icuSources/data/translit/mk_mk_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: mk_mk_Latn_BGN.txt
   8 # Generated from CLDR
   9 #
  10
  11 #
  12 ########################################################################
  13 # BGN/PCGN 1981 System
  14 #
  15 # Macedonian was official established as a literary language in
  16 # Yugoslavia during World War II and is now the official language
  17 # of Macedonia. Its alphabet is identical to Serbian, except
  18 # that the letters Ђђ and Ћћ are replaced by Ѓѓ and Ќќ, and
  19 # the letter Ѕѕ and the apostrophe are added.
  20 #
  21 # The Macedonian Alphabet as defined by the BGN (Page 69):
  22 #
  23 #   АБВГДЃЕЖЗЅИЈКЛЉМНЊОПРСТЌУФХЦЧЏШ
  24 #   абвгдѓежзѕијклљмнњопрстќуфхцчџш’
  25 #
  26 # Originally prepared by Michael Everson <everson@evertype.com>
  27 ########################################################################
  28 #
  29 # MINIMAL FILTER: Macedonian-Latin
  30 #
  31 :: [АБВГДЃЕЖЗЅИЈКЛЉМНЊОПРСТЌУФХЦЧЏШабвгдѓежзѕијклљмнњопрстќуфхцчџш’] ;
  32 :: NFD (NFC) ;
  33 #
  34 #
  35 ########################################################################
  36 #
  37 ########################################################################
  38 #
  39 # Define All Transformation Variables
  40 #
  41 ########################################################################
  42 #
  43 $prime  = ’ ;
  44 $upperConsonants = [БВГДЃЖЗЅЈКЛЉМНЊПРСТЌФХЦЧЏШ] ;
  45 $lowerConsonants = [бвгдѓжзѕјклљмнњпрстќфхцчџш’] ;
  46 $consonants = [$upperConsonants $lowerConsonants] ;
  47 $upperVowels = [АЕИОУ] ;
  48 $lowerVowels = [аеиоу] ;
  49 $vowels = [$upperVowels $lowerVowels] ;
  50 $lower = [$lowerConsonants $lowerVowels] ;
  51 #
  52 #
  53 # Use this $wordBoundary until bug 2034 is fixed in ICU:
  54 # http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest
  55 #
  56 $wordBoundary =  [^[:L:][:M:][:N:]] ;
  57 #
  58 #
  59 ########################################################################
  60 #
  61 ########################################################################
  62 #
  63 # Start of Alphabetic Transformations
  64 #
  65 ########################################################################
  66 #
  67 А → A ; # CYRILLIC CAPITAL LETTER A
  68 а → a ; # CYRILLIC SMALL LETTER A
  69 Б → B ; # CYRILLIC CAPITAL LETTER BE
  70 б → b ; # CYRILLIC SMALL LETTER BE
  71 В → V ; # CYRILLIC CAPITAL LETTER VE
  72 в → v ; # CYRILLIC SMALL LETTER VE
  73 Г → G ; # CYRILLIC CAPITAL LETTER GHE
  74 г → g ; # CYRILLIC SMALL LETTER GHE
  75 Д → D ; # CYRILLIC CAPITAL LETTER DE
  76 д → d ; # CYRILLIC SMALL LETTER DE
  77 #
  78 #
  79 ########################################################################
  80 #
  81 # BGN Page 70 Rule 1:
  82 #
  83 # The character ѓ should be romanized g when it occurs before е
  84 # and и. In all other instances, it should be romanized đ (Đ).
