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 );
 142 %utf8::InPat = (
 143 'alp' => {
 144         'Alphabetic[- _]?Presentation[- _]?Forms' => 'Alphabetic Presentation Forms',
 145 },
 146 'ara' => {
 147         'ARABIC' => 'ARABIC',
 148         'Arabic[- _]?Block' => 'Arabic Block',
 149         'Arabic[- _]?Presentation[- _]?Forms[- _]?A' => 'Arabic Presentation Forms-A',
 150         'Arabic[- _]?Presentation[- _]?Forms[- _]?B' => 'Arabic Presentation Forms-B',
 151 },
 152 'arm' => {
 153         'ARMENIAN' => 'ARMENIAN',
 154         'Armenian[- _]?Block' => 'Armenian Block',
 155 },
 156 'arr' => {
 157         'Arrows' => 'Arrows',
 158 },
 159 'bas' => {
 160         'Basic[- _]?Latin' => 'Basic Latin',
 161 },
 162 'ben' => {
 163         'BENGALI' => 'BENGALI',
 164         'Bengali[- _]?Block' => 'Bengali Block',
 165 },
 166 'blo' => {
 167         'Block[- _]?Elements' => 'Block Elements',
 168 },
 169 'bop' => {
 170         'BOPOMOFO' => 'BOPOMOFO',
 171         'Bopomofo[- _]?Block' => 'Bopomofo Block',
 172         'Bopomofo[- _]?Extended' => 'Bopomofo Extended',
 173 },
 174 'box' => {
 175         'Box[- _]?Drawing' => 'Box Drawing',
 176 },
 177 'bra' => {
 178         'Braille[- _]?Patterns' => 'Braille Patterns',
 179 },
 180 'byz' => {
 181         'Byzantine[- _]?Musical[- _]?Symbols' => 'Byzantine Musical Symbols',
 182 },
 183 'can' => {
 184         'CANADIAN[- _]?ABORIGINAL' => 'CANADIAN-ABORIGINAL',
 185 },
 186 'che' => {
 187         'CHEROKEE' => 'CHEROKEE',
 188         'Cherokee[- _]?Block' => 'Cherokee Block',
 189 },
 190 'cjk' => {
 191         'CJK[- _]?Radicals[- _]?Supplement' => 'CJK Radicals Supplement',
 192         'CJK[- _]?Symbols[- _]?and[- _]?Punctuation' => 'CJK Symbols and Punctuation',
 193         'CJK[- _]?Compatibility' => 'CJK Compatibility',
 194         'CJK[- _]?Unified[- _]?Ideographs[- _]?Extension[- _]?A' => 'CJK Unified Ideographs Extension A',
 195         'CJK[- _]?Unified[- _]?Ideographs' => 'CJK Unified Ideographs',
 196         'CJK[- _]?Compatibility[- _]?Ideographs' => 'CJK Compatibility Ideographs',
 197         'CJK[- _]?Compatibility[- _]?Forms' => 'CJK Compatibility Forms',
 198         'CJK[- _]?Unified[- _]?Ideographs[- _]?Extension[- _]?B' => 'CJK Unified Ideographs Extension B',
 199         'CJK[- _]?Compatibility[- _]?Ideographs[- _]?Supplement' => 'CJK Compatibility Ideographs Supplement',
 200 },
 201 'com' => {
 202         'Combining[- _]?Diacritical[- _]?Marks' => 'Combining Diacritical Marks',
 203         'Combining[- _]?Marks[- _]?for[- _]?Symbols' => 'Combining Marks for Symbols',
 204         'Combining[- _]?Half[- _]?Marks' => 'Combining Half Marks',
 205 },
 206 'con' => {
 207         'Control[- _]?Pictures' => 'Control Pictures',
 208 },
 209 'cur' => {
 210         'Currency[- _]?Symbols' => 'Currency Symbols',
 211 },
 212 'cyr' => {
 213         'CYRILLIC' => 'CYRILLIC',
 214         'Cyrillic[- _]?Block' => 'Cyrillic Block',
 215 },
 216 'des' => {
 217         'DESERET' => 'DESERET',
 218         'Deseret[- _]?Block' => 'Deseret Block',
 219 },
 220 'dev' => {
 221         'DEVANAGARI' => 'DEVANAGARI',
