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