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 'Any'                                         => 137,
 143 'White_space'                                 => 138,
 144 'Bidi_Control'                                => 139,
 145 'Join_Control'                                => 140,
 146 'Dash'                                        => 141,
 147 'Hyphen'                                      => 142,
 148 'Quotation_Mark'                              => 143,
 149 'Terminal_Punctuation'                        => 144,
 150 'Other_Math'                                  => 145,
 151 'Hex_Digit'                                   => 146,
 152 'ASCII_Hex_Digit'                             => 147,
 153 'Other_Alphabetic'                            => 148,
 154 'Ideographic'                                 => 149,
 155 'Diacritic'                                   => 150,
 156 'Extender'                                    => 151,
 157 'Other_Lowercase'                             => 152,
 158 'Other_Uppercase'                             => 153,
 159 'Noncharacter_Code_Point'                     => 154,
 160 'Assigned'                                    => 155,
 161 'Alphabetic'                                  => 156,
 162 'Lowercase'                                   => 157,
 163 'Uppercase'                                   => 158,
 164 'Math'                                        => 159,
 165 'Lampersand'                                  => 160,
 166 'ID_Start'                                    => 161,
 167 'ID_Continue'                                 => 162,
 168 );
 169 %utf8::InPat = (
 170 'al' => {
 171         'Alphabetic(?:[-_]|\s+)?Presentation(?:[-_]|\s+)?Forms' => 'Alphabetic Presentation Forms',
 172         'Alphabetic' => 'Alphabetic',
 173 },
 174 'an' => {
 175         'Any' => 'Any',
 176 },
 177 'ar' => {
 178         'ARMENIAN' => 'ARMENIAN',
 179         'ARABIC' => 'ARABIC',
 180         'Armenian(?:[-_]|\s+)?Block' => 'Armenian Block',
 181         'Arabic(?:[-_]|\s+)?Block' => 'Arabic Block',
 182         'Arrows' => 'Arrows',
 183         'Arabic(?:[-_]|\s+)?Presentation(?:[-_]|\s+)?Forms(?:[-_]|\s+)?A' => 'Arabic Presentation Forms-A',
 184         'Arabic(?:[-_]|\s+)?Presentation(?:[-_]|\s+)?Forms(?:[-_]|\s+)?B' => 'Arabic Presentation Forms-B',
 185 },
 186 'as' => {
 187         'ASCII(?:[-_]|\s+)?Hex(?:[-_]|\s+)?Digit' => 'ASCII_Hex_Digit',
 188         'Assigned' => 'Assigned',
 189 },
 190 'ba' => {
 191         'Basic(?:[-_]|\s+)?Latin' => 'Basic Latin',
 192 },
 193 'be' => {
 194         'BENGALI' => 'BENGALI',
 195         'Bengali(?:[-_]|\s+)?Block' => 'Bengali Block',
 196 },
 197 'bi' => {
 198         'Bidi(?:[-_]|\s+)?Control' => 'Bidi_Control',
 199 },
 200 'bl' => {
 201         'Block(?:[-_]|\s+)?Elements' => 'Block Elements',
 202 },
 203 'bo' => {
 204         'BOPOMOFO' => 'BOPOMOFO',
 205         'Box(?:[-_]|\s+)?Drawing' => 'Box Drawing',
 206         'Bopomofo(?:[-_]|\s+)?Block' => 'Bopomofo Block',
 207         'Bopomofo(?:[-_]|\s+)?Extended' => 'Bopomofo Extended',
 208 },
 209 'br' => {
 210         'Braille(?:[-_]|\s+)?Patterns' => 'Braille Patterns',
 211 },
 212 'by' => {
 213         'Byzantine(?:[-_]|\s+)?Musical(?:[-_]|\s+)?Symbols' => 'Byzantine Musical Symbols',
 214 },
 215 'ca' => {
