   9 #
  10
  11 # BGN/PCGN 1947 System
  12 #
  13 # The BGN/PCGN system for Russian was adopted by the BGN in 1944 and
  14 # by the PCGN in 1947 for use in romanizing names written in the
  15 # Russian Cyrillic alphabet.
  16 #
  17 # The Russian Alphabet as defined by the BGN (Page 93):
  18 #     АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯ
  19 #     абвгдеёжзийклмнопрстуфхцчшщъыьэюя
  20 #
  21 # Originally prepared by Michael Everson everson@evertype.com
  22 # Fixed by Frank Yung-Fong Tang ftang@google.com
  23 #
  24 # Test Data from http://en.wikipedia.org/wiki/BGN/PCGN_romanization_of_Russian
  25 ########################################################################
  26 # MINIMAL FILTER: Russian-Latin
  27 ::[АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯабвгдеёжзийклмнопрстуфхцчшщъыьэюя];
  28 # BUG(ftang) remove the following line. Otherwise the rule for
  29 # Й й Ё ё will break since the rule is written in NFC but
  30 # the line decomposes the text.
  31 # :: NFD (NFC) ;
  32 ########################################################################
  33 # Define All Transformation Variables
  34 ########################################################################
  35 $prime = ʹ ;
  36 $doublePrime = ʺ ;
  37 $wordBoundary = [^[:L:][:M:][:N:]] ;
  38 $upperConsonants = [БВГДЖЙКЛМНПРСТФХЦЧШЩЭ] ;
  39 $lowerConsonants = [бвгджйклмнпрстфхцчшщэ] ;
  40 $consonants = [$upperConsonants $lowerConsonants] ;
  41 $upperVowels = [АЕЁЭИОУЫЮЯ] ;
  42 $lowerVowels = [аеёэиоуыюя] ;
  43 $vowels = [$upperVowels $lowerVowels] ;
  44 $lower = [$lowerConsonants $lowerVowels] ;
  45 $upper = [$upperConsonants $upperVowels] ;
  46 ########################################################################
  47 # Rules moved to front to avoid masking
  48 ########################################################################
  49 $lowerVowels { ы → ·y ;
  50 $upperVowels { [Ыы] → ·Y ;
  51 [$consonants - [Йй]]{Э → ·E ;
  52 [$consonants - [Йй]]{э → ·e ;
  53 [$upperVowels [ЙЪЬ]] { Е } $upper → YE ; # CYRILLIC CAPITAL LETTER IE
  54 [$upperVowels [ЙЪЬ]] { Е → Ye ; # CYRILLIC CAPITAL LETTER IE
  55 [$upperVowels $lowerVowels [ЙйЪъЬь]] { е → ye ; # CYRILLIC SMALL LETTER IE
  56 [$upperVowels [ЙЪЬ]] { Ё } $upper → YË ; # CYRILLIC CAPITAL LETTER IO
  57 [$upperVowels [ЙЪЬ]] { Ё → Yë ; # CYRILLIC CAPITAL LETTER IO
  58 [$upperVowels $lowerVowels [ЙйЪъЬь]] { ё → yë ; # CYRILLIC SMALL LETTER IO
  59 # Since in the above rule we look use the context before the characters,
  60 # we have to perform them in a separate pass before we change the vowels
  61 # the ::Null forces a separate pass.
  62 ::Null;
  63 ########################################################################
  64 # Start of Alphabetic Transformations
  65 ########################################################################
  66 А → A ; # CYRILLIC CAPITAL LETTER A
  67 а → a ; # CYRILLIC SMALL LETTER A
  68 Б → B ; # CYRILLIC CAPITAL LETTER BE
  69 б → b ; # CYRILLIC SMALL LETTER BE
  70 В → V ; # CYRILLIC CAPITAL LETTER VE
  71 в → v ; # CYRILLIC SMALL LETTER VE
  72 Г → G ; # CYRILLIC CAPITAL LETTER GHE
  73 г → g ; # CYRILLIC SMALL LETTER GHE
  74 Д → D ; # CYRILLIC CAPITAL LETTER DE
  75 д → d ; # CYRILLIC SMALL LETTER DE
  76 ########################################################################
  77 # BGN Page 94 Rule 1:
  78 # # The character e should be romanized ye
  79 # initially, after the vowel # characters a, e, ё, и, о, у, ы, э, ю,
  80 # and я, and after й, ъ, and ь.
  81 # In all other instances, it should
  82 # be romanized e.