 222         'Devanagari[- _]?Block' => 'Devanagari Block',
 223 },
 224 'din' => {
 225         'Dingbats' => 'Dingbats',
 226 },
 227 'enc' => {
 228         'Enclosed[- _]?Alphanumerics' => 'Enclosed Alphanumerics',
 229         'Enclosed[- _]?CJK[- _]?Letters[- _]?and[- _]?Months' => 'Enclosed CJK Letters and Months',
 230 },
 231 'eth' => {
 232         'ETHIOPIC' => 'ETHIOPIC',
 233         'Ethiopic[- _]?Block' => 'Ethiopic Block',
 234 },
 235 'gen' => {
 236         'General[- _]?Punctuation' => 'General Punctuation',
 237 },
 238 'geo' => {
 239         'GEORGIAN' => 'GEORGIAN',
 240         'Georgian[- _]?Block' => 'Georgian Block',
 241         'Geometric[- _]?Shapes' => 'Geometric Shapes',
 242 },
 243 'got' => {
 244         'GOTHIC' => 'GOTHIC',
 245         'Gothic[- _]?Block' => 'Gothic Block',
 246 },
 247 'gre' => {
 248         'GREEK' => 'GREEK',
 249         'Greek[- _]?Block' => 'Greek Block',
 250         'Greek[- _]?Extended' => 'Greek Extended',
 251 },
 252 'guj' => {
 253         'GUJARATI' => 'GUJARATI',
 254         'Gujarati[- _]?Block' => 'Gujarati Block',
 255 },
 256 'gur' => {
 257         'GURMUKHI' => 'GURMUKHI',
 258         'Gurmukhi[- _]?Block' => 'Gurmukhi Block',
 259 },
 260 'hal' => {
 261         'Halfwidth[- _]?and[- _]?Fullwidth[- _]?Forms' => 'Halfwidth and Fullwidth Forms',
 262 },
 263 'han' => {
 264         'HANGUL' => 'HANGUL',
 265         'HAN' => 'HAN',
 266         'Hangul[- _]?Jamo' => 'Hangul Jamo',
 267         'Hangul[- _]?Compatibility[- _]?Jamo' => 'Hangul Compatibility Jamo',
 268         'Hangul[- _]?Syllables' => 'Hangul Syllables',
 269 },
 270 'heb' => {
 271         'HEBREW' => 'HEBREW',
 272         'Hebrew[- _]?Block' => 'Hebrew Block',
 273 },
 274 'hig' => {
 275         'High[- _]?Surrogates' => 'High Surrogates',
 276         'High[- _]?Private[- _]?Use[- _]?Surrogates' => 'High Private Use Surrogates',
 277 },
 278 'hir' => {
 279         'HIRAGANA' => 'HIRAGANA',
 280         'Hiragana[- _]?Block' => 'Hiragana Block',
 281 },
 282 'ide' => {
 283         'Ideographic[- _]?Description[- _]?Characters' => 'Ideographic Description Characters',
 284 },
 285 'inh' => {
 286         'INHERITED' => 'INHERITED',
 287 },
 288 'ipa' => {
 289         'IPA[- _]?Extensions' => 'IPA Extensions',
 290 },
 291 'kan' => {
 292         'KANNADA' => 'KANNADA',
 293         'Kannada[- _]?Block' => 'Kannada Block',
 294         'Kangxi[- _]?Radicals' => 'Kangxi Radicals',
 295         'Kanbun' => 'Kanbun',
 296 },
 297 'kat' => {
 298         'KATAKANA' => 'KATAKANA',
 299         'Katakana[- _]?Block' => 'Katakana Block',
 300 },
 301 'khm' => {
 302         'KHMER' => 'KHMER',
 303         'Khmer[- _]?Block' => 'Khmer Block',
 304 },
 305 'lao' => {
 306         'LAO' => 'LAO',
 307         'Lao[- _]?Block' => 'Lao Block',
 308 },
 309 'lat' => {
 310         'LATIN' => 'LATIN',
 311         'Latin[- _]?1[- _]?Supplement' => 'Latin-1 Supplement',
 312         'Latin[- _]?Extended[- _]?A' => 'Latin Extended-A',
 313         'Latin[- _]?Extended[- _]?B' => 'Latin Extended-B',
 314         'Latin[- _]?Extended[- _]?Additional' => 'Latin Extended Additional',
 315 },
