lib/unicore/In.pl

   1 # !!!!!!!   DO NOT EDIT THIS FILE   !!!!!!!
   2 # This file is built by mktables from e.g. Unicode.txt.
   3 # Any changes made here will be lost!
   4 %utf8::In =
   5 (
   6 'Alphabetic'                                  => '164',
   7 'Alphabetic Presentation Forms'               => '129',
   8 'Any'                                         => '171',
   9 'ARABIC'                                      => '16',
  10 'Arabic Block'                                => '62',
  11 'Arabic Presentation Forms-A'                 => '130',
  12 'Arabic Presentation Forms-B'                 => '134',
  13 'ARMENIAN'                                    => '14',
  14 'Armenian Block'                              => '60',
  15 'Arrows'                                      => '96',
  16 'ASCII_Hex_Digit'                             => '152',
  17 'Assigned'                                    => '163',
  18 'Basic Latin'                                 => '51',
  19 'BENGALI'                                     => '20',
  20 'Bengali Block'                               => '66',
  21 'Bidi_Control'                                => '159',
  22 'Block Elements'                              => '103',
  23 'BOPOMOFO'                                    => '45',
  24 'Bopomofo Block'                              => '114',
  25 'Bopomofo Extended'                           => '117',
  26 'Box Drawing'                                 => '102',
  27 'Braille Patterns'                            => '107',
  28 'Byzantine Musical Symbols'                   => '140',
  29 'CANADIAN-ABORIGINAL'                         => '37',
  30 'CHEROKEE'                                    => '36',
  31 'Cherokee Block'                              => '82',
  32 'CJK Compatibility'                           => '119',
  33 'CJK Compatibility Forms'                     => '132',
  34 'CJK Compatibility Ideographs'                => '128',
  35 'CJK Compatibility Ideographs Supplement'     => '144',
  36 'CJK Ideograph'                               => '1',
  37 'CJK Ideograph Extension A'                   => '0',
  38 'CJK Ideograph Extension B'                   => '7',
  39 'CJK Radicals Supplement'                     => '108',
  40 'CJK Symbols and Punctuation'                 => '111',
  41 'CJK Unified Ideographs'                      => '121',
  42 'CJK Unified Ideographs Extension A'          => '120',
  43 'CJK Unified Ideographs Extension B'          => '143',
  44 'Combining Diacritical Marks'                 => '57',
  45 'Combining Half Marks'                        => '131',
  46 'Combining Marks for Symbols'                 => '93',
  47 'Common'                                      => '50',
  48 'Control Pictures'                            => '99',
  49 'Currency Symbols'                            => '92',
  50 'CYRILLIC'                                    => '13',
  51 'Cyrillic Block'                              => '59',
  52 'Dash'                                        => '151',
  53 'DESERET'                                     => '49',
  54 'Deseret Block'                               => '139',
  55 'DEVANAGARI'                                  => '19',
  56 'Devanagari Block'                            => '65',
  57 'Diacritic'                                   => '154',
  58 'Dingbats'                                    => '106',
  59 'Enclosed Alphanumerics'                      => '101',
  60 'Enclosed CJK Letters and Months'             => '118',
  61 'ETHIOPIC'                                    => '35',
  62 'Ethiopic Block'                              => '81',