 216         'CANADIAN(?:[-_]|\s+)?ABORIGINAL' => 'CANADIAN-ABORIGINAL',
 217 },
 218 'ch' => {
 219         'CHEROKEE' => 'CHEROKEE',
 220         'Cherokee(?:[-_]|\s+)?Block' => 'Cherokee Block',
 221 },
 222 'cj' => {
 223         'CJK(?:[-_]|\s+)?Radicals(?:[-_]|\s+)?Supplement' => 'CJK Radicals Supplement',
 224         'CJK(?:[-_]|\s+)?Symbols(?:[-_]|\s+)?and(?:[-_]|\s+)?Punctuation' => 'CJK Symbols and Punctuation',
 225         'CJK(?:[-_]|\s+)?Compatibility' => 'CJK Compatibility',
 226         'CJK(?:[-_]|\s+)?Unified(?:[-_]|\s+)?Ideographs(?:[-_]|\s+)?Extension(?:[-_]|\s+)?A' => 'CJK Unified Ideographs Extension A',
 227         'CJK(?:[-_]|\s+)?Unified(?:[-_]|\s+)?Ideographs' => 'CJK Unified Ideographs',
 228         'CJK(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Ideographs' => 'CJK Compatibility Ideographs',
 229         'CJK(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Forms' => 'CJK Compatibility Forms',
 230         'CJK(?:[-_]|\s+)?Unified(?:[-_]|\s+)?Ideographs(?:[-_]|\s+)?Extension(?:[-_]|\s+)?B' => 'CJK Unified Ideographs Extension B',
 231         'CJK(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Ideographs(?:[-_]|\s+)?Supplement' => 'CJK Compatibility Ideographs Supplement',
 232 },
 233 'co' => {
 234         'Combining(?:[-_]|\s+)?Diacritical(?:[-_]|\s+)?Marks' => 'Combining Diacritical Marks',
 235         'Combining(?:[-_]|\s+)?Marks(?:[-_]|\s+)?for(?:[-_]|\s+)?Symbols' => 'Combining Marks for Symbols',
 236         'Control(?:[-_]|\s+)?Pictures' => 'Control Pictures',
 237         'Combining(?:[-_]|\s+)?Half(?:[-_]|\s+)?Marks' => 'Combining Half Marks',
 238         'Common' => 'Common',
 239 },
 240 'cu' => {
 241         'Currency(?:[-_]|\s+)?Symbols' => 'Currency Symbols',
 242 },
 243 'cy' => {
 244         'CYRILLIC' => 'CYRILLIC',
 245         'Cyrillic(?:[-_]|\s+)?Block' => 'Cyrillic Block',
 246 },
 247 'da' => {
 248         'Dash' => 'Dash',
 249 },
 250 'de' => {
 251         'DEVANAGARI' => 'DEVANAGARI',
 252         'DESERET' => 'DESERET',
 253         'Devanagari(?:[-_]|\s+)?Block' => 'Devanagari Block',
 254         'Deseret(?:[-_]|\s+)?Block' => 'Deseret Block',
 255 },
 256 'di' => {
 257         'Dingbats' => 'Dingbats',
 258         'Diacritic' => 'Diacritic',
 259 },
 260 'en' => {
 261         'Enclosed(?:[-_]|\s+)?Alphanumerics' => 'Enclosed Alphanumerics',
 262         'Enclosed(?:[-_]|\s+)?CJK(?:[-_]|\s+)?Letters(?:[-_]|\s+)?and(?:[-_]|\s+)?Months' => 'Enclosed CJK Letters and Months',
 263 },
 264 'et' => {
 265         'ETHIOPIC' => 'ETHIOPIC',
 266         'Ethiopic(?:[-_]|\s+)?Block' => 'Ethiopic Block',
 267 },
 268 'ex' => {
 269         'Extender' => 'Extender',
 270 },
 271 'ge' => {
 272         'GEORGIAN' => 'GEORGIAN',
 273         'Georgian(?:[-_]|\s+)?Block' => 'Georgian Block',
 274         'General(?:[-_]|\s+)?Punctuation' => 'General Punctuation',
 275         'Geometric(?:[-_]|\s+)?Shapes' => 'Geometric Shapes',
 276 },
 277 'go' => {
 278         'GOTHIC' => 'GOTHIC',
 279         'Gothic(?:[-_]|\s+)?Block' => 'Gothic Block',
 280 },
 281 'gr' => {
 282         'GREEK' => 'GREEK',
 283         'Greek(?:[-_]|\s+)?Block' => 'Greek Block',
 284         'Greek(?:[-_]|\s+)?Extended' => 'Greek Extended',
 285 },
 286 'gu' => {
 287         'GURMUKHI' => 'GURMUKHI',
 288         'GUJARATI' => 'GUJARATI',