 316 'let' => {
 317         'Letterlike[- _]?Symbols' => 'Letterlike Symbols',
 318 },
 319 'low' => {
 320         'Low[- _]?Surrogates' => 'Low Surrogates',
 321 },
 322 'mal' => {
 323         'MALAYALAM' => 'MALAYALAM',
 324         'Malayalam[- _]?Block' => 'Malayalam Block',
 325 },
 326 'mat' => {
 327         'Mathematical[- _]?Operators' => 'Mathematical Operators',
 328         'Mathematical[- _]?Alphanumeric[- _]?Symbols' => 'Mathematical Alphanumeric Symbols',
 329 },
 330 'mis' => {
 331         'Miscellaneous[- _]?Technical' => 'Miscellaneous Technical',
 332         'Miscellaneous[- _]?Symbols' => 'Miscellaneous Symbols',
 333 },
 334 'mon' => {
 335         'MONGOLIAN' => 'MONGOLIAN',
 336         'Mongolian[- _]?Block' => 'Mongolian Block',
 337 },
 338 'mus' => {
 339         'Musical[- _]?Symbols' => 'Musical Symbols',
 340 },
 341 'mya' => {
 342         'MYANMAR' => 'MYANMAR',
 343         'Myanmar[- _]?Block' => 'Myanmar Block',
 344 },
 345 'num' => {
 346         'Number[- _]?Forms' => 'Number Forms',
 347 },
 348 'ogh' => {
 349         'OGHAM' => 'OGHAM',
 350         'Ogham[- _]?Block' => 'Ogham Block',
 351 },
 352 'old' => {
 353         'OLD[- _]?ITALIC' => 'OLD-ITALIC',
 354         'Old[- _]?Italic' => 'Old Italic',
 355 },
 356 'opt' => {
 357         'Optical[- _]?Character[- _]?Recognition' => 'Optical Character Recognition',
 358 },
 359 'ori' => {
 360         'ORIYA' => 'ORIYA',
 361         'Oriya[- _]?Block' => 'Oriya Block',
 362 },
 363 'pri' => {
 364         'Private[- _]?Use' => 'Private Use',
 365 },
 366 'run' => {
 367         'RUNIC' => 'RUNIC',
 368         'Runic[- _]?Block' => 'Runic Block',
 369 },
 370 'sin' => {
 371         'SINHALA' => 'SINHALA',
 372         'Sinhala[- _]?Block' => 'Sinhala Block',
 373 },
 374 'sma' => {
 375         'Small[- _]?Form[- _]?Variants' => 'Small Form Variants',
 376 },
 377 'spa' => {
 378         'Spacing[- _]?Modifier[- _]?Letters' => 'Spacing Modifier Letters',
 379 },
 380 'spe' => {
 381         'Specials' => 'Specials',
 382 },
 383 'sup' => {
 384         'Superscripts[- _]?and[- _]?Subscripts' => 'Superscripts and Subscripts',
 385 },
 386 'syr' => {
 387         'SYRIAC' => 'SYRIAC',
 388         'Syriac[- _]?Block' => 'Syriac Block',
 389 },
 390 'tag' => {
 391         'Tags' => 'Tags',
 392 },
 393 'tam' => {
 394         'TAMIL' => 'TAMIL',
 395         'Tamil[- _]?Block' => 'Tamil Block',
 396 },
 397 'tel' => {
 398         'TELUGU' => 'TELUGU',
 399         'Telugu[- _]?Block' => 'Telugu Block',
 400 },
 401 'tha' => {
 402         'THAANA' => 'THAANA',
 403         'THAI' => 'THAI',
 404         'Thaana[- _]?Block' => 'Thaana Block',
 405         'Thai[- _]?Block' => 'Thai Block',
 406 },
 407 'tib' => {
 408         'TIBETAN' => 'TIBETAN',
 409         'Tibetan[- _]?Block' => 'Tibetan Block',
 410 },
 411 'uni' => {
 412         'Unified[- _]?Canadian[- _]?Aboriginal[- _]?Syllabics' => 'Unified Canadian Aboriginal Syllabics',
 413 },
 414 'yi' => {
 415         'YI' => 'YI',
 416 },
 417 'yi ' => {
 418         'Yi[- _]?Syllables' => 'Yi Syllables',
 419         'Yi[- _]?Radicals' => 'Yi Radicals',
 420 },
 421 );