icuSources/data/translit/uz_Cyrl_uz_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: uz_Cyrl_uz_BGN.txt
   8 # Generated from CLDR
   9 #
  10
  11 #
  12 ########################################################################
  13 # BGN/PCGN 1979 System
  14 #
  15 # The BGN/PCGN system for Uzbek was designed for use in
  16 # romanizing names written in the Uzbek alphabet.
  17 # The Uzbek alphabet contains four letters not present
  18 # in the Russian alphabet: Ўў, Ққ, Ғғ, and Ҳҳ.
  19 #
  20 # The Uzbek Alphabet as defined by the BGN (Page 107):
  21 #
  22 #   АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЪЬЭЮЯЎҚҒҲ
  23 #   абвгдеёжзийклмнопрстуфхцчшъьэюяўқғҳ
  24 #
  25 # Originally prepared by Michael Everson <everson@evertype.com>
  26 ########################################################################
  27 #
  28 # MINIMAL FILTER: Uzbek-Latin
  29 #
  30 :: [АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЪЬЭЮЯЎҚҒҲабвгдеёжзийклмнопрстуфхцчшъьэюяўқғҳ] ;
  31 :: NFD (NFC) ;
  32 #
  33 #
  34 ########################################################################
  35 #
  36 ########################################################################
  37 #
  38 # Define All Transformation Variables
  39 #
  40 ########################################################################
  41 #
  42 $prime  = ʹ ;
  43 $doublePrime = ʺ ;
  44 $upperConsonants = [БВГДЖЗЙКЛМНПРСТФХЦЧШЪЬҚҒҲ] ;
  45 $lowerConsonants = [бвгджзйклмнпрстфхцчшъьқғҳ] ;
  46 $consonants = [$upperConsonants $lowerConsonants] ;
  47 $upperVowels = [АЕЁИОУЭЮЯЎ] ;
  48 $lowerVowels = [аеёиоуэюяў] ;
  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 # Rules moved to front to avoid masking
  64 #
  65 ########################################################################
  66 #
  67 $lowerVowels ы → y ;
  68 $upperVowels[Ыы] → Y ;
  69 #
  70 #
  71 ########################################################################
  72 #
  73 # Start of Alphabetic Transformations
  74 #
  75 ########################################################################
  76 #
  77 А → A ; # CYRILLIC CAPITAL LETTER A
  78 а → a ; # CYRILLIC SMALL LETTER A
  79 Б → B ; # CYRILLIC CAPITAL LETTER BE
  80 б → b ; # CYRILLIC SMALL LETTER BE
  81 В → W ; # CYRILLIC CAPITAL LETTER VE
  82 в → w ; # CYRILLIC SMALL LETTER VE
  83 #
  84 #
  85 ########################################################################
  86 #
  87 # BGN Page 108 Rule 2
  88 #
  89 # The character sequences гҳ, кҳ, сҳ, and цҳ may be romanized g·h,
  90 # k·h, s·h, and ts·h in order to differentiate those romanizations from
  91 # the digraphs gh, kh, sh, and the letter sequence tsh, which are used
  92 # to render the chаracters г, х, ш, and the character sequence тш.
  93 #
  94 ########################################################################
  95 #
  96 ГҲ → G·H ; # CYRILLIC CAPITAL LETTER GHE
  97 Гҳ → G·h ; # CYRILLIC CAPITAL LETTER GHE
  98 гҳ → g·h ; # CYRILLIC SMALL LETTER GHE
  99 Г → G ; # CYRILLIC CAPITAL LETTER GHE
 100 г → g ; # CYRILLIC SMALL LETTER GHE
 101 #
 102 #
 103 ########################################################################
 104 #
 105 # End Rule 2
 106 #
 107 ########################################################################
 108 #
 109 Д → D ; # CYRILLIC CAPITAL LETTER DE
 110 д → d ; # CYRILLIC SMALL LETTER DE
 111 #
 112 #
 113 ########################################################################
 114 #
 115 # BGN Page 108 Rule 1:
 116 #
 117 # The character e should be romanized ye initially, after the vowel
 118 # characters a, e, ё, и, о, у, э, ю, я, and ў, and after й and ь.
