lib/unicore/In.pl

   1 # !!!!!!!   DO NOT EDIT THIS FILE   !!!!!!!
   2 # This file is built by mktables.PL from e.g. Unicode.txt.
   3 # Any changes made here will be lost!
   4 %utf8::In = (
   5 'LATIN'                                       =>   0,
   6 'GREEK'                                       =>   1,
   7 'CYRILLIC'                                    =>   2,
   8 'ARMENIAN'                                    =>   3,
   9 'HEBREW'                                      =>   4,
  10 'ARABIC'                                      =>   5,
  11 'SYRIAC'                                      =>   6,
  12 'THAANA'                                      =>   7,
  13 'DEVANAGARI'                                  =>   8,
  14 'BENGALI'                                     =>   9,
  15 'GURMUKHI'                                    =>  10,
  16 'GUJARATI'                                    =>  11,
  17 'ORIYA'                                       =>  12,
  18 'TAMIL'                                       =>  13,
  19 'TELUGU'                                      =>  14,
  20 'KANNADA'                                     =>  15,
  21 'MALAYALAM'                                   =>  16,
  22 'SINHALA'                                     =>  17,
  23 'THAI'                                        =>  18,
  24 'LAO'                                         =>  19,
  25 'TIBETAN'                                     =>  20,
  26 'MYANMAR'                                     =>  21,
  27 'GEORGIAN'                                    =>  22,
  28 'HANGUL'                                      =>  23,
  29 'ETHIOPIC'                                    =>  24,
  30 'CHEROKEE'                                    =>  25,
  31 'CANADIAN-ABORIGINAL'                         =>  26,
  32 'OGHAM'                                       =>  27,
  33 'RUNIC'                                       =>  28,
  34 'KHMER'                                       =>  29,
  35 'MONGOLIAN'                                   =>  30,
  36 'HIRAGANA'                                    =>  31,
  37 'KATAKANA'                                    =>  32,
  38 'BOPOMOFO'                                    =>  33,
  39 'HAN'                                         =>  34,
  40 'YI'                                          =>  35,
  41 'OLD-ITALIC'                                  =>  36,
  42 'GOTHIC'                                      =>  37,
  43 'DESERET'                                     =>  38,
  44 'INHERITED'                                   =>  39,
  45 'Basic Latin'                                 =>  40,
  46 'Latin-1 Supplement'                          =>  41,
  47 'Latin Extended-A'                            =>  42,
  48 'Latin Extended-B'                            =>  43,
  49 'IPA Extensions'                              =>  44,
  50 'Spacing Modifier Letters'                    =>  45,
  51 'Combining Diacritical Marks'                 =>  46,
  52 'Greek Block'                                 =>  47,
  53 'Cyrillic Block'                              =>  48,
  54 'Armenian Block'                              =>  49,
  55 'Hebrew Block'                                =>  50,
  56 'Arabic Block'                                =>  51,
  57 'Syriac Block'                                =>  52,
  58 'Thaana Block'                                =>  53,
  59 'Devanagari Block'                            =>  54,
  60 'Bengali Block'                               =>  55,
  61 'Gurmukhi Block'                              =>  56,
  62 'Gujarati Block'                              =>  57,
  63 'Oriya Block'                                 =>  58,
  64 'Tamil Block'                                 =>  59,