  85 #
  86 ########################################################################
  87 #
  88 Ѓ}[ЕеИи] → G ; # CYRILLIC CAPITAL LETTER GJE
  89 ѓ}[ЕеИи] → g ; # CYRILLIC SMALL LETTER GJE
  90 Ѓ → Đ ; # CYRILLIC CAPITAL LETTER GJE
  91 ѓ → đ ; # CYRILLIC SMALL LETTER GJE
  92 #
  93 #
  94 ########################################################################
  95 #
  96 # End Rule 1
  97 #
  98 ########################################################################
  99 #
 100 Е → E ; # CYRILLIC CAPITAL LETTER DE
 101 е → e ; # CYRILLIC SMALL LETTER DE
 102 Ж → Ž ; # CYRILLIC CAPITAL LETTER ZHE
 103 ж → ž ; # CYRILLIC SMALL LETTER ZHE
 104 З → Z ; # CYRILLIC CAPITAL LETTER ZE
 105 з → z ; # CYRILLIC SMALL LETTER ZE
 106 Ѕ} $lower → Dz ; # CYRILLIC CAPITAL LETTER DZE
 107 Ѕ → DZ ; # CYRILLIC CAPITAL LETTER DZE
 108 ѕ → dz ; # CYRILLIC SMALL LETTER DZE
 109 И → I ; # CYRILLIC CAPITAL LETTER I
 110 и → i ; # CYRILLIC SMALL LETTER I
 111 Ј → J ; # CYRILLIC CAPITAL LETTER JE
 112 ј → j ; # CYRILLIC SMALL LETTER JE
 113 К → K ; # CYRILLIC CAPITAL LETTER KA
 114 к → k ; # CYRILLIC SMALL LETTER KA
 115 Л → L ; # CYRILLIC CAPITAL LETTER EL
 116 л → l ; # CYRILLIC SMALL LETTER EL
 117 Љ} $lower → Lj ; # CYRILLIC CAPITAL LETTER LJE
 118 Љ → LJ ; # CYRILLIC CAPITAL LETTER LJE
 119 љ → lj ; # CYRILLIC SMALL LETTER LJE
 120 М → M ; # CYRILLIC CAPITAL LETTER EM
 121 м → m ; # CYRILLIC SMALL LETTER EM
 122 Н → N ; # CYRILLIC CAPITAL LETTER EN
 123 н → n ; # CYRILLIC SMALL LETTER EN
 124 Њ} $lower → Nj ; # CYRILLIC CAPITAL LETTER NJE
 125 Њ → NJ ; # CYRILLIC CAPITAL LETTER NJE
 126 њ → nj ; # CYRILLIC SMALL LETTER NJE
 127 О → O ; # CYRILLIC CAPITAL LETTER O
 128 о → o ; # CYRILLIC SMALL LETTER O
 129 П → P ; # CYRILLIC CAPITAL LETTER PE
 130 п → p ; # CYRILLIC SMALL LETTER PE
 131 Р → R ; # CYRILLIC CAPITAL LETTER ER
 132 р → r ; # CYRILLIC SMALL LETTER ER
 133 С → S ; # CYRILLIC CAPITAL LETTER ES
 134 с → s ; # CYRILLIC SMALL LETTER ES
 135 Т → T ; # CYRILLIC CAPITAL LETTER TE
 136 т → t ; # CYRILLIC SMALL LETTER TE
 137 #
 138 #
 139 ########################################################################
 140 #
 141 # BGN Page 70 Rule 2:
 142 #
 143 # The character ќ should be romanized k when it occurs before е
 144 # and и. In all other instances, it should be romanized c\u0301.
 145 #
 146 ########################################################################
 147 #
 148 Ќ}[ЕеИи] → K ; # CYRILLIC CAPITAL LETTER KJE
 149 ќ}[ЕеИи] → k ; # CYRILLIC SMALL LETTER KJE
 150 Ќ → C\u0301 ; # CYRILLIC CAPITAL LETTER KJE
 151 ќ → c\u0301 ; # CYRILLIC SMALL LETTER KJE
 152 #
 153 #
 154 ########################################################################
 155 #
 156 # End Rule 2
 157 #
 158 ########################################################################
 159 #
 160 У → U ; # CYRILLIC CAPITAL LETTER U
 161 у → u ; # CYRILLIC SMALL LETTER U
 162 Ф → F ; # CYRILLIC CAPITAL LETTER EF
 163 ф → f ; # CYRILLIC SMALL LETTER EF
 164 Х → H ; # CYRILLIC CAPITAL LETTER HA
 165 х → h ; # CYRILLIC SMALL LETTER HA
 166 Ц → C ; # CYRILLIC CAPITAL LETTER TSE
 167 ц → c ; # CYRILLIC SMALL LETTER TSE
 168 Ч → Č ; # CYRILLIC CAPITAL LETTER CHE
 169 ч → č ; # CYRILLIC SMALL LETTER CHE
 170 Џ} $lower → Dž ; # CYRILLIC CAPITAL LETTER SHA
 171 Џ → DŽ ; # CYRILLIC CAPITAL LETTER SHA
 172 џ → dž ; # CYRILLIC SMALL LETTER SHA
 173 Ш → Š ; # CYRILLIC CAPITAL LETTER SHA
 174 ш → š ; # CYRILLIC SMALL LETTER SHA
 175 #
 176 #
 177 ########################################################################
 178 #
 179 # BGN Page 69 Rule 32, maps the symbol onto itself and
 180 # is ignored here for computational efficiency.
 181 #
 182 # $prime → $prime ; # RIGHT SINGLE QUOTATION MARK
 183 #
 184 ########################################################################
 185