icuSources/data/translit/mn_mn_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: mn_mn_Latn_BGN.txt
   8 # Generated from CLDR
   9 #
  10
  11 ########################################################################
  12 # BGN/PCGN 1964 System
  13 #
  14 # The BGN/PCGN system for Mongolian was adopted by the BGN in 1957
  15 # and by the PCGN in 1964 for use in romanizing names written in
  16 # the Mongolian Cyrillic alphabet. The Mongolian Cyrillic alphabet
  17 # contains two letters not present in the Russian alphabet, Өө
  18 # and Үү. Names written in the indigenous Mongolian alphabet, which
  19 # is still utilized in the Inner Mongolia Autonomous Region of China,
  20 # are not romanized by BGN and PCGN. Instead, for such names,
  21 # BGN and PCGN utilize the Roman-script spellings appearing in
  22 # official sources published by the People's Republic of China
  23 #
  24 # The Mongolian Alphabet as defined by the BGN (Page 73):
  25 #
  26 #   АБВГДЕЁЖЗИЙКЛМНОӨПРСТУҮФХЦЧШЩЪЫЬЭЮЯ
  27 #   абвгдеёжзийклмноөпрстуүфхцчшщъыьэюя
  28 #
  29 # Originally prepared by Michael Everson <everson@evertype.com>
  30 ########################################################################
  31 #
  32 # MINIMAL FILTER: Mongolian-Latin, works both in NFC and NFD
  33 ::[АБВГДЕЁЖЗИЙКЛМНОӨПРСТУҮФХЦЧШЩЪЫЬЭЮЯабвгдеёжзийклмноөпрстуүфхцчшщъыьэюя\u0308];
  34 ::NFC;
  35 ########################################################################
  36 #
  37 # Define All Transformation Variables
  38 #
  39 ########################################################################
  40 #
  41 $prime  = ʹ ;
  42 $upperConsonants = [БВГДЖЙКЛМНПРСТФХЦЧШЩЭ] ;
  43 $lowerConsonants = [бвгджйклмнпрстфхцчшщэ] ;
  44 $consonants = [$upperConsonants $lowerConsonants] ;
  45 $upperVowels = [АЕЁЭИОУЫЮЯ] ;
  46 $lowerVowels = [аеёэиоуыюя] ;
  47 $vowels = [$upperVowels $lowerVowels] ;
  48 $lower = [$lowerConsonants $lowerVowels] ;
  49 # Use this $wordBoundary until bug 2034 is fixed in ICU:
  50 # http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest
  51 $wordBoundary =  [^[:L:][:M:][:N:]] ;
  52 ########################################################################
  53 # Start of Alphabetic Transformations
  54 ########################################################################
  55 А → A ; # CYRILLIC CAPITAL LETTER A
  56 а → a ; # CYRILLIC SMALL LETTER A
  57 Б → B ; # CYRILLIC CAPITAL LETTER BE
  58 б → b ; # CYRILLIC SMALL LETTER BE
  59 В → V ; # CYRILLIC CAPITAL LETTER VE
  60 в → v ; # CYRILLIC SMALL LETTER VE
  61 Г → G ; # CYRILLIC CAPITAL LETTER GHE
  62 г → g ; # CYRILLIC SMALL LETTER GHE
  63 Д → D ; # CYRILLIC CAPITAL LETTER DE
  64 д → d ; # CYRILLIC SMALL LETTER DE
  65 Е} $lower → Yö ; # CYRILLIC CAPITAL LETTER IE
  66 Е → YÖ ; # CYRILLIC CAPITAL LETTER IE
  67 е → yö ; # CYRILLIC SMALL LETTER IE
  68 Ё} $lower → Yo ; # CYRILLIC CAPITAL LETTER IO
  69 Ё → YO ; # CYRILLIC CAPITAL LETTER IO
  70 ё → yo ; # CYRILLIC SMALL LETTER IO
  71 Ж → J ; # CYRILLIC CAPITAL LETTER ZHE
  72 ж → j ; # CYRILLIC SMALL LETTER ZHE
  73 З} $lower → Dz ; # CYRILLIC CAPITAL LETTER ZE
  74 З → DZ ; # CYRILLIC CAPITAL LETTER ZE
  75 з → dz ; # CYRILLIC SMALL LETTER ZE
  76 И → I ; # CYRILLIC CAPITAL LETTER I
  77 и → i ; # CYRILLIC SMALL LETTER I
  78 Й → Y ; # CYRILLIC CAPITAL LETTER I
  79 й → y ; # CYRILLIC SMALL LETTER I
  80 К → K ; # CYRILLIC CAPITAL LETTER KA
  81 к → k ; # CYRILLIC SMALL LETTER KA
  82 Л → L ; # CYRILLIC CAPITAL LETTER EL