  63 'Extender'                                    => '155',
  64 'General Punctuation'                         => '90',
  65 'Geometric Shapes'                            => '104',
  66 'GEORGIAN'                                    => '33',
  67 'Georgian Block'                              => '79',
  68 'GOTHIC'                                      => '48',
  69 'Gothic Block'                                => '138',
  70 'GREEK'                                       => '11',
  71 'Greek Block'                                 => '58',
  72 'Greek Extended'                              => '89',
  73 'GUJARATI'                                    => '22',
  74 'Gujarati Block'                              => '68',
  75 'GURMUKHI'                                    => '21',
  76 'Gurmukhi Block'                              => '67',
  77 'Halfwidth and Fullwidth Forms'               => '136',
  78 'HAN'                                         => '42',
  79 'HANGUL'                                      => '34',
  80 'Hangul Compatibility Jamo'                   => '115',
  81 'Hangul Jamo'                                 => '80',
  82 'Hangul Syllable'                             => '2',
  83 'Hangul Syllables'                            => '124',
  84 'HEBREW'                                      => '15',
  85 'Hebrew Block'                                => '61',
  86 'Hex_Digit'                                   => '153',
  87 'High Private Use Surrogates'                 => '126',
  88 'High Surrogates'                             => '125',
  89 'HIRAGANA'                                    => '43',
  90 'Hiragana Block'                              => '112',
  91 'Hyphen'                                      => '150',
  92 'ID_Continue'                                 => '170',
  93 'ID_Start'                                    => '169',
  94 'Ideographic'                                 => '161',
  95 'Ideographic Description Characters'          => '110',
  96 'INHERITED'                                   => '12',
  97 'IPA Extensions'                              => '55',
  98 'Join_Control'                                => '158',
  99 'Kanbun'                                      => '116',
 100 'Kangxi Radicals'                             => '109',
 101 'KANNADA'                                     => '26',
 102 'Kannada Block'                               => '72',
 103 'KATAKANA'                                    => '44',
 104 'Katakana Block'                              => '113',
 105 'KHMER'                                       => '40',
 106 'Khmer Block'                                 => '86',
 107 'Lampersand'                                  => '168',
 108 'LAO'                                         => '30',
 109 'Lao Block'                                   => '76',
 110 'LATIN'                                       => '10',
 111 'Latin Extended Additional'                   => '88',
 112 'Latin Extended-A'                            => '53',
 113 'Latin Extended-B'                            => '54',
 114 'Latin-1 Supplement'                          => '52',
 115 'Letterlike Symbols'                          => '94',
 116 'Low Surrogate'                               => '5',
 117 'Low Surrogates'                              => '127',
 118 'Lowercase'                                   => '165',
 119 'MALAYALAM'                                   => '27',
 120 'Malayalam Block'                             => '73',
 121 'Math'                                        => '167',
