projects / apple / icu.git / blame
| Commit | Line | Data |
|---|---|---|
| 374ca955 A |
1 | #-------------------------------------------------------------------- |
| 2 | # Copyright (c) 1999-2004, International Business Machines | |
| 3 | # Corporation and others. All Rights Reserved. | |
| 4 | #-------------------------------------------------------------------- | |
| 5 | # TODO: add remaining characters | |
| 6 | # Should add variants for Russian-English, Russian-German | |
| 7 | # Those can use this as a base, and then remap cases | |
| 8 | # like a $hat to ya or ja. | |
| 9 | ||
| 10 | # :: [\u0000-\u007E \u02B9 \u02BA [:Cyrillic:] [:Latin:] [:nonspacing mark:]] ; | |
| 11 | ### WARNING, \u0308 must be added to the generated filters, in both directions ### | |
| 12 | # MINIMAL FILTER | |
| 13 | :: [\u0308\u0102-\u0103\u0114-\u0115\u011E-\u011F\u012C-\u012D\u014E-\u014F\u016C-\u016D\u0306\u0400-\u045F\u0490-\u0495\u0498-\u0499\u04C1-\u04C2\u04D0-\u04DF\u04E2-\u04E7\u04EC-\u04F5\u04F8-\u04F9\u1E1C-\u1E1D\u1EAE-\u1EB7\u1FB0\u1FB8\u1FD0\u1FD8\u1FE0\u1FE8] ; | |
| 14 | :: NFD (NFC) ; | |
| 15 | ||
| 16 | $modprime = \u02B9; | |
| 17 | $modprime2 = \u02BA; | |
| 18 | ||
| 19 | $grave = \u0300; | |
| 20 | $acute = \u0301; | |
| 21 | $hat = \u0302; | |
| 22 | $breve = \u0306 ; | |
| 23 | $dot = \u0307 ; | |
| 24 | $caron = \u030C ; | |
| 25 | $comma = \u0326 ; | |
| 26 | $under = \u0331 ; | |
| 27 | ||
| 28 | # move up so not masked | |
| 29 | ||
| 30 | я <> a $hat ; # CYRILLIC SMALL LETTER YA | |
| 31 | Я <> A $hat ; # CYRILLIC CAPITAL LETTER YA | |
| 32 | ||
| 33 | ч <> c $caron ; # CYRILLIC SMALL LETTER CHE | |
| 34 | Ч <> C $caron; # CYRILLIC CAPITAL LETTER CHE | |
| 35 | # ҷ <> XXX ; # CYRILLIC SMALL LETTER CHE WITH DESCENDER | |
| 36 | # Ҷ <> XXX ; # CYRILLIC CAPITAL LETTER CHE WITH DESCENDER | |
| 37 | # ӌ <> XXX ; # CYRILLIC SMALL LETTER KHAKASSIAN CHE | |
| 38 | # Ӌ <> XXX ; # CYRILLIC CAPITAL LETTER KHAKASSIAN CHE | |
| 39 | # ҹ <> XXX ; # CYRILLIC SMALL LETTER CHE WITH VERTICAL STROKE | |
| 40 | # Ҹ <> XXX ; # CYRILLIC CAPITAL LETTER CHE WITH VERTICAL STROKE | |
| 41 | ||
| 42 | э <> e $acute; # CYRILLIC SMALL LETTER E | |
| 43 | Э <> E $acute; # CYRILLIC CAPITAL LETTER E | |
| 44 | є <> e $hat; # CYRILLIC SMALL LETTER UKRAINIAN IE | |
| 45 | Є <> E $hat; # CYRILLIC CAPITAL LETTER UKRAINIAN IE | |
| 46 | ||
| 47 | ш <> s $caron ; # CYRILLIC SMALL LETTER SHA | |
