1 <?xml version="1.0" encoding="UTF-8" ?> 2 <!DOCTYPE supplementalData SYSTEM "../../common/dtd/ldmlSupplemental.dtd"> 3 <!-- 4 Copyright 1991-2013 Unicode, Inc. 5 CLDR data files are interpreted according to the LDML specification (http://unicode.org/reports/tr35/) 6 For terms of use, see http://www.unicode.org/copyright.html 7 --> 8 <supplementalData> 9 <version number="$Revision: 12263 $"/> 10 <transforms> 11 <transform source="Cyrl" target="Latn" direction="both" alias="Cyrillic-Latin und-Latn-t-und-cyrl" backwardAlias="Latin-Cyrillic und-Cyrl-t-und-latn"> 12 <tRule> 13 # TODO: add remaining characters 14 # Should add variants for Russian-English, Russian-German 15 # Those can use this as a base, and then remap cases 16 # like a $hat to ya or ja. 17 # :: [\u0000-\u007E [:Cyrillic:] [:Latin:] [:nonspacing mark:]] ; 18 ### WARNING, must be added to the generated filters, in both directions ### 19 # MINIMAL FILTER 20 :: [----------------] ; 21 :: NFD (NFC) ; 22 $modprime = ; 23 $modprime2 = ; 24 $grave = ; 25 $acute = ; 26 $hat = ; 27 $breve = ; 28 $dot = ; 29 $caron = ; 30 $comma = ; 31 $under = ; 32 $descender = ; 33 # move up so not masked 34 a $hat ; # CYRILLIC SMALL LETTER YA 35 A $hat ; # CYRILLIC CAPITAL LETTER YA 36 c $caron ; # CYRILLIC SMALL LETTER CHE 37 C $caron; # CYRILLIC CAPITAL LETTER CHE 38 # XXX ; # CYRILLIC SMALL LETTER CHE WITH DESCENDER 39 # XXX ; # CYRILLIC CAPITAL LETTER CHE WITH DESCENDER 40 # XXX ; # CYRILLIC SMALL LETTER KHAKASSIAN CHE 41 # XXX ; # CYRILLIC CAPITAL LETTER KHAKASSIAN CHE 42 # XXX ; # CYRILLIC SMALL LETTER CHE WITH VERTICAL STROKE 43 # XXX ; # CYRILLIC CAPITAL LETTER CHE WITH VERTICAL STROKE 44 e $acute; # CYRILLIC SMALL LETTER E 45 E $acute; # CYRILLIC CAPITAL LETTER E 46 e $hat; # CYRILLIC SMALL LETTER UKRAINIAN IE 47 E $hat; # CYRILLIC CAPITAL LETTER UKRAINIAN IE 48 s $caron ; # CYRILLIC SMALL LETTER SHA 49 S $caron ; # CYRILLIC CAPITAL LETTER SHA 50 s $hat ; # CYRILLIC SMALL LETTER SHCHA 51 S $hat; # CYRILLIC CAPITAL LETTER SHCHA 52 z $hat ; # CYRILLIC SMALL LETTER DZE 53 Z $hat; # CYRILLIC CAPITAL LETTER DZE 54 # XXX ; # CYRILLIC SMALL LETTER ABKHASIAN DZE 55 # XXX ; # CYRILLIC CAPITAL LETTER ABKHASIAN DZE 56 u $hat ; # CYRILLIC SMALL LETTER YU 57 U $hat ; # CYRILLIC CAPITAL LETTER YU 58 i $acute; # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I 59 I $acute; # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I 60 j $caron; # CYRILLIC