 122 'Mathematical Alphanumeric Symbols'           => '142',
 123 'Mathematical Operators'                      => '97',
 124 'Miscellaneous Symbols'                       => '105',
 125 'Miscellaneous Technical'                     => '98',
 126 'MONGOLIAN'                                   => '41',
 127 'Mongolian Block'                             => '87',
 128 'Musical Symbols'                             => '141',
 129 'MYANMAR'                                     => '32',
 130 'Myanmar Block'                               => '78',
 131 'Non Private Use High Surrogate'              => '3',
 132 'Noncharacter_Code_Point'                     => '162',
 133 'Number Forms'                                => '95',
 134 'OGHAM'                                       => '38',
 135 'Ogham Block'                                 => '84',
 136 'Old Italic Block'                            => '137',
 137 'OLD-ITALIC'                                  => '47',
 138 'Optical Character Recognition'               => '100',
 139 'ORIYA'                                       => '23',
 140 'Oriya Block'                                 => '69',
 141 'Other_Alphabetic'                            => '157',
 142 'Other_Lowercase'                             => '156',
 143 'Other_Math'                                  => '149',
 144 'Other_Uppercase'                             => '160',
 145 'Plane 15 Private Use'                        => '8',
 146 'Plane 16 Private Use'                        => '9',
 147 'Private Use'                                 => '6',
 148 'Private Use High Surrogate'                  => '4',
 149 'Quotation_Mark'                              => '148',
 150 'RUNIC'                                       => '39',
 151 'Runic Block'                                 => '85',
 152 'SINHALA'                                     => '28',
 153 'Sinhala Block'                               => '74',
 154 'Small Form Variants'                         => '133',
 155 'Spacing Modifier Letters'                    => '56',
 156 'Specials'                                    => '135',
 157 'Superscripts and Subscripts'                 => '91',
 158 'SYRIAC'                                      => '17',
 159 'Syriac Block'                                => '63',
 160 'Tags'                                        => '145',
 161 'TAMIL'                                       => '24',
 162 'Tamil Block'                                 => '70',
 163 'TELUGU'                                      => '25',
 164 'Telugu Block'                                => '71',
 165 'Terminal_Punctuation'                        => '147',
 166 'THAANA'                                      => '18',
 167 'Thaana Block'                                => '64',
 168 'THAI'                                        => '29',
 169 'Thai Block'                                  => '75',
 170 'TIBETAN'                                     => '31',
 171 'Tibetan Block'                               => '77',
 172 'Unified Canadian Aboriginal Syllabics'       => '83',
 173 'Uppercase'                                   => '166',
 174 'White_space'                                 => '146',
 175 'YI'                                          => '46',
 176 'Yi Radicals'                                 => '123',
 177 'Yi Syllables'                                => '122',
 178 );
 179 %utf8::InPat =
 180 (
 181 'al' => {
 182         'Alphabetic' => '164',
 183         'Alphabetic(?:[-_]|\s+)?Presentation(?:[-_]|\s+)?Forms' => '129',
 184 },
 185 'an' => {
 186         'Any' => '171',
 187 },
 188 'ar' => {
 189         'ARABIC' => '16',