| 48 | Ш <> S $caron ; # CYRILLIC CAPITAL LETTER SHA | |
| 49 | щ <> s $hat ; # CYRILLIC SMALL LETTER SHCHA | |
| 50 | Щ <> S $hat; # CYRILLIC CAPITAL LETTER SHCHA | |
| 51 | ||
| 52 | ѕ <> z $hat ; # CYRILLIC SMALL LETTER DZE | |
| 53 | Ѕ <> Z $hat; # CYRILLIC CAPITAL LETTER DZE | |
| 54 | # ӡ <> XXX ; # CYRILLIC SMALL LETTER ABKHASIAN DZE | |
| 55 | # Ӡ <> XXX ; # CYRILLIC CAPITAL LETTER ABKHASIAN DZE | |
| 56 | ||
| 57 | ю <> u $hat ; # CYRILLIC SMALL LETTER YU | |
| 58 | Ю <> U $hat ; # CYRILLIC CAPITAL LETTER YU | |
| 59 | ||
| 60 | і <> i $acute; # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I | |
| 61 | І <> I $acute; # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I | |
| 62 | ј <> j $caron; # CYRILLIC SMALL LETTER JE | |
| 63 | Ј <> J $caron; # CYRILLIC CAPITAL LETTER JE | |
| 64 | ||
| 65 | љ <> l $hat ; # CYRILLIC SMALL LETTER LJE | |
| 66 | Љ <> L $hat ; # CYRILLIC CAPITAL LETTER LJE | |
| 67 | њ <> n $hat ; # CYRILLIC SMALL LETTER NJE | |
| 68 | Њ <> N $hat ; # CYRILLIC CAPITAL LETTER NJE | |
| 69 | ||
| 70 | ћ <> c $acute ; # CYRILLIC SMALL LETTER TSHE | |
| 71 | Ћ <> C $acute ; # CYRILLIC CAPITAL LETTER TSHE | |
| 72 | ||
| 73 | џ <> d $hat ; # CYRILLIC SMALL LETTER DZHE | |
| 74 | Џ <> D $hat ; # CYRILLIC CAPITAL LETTER DZHE | |
| 75 | ||
| 76 | # Normal order | |
| 77 | ||
| 78 | а <> a ; # CYRILLIC SMALL LETTER A | |
| 79 | А <> A ; # CYRILLIC CAPITAL LETTER A | |
| 80 | ә <> \u0259 ; # CYRILLIC SMALL LETTER SCHWA | |
| 81 | Ә <> \u018F ; # CYRILLIC CAPITAL LETTER SCHWA | |
| 82 | ӕ <> \u00E6 ; # CYRILLIC SMALL LIGATURE A IE | |
| 83 | Ӕ <> \u00C6 ; # CYRILLIC CAPITAL LIGATURE A IE | |
| 84 | б <> b ; # CYRILLIC SMALL LETTER BE | |
| 85 | Б <> B ; # CYRILLIC CAPITAL LETTER BE | |
| 86 | в <> v ; # CYRILLIC SMALL LETTER VE | |
| 87 | В <> V ; # CYRILLIC CAPITAL LETTER VE | |
| 88 | ||
| 89 | ґ <> g $grave ; # CYRILLIC SMALL LETTER GHE WITH UPTURN | |
| 90 | Ґ <> G $grave ; # CYRILLIC CAPITAL LETTER GHE WITH UPTURN | |
| 91 | ғ <> g $dot ; # CYRILLIC SMALL LETTER GHE WITH STROKE | |
| 92 | Ғ <> G $dot; # CYRILLIC CAPITAL LETTER GHE WITH STROKE | |
| 93 | ҕ <> g $breve; # CYRILLIC SMALL LETTER GHE WITH MIDDLE HOOK | |
| 94 | Ҕ <> G $breve; # CYRILLIC CAPITAL LETTER GHE WITH MIDDLE HOOK | |
| 95 | г <> g ; # CYRILLIC SMALL LETTER GHE | |
| 96 | Г <> G ; # CYRILLIC CAPITAL LETTER GHE | |
| 97 | ||
| 98 | д <> d; # CYRILLIC SMALL LETTER DE | |