SMALL LETTER JE 61 J $caron; # CYRILLIC CAPITAL LETTER JE 62 l $hat ; # CYRILLIC SMALL LETTER LJE 63 L $hat ; # CYRILLIC CAPITAL LETTER LJE 64 n $hat ; # CYRILLIC SMALL LETTER NJE 65 N $hat ; # CYRILLIC CAPITAL LETTER NJE 66 c $acute ; # CYRILLIC SMALL LETTER TSHE 67 C $acute ; # CYRILLIC CAPITAL LETTER TSHE 68 d $hat ; # CYRILLIC SMALL LETTER DZHE 69 D $hat ; # CYRILLIC CAPITAL LETTER DZHE 70 # Normal order 71 a ; # CYRILLIC SMALL LETTER A 72 A ; # CYRILLIC CAPITAL LETTER A 73 ; # CYRILLIC SMALL LETTER SCHWA 74 ; # CYRILLIC CAPITAL LETTER SCHWA 75 ; # CYRILLIC SMALL LIGATURE A IE 76 ; # CYRILLIC CAPITAL LIGATURE A IE 77 b ; # CYRILLIC SMALL LETTER BE 78 B ; # CYRILLIC CAPITAL LETTER BE 79 v ; # CYRILLIC SMALL LETTER VE 80 V ; # CYRILLIC CAPITAL LETTER VE 81 g $grave ; # CYRILLIC SMALL LETTER GHE WITH UPTURN 82 G $grave ; # CYRILLIC CAPITAL LETTER GHE WITH UPTURN 83 g $dot ; # CYRILLIC SMALL LETTER GHE WITH STROKE 84 G $dot; # CYRILLIC CAPITAL LETTER GHE WITH STROKE 85 g $breve; # CYRILLIC SMALL LETTER GHE WITH MIDDLE HOOK 86 G $breve; # CYRILLIC CAPITAL LETTER GHE WITH MIDDLE HOOK 87 g ; # CYRILLIC SMALL LETTER GHE 88 G ; # CYRILLIC CAPITAL LETTER GHE 89 d; # CYRILLIC SMALL LETTER DE 90 D; # CYRILLIC CAPITAL LETTER DE 91 ; # CYRILLIC SMALL LETTER DJE 92 ; # CYRILLIC CAPITAL LETTER DJE 93 z $comma ; # CYRILLIC SMALL LETTER ZE WITH DESCENDER 94 Z $comma ; # CYRILLIC CAPITAL LETTER ZE WITH DESCENDER 95 e ; # CYRILLIC SMALL LETTER IE 96 E; # CYRILLIC CAPITAL LETTER IE 97 z $caron; # CYRILLIC SMALL LETTER ZHE 98 Z $caron; # CYRILLIC CAPITAL LETTER ZHE 99 # XXX ; # CYRILLIC SMALL LETTER ZHE WITH DESCENDER 100 # XXX ; # CYRILLIC CAPITAL LETTER ZHE WITH DESCENDER 101 z ; # CYRILLIC SMALL LETTER ZE 102 Z; # CYRILLIC CAPITAL LETTER ZE 103 j ; # CYRILLIC SMALL LETTER I 104 J ; # CYRILLIC CAPITAL LETTER I 105 i ; # CYRILLIC SMALL LETTER I 106 I ; # CYRILLIC CAPITAL LETTER I 107 k $descender ; # CYRILLIC SMALL LETTER KA WITH DESCENDER 108 K $descender ; # CYRILLIC CAPITAL LETTER KA WITH DESCENDER 109 k ; # CYRILLIC SMALL LETTER KA 110 K; # CYRILLIC CAPITAL LETTER KA 111 # XXX ; # CYRILLIC SMALL LETTER KA WITH HOOK 112 # XXX ; # CYRILLIC CAPITAL LETTER KA WITH HOOK 113 # XXX ; # CYRILLIC SMALL LETTER BASHKIR KA 114 # XXX ; # CYRILLIC CAPITAL LETTER BASHKIR KA 115 # XXX ; # CYRILLIC SMALL LETTER KA WITH STROKE 116 # XXX ; # CYRILLIC CAPITAL LETTER KA WITH STROKE 117 # XXX ; # CYRILLIC SMALL