  83 л → l ; # CYRILLIC SMALL LETTER EL
  84 М → M ; # CYRILLIC CAPITAL LETTER EM
  85 м → m ; # CYRILLIC SMALL LETTER EM
  86 Н → N ; # CYRILLIC CAPITAL LETTER EN
  87 н → n ; # CYRILLIC SMALL LETTER EN
  88 О → O ; # CYRILLIC CAPITAL LETTER O
  89 о → o ; # CYRILLIC SMALL LETTER O
  90 Ө → Ö ; # CYRILLIC CAPITAL LETTER BARRED O
  91 ө → ö ; # CYRILLIC SMALL LETTER BARRED O
  92 П → P ; # CYRILLIC CAPITAL LETTER PE
  93 п → p ; # CYRILLIC SMALL LETTER PE
  94 Р → R ; # CYRILLIC CAPITAL LETTER ER
  95 р → r ; # CYRILLIC SMALL LETTER ER
  96 С → S ; # CYRILLIC CAPITAL LETTER ES
  97 с → s ; # CYRILLIC SMALL LETTER ES
  98 Т → T ; # CYRILLIC CAPITAL LETTER TE
  99 т → t ; # CYRILLIC SMALL LETTER TE
 100 У → U ; # CYRILLIC CAPITAL LETTER U
 101 у → u ; # CYRILLIC SMALL LETTER U
 102 Ү → Ü ; # CYRILLIC CAPITAL LETTER STRAIGHT U
 103 ү → ü ; # CYRILLIC SMALL LETTER STRAIGHT U
 104 Ф → F ; # CYRILLIC CAPITAL LETTER EF
 105 ф → f ; # CYRILLIC SMALL LETTER EF
 106 Х → H ; # CYRILLIC CAPITAL LETTER HA
 107 х → h ; # CYRILLIC SMALL LETTER HA
 108 Ц} $lower → Ts ; # CYRILLIC CAPITAL LETTER TSE
 109 Ц → TS ; # CYRILLIC CAPITAL LETTER TSE
 110 ц → ts ; # CYRILLIC SMALL LETTER TSE
 111 Ч} $lower → Ch ; # CYRILLIC CAPITAL LETTER CHE
 112 Ч → CH ; # CYRILLIC CAPITAL LETTER CHE
 113 ч → ch ; # CYRILLIC SMALL LETTER CHE
 114 ########################################################################
 115 # Implied rule from BGN Russian-Latin transliteration (Page 94 Note 3.6).
 116 # шч becomes sh·ch
 117 ########################################################################
 118 ШЧ → SH·CH ; # CYRILLIC CAPITAL LETTER SHA
 119 Шч → Sh·ch ; # CYRILLIC CAPITAL LETTER SHA
 120 шч → sh·ch ; # CYRILLIC SMALL LETTER SHA
 121 Ш} $lower → Sh ; # CYRILLIC CAPITAL LETTER SHA
 122 Ш → SH ; # CYRILLIC CAPITAL LETTER SHA
 123 ш → sh ; # CYRILLIC SMALL LETTER SHA
 124 Щ} $lower → Shch ; # CYRILLIC CAPITAL LETTER SHCHA
 125 Щ → SHCH ; # CYRILLIC CAPITAL LETTER SHCHA
 126 щ → shch ; # CYRILLIC SMALL LETTER SHCHA
 127 ########################################################################
 128 # End Implied rule
 129 ########################################################################
 130 Ъ → $prime ; # CYRILLIC CAPITAL LETTER HARD SIGN
 131 ъ → $prime ; # CYRILLIC SMALL LETTER HARD SIGN
 132 Ы → Ï ; # CYRILLIC CAPITAL LETTER YERU
 133 ы → ï ; # CYRILLIC SMALL LETTER YERU
 134 Ь → Ĭ ; # CYRILLIC CAPITAL LETTER SOFT SIGN
 135 ь → ĭ ; # CYRILLIC SMALL LETTER SOFT SIGN
 136 Э → E ; # CYRILLIC CAPITAL LETTER E
 137 э → e ; # CYRILLIC SMALL LETTER E
 138 ########################################################################
 139 #
 140 # BGN Page 74 Rule 7
 141 #
 142 # In monosyllables, the character ю is romanized yu or yü depending on
 143 # pronunciation; in polysyllables, it is romanized yu when followed by
 144 # a, o, or u, buy yü when followed by i, e, ö, or ü.
 145 #
 146 # This rule is lexical and has not been implemented in this file.
 147 #
 148 ########################################################################
 149 Ю} $lower → Yu ; # CYRILLIC CAPITAL LETTER YU
 150 Ю → YU ; # CYRILLIC CAPITAL LETTER YU
 151 ю → yu ; # CYRILLIC SMALL LETTER YU
 152 ########################################################################
 153 #
 154 # End Rule 7
 155 #
 156 ########################################################################
 157 Я} $lower → Ya ; # CYRILLIC CAPITAL LETTER YA
 158 Я → YA ; # CYRILLIC CAPITAL LETTER YA
 159 я → ya ; # CYRILLIC SMALL LETTER YA
 160