| 99 | Д <> D; # CYRILLIC CAPITAL LETTER DE | |
| 100 | ђ <> đ ; # CYRILLIC SMALL LETTER DJE | |
| 101 | Ђ <> Đ ; # CYRILLIC CAPITAL LETTER DJE | |
| 102 | ҙ <> z $comma ; # CYRILLIC SMALL LETTER ZE WITH DESCENDER | |
| 103 | Ҙ <> Z $comma ; # CYRILLIC CAPITAL LETTER ZE WITH DESCENDER | |
| 104 | е <> e ; # CYRILLIC SMALL LETTER IE | |
| 105 | Е <> E; # CYRILLIC CAPITAL LETTER IE | |
| 106 | ||
| 107 | ж <> z $caron; # CYRILLIC SMALL LETTER ZHE | |
| 108 | Ж <> Z $caron; # CYRILLIC CAPITAL LETTER ZHE | |
| 109 | ||
| 110 | # җ <> XXX ; # CYRILLIC SMALL LETTER ZHE WITH DESCENDER | |
| 111 | # Җ <> XXX ; # CYRILLIC CAPITAL LETTER ZHE WITH DESCENDER | |
| 112 | ||
| 113 | з <> z ; # CYRILLIC SMALL LETTER ZE | |
| 114 | З <> Z; # CYRILLIC CAPITAL LETTER ZE | |
| 115 | ||
| 116 | й <> j ; # CYRILLIC SMALL LETTER I | |
| 117 | Й <> J ; # CYRILLIC CAPITAL LETTER I | |
| 118 | и <> i ; # CYRILLIC SMALL LETTER I | |
| 119 | И <> I ; # CYRILLIC CAPITAL LETTER I | |
| 120 | ||
| 121 | к <> k ; # CYRILLIC SMALL LETTER KA | |
| 122 | К <> K; # CYRILLIC CAPITAL LETTER KA | |
| 123 | ||
| 124 | # қ <> XXX ; # CYRILLIC SMALL LETTER KA WITH DESCENDER | |
| 125 | # Қ <> XXX ; # CYRILLIC CAPITAL LETTER KA WITH DESCENDER | |
| 126 | # ӄ <> XXX ; # CYRILLIC SMALL LETTER KA WITH HOOK | |
| 127 | # Ӄ <> XXX ; # CYRILLIC CAPITAL LETTER KA WITH HOOK | |
| 128 | # ҡ <> XXX ; # CYRILLIC SMALL LETTER BASHKIR KA | |
| 129 | # Ҡ <> XXX ; # CYRILLIC CAPITAL LETTER BASHKIR KA | |
| 130 | # ҟ <> XXX ; # CYRILLIC SMALL LETTER KA WITH STROKE | |
| 131 | # Ҟ <> XXX ; # CYRILLIC CAPITAL LETTER KA WITH STROKE | |
| 132 | # ҝ <> XXX ; # CYRILLIC SMALL LETTER KA WITH VERTICAL STROKE | |
| 133 | # Ҝ <> XXX ; # CYRILLIC CAPITAL LETTER KA WITH VERTICAL STROKE | |
| 134 | л <> l ; # CYRILLIC SMALL LETTER EL | |
| 135 | Л <> L; # CYRILLIC CAPITAL LETTER EL | |
| 136 | ||
| 137 | м <> m ; # CYRILLIC SMALL LETTER EM | |
| 138 | М <> M ; # CYRILLIC CAPITAL LETTER EM | |
| 139 | н <> n ; # CYRILLIC SMALL LETTER EN | |
| 140 | Н <> N; # CYRILLIC CAPITAL LETTER EN | |
| 141 | # ң <> XXX ; # CYRILLIC SMALL LETTER EN WITH DESCENDER | |
| 142 | # Ң <> XXX ; # CYRILLIC CAPITAL LETTER EN WITH DESCENDER | |
| 143 | # ӈ <> XXX ; # CYRILLIC SMALL LETTER EN WITH HOOK | |
| 144 | # Ӈ <> XXX ; # CYRILLIC CAPITAL LETTER EN WITH HOOK | |
| 145 | # ҥ <> XXX ; # CYRILLIC SMALL LIGATURE EN GHE | |
| 146 | # Ҥ <> XXX ; # CYRILLIC CAPITAL LIGATURE EN GHE | |