 190         'Arabic(?:[-_]|\s+)?Block' => '62',
 191         'Arabic(?:[-_]|\s+)?Presentation(?:[-_]|\s+)?Forms(?:[-_]|\s+)?A' => '130',
 192         'Arabic(?:[-_]|\s+)?Presentation(?:[-_]|\s+)?Forms(?:[-_]|\s+)?B' => '134',
 193         'ARMENIAN' => '14',
 194         'Armenian(?:[-_]|\s+)?Block' => '60',
 195         'Arrows' => '96',
 196 },
 197 'as' => {
 198         'ASCII(?:[-_]|\s+)?Hex(?:[-_]|\s+)?Digit' => '152',
 199         'Assigned' => '163',
 200 },
 201 'ba' => {
 202         'Basic(?:[-_]|\s+)?Latin' => '51',
 203 },
 204 'be' => {
 205         'BENGALI' => '20',
 206         'Bengali(?:[-_]|\s+)?Block' => '66',
 207 },
 208 'bi' => {
 209         'Bidi(?:[-_]|\s+)?Control' => '159',
 210 },
 211 'bl' => {
 212         'Block(?:[-_]|\s+)?Elements' => '103',
 213 },
 214 'bo' => {
 215         'BOPOMOFO' => '45',
 216         'Bopomofo(?:[-_]|\s+)?Block' => '114',
 217         'Bopomofo(?:[-_]|\s+)?Extended' => '117',
 218         'Box(?:[-_]|\s+)?Drawing' => '102',
 219 },
 220 'br' => {
 221         'Braille(?:[-_]|\s+)?Patterns' => '107',
 222 },
 223 'by' => {
 224         'Byzantine(?:[-_]|\s+)?Musical(?:[-_]|\s+)?Symbols' => '140',
 225 },
 226 'ca' => {
 227         'CANADIAN(?:[-_]|\s+)?ABORIGINAL' => '37',
 228 },
 229 'ch' => {
 230         'CHEROKEE' => '36',
 231         'Cherokee(?:[-_]|\s+)?Block' => '82',
 232 },
 233 'cj' => {
 234         'CJK(?:[-_]|\s+)?Compatibility' => '119',
 235         'CJK(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Forms' => '132',
 236         'CJK(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Ideographs' => '128',
 237         'CJK(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Ideographs(?:[-_]|\s+)?Supplement' => '144',
 238         'CJK(?:[-_]|\s+)?Ideograph' => '1',
 239         'CJK(?:[-_]|\s+)?Ideograph(?:[-_]|\s+)?Extension(?:[-_]|\s+)?A' => '0',
 240         'CJK(?:[-_]|\s+)?Ideograph(?:[-_]|\s+)?Extension(?:[-_]|\s+)?B' => '7',
 241         'CJK(?:[-_]|\s+)?Radicals(?:[-_]|\s+)?Supplement' => '108',
 242         'CJK(?:[-_]|\s+)?Symbols(?:[-_]|\s+)?and(?:[-_]|\s+)?Punctuation' => '111',
 243         'CJK(?:[-_]|\s+)?Unified(?:[-_]|\s+)?Ideographs' => '121',
 244         'CJK(?:[-_]|\s+)?Unified(?:[-_]|\s+)?Ideographs(?:[-_]|\s+)?Extension(?:[-_]|\s+)?A' => '120',
 245         'CJK(?:[-_]|\s+)?Unified(?:[-_]|\s+)?Ideographs(?:[-_]|\s+)?Extension(?:[-_]|\s+)?B' => '143',
 246 },
 247 'co' => {
 248         'Combining(?:[-_]|\s+)?Diacritical(?:[-_]|\s+)?Marks' => '57',
 249         'Combining(?:[-_]|\s+)?Half(?:[-_]|\s+)?Marks' => '131',
 250         'Combining(?:[-_]|\s+)?Marks(?:[-_]|\s+)?for(?:[-_]|\s+)?Symbols' => '93',
 251         'Common' => '50',
 252         'Control(?:[-_]|\s+)?Pictures' => '99',
 253 },
 254 'cu' => {
 255         'Currency(?:[-_]|\s+)?Symbols' => '92',
 256 },
 257 'cy' => {
 258         'CYRILLIC' => '13',
 259         'Cyrillic(?:[-_]|\s+)?Block' => '59',
 260 },
 261 'da' => {
 262         'Dash' => '151',
 263 },
 264 'de' => {
 265         'DESERET' => '49',
 266         'Deseret(?:[-_]|\s+)?Block' => '139',
 267         'DEVANAGARI' => '19',
 268         'Devanagari(?:[-_]|\s+)?Block' => '65',
 269 },
 270 'di' => {
 271         'Diacritic' => '154',
 272         'Dingbats' => '106',
 273 },
 274 'en' => {
 275         'Enclosed(?:[-_]|\s+)?Alphanumerics' => '101',
 276         'Enclosed(?:[-_]|\s+)?CJK(?:[-_]|\s+)?Letters(?:[-_]|\s+)?and(?:[-_]|\s+)?Months' => '118',
 277 },
 278 'et' => {
 279         'ETHIOPIC' => '35',
 280         'Ethiopic(?:[-_]|\s+)?Block' => '81',
 281 },
 282 'ex' => {
 283         'Extender' => '155',