 119 # In all other instances, it should be romanized e.
 120 #
 121 ########################################################################
 122 #
 123 Е}[$upperVowels [ЙЬ]] → YE ; # CYRILLIC CAPITAL LETTER IE
 124 Е}[$lowerVowels [йь]] → Ye ; # CYRILLIC CAPITAL LETTER IE
 125 $wordBoundary{Е → Ye ; # CYRILLIC CAPITAL LETTER IE
 126 Е →  E ; # CYRILLIC CAPITAL LETTER IE
 127 е}[$upperVowels $lowerVowels [ЙйЬь]] → ye ; # CYRILLIC SMALL LETTER IE
 128 $wordBoundary{е → ye ; # CYRILLIC SMALL LETTER IE
 129 е →  e ; # CYRILLIC SMALL LETTER IE
 130 #
 131 #
 132 ########################################################################
 133 #
 134 # End of Rule 1
 135 #
 136 ########################################################################
 137 #
 138 Ё} $lower → Yo ; # CYRILLIC CAPITAL LETTER IO
 139 Ё → YO ; # CYRILLIC CAPITAL LETTER IO
 140 ё → yo ; # CYRILLIC SMALL LETTER IO
 141 Ж → J ; # CYRILLIC CAPITAL LETTER ZHE
 142 ж → j ; # CYRILLIC SMALL LETTER ZHE
 143 З → Z ; # CYRILLIC CAPITAL LETTER ZE
 144 з → z ; # CYRILLIC SMALL LETTER ZE
 145 И → I ; # CYRILLIC CAPITAL LETTER I
 146 и → i ; # CYRILLIC SMALL LETTER I
 147 Й → Y ; # CYRILLIC CAPITAL LETTER I
 148 й → y ; # CYRILLIC SMALL LETTER I
 149 #
 150 #
 151 ########################################################################
 152 #
 153 # BGN Page 108 Rule 2
 154 #
 155 # кҳ becomes k·h
 156 #
 157 ########################################################################
 158 #
 159 КҲ → K·H ; # CYRILLIC CAPITAL LETTER KA
 160 Кҳ → K·h ; # CYRILLIC CAPITAL LETTER KA
 161 кҳ → k·h ; # CYRILLIC SMALL LETTER KA
 162 К → K ; # CYRILLIC CAPITAL LETTER KA
 163 к → k ; # CYRILLIC SMALL LETTER KA
 164 #
 165 #
 166 ########################################################################
 167 #
 168 # End Rule 2
 169 #
 170 ########################################################################
 171 #
 172 Л → L ; # CYRILLIC CAPITAL LETTER EL
 173 л → l ; # CYRILLIC SMALL LETTER EL
 174 М → M ; # CYRILLIC CAPITAL LETTER EM
 175 м → m ; # CYRILLIC SMALL LETTER EM
 176 Н → N ; # CYRILLIC CAPITAL LETTER EN
 177 н → n ; # CYRILLIC SMALL LETTER EN
 178 О → O ; # CYRILLIC CAPITAL LETTER O
 179 о → o ; # CYRILLIC SMALL LETTER O
 180 П → P ; # CYRILLIC CAPITAL LETTER PE
 181 п → p ; # CYRILLIC SMALL LETTER PE
 182 Р → R ; # CYRILLIC CAPITAL LETTER ER
 183 р → r ; # CYRILLIC SMALL LETTER ER
 184 #
 185 #
 186 ########################################################################
 187 #
 188 # BGN Page 108 Rule 2
 189 #
 190 # сҳ becomes s·h
 191 #
 192 ########################################################################
 193 #
 194 СҲ → S·H ; # CYRILLIC CAPITAL LETTER ES
 195 Сҳ → S·h ; # CYRILLIC CAPITAL LETTER ES
 196 сҳ → s·h ; # CYRILLIC SMALL LETTER ES
 197 С → S ; # CYRILLIC CAPITAL LETTER ES