| 147 | ||
| 148 | о <> o ; # CYRILLIC SMALL LETTER O | |
| 149 | О <> O ; # CYRILLIC CAPITAL LETTER O | |
| 150 | # ө <> XXX ; # CYRILLIC SMALL LETTER BARRED O | |
| 151 | # Ө <> XXX ; # CYRILLIC CAPITAL LETTER BARRED O | |
| 152 | п <> p ; # CYRILLIC SMALL LETTER PE | |
| 153 | П <> P ; # CYRILLIC CAPITAL LETTER PE | |
| 154 | # ҧ <> XXX ; # CYRILLIC SMALL LETTER PE WITH MIDDLE HOOK | |
| 155 | # Ҧ <> XXX ; # CYRILLIC CAPITAL LETTER PE WITH MIDDLE HOOK | |
| 156 | # ҁ <> XXX ; # CYRILLIC SMALL LETTER KOPPA | |
| 157 | # Ҁ <> XXX ; # CYRILLIC CAPITAL LETTER KOPPA | |
| 158 | р <> r ; # CYRILLIC SMALL LETTER ER | |
| 159 | Р <> R ; # CYRILLIC CAPITAL LETTER ER | |
| 160 | # ҏ <> XXX ; # CYRILLIC SMALL LETTER ER WITH TICK | |
| 161 | # Ҏ <> XXX ; # CYRILLIC CAPITAL LETTER ER WITH TICK | |
| 162 | с <> s ; # CYRILLIC SMALL LETTER ES | |
| 163 | С <> S ; # CYRILLIC CAPITAL LETTER ES | |
| 164 | # ҫ <> XXX ; # CYRILLIC SMALL LETTER ES WITH DESCENDER | |
| 165 | # Ҫ <> XXX ; # CYRILLIC CAPITAL LETTER ES WITH DESCENDER | |
| 166 | т <> t ; # CYRILLIC SMALL LETTER TE | |
| 167 | Т <> T ; # CYRILLIC CAPITAL LETTER TE | |
| 168 | # ҭ <> XXX ; # CYRILLIC SMALL LETTER TE WITH DESCENDER | |
| 169 | # Ҭ <> XXX ; # CYRILLIC CAPITAL LETTER TE WITH DESCENDER | |
| 170 | ||
| 171 | у <> u ; # CYRILLIC SMALL LETTER U | |
| 172 | У <> U ; # CYRILLIC CAPITAL LETTER U | |
| 173 | # ү <> XXX ; # CYRILLIC SMALL LETTER STRAIGHT U | |
| 174 | # Ү <> XXX ; # CYRILLIC CAPITAL LETTER STRAIGHT U | |
| 175 | # ұ <> XXX ; # CYRILLIC SMALL LETTER STRAIGHT U WITH STROKE | |
| 176 | # Ұ <> XXX ; # CYRILLIC CAPITAL LETTER STRAIGHT U WITH STROKE | |
| 177 | # ѹ <> XXX ; # CYRILLIC SMALL LETTER UK | |
| 178 | # Ѹ <> XXX ; # CYRILLIC CAPITAL LETTER UK | |
| 179 | ф <> f ; # CYRILLIC SMALL LETTER EF | |
| 180 | Ф <> F ; # CYRILLIC CAPITAL LETTER EF | |
| 181 | х <> h ; # CYRILLIC SMALL LETTER HA | |
| 182 | Х <> H; # CYRILLIC CAPITAL LETTER HA | |
| 183 | # ҳ <> XXX ; # CYRILLIC SMALL LETTER HA WITH DESCENDER | |
| 184 | # Ҳ <> XXX ; # CYRILLIC CAPITAL LETTER HA WITH DESCENDER | |
| 185 | # һ <> XXX ; # CYRILLIC SMALL LETTER SHHA | |
| 186 | # Һ <> XXX ; # CYRILLIC CAPITAL LETTER SHHA | |
| 187 | # ѡ <> XXX ; # CYRILLIC SMALL LETTER OMEGA | |
| 188 | # Ѡ <> XXX ; # CYRILLIC CAPITAL LETTER OMEGA | |
| 189 | # ѿ <> XXX ; # CYRILLIC SMALL LETTER OT | |
| 190 | # Ѿ <> XXX ; # CYRILLIC CAPITAL LETTER OT | |