 284 },
 285 'ge' => {
 286         'General(?:[-_]|\s+)?Punctuation' => '90',
 287         'Geometric(?:[-_]|\s+)?Shapes' => '104',
 288         'GEORGIAN' => '33',
 289         'Georgian(?:[-_]|\s+)?Block' => '79',
 290 },
 291 'go' => {
 292         'GOTHIC' => '48',
 293         'Gothic(?:[-_]|\s+)?Block' => '138',
 294 },
 295 'gr' => {
 296         'GREEK' => '11',
 297         'Greek(?:[-_]|\s+)?Block' => '58',
 298         'Greek(?:[-_]|\s+)?Extended' => '89',
 299 },
 300 'gu' => {
 301         'GUJARATI' => '22',
 302         'Gujarati(?:[-_]|\s+)?Block' => '68',
 303         'GURMUKHI' => '21',
 304         'Gurmukhi(?:[-_]|\s+)?Block' => '67',
 305 },
 306 'ha' => {
 307         'Halfwidth(?:[-_]|\s+)?and(?:[-_]|\s+)?Fullwidth(?:[-_]|\s+)?Forms' => '136',
 308         'HAN' => '42',
 309         'HANGUL' => '34',
 310         'Hangul(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Jamo' => '115',
 311         'Hangul(?:[-_]|\s+)?Jamo' => '80',
 312         'Hangul(?:[-_]|\s+)?Syllable' => '2',
 313         'Hangul(?:[-_]|\s+)?Syllables' => '124',
 314 },
 315 'he' => {
 316         'HEBREW' => '15',
 317         'Hebrew(?:[-_]|\s+)?Block' => '61',
 318         'Hex(?:[-_]|\s+)?Digit' => '153',
 319 },
 320 'hi' => {
 321         'High(?:[-_]|\s+)?Private(?:[-_]|\s+)?Use(?:[-_]|\s+)?Surrogates' => '126',
 322         'High(?:[-_]|\s+)?Surrogates' => '125',
 323         'HIRAGANA' => '43',
 324         'Hiragana(?:[-_]|\s+)?Block' => '112',
 325 },
 326 'hy' => {
 327         'Hyphen' => '150',
 328 },
 329 'id' => {
 330         'ID(?:[-_]|\s+)?Continue' => '170',
 331         'ID(?:[-_]|\s+)?Start' => '169',
 332         'Ideographic' => '161',
 333         'Ideographic(?:[-_]|\s+)?Description(?:[-_]|\s+)?Characters' => '110',
 334 },
 335 'in' => {
 336         'INHERITED' => '12',
 337 },
 338 'ip' => {
 339         'IPA(?:[-_]|\s+)?Extensions' => '55',
 340 },
 341 'jo' => {
 342         'Join(?:[-_]|\s+)?Control' => '158',
 343 },
 344 'ka' => {
 345         'Kanbun' => '116',
 346         'Kangxi(?:[-_]|\s+)?Radicals' => '109',
 347         'KANNADA' => '26',
 348         'Kannada(?:[-_]|\s+)?Block' => '72',
 349         'KATAKANA' => '44',
 350         'Katakana(?:[-_]|\s+)?Block' => '113',
 351 },
 352 'kh' => {
 353         'KHMER' => '40',
 354         'Khmer(?:[-_]|\s+)?Block' => '86',
 355 },
 356 'la' => {
 357         'Lampersand' => '168',
 358         'LAO' => '30',
 359         'Lao(?:[-_]|\s+)?Block' => '76',
 360         'LATIN' => '10',
 361         'Latin(?:[-_]|\s+)?Extended(?:[-_]|\s+)?Additional' => '88',
 362         'Latin(?:[-_]|\s+)?Extended(?:[-_]|\s+)?A' => '53',
 363         'Latin(?:[-_]|\s+)?Extended(?:[-_]|\s+)?B' => '54',
 364         'Latin(?:[-_]|\s+)?1(?:[-_]|\s+)?Supplement' => '52',
 365 },
 366 'le' => {
 367         'Letterlike(?:[-_]|\s+)?Symbols' => '94',
 368 },
 369 'lo' => {
 370         'Low(?:[-_]|\s+)?Surrogate' => '5',
 371         'Low(?:[-_]|\s+)?Surrogates' => '127',
 372         'Lowercase' => '165',
 373 },
 374 'ma' => {
 375         'MALAYALAM' => '27',
 376         'Malayalam(?:[-_]|\s+)?Block' => '73',
 377         'Math' => '167',
 378         'Mathematical(?:[-_]|\s+)?Alphanumeric(?:[-_]|\s+)?Symbols' => '142',
 379         'Mathematical(?:[-_]|\s+)?Operators' => '97',
 380 },
 381 'mi' => {
 382         'Miscellaneous(?:[-_]|\s+)?Symbols' => '105',
 383         'Miscellaneous(?:[-_]|\s+)?Technical' => '98',
 384 },
 385 'mo' => {
 386         'MONGOLIAN' => '41',
 387         'Mongolian(?:[-_]|\s+)?Block' => '87',
 388 },
 389 'mu' => {
 390         'Musical(?:[-_]|\s+)?Symbols' => '141',