 198 с → s ; # CYRILLIC SMALL LETTER ES
 199 #
 200 #
 201 ########################################################################
 202 #
 203 # End Rule 2
 204 #
 205 ########################################################################
 206 #
 207 Т → T ; # CYRILLIC CAPITAL LETTER TE
 208 т → t ; # CYRILLIC SMALL LETTER TE
 209 У → Ū ; # CYRILLIC CAPITAL LETTER U
 210 у → ū ; # CYRILLIC SMALL LETTER U
 211 Ф → F ; # CYRILLIC CAPITAL LETTER EF
 212 ф → f ; # CYRILLIC SMALL LETTER EF
 213 Х} $lower → Kh ; # CYRILLIC CAPITAL LETTER HA
 214 Х → KH ; # CYRILLIC CAPITAL LETTER HA
 215 х → kh ; # CYRILLIC SMALL LETTER HA
 216 #
 217 #
 218 ########################################################################
 219 #
 220 # BGN Page 108 Rule 2
 221 #
 222 # цҳ becomes ts·h
 223 #
 224 ########################################################################
 225 #
 226 ЦҲ → TS·H ; # CYRILLIC CAPITAL LETTER GHE
 227 Цҳ → Ts·h ; # CYRILLIC CAPITAL LETTER GHE
 228 цҳ → ts·h ; # CYRILLIC SMALL LETTER GHE
 229 Ц} $lower → Ts ; # CYRILLIC CAPITAL LETTER TSE
 230 Ц → TS ; # CYRILLIC CAPITAL LETTER TSE
 231 ц → ts ; # CYRILLIC SMALL LETTER TSE
 232 #
 233 #
 234 ########################################################################
 235 #
 236 # End Rule 2
 237 #
 238 ########################################################################
 239 #
 240 Ч} $lower → Ch ; # CYRILLIC CAPITAL LETTER CHE
 241 Ч → CH ; # CYRILLIC CAPITAL LETTER CHE
 242 ч → ch ; # CYRILLIC SMALL LETTER CHE
 243 Ш} $lower → Sh ; # CYRILLIC CAPITAL LETTER SHA
 244 Ш → SH ; # CYRILLIC CAPITAL LETTER SHA
 245 ш → sh ; # CYRILLIC SMALL LETTER SHA
 246 Ъ → $prime ; # CYRILLIC CAPITAL LETTER HARD SIGN
 247 ъ → $prime ; # CYRILLIC SMALL LETTER HARD SIGN
 248 Ь → $prime ; # CYRILLIC CAPITAL LETTER SOFT SIGN
 249 ь → $prime ; # CYRILLIC SMALL LETTER SOFT SIGN
 250 Э → e ; # CYRILLIC CAPITAL LETTER E
 251 э → e ; # CYRILLIC SMALL LETTER E
 252 Ю} $lower → Yu ; # CYRILLIC CAPITAL LETTER YU
 253 Ю → YU ; # CYRILLIC CAPITAL LETTER YU
 254 ю → yu ; # CYRILLIC SMALL LETTER YU
 255 Я} $lower → Ya ; # CYRILLIC CAPITAL LETTER YA
 256 Я → YA ; # CYRILLIC CAPITAL LETTER YA
 257 я → ya ; # CYRILLIC SMALL LETTER YA
 258 Ў → Ŭ ; # CYRILLIC CAPITAL LETTER SHORT U
 259 ў → ŭ ; # CYRILLIC SMALL LETTER SHORT U
 260 Қ → Q ; # CYRILLIC CAPITAL LETTER KA WITH DESCENDER
 261 қ → q ; # CYRILLIC SMALL LETTER KA WITH DESCENDER
 262 Ғ} $lower → Gh ; # CYRILLIC CAPITAL LETTER GHE WITH STROKE
 263 Ғ → GH ; # CYRILLIC CAPITAL LETTER GHE WITH STROKE
 264 ғ → gh ; # CYRILLIC SMALL LETTER GHE WITH STROKE
 265 Ҳ → H ; # CYRILLIC CAPITAL LETTER HA WITH DESCENDER
 266 ҳ → h ; # CYRILLIC SMALL LETTER HA WITH DESCENDER
 267 #
 268 #
 269 ########################################################################
 270