| 191 | # ѽ <> XXX ; # CYRILLIC SMALL LETTER OMEGA WITH TITLO | |
| 192 | # Ѽ <> XXX ; # CYRILLIC CAPITAL LETTER OMEGA WITH TITLO | |
| 193 | # ѻ <> XXX ; # CYRILLIC SMALL LETTER ROUND OMEGA | |
| 194 | # Ѻ <> XXX ; # CYRILLIC CAPITAL LETTER ROUND OMEGA | |
| 195 | ц <> c ; # CYRILLIC SMALL LETTER TSE | |
| 196 | Ц <> C; # CYRILLIC CAPITAL LETTER TSE | |
| 197 | # ҵ <> XXX ; # CYRILLIC SMALL LIGATURE TE TSE | |
| 198 | # Ҵ <> XXX ; # CYRILLIC CAPITAL LIGATURE TE TSE | |
| 199 | ||
| 200 | # ҽ <> XXX ; # CYRILLIC SMALL LETTER ABKHASIAN CHE | |
| 201 | # Ҽ <> XXX ; # CYRILLIC CAPITAL LETTER ABKHASIAN CHE | |
| 202 | # ҿ <> XXX ; # CYRILLIC SMALL LETTER ABKHASIAN CHE WITH DESCENDER | |
| 203 | # Ҿ <> XXX ; # CYRILLIC CAPITAL LETTER ABKHASIAN CHE WITH DESCENDER | |
| 204 | ||
| 205 | ||
| 206 | Ъ <> $modprime2 $under ; # CYRILLIC CAPITAL LETTER HARD SIGN | |
| 207 | ъ <> $modprime2 ; # CYRILLIC SMALL LETTER HARD SIGN | |
| 208 | Ь <> $modprime $under ; # CYRILLIC CAPITAL LETTER SOFT SIGN | |
| 209 | ь <> $modprime ; # CYRILLIC SMALL LETTER SOFT SIGN | |
| 210 | ||
| 211 | ы <> y ; # CYRILLIC SMALL LETTER YERU | |
| 212 | Ы <> Y ; # CYRILLIC CAPITAL LETTER YERU | |
| 213 | ||
| 214 | # ҍ <> XXX ; # CYRILLIC SMALL LETTER SEMISOFT SIGN | |
| 215 | # Ҍ <> XXX ; # CYRILLIC CAPITAL LETTER SEMISOFT SIGN | |
| 216 | # ѣ <> XXX ; # CYRILLIC SMALL LETTER YAT | |
| 217 | # Ѣ <> XXX ; # CYRILLIC CAPITAL LETTER YAT | |
| 218 | ||
| 219 | # ѥ <> XXX ; # CYRILLIC SMALL LETTER IOTIFIED E | |
| 220 | # Ѥ <> XXX ; # CYRILLIC CAPITAL LETTER IOTIFIED E | |
| 221 | # ѧ <> XXX ; # CYRILLIC SMALL LETTER LITTLE YUS | |
| 222 | # Ѧ <> XXX ; # CYRILLIC CAPITAL LETTER LITTLE YUS | |
| 223 | # ѫ <> XXX ; # CYRILLIC SMALL LETTER BIG YUS | |
| 224 | # Ѫ <> XXX ; # CYRILLIC CAPITAL LETTER BIG YUS | |
| 225 | # ѩ <> XXX ; # CYRILLIC SMALL LETTER IOTIFIED LITTLE YUS | |
| 226 | # Ѩ <> XXX ; # CYRILLIC CAPITAL LETTER IOTIFIED LITTLE YUS | |
| 227 | # ѭ <> XXX ; # CYRILLIC SMALL LETTER IOTIFIED BIG YUS | |
| 228 | # Ѭ <> XXX ; # CYRILLIC CAPITAL LETTER IOTIFIED BIG YUS | |
| 229 | # ѯ <> XXX ; # CYRILLIC SMALL LETTER KSI | |
| 230 | # Ѯ <> XXX ; # CYRILLIC CAPITAL LETTER KSI | |
| 231 | # ѱ <> XXX ; # CYRILLIC SMALL LETTER PSI | |
| 232 | # Ѱ <> XXX ; # CYRILLIC CAPITAL LETTER PSI | |
| 233 | # ѳ <> XXX ; # CYRILLIC SMALL LETTER FITA | |
| 234 | # Ѳ <> XXX ; # CYRILLIC CAPITAL LETTER FITA | |