  83 ########################################################################
  84 # BUG(ftang)- the following two lines said BEFORE the vowels, instead of AFTER
  85 # Е}[$upperVowels [ЙЪЬ]] → YE ; # CYRILLIC CAPITAL LETTER IE
  86 # Е}[$lowerVowels [йъь]] → Ye ; # CYRILLIC CAPITAL LETTER IE
  87 $wordBoundary{Е} $upper → YE ; # CYRILLIC CAPITAL LETTER IE
  88 $wordBoundary{Е → Ye ; # CYRILLIC CAPITAL LETTER IE
  89 Е → E ; # CYRILLIC CAPITAL LETTER IE
  90 #
  91 # BUG(ftang)- the following line said BEFORE the vowels, instead of AFTER
  92 # е}[$upperVowels $lowerVowels [ЙйЪъЬь]] → ye ; # CYRILLIC SMALL LETTER IE
  93 $wordBoundary{е → ye ; # CYRILLIC SMALL LETTER IE
  94 е → e ; # CYRILLIC SMALL LETTER IE
  95 ########################################################################
  96 # End of Rule 1
  97 ########################################################################
  98 ########################################################################
  99 # BGN Page 94 Rule 2:
 100 #
 101 # The character ё is not considered a separate character of the
 102 # Russian alphabet and the dieresis is generally not shown. When the
 103 # dieresis is shown, the character should be romanized yë initially,
 104 # after the vowel characters a, e, ё, и, о, у, ы, э, ю, and я, and
 105 # after й, ъ, and ь, In all other instances, it should be romanized
 106 # ё. When the dieresis is not shown, the character may still be
 107 # romanized in the preceding manner or, alternatively, in accordance
 108 # with note 1.
 109 ########################################################################
 110 # BUG(ftang)- the following two lines said BEFORE the vowels, instead of AFTER
 111 # Ё}[$upperVowels [ЙЪЬ]] → YË ; # CYRILLIC CAPITAL LETTER IO
 112 # Ё}[$lowerVowels [йъь]] → Yë ; # CYRILLIC CAPITAL LETTER IO
 113 $wordBoundary {Ё} $upper → YË ; # CYRILLIC CAPITAL LETTER IO
 114 $wordBoundary {Ё} $lower → Yë ; # CYRILLIC CAPITAL LETTER IO
 115 Ё → Ë ; # CYRILLIC CAPITAL LETTER IO
 116 # BUG(ftang)- the following line said BEFORE the vowels, instead of AFTER
 117 # ё}[$upperVowels $lowerVowels [ЙйЪъЬь]] → yë ; # CYRILLIC SMALL LETTER IO
 118 $wordBoundary{ё → yë ; # CYRILLIC SMALL LETTER IO
 119 ё → ë ; # CYRILLIC SMALL LETTER IO
 120 ########################################################################
 121 # End of Rule 2
 122 ########################################################################
 123 Ж} $lower → Zh ; # CYRILLIC CAPITAL LETTER ZHE
 124 Ж → ZH ; # CYRILLIC CAPITAL LETTER ZHE
 125 ж → zh ; # CYRILLIC SMALL LETTER ZHE
 126 ########################################################################
 127 # BGN Page 94 Rule 3.4
 128 # э after any consonant character except
 129 # й becomes ·е
 130 ########################################################################
 131 З → Z ; # CYRILLIC CAPITAL LETTER ZE
 132 з → z ; # CYRILLIC SMALL LETTER ZE
 133 # BUG(ftang) The following two lines said those consonant becomes ·е
 134 # [$consonants - [Йй]]}Э → ·Е ;
 135 # [$consonants - [Йй]]}э → ·е ;
 136 ########################################################################
 137 # End of Rule 3.4
 138 ########################################################################
 139 И → I ; # CYRILLIC CAPITAL LETTER I
 140 и → i ; # CYRILLIC SMALL LETTER I
 141 ########################################################################
 142 # BGN Page 94 Rule 3:
 143 #
 144 # Unusual Russian character sequences occurring primarily in
 145 # non-Russian-language names may be romanized as shown below in order
 146 # to provide differentiation from regularly-occurring digraphs and
 147 # character sequences.