  65 'Telugu Block'                                =>  60,
  66 'Kannada Block'                               =>  61,
  67 'Malayalam Block'                             =>  62,
  68 'Sinhala Block'                               =>  63,
  69 'Thai Block'                                  =>  64,
  70 'Lao Block'                                   =>  65,
  71 'Tibetan Block'                               =>  66,
  72 'Myanmar Block'                               =>  67,
  73 'Georgian Block'                              =>  68,
  74 'Hangul Jamo'                                 =>  69,
  75 'Ethiopic Block'                              =>  70,
  76 'Cherokee Block'                              =>  71,
  77 'Unified Canadian Aboriginal Syllabics'       =>  72,
  78 'Ogham Block'                                 =>  73,
  79 'Runic Block'                                 =>  74,
  80 'Khmer Block'                                 =>  75,
  81 'Mongolian Block'                             =>  76,
  82 'Latin Extended Additional'                   =>  77,
  83 'Greek Extended'                              =>  78,
  84 'General Punctuation'                         =>  79,
  85 'Superscripts and Subscripts'                 =>  80,
  86 'Currency Symbols'                            =>  81,
  87 'Combining Marks for Symbols'                 =>  82,
  88 'Letterlike Symbols'                          =>  83,
  89 'Number Forms'                                =>  84,
  90 'Arrows'                                      =>  85,
  91 'Mathematical Operators'                      =>  86,
  92 'Miscellaneous Technical'                     =>  87,
  93 'Control Pictures'                            =>  88,
  94 'Optical Character Recognition'               =>  89,
  95 'Enclosed Alphanumerics'                      =>  90,
  96 'Box Drawing'                                 =>  91,
  97 'Block Elements'                              =>  92,
  98 'Geometric Shapes'                            =>  93,
  99 'Miscellaneous Symbols'                       =>  94,
 100 'Dingbats'                                    =>  95,
 101 'Braille Patterns'                            =>  96,
 102 'CJK Radicals Supplement'                     =>  97,
 103 'Kangxi Radicals'                             =>  98,
 104 'Ideographic Description Characters'          =>  99,
 105 'CJK Symbols and Punctuation'                 => 100,
 106 'Hiragana Block'                              => 101,
 107 'Katakana Block'                              => 102,
 108 'Bopomofo Block'                              => 103,
 109 'Hangul Compatibility Jamo'                   => 104,
 110 'Kanbun'                                      => 105,
 111 'Bopomofo Extended'                           => 106,
 112 'Enclosed CJK Letters and Months'             => 107,
 113 'CJK Compatibility'                           => 108,
 114 'CJK Unified Ideographs Extension A'          => 109,
 115 'CJK Unified Ideographs'                      => 110,
 116 'Yi Syllables'                                => 111,
 117 'Yi Radicals'                                 => 112,
 118 'Hangul Syllables'                            => 113,
 119 'High Surrogates'                             => 114,
 120 'High Private Use Surrogates'                 => 115,
 121 'Low Surrogates'                              => 116,
 122 'Private Use'                                 => 117,
 123 'CJK Compatibility Ideographs'                => 118,
 124 'Alphabetic Presentation Forms'               => 119,
 125 'Arabic Presentation Forms-A'                 => 120,
 126 'Combining Half Marks'                        => 121,
 127 'CJK Compatibility Forms'                     => 122,
 128 'Small Form Variants'                         => 123,