LETTER KA WITH VERTICAL STROKE 118 # XXX ; # CYRILLIC CAPITAL LETTER KA WITH VERTICAL STROKE 119 l ; # CYRILLIC SMALL LETTER EL 120 L; # CYRILLIC CAPITAL LETTER EL 121 m ; # CYRILLIC SMALL LETTER EM 122 M ; # CYRILLIC CAPITAL LETTER EM 123 n ; # CYRILLIC SMALL LETTER EN 124 N; # CYRILLIC CAPITAL LETTER EN 125 # XXX ; # CYRILLIC SMALL LETTER EN WITH DESCENDER 126 # XXX ; # CYRILLIC CAPITAL LETTER EN WITH DESCENDER 127 # XXX ; # CYRILLIC SMALL LETTER EN WITH HOOK 128 # XXX ; # CYRILLIC CAPITAL LETTER EN WITH HOOK 129 # XXX ; # CYRILLIC SMALL LIGATURE EN GHE 130 # XXX ; # CYRILLIC CAPITAL LIGATURE EN GHE 131 o ; # CYRILLIC SMALL LETTER O 132 O ; # CYRILLIC CAPITAL LETTER O 133 # XXX ; # CYRILLIC SMALL LETTER BARRED O 134 # XXX ; # CYRILLIC CAPITAL LETTER BARRED O 135 p ; # CYRILLIC SMALL LETTER PE 136 P ; # CYRILLIC CAPITAL LETTER PE 137 # XXX ; # CYRILLIC SMALL LETTER PE WITH MIDDLE HOOK 138 # XXX ; # CYRILLIC CAPITAL LETTER PE WITH MIDDLE HOOK 139 # XXX ; # CYRILLIC SMALL LETTER KOPPA 140 # XXX ; # CYRILLIC CAPITAL LETTER KOPPA 141 r ; # CYRILLIC SMALL LETTER ER 142 R ; # CYRILLIC CAPITAL LETTER ER 143 # XXX ; # CYRILLIC SMALL LETTER ER WITH TICK 144 # XXX ; # CYRILLIC CAPITAL LETTER ER WITH TICK 145 s ; # CYRILLIC SMALL LETTER ES 146 S ; # CYRILLIC CAPITAL LETTER ES 147 # XXX ; # CYRILLIC SMALL LETTER ES WITH DESCENDER 148 # XXX ; # CYRILLIC CAPITAL LETTER ES WITH DESCENDER 149 t ; # CYRILLIC SMALL LETTER TE 150 T ; # CYRILLIC CAPITAL LETTER TE 151 # XXX ; # CYRILLIC SMALL LETTER TE WITH DESCENDER 152 # XXX ; # CYRILLIC CAPITAL LETTER TE WITH DESCENDER 153 u ; # CYRILLIC SMALL LETTER U 154 U ; # CYRILLIC CAPITAL LETTER U 155 # XXX ; # CYRILLIC SMALL LETTER STRAIGHT U 156 # XXX ; # CYRILLIC CAPITAL LETTER STRAIGHT U 157 # XXX ; # CYRILLIC SMALL LETTER STRAIGHT U WITH STROKE 158 # XXX ; # CYRILLIC CAPITAL LETTER STRAIGHT U WITH STROKE 159 # XXX ; # CYRILLIC SMALL LETTER UK 160 # XXX ; # CYRILLIC CAPITAL LETTER UK 161 f ; # CYRILLIC SMALL LETTER EF 162 F ; # CYRILLIC CAPITAL LETTER EF 163 h ; # CYRILLIC SMALL LETTER HA 164 H; # CYRILLIC CAPITAL LETTER HA 165 # XXX ; # CYRILLIC SMALL LETTER HA WITH DESCENDER 166 # XXX ; # CYRILLIC CAPITAL LETTER HA WITH DESCENDER 167 # XXX ; # CYRILLIC SMALL LETTER SHHA 168 # XXX ; # CYRILLIC CAPITAL LETTER SHHA 169 # XXX ; # CYRILLIC SMALL LETTER OMEGA 170 # XXX ; # CYRILLIC CAPITAL LETTER OMEGA 171 # XXX ; # CYRILLIC SMALL LETTER OT 172 # XXX ; # CYRILLIC