 148 #
 149 # BGN Page 94 Rule 3.1
 150 # й before а, у, ы, or э becomes у·
 151 ########################################################################
 152 Й}[АаУуЫыЭэ] → Y· ; # CYRILLIC CAPITAL LETTER I
 153 й}[АаУуЫыЭэ] → y· ; # CYRILLIC SMALL LETTER I
 154 Й → Y ; # CYRILLIC CAPITAL LETTER I
 155 й → y ; # CYRILLIC SMALL LETTER I
 156 ########################################################################
 157 # End Rule 3.1
 158 ########################################################################
 159 К → K ; # CYRILLIC CAPITAL LETTER KA
 160 к → k ; # CYRILLIC SMALL LETTER KA
 161 Л → L ; # CYRILLIC CAPITAL LETTER EL
 162 л → l ; # CYRILLIC SMALL LETTER EL
 163 М → M ; # CYRILLIC CAPITAL LETTER EM
 164 м → m ; # CYRILLIC SMALL LETTER EM
 165 Н → N ; # CYRILLIC CAPITAL LETTER EN
 166 н → n ; # CYRILLIC SMALL LETTER EN
 167 О → O ; # CYRILLIC CAPITAL LETTER O
 168 о → o ; # CYRILLIC SMALL LETTER O
 169 П → P ; # CYRILLIC CAPITAL LETTER PE
 170 п → p ; # CYRILLIC SMALL LETTER PE
 171 Р → R ; # CYRILLIC CAPITAL LETTER ER
 172 р → r ; # CYRILLIC SMALL LETTER ER
 173 С → S ; # CYRILLIC CAPITAL LETTER ES
 174 с → s ; # CYRILLIC SMALL LETTER ES
 175 ########################################################################
 176 # BGN Page 94 Rule 3.5
 177 # тс becomes t·s
 178 ########################################################################
 179 ТС → T·S ; # CYRILLIC CAPITAL LETTER TE
 180 Тс → T·s ; # CYRILLIC CAPITAL LETTER TE
 181 тс → t·s ; # CYRILLIC SMALL LETTER TE
 182 Т → T ; # CYRILLIC CAPITAL LETTER TE
 183 т → t ; # CYRILLIC SMALL LETTER TE
 184 ########################################################################
 185 # End Rule 3.5
 186 ########################################################################
 187 У → U ; # CYRILLIC CAPITAL LETTER U
 188 у → u ; # CYRILLIC SMALL LETTER U
 189 Ф → F ; # CYRILLIC CAPITAL LETTER EF
 190 ф → f ; # CYRILLIC SMALL LETTER EF
 191 Х} $lower → Kh ; # CYRILLIC CAPITAL LETTER HA
 192 Х → KH ; # CYRILLIC CAPITAL LETTER HA
 193 х → kh ; # CYRILLIC SMALL LETTER HA
 194 Ц} $lower → Ts ; # CYRILLIC CAPITAL LETTER TSE
 195 Ц → TS ; # CYRILLIC CAPITAL LETTER TSE
 196 ц → ts ; # CYRILLIC SMALL LETTER TSE
 197 Ч} $lower → Ch ; # CYRILLIC CAPITAL LETTER CHE
 198 Ч → CH ; # CYRILLIC CAPITAL LETTER CHE
 199 ч → ch ; # CYRILLIC SMALL LETTER CHE
 200 ########################################################################
 201 # BGN Page 94 Rule 3.6
 202 # шч becomes sh·ch
 203 ########################################################################
 204 ШЧ → SH·CH ; # CYRILLIC CAPITAL LETTER SHA
 205 Шч → Sh·ch ; # CYRILLIC CAPITAL LETTER SHA
 206 шч → sh·ch ; # CYRILLIC SMALL LETTER SHA
 207 Ш} $lower → Sh ; # CYRILLIC CAPITAL LETTER SHA
 208 Ш → SH ; # CYRILLIC CAPITAL LETTER SHA
 209 ш → sh ; # CYRILLIC SMALL LETTER SHA
 210 Щ} $lower → Shch ; # CYRILLIC CAPITAL LETTER SHCHA
 211 Щ → SHCH ; # CYRILLIC CAPITAL LETTER SHCHA
 212 щ → shch ; # CYRILLIC SMALL LETTER SHCHA
 213 ########################################################################
 214 # End Rule 3.6
 215 ########################################################################
 216 Ъ → $doublePrime ; # CYRILLIC CAPITAL LETTER HARD SIGN
 217 ъ → $doublePrime ; # CYRILLIC SMALL LETTER HARD SIGN
 218 ########################################################################
 219 # BGN Page 94 Rule 3.2
 220 # ы before а, у, ы, or э becomes у·
 221 #
 222 # BGN Page 94 Rule 3.3
 223 # ы after any vowel character becomes ·у
 224 ########################################################################
 225 #
 226 # BUG(ftang) the following line said the vowels will change
 227 # $vowels}Ы → ·Y ; # CYRILLIC CAPITAL LETTER I
 228 # $vowels}ы → ·y ; # CYRILLIC CAPITAL LETTER I
 229 Ы}[АаУуЫыЭэ] → Y· ; # CYRILLIC CAPITAL LETTER YERU
 230 ы}[ауыэ] → y· ; # CYRILLIC SMALL LETTER YERU
 231 Ы → Y ; # CYRILLIC CAPITAL LETTER YERU
 232 ы → y ; # CYRILLIC SMALL LETTER YERU
 233 ########################################################################
 234 # End Rule 3.2 and 3.3
 235 ########################################################################
 236 Ь → $prime ; # CYRILLIC CAPITAL LETTER SOFT SIGN
 237 ь → $prime ; # CYRILLIC SMALL LETTER SOFT SIGN
 238 Э → E ; # CYRILLIC CAPITAL LETTER E
 239 э → e ; # CYRILLIC SMALL LETTER E
 240 Ю} $lower → Yu ; # CYRILLIC CAPITAL LETTER YU
 241 Ю → YU ; # CYRILLIC CAPITAL LETTER YU
 242 ю → yu ; # CYRILLIC SMALL LETTER YU
 243 Я} $lower → Ya ; # CYRILLIC CAPITAL LETTER YA
 244 Я → YA ; # CYRILLIC CAPITAL LETTER YA
 245 я → ya ; # CYRILLIC SMALL LETTER YA
 246