 129 'Arabic Presentation Forms-B'                 => 124,
 130 'Specials'                                    => 125,
 131 'Halfwidth and Fullwidth Forms'               => 126,
 132 'Old Italic'                                  => 127,
 133 'Gothic Block'                                => 128,
 134 'Deseret Block'                               => 129,
 135 'Byzantine Musical Symbols'                   => 130,
 136 'Musical Symbols'                             => 131,
 137 'Mathematical Alphanumeric Symbols'           => 132,
 138 'CJK Unified Ideographs Extension B'          => 133,
 139 'CJK Compatibility Ideographs Supplement'     => 134,
 140 'Tags'                                        => 135,
 141 'Common'                                      => 136,
 142 );
 143 %utf8::InPat = (
 144 'alp' => {
 145         'Alphabetic(?:[-_]|\s+)?Presentation(?:[-_]|\s+)?Forms' => 'Alphabetic Presentation Forms',
 146 },
 147 'ara' => {
 148         'ARABIC' => 'ARABIC',
 149         'Arabic(?:[-_]|\s+)?Block' => 'Arabic Block',
 150         'Arabic(?:[-_]|\s+)?Presentation(?:[-_]|\s+)?Forms(?:[-_]|\s+)?A' => 'Arabic Presentation Forms-A',
 151         'Arabic(?:[-_]|\s+)?Presentation(?:[-_]|\s+)?Forms(?:[-_]|\s+)?B' => 'Arabic Presentation Forms-B',
 152 },
 153 'arm' => {
 154         'ARMENIAN' => 'ARMENIAN',
 155         'Armenian(?:[-_]|\s+)?Block' => 'Armenian Block',
 156 },
 157 'arr' => {
 158         'Arrows' => 'Arrows',
 159 },
 160 'bas' => {
 161         'Basic(?:[-_]|\s+)?Latin' => 'Basic Latin',
 162 },
 163 'ben' => {
 164         'BENGALI' => 'BENGALI',
 165         'Bengali(?:[-_]|\s+)?Block' => 'Bengali Block',
 166 },
 167 'blo' => {
 168         'Block(?:[-_]|\s+)?Elements' => 'Block Elements',
 169 },
 170 'bop' => {
 171         'BOPOMOFO' => 'BOPOMOFO',
 172         'Bopomofo(?:[-_]|\s+)?Block' => 'Bopomofo Block',
 173         'Bopomofo(?:[-_]|\s+)?Extended' => 'Bopomofo Extended',
 174 },
 175 'box' => {
 176         'Box(?:[-_]|\s+)?Drawing' => 'Box Drawing',
 177 },
 178 'bra' => {
 179         'Braille(?:[-_]|\s+)?Patterns' => 'Braille Patterns',
 180 },
 181 'byz' => {
 182         'Byzantine(?:[-_]|\s+)?Musical(?:[-_]|\s+)?Symbols' => 'Byzantine Musical Symbols',
 183 },
 184 'can' => {
 185         'CANADIAN(?:[-_]|\s+)?ABORIGINAL' => 'CANADIAN-ABORIGINAL',
 186 },
 187 'che' => {
 188         'CHEROKEE' => 'CHEROKEE',
 189         'Cherokee(?:[-_]|\s+)?Block' => 'Cherokee Block',
 190 },
 191 'cjk' => {
 192         'CJK(?:[-_]|\s+)?Radicals(?:[-_]|\s+)?Supplement' => 'CJK Radicals Supplement',
 193         'CJK(?:[-_]|\s+)?Symbols(?:[-_]|\s+)?and(?:[-_]|\s+)?Punctuation' => 'CJK Symbols and Punctuation',
 194         'CJK(?:[-_]|\s+)?Compatibility' => 'CJK Compatibility',
 195         'CJK(?:[-_]|\s+)?Unified(?:[-_]|\s+)?Ideographs(?:[-_]|\s+)?Extension(?:[-_]|\s+)?A' => 'CJK Unified Ideographs Extension A',
 196         'CJK(?:[-_]|\s+)?Unified(?:[-_]|\s+)?Ideographs' => 'CJK Unified Ideographs',
 197         'CJK(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Ideographs' => 'CJK Compatibility Ideographs',
 198         'CJK(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Forms' => 'CJK Compatibility Forms',
 199         'CJK(?:[-_]|\s+)?Unified(?:[-_]|\s+)?Ideographs(?:[-_]|\s+)?Extension(?:[-_]|\s+)?B' => 'CJK Unified Ideographs Extension B',
 200         'CJK(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Ideographs(?:[-_]|\s+)?Supplement' => 'CJK Compatibility Ideographs Supplement',
 201 },
 202 'com' => {
 203         'Combining(?:[-_]|\s+)?Diacritical(?:[-_]|\s+)?Marks' => 'Combining Diacritical Marks',
 204         'Combining(?:[-_]|\s+)?Marks(?:[-_]|\s+)?for(?:[-_]|\s+)?Symbols' => 'Combining Marks for Symbols',