 289         'Gurmukhi(?:[-_]|\s+)?Block' => 'Gurmukhi Block',
 290         'Gujarati(?:[-_]|\s+)?Block' => 'Gujarati Block',
 291 },
 292 'ha' => {
 293         'HANGUL' => 'HANGUL',
 294         'HAN' => 'HAN',
 295         'Hangul(?:[-_]|\s+)?Jamo' => 'Hangul Jamo',
 296         'Hangul(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Jamo' => 'Hangul Compatibility Jamo',
 297         'Hangul(?:[-_]|\s+)?Syllables' => 'Hangul Syllables',
 298         'Halfwidth(?:[-_]|\s+)?and(?:[-_]|\s+)?Fullwidth(?:[-_]|\s+)?Forms' => 'Halfwidth and Fullwidth Forms',
 299 },
 300 'he' => {
 301         'HEBREW' => 'HEBREW',
 302         'Hebrew(?:[-_]|\s+)?Block' => 'Hebrew Block',
 303         'Hex(?:[-_]|\s+)?Digit' => 'Hex_Digit',
 304 },
 305 'hi' => {
 306         'HIRAGANA' => 'HIRAGANA',
 307         'Hiragana(?:[-_]|\s+)?Block' => 'Hiragana Block',
 308         'High(?:[-_]|\s+)?Surrogates' => 'High Surrogates',
 309         'High(?:[-_]|\s+)?Private(?:[-_]|\s+)?Use(?:[-_]|\s+)?Surrogates' => 'High Private Use Surrogates',
 310 },
 311 'hy' => {
 312         'Hyphen' => 'Hyphen',
 313 },
 314 'id' => {
 315         'Ideographic(?:[-_]|\s+)?Description(?:[-_]|\s+)?Characters' => 'Ideographic Description Characters',
 316         'Ideographic' => 'Ideographic',
 317         'ID(?:[-_]|\s+)?Start' => 'ID_Start',
 318         'ID(?:[-_]|\s+)?Continue' => 'ID_Continue',
 319 },
 320 'in' => {
 321         'INHERITED' => 'INHERITED',
 322 },
 323 'ip' => {
 324         'IPA(?:[-_]|\s+)?Extensions' => 'IPA Extensions',
 325 },
 326 'jo' => {
 327         'Join(?:[-_]|\s+)?Control' => 'Join_Control',
 328 },
 329 'ka' => {
 330         'KANNADA' => 'KANNADA',
 331         'KATAKANA' => 'KATAKANA',
 332         'Kannada(?:[-_]|\s+)?Block' => 'Kannada Block',
 333         'Kangxi(?:[-_]|\s+)?Radicals' => 'Kangxi Radicals',
 334         'Katakana(?:[-_]|\s+)?Block' => 'Katakana Block',
 335         'Kanbun' => 'Kanbun',
 336 },
 337 'kh' => {
 338         'KHMER' => 'KHMER',
 339         'Khmer(?:[-_]|\s+)?Block' => 'Khmer Block',
 340 },
 341 'la' => {
 342         'LATIN' => 'LATIN',
 343         'LAO' => 'LAO',
 344         'Latin(?:[-_]|\s+)?1(?:[-_]|\s+)?Supplement' => 'Latin-1 Supplement',
 345         'Latin(?:[-_]|\s+)?Extended(?:[-_]|\s+)?A' => 'Latin Extended-A',
 346         'Latin(?:[-_]|\s+)?Extended(?:[-_]|\s+)?B' => 'Latin Extended-B',
 347         'Lao(?:[-_]|\s+)?Block' => 'Lao Block',
 348         'Latin(?:[-_]|\s+)?Extended(?:[-_]|\s+)?Additional' => 'Latin Extended Additional',
 349         'Lampersand' => 'Lampersand',
 350 },
 351 'le' => {
 352         'Letterlike(?:[-_]|\s+)?Symbols' => 'Letterlike Symbols',
 353 },
 354 'lo' => {
 355         'Low(?:[-_]|\s+)?Surrogates' => 'Low Surrogates',
 356         'Lowercase' => 'Lowercase',
 357 },
 358 'ma' => {
 359         'MALAYALAM' => 'MALAYALAM',
 360         'Malayalam(?:[-_]|\s+)?Block' => 'Malayalam Block',
 361         'Mathematical(?:[-_]|\s+)?Operators' => 'Mathematical Operators',
 362         'Mathematical(?:[-_]|\s+)?Alphanumeric(?:[-_]|\s+)?Symbols' => 'Mathematical Alphanumeric Symbols',
 363         'Math' => 'Math',
 364 },
 365 'mi' => {
 366         'Miscellaneous(?:[-_]|\s+)?Technical' => 'Miscellaneous Technical',
 367         'Miscellaneous(?:[-_]|\s+)?Symbols' => 'Miscellaneous Symbols',
 368 },