| 235 | # ѵ <> XXX ; # CYRILLIC SMALL LETTER IZHITSA | |
| 236 | # Ѵ <> XXX ; # CYRILLIC CAPITAL LETTER IZHITSA | |
| 237 | # ҩ <> XXX ; # CYRILLIC SMALL LETTER ABKHASIAN HA | |
| 238 | # Ҩ <> XXX ; # CYRILLIC CAPITAL LETTER ABKHASIAN HA | |
| 239 | # Ӏ <> XXX ; # CYRILLIC LETTER PALOCHKA | |
| 240 | ### ӑ <> XXX ; # CYRILLIC SMALL LETTER A | |
| 241 | ### Ӑ <> XXX ; # CYRILLIC CAPITAL LETTER A | |
| 242 | ### ӓ <> XXX ; # CYRILLIC SMALL LETTER A | |
| 243 | ### Ӓ <> XXX ; # CYRILLIC CAPITAL LETTER A | |
| 244 | ### ӛ <> XXX ; # CYRILLIC SMALL LETTER SCHWA | |
| 245 | ### Ӛ <> XXX ; # CYRILLIC CAPITAL LETTER SCHWA | |
| 246 | ### ѓ <> XXX ; # CYRILLIC SMALL LETTER GHE | |
| 247 | ### Ѓ <> XXX ; # CYRILLIC CAPITAL LETTER GHE | |
| 248 | ### ѐ <> XXX ; # CYRILLIC SMALL LETTER IE | |
| 249 | ### Ѐ <> XXX ; # CYRILLIC CAPITAL LETTER IE | |
| 250 | ### ё <> XXX ; # CYRILLIC SMALL LETTER IE | |
| 251 | ### Ё <> XXX ; # CYRILLIC CAPITAL LETTER IE | |
| 252 | ### ӗ <> XXX ; # CYRILLIC SMALL LETTER IE | |
| 253 | ### Ӗ <> XXX ; # CYRILLIC CAPITAL LETTER IE | |
| 254 | ### ӂ <> XXX ; # CYRILLIC SMALL LETTER ZHE | |
| 255 | ### Ӂ <> XXX ; # CYRILLIC CAPITAL LETTER ZHE | |
| 256 | ### ӝ <> XXX ; # CYRILLIC SMALL LETTER ZHE | |
| 257 | ### Ӝ <> XXX ; # CYRILLIC CAPITAL LETTER ZHE | |
| 258 | ### ӟ <> XXX ; # CYRILLIC SMALL LETTER ZE | |
| 259 | ### Ӟ <> XXX ; # CYRILLIC CAPITAL LETTER ZE | |
| 260 | ### ѝ <> XXX ; # CYRILLIC SMALL LETTER I | |
| 261 | ### Ѝ <> XXX ; # CYRILLIC CAPITAL LETTER I | |
| 262 | ### ӣ <> XXX ; # CYRILLIC SMALL LETTER I | |
| 263 | ### Ӣ <> XXX ; # CYRILLIC CAPITAL LETTER I | |
| 264 | ### ӥ <> XXX ; # CYRILLIC SMALL LETTER I | |
| 265 | ### Ӥ <> XXX ; # CYRILLIC CAPITAL LETTER I | |
| 266 | ### ї <> XXX ; # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I | |
| 267 | ### Ї <> XXX ; # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I | |
| 268 | ### ӧ <> XXX ; # CYRILLIC SMALL LETTER O | |
| 269 | ### Ӧ <> XXX ; # CYRILLIC CAPITAL LETTER O | |
| 270 | ### ӫ <> XXX ; # CYRILLIC SMALL LETTER BARRED O | |
| 271 | ### Ӫ <> XXX ; # CYRILLIC CAPITAL LETTER BARRED O | |
| 272 | ### ќ <> XXX ; # CYRILLIC SMALL LETTER KA | |
| 273 | ### Ќ <> XXX ; # CYRILLIC CAPITAL LETTER KA | |
| 274 | ### ӯ <> XXX ; # CYRILLIC SMALL LETTER U | |
| 275 | ### Ӯ <> XXX ; # CYRILLIC CAPITAL LETTER U | |
| 276 | ### ў <> XXX ; # CYRILLIC SMALL LETTER U | |