 205         'Combining(?:[-_]|\s+)?Half(?:[-_]|\s+)?Marks' => 'Combining Half Marks',
 206         'Common' => 'Common',
 207 },
 208 'con' => {
 209         'Control(?:[-_]|\s+)?Pictures' => 'Control Pictures',
 210 },
 211 'cur' => {
 212         'Currency(?:[-_]|\s+)?Symbols' => 'Currency Symbols',
 213 },
 214 'cyr' => {
 215         'CYRILLIC' => 'CYRILLIC',
 216         'Cyrillic(?:[-_]|\s+)?Block' => 'Cyrillic Block',
 217 },
 218 'des' => {
 219         'DESERET' => 'DESERET',
 220         'Deseret(?:[-_]|\s+)?Block' => 'Deseret Block',
 221 },
 222 'dev' => {
 223         'DEVANAGARI' => 'DEVANAGARI',
 224         'Devanagari(?:[-_]|\s+)?Block' => 'Devanagari Block',
 225 },
 226 'din' => {
 227         'Dingbats' => 'Dingbats',
 228 },
 229 'enc' => {
 230         'Enclosed(?:[-_]|\s+)?Alphanumerics' => 'Enclosed Alphanumerics',
 231         'Enclosed(?:[-_]|\s+)?CJK(?:[-_]|\s+)?Letters(?:[-_]|\s+)?and(?:[-_]|\s+)?Months' => 'Enclosed CJK Letters and Months',
 232 },
 233 'eth' => {
 234         'ETHIOPIC' => 'ETHIOPIC',
 235         'Ethiopic(?:[-_]|\s+)?Block' => 'Ethiopic Block',
 236 },
 237 'gen' => {
 238         'General(?:[-_]|\s+)?Punctuation' => 'General Punctuation',
 239 },
 240 'geo' => {
 241         'GEORGIAN' => 'GEORGIAN',
 242         'Georgian(?:[-_]|\s+)?Block' => 'Georgian Block',
 243         'Geometric(?:[-_]|\s+)?Shapes' => 'Geometric Shapes',
 244 },
 245 'got' => {
 246         'GOTHIC' => 'GOTHIC',
 247         'Gothic(?:[-_]|\s+)?Block' => 'Gothic Block',
 248 },
 249 'gre' => {
 250         'GREEK' => 'GREEK',
 251         'Greek(?:[-_]|\s+)?Block' => 'Greek Block',
 252         'Greek(?:[-_]|\s+)?Extended' => 'Greek Extended',
 253 },
 254 'guj' => {
 255         'GUJARATI' => 'GUJARATI',
 256         'Gujarati(?:[-_]|\s+)?Block' => 'Gujarati Block',
 257 },
 258 'gur' => {
 259         'GURMUKHI' => 'GURMUKHI',
 260         'Gurmukhi(?:[-_]|\s+)?Block' => 'Gurmukhi Block',
 261 },
 262 'hal' => {
 263         'Halfwidth(?:[-_]|\s+)?and(?:[-_]|\s+)?Fullwidth(?:[-_]|\s+)?Forms' => 'Halfwidth and Fullwidth Forms',
 264 },
 265 'han' => {
 266         'HANGUL' => 'HANGUL',
 267         'HAN' => 'HAN',
 268         'Hangul(?:[-_]|\s+)?Jamo' => 'Hangul Jamo',
 269         'Hangul(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Jamo' => 'Hangul Compatibility Jamo',
 270         'Hangul(?:[-_]|\s+)?Syllables' => 'Hangul Syllables',
 271 },
 272 'heb' => {
 273         'HEBREW' => 'HEBREW',
 274         'Hebrew(?:[-_]|\s+)?Block' => 'Hebrew Block',
 275 },
 276 'hig' => {
 277         'High(?:[-_]|\s+)?Surrogates' => 'High Surrogates',
 278         'High(?:[-_]|\s+)?Private(?:[-_]|\s+)?Use(?:[-_]|\s+)?Surrogates' => 'High Private Use Surrogates',
 279 },
 280 'hir' => {
 281         'HIRAGANA' => 'HIRAGANA',
 282         'Hiragana(?:[-_]|\s+)?Block' => 'Hiragana Block',
 283 },
 284 'ide' => {
 285         'Ideographic(?:[-_]|\s+)?Description(?:[-_]|\s+)?Characters' => 'Ideographic Description Characters',
 286 },
 287 'inh' => {
 288         'INHERITED' => 'INHERITED',
 289 },
 290 'ipa' => {
 291         'IPA(?:[-_]|\s+)?Extensions' => 'IPA Extensions',
 292 },
 293 'kan' => {
 294         'KANNADA' => 'KANNADA',
 295         'Kannada(?:[-_]|\s+)?Block' => 'Kannada Block',
 296         'Kangxi(?:[-_]|\s+)?Radicals' => 'Kangxi Radicals',
 297         'Kanbun' => 'Kanbun',
 298 },
 299 'kat' => {
 300         'KATAKANA' => 'KATAKANA',
 301         'Katakana(?:[-_]|\s+)?Block' => 'Katakana Block',
 302 },
 303 'khm' => {
 304         'KHMER' => 'KHMER',
 305         'Khmer(?:[-_]|\s+)?Block' => 'Khmer Block',
 306 },
 307 'lao' => {
 308         'LAO' => 'LAO',
 309         'Lao(?:[-_]|\s+)?Block' => 'Lao Block',
 310 },
 311 'lat' => {
 312         'LATIN' => 'LATIN',