CAPITAL LETTER OT 173 # XXX ; # CYRILLIC SMALL LETTER OMEGA WITH TITLO 174 # XXX ; # CYRILLIC CAPITAL LETTER OMEGA WITH TITLO 175 # XXX ; # CYRILLIC SMALL LETTER ROUND OMEGA 176 # XXX ; # CYRILLIC CAPITAL LETTER ROUND OMEGA 177 c ; # CYRILLIC SMALL LETTER TSE 178 C; # CYRILLIC CAPITAL LETTER TSE 179 # XXX ; # CYRILLIC SMALL LIGATURE TE TSE 180 # XXX ; # CYRILLIC CAPITAL LIGATURE TE TSE 181 # XXX ; # CYRILLIC SMALL LETTER ABKHASIAN CHE 182 # XXX ; # CYRILLIC CAPITAL LETTER ABKHASIAN CHE 183 # XXX ; # CYRILLIC SMALL LETTER ABKHASIAN CHE WITH DESCENDER 184 # XXX ; # CYRILLIC CAPITAL LETTER ABKHASIAN CHE WITH DESCENDER 185 $modprime2 $under ; # CYRILLIC CAPITAL LETTER HARD SIGN 186 $modprime2 ; # CYRILLIC SMALL LETTER HARD SIGN 187 $modprime $under ; # CYRILLIC CAPITAL LETTER SOFT SIGN 188 $modprime ; # CYRILLIC SMALL LETTER SOFT SIGN 189 y ; # CYRILLIC SMALL LETTER YERU 190 Y ; # CYRILLIC CAPITAL LETTER YERU 191 # XXX ; # CYRILLIC SMALL LETTER SEMISOFT SIGN 192 # XXX ; # CYRILLIC CAPITAL LETTER SEMISOFT SIGN 193 # XXX ; # CYRILLIC SMALL LETTER YAT 194 # XXX ; # CYRILLIC CAPITAL LETTER YAT 195 # XXX ; # CYRILLIC SMALL LETTER IOTIFIED E 196 # XXX ; # CYRILLIC CAPITAL LETTER IOTIFIED E 197 # XXX ; # CYRILLIC SMALL LETTER LITTLE YUS 198 # XXX ; # CYRILLIC CAPITAL LETTER LITTLE YUS 199 # XXX ; # CYRILLIC SMALL LETTER BIG YUS 200 # XXX ; # CYRILLIC CAPITAL LETTER BIG YUS 201 # XXX ; # CYRILLIC SMALL LETTER IOTIFIED LITTLE YUS 202 # XXX ; # CYRILLIC CAPITAL LETTER IOTIFIED LITTLE YUS 203 # XXX ; # CYRILLIC SMALL LETTER IOTIFIED BIG YUS 204 # XXX ; # CYRILLIC CAPITAL LETTER IOTIFIED BIG YUS 205 # XXX ; # CYRILLIC SMALL LETTER KSI 206 # XXX ; # CYRILLIC CAPITAL LETTER KSI 207 # XXX ; # CYRILLIC SMALL LETTER PSI 208 # XXX ; # CYRILLIC CAPITAL LETTER PSI 209 # XXX ; # CYRILLIC SMALL LETTER FITA 210 # XXX ; # CYRILLIC CAPITAL LETTER FITA 211 # XXX ; # CYRILLIC SMALL LETTER IZHITSA 212 # XXX ; # CYRILLIC CAPITAL LETTER IZHITSA 213 # XXX ; # CYRILLIC SMALL LETTER ABKHASIAN HA 214 # XXX ; # CYRILLIC CAPITAL LETTER ABKHASIAN HA 215 # XXX ; # CYRILLIC LETTER PALOCHKA 216 ### XXX ; # CYRILLIC SMALL LETTER A 217 ### XXX ; # CYRILLIC CAPITAL LETTER A 218 ### XXX ; # CYRILLIC SMALL LETTER A 219 ### XXX ; # CYRILLIC CAPITAL LETTER A 220 ### XXX ; # CYRILLIC SMALL LETTER SCHWA 221 ### XXX ; # CYRILLIC CAPITAL LETTER SCHWA 222 ### XXX ; # CYRILLIC SMALL LETTER GHE 223 ### XXX ; # CYRILLIC CAPITAL LETTER