 369 'mo' => {
 370         'MONGOLIAN' => 'MONGOLIAN',
 371         'Mongolian(?:[-_]|\s+)?Block' => 'Mongolian Block',
 372 },
 373 'mu' => {
 374         'Musical(?:[-_]|\s+)?Symbols' => 'Musical Symbols',
 375 },
 376 'my' => {
 377         'MYANMAR' => 'MYANMAR',
 378         'Myanmar(?:[-_]|\s+)?Block' => 'Myanmar Block',
 379 },
 380 'no' => {
 381         'Noncharacter(?:[-_]|\s+)?Code(?:[-_]|\s+)?Point' => 'Noncharacter_Code_Point',
 382 },
 383 'nu' => {
 384         'Number(?:[-_]|\s+)?Forms' => 'Number Forms',
 385 },
 386 'og' => {
 387         'OGHAM' => 'OGHAM',
 388         'Ogham(?:[-_]|\s+)?Block' => 'Ogham Block',
 389 },
 390 'ol' => {
 391         'OLD(?:[-_]|\s+)?ITALIC' => 'OLD-ITALIC',
 392         'Old(?:[-_]|\s+)?Italic' => 'Old Italic',
 393 },
 394 'op' => {
 395         'Optical(?:[-_]|\s+)?Character(?:[-_]|\s+)?Recognition' => 'Optical Character Recognition',
 396 },
 397 'or' => {
 398         'ORIYA' => 'ORIYA',
 399         'Oriya(?:[-_]|\s+)?Block' => 'Oriya Block',
 400 },
 401 'ot' => {
 402         'Other(?:[-_]|\s+)?Math' => 'Other_Math',
 403         'Other(?:[-_]|\s+)?Alphabetic' => 'Other_Alphabetic',
 404         'Other(?:[-_]|\s+)?Lowercase' => 'Other_Lowercase',
 405         'Other(?:[-_]|\s+)?Uppercase' => 'Other_Uppercase',
 406 },
 407 'pr' => {
 408         'Private(?:[-_]|\s+)?Use' => 'Private Use',
 409 },
 410 'qu' => {
 411         'Quotation(?:[-_]|\s+)?Mark' => 'Quotation_Mark',
 412 },
 413 'ru' => {
 414         'RUNIC' => 'RUNIC',
 415         'Runic(?:[-_]|\s+)?Block' => 'Runic Block',
 416 },
 417 'si' => {
 418         'SINHALA' => 'SINHALA',
 419         'Sinhala(?:[-_]|\s+)?Block' => 'Sinhala Block',
 420 },
 421 'sm' => {
 422         'Small(?:[-_]|\s+)?Form(?:[-_]|\s+)?Variants' => 'Small Form Variants',
 423 },
 424 'sp' => {
 425         'Spacing(?:[-_]|\s+)?Modifier(?:[-_]|\s+)?Letters' => 'Spacing Modifier Letters',
 426         'Specials' => 'Specials',
 427 },
 428 'su' => {
 429         'Superscripts(?:[-_]|\s+)?and(?:[-_]|\s+)?Subscripts' => 'Superscripts and Subscripts',
 430 },
 431 'sy' => {
 432         'SYRIAC' => 'SYRIAC',
 433         'Syriac(?:[-_]|\s+)?Block' => 'Syriac Block',
 434 },
 435 'ta' => {
 436         'TAMIL' => 'TAMIL',
 437         'Tamil(?:[-_]|\s+)?Block' => 'Tamil Block',
 438         'Tags' => 'Tags',
 439 },
 440 'te' => {
 441         'TELUGU' => 'TELUGU',
 442         'Telugu(?:[-_]|\s+)?Block' => 'Telugu Block',
 443         'Terminal(?:[-_]|\s+)?Punctuation' => 'Terminal_Punctuation',
 444 },
 445 'th' => {
 446         'THAANA' => 'THAANA',
 447         'THAI' => 'THAI',
 448         'Thaana(?:[-_]|\s+)?Block' => 'Thaana Block',
 449         'Thai(?:[-_]|\s+)?Block' => 'Thai Block',
 450 },
 451 'ti' => {
 452         'TIBETAN' => 'TIBETAN',
 453         'Tibetan(?:[-_]|\s+)?Block' => 'Tibetan Block',
 454 },
 455 'un' => {
 456         'Unified(?:[-_]|\s+)?Canadian(?:[-_]|\s+)?Aboriginal(?:[-_]|\s+)?Syllabics' => 'Unified Canadian Aboriginal Syllabics',
 457 },
 458 'up' => {
 459         'Uppercase' => 'Uppercase',
 460 },
 461 'wh' => {
 462         'White(?:[-_]|\s+)?space' => 'White_space',
 463 },
 464 'yi' => {
 465         'YI' => 'YI',
 466         'Yi(?:[-_]|\s+)?Syllables' => 'Yi Syllables',
 467         'Yi(?:[-_]|\s+)?Radicals' => 'Yi Radicals',
 468 },
 469 );