 313         'Latin(?:[-_]|\s+)?1(?:[-_]|\s+)?Supplement' => 'Latin-1 Supplement',
 314         'Latin(?:[-_]|\s+)?Extended(?:[-_]|\s+)?A' => 'Latin Extended-A',
 315         'Latin(?:[-_]|\s+)?Extended(?:[-_]|\s+)?B' => 'Latin Extended-B',
 316         'Latin(?:[-_]|\s+)?Extended(?:[-_]|\s+)?Additional' => 'Latin Extended Additional',
 317 },
 318 'let' => {
 319         'Letterlike(?:[-_]|\s+)?Symbols' => 'Letterlike Symbols',
 320 },
 321 'low' => {
 322         'Low(?:[-_]|\s+)?Surrogates' => 'Low Surrogates',
 323 },
 324 'mal' => {
 325         'MALAYALAM' => 'MALAYALAM',
 326         'Malayalam(?:[-_]|\s+)?Block' => 'Malayalam Block',
 327 },
 328 'mat' => {
 329         'Mathematical(?:[-_]|\s+)?Operators' => 'Mathematical Operators',
 330         'Mathematical(?:[-_]|\s+)?Alphanumeric(?:[-_]|\s+)?Symbols' => 'Mathematical Alphanumeric Symbols',
 331 },
 332 'mis' => {
 333         'Miscellaneous(?:[-_]|\s+)?Technical' => 'Miscellaneous Technical',
 334         'Miscellaneous(?:[-_]|\s+)?Symbols' => 'Miscellaneous Symbols',
 335 },
 336 'mon' => {
 337         'MONGOLIAN' => 'MONGOLIAN',
 338         'Mongolian(?:[-_]|\s+)?Block' => 'Mongolian Block',
 339 },
 340 'mus' => {
 341         'Musical(?:[-_]|\s+)?Symbols' => 'Musical Symbols',
 342 },
 343 'mya' => {
 344         'MYANMAR' => 'MYANMAR',
 345         'Myanmar(?:[-_]|\s+)?Block' => 'Myanmar Block',
 346 },
 347 'num' => {
 348         'Number(?:[-_]|\s+)?Forms' => 'Number Forms',
 349 },
 350 'ogh' => {
 351         'OGHAM' => 'OGHAM',
 352         'Ogham(?:[-_]|\s+)?Block' => 'Ogham Block',
 353 },
 354 'old' => {
 355         'OLD(?:[-_]|\s+)?ITALIC' => 'OLD-ITALIC',
 356         'Old(?:[-_]|\s+)?Italic' => 'Old Italic',
 357 },
 358 'opt' => {
 359         'Optical(?:[-_]|\s+)?Character(?:[-_]|\s+)?Recognition' => 'Optical Character Recognition',
 360 },
 361 'ori' => {
 362         'ORIYA' => 'ORIYA',
 363         'Oriya(?:[-_]|\s+)?Block' => 'Oriya Block',
 364 },
 365 'pri' => {
 366         'Private(?:[-_]|\s+)?Use' => 'Private Use',
 367 },
 368 'run' => {
 369         'RUNIC' => 'RUNIC',
 370         'Runic(?:[-_]|\s+)?Block' => 'Runic Block',
 371 },
 372 'sin' => {
 373         'SINHALA' => 'SINHALA',
 374         'Sinhala(?:[-_]|\s+)?Block' => 'Sinhala Block',
 375 },
 376 'sma' => {
 377         'Small(?:[-_]|\s+)?Form(?:[-_]|\s+)?Variants' => 'Small Form Variants',
 378 },
 379 'spa' => {
 380         'Spacing(?:[-_]|\s+)?Modifier(?:[-_]|\s+)?Letters' => 'Spacing Modifier Letters',
 381 },
 382 'spe' => {
 383         'Specials' => 'Specials',
 384 },
 385 'sup' => {
 386         'Superscripts(?:[-_]|\s+)?and(?:[-_]|\s+)?Subscripts' => 'Superscripts and Subscripts',
 387 },
 388 'syr' => {
 389         'SYRIAC' => 'SYRIAC',
 390         'Syriac(?:[-_]|\s+)?Block' => 'Syriac Block',
 391 },
 392 'tag' => {
 393         'Tags' => 'Tags',
 394 },
 395 'tam' => {
 396         'TAMIL' => 'TAMIL',
 397         'Tamil(?:[-_]|\s+)?Block' => 'Tamil Block',
 398 },
 399 'tel' => {
 400         'TELUGU' => 'TELUGU',
 401         'Telugu(?:[-_]|\s+)?Block' => 'Telugu Block',
 402 },
 403 'tha' => {
 404         'THAANA' => 'THAANA',
 405         'THAI' => 'THAI',
 406         'Thaana(?:[-_]|\s+)?Block' => 'Thaana Block',
 407         'Thai(?:[-_]|\s+)?Block' => 'Thai Block',
 408 },
 409 'tib' => {
 410         'TIBETAN' => 'TIBETAN',
 411         'Tibetan(?:[-_]|\s+)?Block' => 'Tibetan Block',
 412 },
 413 'uni' => {
 414         'Unified(?:[-_]|\s+)?Canadian(?:[-_]|\s+)?Aboriginal(?:[-_]|\s+)?Syllabics' => 'Unified Canadian Aboriginal Syllabics',
 415 },
 416 'yi' => {
 417         'YI' => 'YI',
 418 },
 419 'yi ' => {
 420         'Yi(?:[-_]|\s+)?Syllables' => 'Yi Syllables',
 421         'Yi(?:[-_]|\s+)?Radicals' => 'Yi Radicals',
 422 },
 423 );