GHE 224 ### XXX ; # CYRILLIC SMALL LETTER IE 225 ### XXX ; # CYRILLIC CAPITAL LETTER IE 226 ### XXX ; # CYRILLIC SMALL LETTER IE 227 ### XXX ; # CYRILLIC CAPITAL LETTER IE 228 ### XXX ; # CYRILLIC SMALL LETTER IE 229 ### XXX ; # CYRILLIC CAPITAL LETTER IE 230 ### XXX ; # CYRILLIC SMALL LETTER ZHE 231 ### XXX ; # CYRILLIC CAPITAL LETTER ZHE 232 ### XXX ; # CYRILLIC SMALL LETTER ZHE 233 ### XXX ; # CYRILLIC CAPITAL LETTER ZHE 234 ### XXX ; # CYRILLIC SMALL LETTER ZE 235 ### XXX ; # CYRILLIC CAPITAL LETTER ZE 236 ### XXX ; # CYRILLIC SMALL LETTER I 237 ### XXX ; # CYRILLIC CAPITAL LETTER I 238 ### XXX ; # CYRILLIC SMALL LETTER I 239 ### XXX ; # CYRILLIC CAPITAL LETTER I 240 ### XXX ; # CYRILLIC SMALL LETTER I 241 ### XXX ; # CYRILLIC CAPITAL LETTER I 242 ### XXX ; # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I 243 ### XXX ; # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I 244 ### XXX ; # CYRILLIC SMALL LETTER O 245 ### XXX ; # CYRILLIC CAPITAL LETTER O 246 ### XXX ; # CYRILLIC SMALL LETTER BARRED O 247 ### XXX ; # CYRILLIC CAPITAL LETTER BARRED O 248 ### XXX ; # CYRILLIC SMALL LETTER KA 249 ### XXX ; # CYRILLIC CAPITAL LETTER KA 250 ### XXX ; # CYRILLIC SMALL LETTER U 251 ### XXX ; # CYRILLIC CAPITAL LETTER U 252 ### XXX ; # CYRILLIC SMALL LETTER U 253 ### XXX ; # CYRILLIC CAPITAL LETTER U 254 ### XXX ; # CYRILLIC SMALL LETTER U 255 ### XXX ; # CYRILLIC CAPITAL LETTER U 256 ### XXX ; # CYRILLIC SMALL LETTER U 257 ### XXX ; # CYRILLIC CAPITAL LETTER U 258 ### XXX ; # CYRILLIC SMALL LETTER CHE 259 ### XXX ; # CYRILLIC CAPITAL LETTER CHE 260 ### XXX ; # CYRILLIC SMALL LETTER YERU 261 ### XXX ; # CYRILLIC CAPITAL LETTER YERU 262 ### XXX ; # CYRILLIC SMALL LETTER E 263 ### XXX ; # CYRILLIC CAPITAL LETTER E 264 ### XXX ; # CYRILLIC SMALL LETTER IZHITSA 265 ### XXX ; # CYRILLIC CAPITAL LETTER IZHITSA 266 # Completeness 267 $ignore = [[:Mark:]''] * ; 268 | k q ; 269 | K Q ; 270 | u w ; 271 | U W ; 272 | KS X } $ignore [:UppercaseLetter:] ; 273 | KS [:UppercaseLetter:] $ignore { X ; 274 | Ks X ; 275 | ks x ; 276 :: NFC (NFD) ; 277 # note: a global filter is more efficient, but MUST include all source chars!! 278 # :: ([\u0000-\u007E [:Cyrillic:] [:Latin:] [:nonspacing mark:]]); 279 # MINIMAL FILTER: Latin-Cyrillic 280 :: ( [A-Za-z----------------------------------------------------------------------] ) ; 281 </tRule> 282 </transform> 283 </transforms> 284 </supplementalData> 285