 391 },
 392 'my' => {
 393         'MYANMAR' => '32',
 394         'Myanmar(?:[-_]|\s+)?Block' => '78',
 395 },
 396 'no' => {
 397         'Non(?:[-_]|\s+)?Private(?:[-_]|\s+)?Use(?:[-_]|\s+)?High(?:[-_]|\s+)?Surrogate' => '3',
 398         'Noncharacter(?:[-_]|\s+)?Code(?:[-_]|\s+)?Point' => '162',
 399 },
 400 'nu' => {
 401         'Number(?:[-_]|\s+)?Forms' => '95',
 402 },
 403 'og' => {
 404         'OGHAM' => '38',
 405         'Ogham(?:[-_]|\s+)?Block' => '84',
 406 },
 407 'ol' => {
 408         'Old(?:[-_]|\s+)?Italic(?:[-_]|\s+)?Block' => '137',
 409         'OLD(?:[-_]|\s+)?ITALIC' => '47',
 410 },
 411 'op' => {
 412         'Optical(?:[-_]|\s+)?Character(?:[-_]|\s+)?Recognition' => '100',
 413 },
 414 'or' => {
 415         'ORIYA' => '23',
 416         'Oriya(?:[-_]|\s+)?Block' => '69',
 417 },
 418 'ot' => {
 419         'Other(?:[-_]|\s+)?Alphabetic' => '157',
 420         'Other(?:[-_]|\s+)?Lowercase' => '156',
 421         'Other(?:[-_]|\s+)?Math' => '149',
 422         'Other(?:[-_]|\s+)?Uppercase' => '160',
 423 },
 424 'pl' => {
 425         'Plane(?:[-_]|\s+)?15(?:[-_]|\s+)?Private(?:[-_]|\s+)?Use' => '8',
 426         'Plane(?:[-_]|\s+)?16(?:[-_]|\s+)?Private(?:[-_]|\s+)?Use' => '9',
 427 },
 428 'pr' => {
 429         'Private(?:[-_]|\s+)?Use' => '6',
 430         'Private(?:[-_]|\s+)?Use(?:[-_]|\s+)?High(?:[-_]|\s+)?Surrogate' => '4',
 431 },
 432 'qu' => {
 433         'Quotation(?:[-_]|\s+)?Mark' => '148',
 434 },
 435 'ru' => {
 436         'RUNIC' => '39',
 437         'Runic(?:[-_]|\s+)?Block' => '85',
 438 },
 439 'si' => {
 440         'SINHALA' => '28',
 441         'Sinhala(?:[-_]|\s+)?Block' => '74',
 442 },
 443 'sm' => {
 444         'Small(?:[-_]|\s+)?Form(?:[-_]|\s+)?Variants' => '133',
 445 },
 446 'sp' => {
 447         'Spacing(?:[-_]|\s+)?Modifier(?:[-_]|\s+)?Letters' => '56',
 448         'Specials' => '135',
 449 },
 450 'su' => {
 451         'Superscripts(?:[-_]|\s+)?and(?:[-_]|\s+)?Subscripts' => '91',
 452 },
 453 'sy' => {
 454         'SYRIAC' => '17',
 455         'Syriac(?:[-_]|\s+)?Block' => '63',
 456 },
 457 'ta' => {
 458         'Tags' => '145',
 459         'TAMIL' => '24',
 460         'Tamil(?:[-_]|\s+)?Block' => '70',
 461 },
 462 'te' => {
 463         'TELUGU' => '25',
 464         'Telugu(?:[-_]|\s+)?Block' => '71',
 465         'Terminal(?:[-_]|\s+)?Punctuation' => '147',
 466 },
 467 'th' => {
 468         'THAANA' => '18',
 469         'Thaana(?:[-_]|\s+)?Block' => '64',
 470         'THAI' => '29',
 471         'Thai(?:[-_]|\s+)?Block' => '75',
 472 },
 473 'ti' => {
 474         'TIBETAN' => '31',
 475         'Tibetan(?:[-_]|\s+)?Block' => '77',
 476 },
 477 'un' => {
 478         'Unified(?:[-_]|\s+)?Canadian(?:[-_]|\s+)?Aboriginal(?:[-_]|\s+)?Syllabics' => '83',
 479 },
 480 'up' => {
 481         'Uppercase' => '166',
 482 },
 483 'wh' => {
 484         'White(?:[-_]|\s+)?space' => '146',
 485 },
 486 'yi' => {
 487         'YI' => '46',
 488         'Yi(?:[-_]|\s+)?Radicals' => '123',
 489         'Yi(?:[-_]|\s+)?Syllables' => '122',
 490 },
 491 );
 492
 493 %utf8::InScript =
 494 (
 495   10 => 'LATIN',
 496   11 => 'GREEK',
 497   12 => 'INHERITED',
 498   13 => 'CYRILLIC',
 499   14 => 'ARMENIAN',
 500   15 => 'HEBREW',
 501   16 => 'ARABIC',
 502   17 => 'SYRIAC',
 503   18 => 'THAANA',
 504   19 => 'DEVANAGARI',
 505   20 => 'BENGALI',
 506   21 => 'GURMUKHI',
 507   22 => 'GUJARATI',
 508   23 => 'ORIYA',
 509   24 => 'TAMIL',
 510   25 => 'TELUGU',
 511   26 => 'KANNADA',
 512   27 => 'MALAYALAM',
 513   28 => 'SINHALA',
 514   29 => 'THAI',