| 277 | ### Ў <> XXX ; # CYRILLIC CAPITAL LETTER U | |
| 278 | ### ӱ <> XXX ; # CYRILLIC SMALL LETTER U | |
| 279 | ### Ӱ <> XXX ; # CYRILLIC CAPITAL LETTER U | |
| 280 | ### ӳ <> XXX ; # CYRILLIC SMALL LETTER U | |
| 281 | ### Ӳ <> XXX ; # CYRILLIC CAPITAL LETTER U | |
| 282 | ### ӵ <> XXX ; # CYRILLIC SMALL LETTER CHE | |
| 283 | ### Ӵ <> XXX ; # CYRILLIC CAPITAL LETTER CHE | |
| 284 | ### ӹ <> XXX ; # CYRILLIC SMALL LETTER YERU | |
| 285 | ### Ӹ <> XXX ; # CYRILLIC CAPITAL LETTER YERU | |
| 286 | ### ӭ <> XXX ; # CYRILLIC SMALL LETTER E | |
| 287 | ### Ӭ <> XXX ; # CYRILLIC CAPITAL LETTER E | |
| 288 | ### ѷ <> XXX ; # CYRILLIC SMALL LETTER IZHITSA | |
| 289 | ### Ѷ <> XXX ; # CYRILLIC CAPITAL LETTER IZHITSA | |
| 290 | ||
| 291 | # Completeness | |
| 292 | $ignore = [[:Mark:]''] * ; | |
| 293 | | k < q ; | |
| 294 | | K < Q ; | |
| 295 | | u < w ; | |
| 296 | | U < W ; | |
| 297 | | KS < X } $ignore [:UppercaseLetter:] ; | |
| 298 | | KS < [:UppercaseLetter:] $ignore { X ; | |
| 299 | | Ks < X ; | |
| 300 | | ks < x ; | |
| 301 | ||
| 302 | :: NFC (NFD) ; | |
| 303 | # note: a global filter is more efficient, but MUST include all source chars!! | |
| 304 | # :: ([\u0000-\u007E \u02B9 \u02BA [:Cyrillic:] [:Latin:] [:nonspacing mark:]]); | |
| 305 | # MINIMAL FILTER: Latin-Cyrillic | |
| 306 | :: ( [\u0308A-Za-z\u00C0-\u00CF\u00D1-\u00D6\u00D9-\u00DD\u00E0-\u00EF\u00F1-\u00F6\u00F9-\u00FD\u00FF-\u0125\u0128-\u0130\u0134-\u0137\u0139-\u013E\u0143-\u0148\u014C-\u0151\u0154-\u0165\u0168-\u017E\u018F\u01A0-\u01A1\u01AF-\u01B0\u01CD-\u01DC\u01DE-\u01E3\u01E6-\u01F0\u01F4-\u01F5\u01F8-\u021B\u021E-\u021F\u0226-\u0233\u0259\u02B9-\u02BA\u0300-\u0302\u0306-\u0307\u030C\u0326\u0331\u0340-\u0341\u0344\u0374\u0385-\u0386\u0388-\u038A\u038C\u038E-\u0390\u03AC-\u03B0\u03CC-\u03CE\u03D3\u0400\u0403\u040C-\u040E\u0419\u0439\u0450\u0453\u045C-\u045E\u04C1-\u04C2\u04D0-\u04D1\u04D6-\u04D7\u1E00-\u1E99\u1E9B\u1EA0-\u1EF9\u1F02-\u1F05\u1F0A-\u1F0D\u1F12-\u1F15\u1F1A-\u1F1D\u1F22-\u1F25\u1F2A-\u1F2D\u1F32-\u1F35\u1F3A-\u1F3D\u1F42-\u1F45\u1F4A-\u1F4D\u1F52-\u1F55\u1F5B\u1F5D\u1F62-\u1F65\u1F6A-\u1F6D\u1F70-\u1F7D\u1F82-\u1F85\u1F8A-\u1F8D\u1F92-\u1F95\u1F9A-\u1F9D\u1FA2-\u1FA5\u1FAA-\u1FAD\u1FB0\u1FB2\u1FB4\u1FB8\u1FBA-\u1FBB\u1FC2\u1FC4\u1FC8-\u1FCB\u1FCD-\u1FCE\u1FD0\u1FD2-\u1FD3\u1FD8\u1FDA-\u1FDB\u1FDD-\u1FDE\u1FE0\u1FE2-\u1FE3\u1FE8\u1FEA-\u1FEB\u1FED-\u1FEE\u1FF2\u1FF4\u1FF8-\u1FFB\u212A-\u212B] ) ; |