 515   30 => 'LAO',
 516   31 => 'TIBETAN',
 517   32 => 'MYANMAR',
 518   33 => 'GEORGIAN',
 519   34 => 'HANGUL',
 520   35 => 'ETHIOPIC',
 521   36 => 'CHEROKEE',
 522   37 => 'CANADIAN-ABORIGINAL',
 523   38 => 'OGHAM',
 524   39 => 'RUNIC',
 525   40 => 'KHMER',
 526   41 => 'MONGOLIAN',
 527   42 => 'HAN',
 528   43 => 'HIRAGANA',
 529   44 => 'KATAKANA',
 530   45 => 'BOPOMOFO',
 531   46 => 'YI',
 532   47 => 'OLD-ITALIC',
 533   48 => 'GOTHIC',
 534   49 => 'DESERET',
 535 );
 536
 537 %utf8::InBlock =
 538 (
 539   51 => 'Basic Latin',
 540   52 => 'Latin-1 Supplement',
 541   53 => 'Latin Extended-A',
 542   54 => 'Latin Extended-B',
 543   55 => 'IPA Extensions',
 544   56 => 'Spacing Modifier Letters',
 545   57 => 'Combining Diacritical Marks',
 546   58 => 'Greek',
 547   59 => 'Cyrillic',
 548   60 => 'Armenian',
 549   61 => 'Hebrew',
 550   62 => 'Arabic',
 551   63 => 'Syriac',
 552   64 => 'Thaana',
 553   65 => 'Devanagari',
 554   66 => 'Bengali',
 555   67 => 'Gurmukhi',
 556   68 => 'Gujarati',
 557   69 => 'Oriya',
 558   70 => 'Tamil',
 559   71 => 'Telugu',
 560   72 => 'Kannada',
 561   73 => 'Malayalam',
 562   74 => 'Sinhala',
 563   75 => 'Thai',
 564   76 => 'Lao',
 565   77 => 'Tibetan',
 566   78 => 'Myanmar',
 567   79 => 'Georgian',
 568   80 => 'Hangul Jamo',
 569   81 => 'Ethiopic',
 570   82 => 'Cherokee',
 571   83 => 'Unified Canadian Aboriginal Syllabics',
 572   84 => 'Ogham',
 573   85 => 'Runic',
 574   86 => 'Khmer',
 575   87 => 'Mongolian',
 576   88 => 'Latin Extended Additional',
 577   89 => 'Greek Extended',
 578   90 => 'General Punctuation',
 579   91 => 'Superscripts and Subscripts',
 580   92 => 'Currency Symbols',
 581   93 => 'Combining Marks for Symbols',
 582   94 => 'Letterlike Symbols',
 583   95 => 'Number Forms',
 584   96 => 'Arrows',
 585   97 => 'Mathematical Operators',
 586   98 => 'Miscellaneous Technical',
 587   99 => 'Control Pictures',
 588  100 => 'Optical Character Recognition',
 589  101 => 'Enclosed Alphanumerics',
 590  102 => 'Box Drawing',
 591  103 => 'Block Elements',
 592  104 => 'Geometric Shapes',
 593  105 => 'Miscellaneous Symbols',
 594  106 => 'Dingbats',
 595  107 => 'Braille Patterns',
 596  108 => 'CJK Radicals Supplement',
 597  109 => 'Kangxi Radicals',
 598  110 => 'Ideographic Description Characters',
 599  111 => 'CJK Symbols and Punctuation',
 600  112 => 'Hiragana',
 601  113 => 'Katakana',
 602  114 => 'Bopomofo',
 603  115 => 'Hangul Compatibility Jamo',
 604  116 => 'Kanbun',
 605  117 => 'Bopomofo Extended',
 606  118 => 'Enclosed CJK Letters and Months',
 607  119 => 'CJK Compatibility',
 608  120 => 'CJK Unified Ideographs Extension A',
 609  121 => 'CJK Unified Ideographs',
 610  122 => 'Yi Syllables',
 611  123 => 'Yi Radicals',
 612  124 => 'Hangul Syllables',
 613  125 => 'High Surrogates',
 614  126 => 'High Private Use Surrogates',
 615  127 => 'Low Surrogates',
 616  128 => 'CJK Compatibility Ideographs',
 617  129 => 'Alphabetic Presentation Forms',
 618  130 => 'Arabic Presentation Forms-A',
 619  131 => 'Combining Half Marks',
 620  132 => 'CJK Compatibility Forms',
 621  133 => 'Small Form Variants',
 622  134 => 'Arabic Presentation Forms-B',
 623  135 => 'Specials',
 624  136 => 'Halfwidth and Fullwidth Forms',
 625  137 => 'Old Italic',
 626  138 => 'Gothic',
 627  139 => 'Deseret',
 628  140 => 'Byzantine Musical Symbols',
 629  141 => 'Musical Symbols',
 630  142 => 'Mathematical Alphanumeric Symbols',
 631  143 => 'CJK Unified Ideographs Extension B',
 632  144 => 'CJK Compatibility Ideographs Supplement',
 633  145 => 'Tags',
 634 );