[p5sagit/p5-mst-13.2.git] / lib / unicore / In.pl

# !!!!!!!   DO NOT EDIT THIS FILE   !!!!!!! 
# This file is built by mktables from e.g. Unicode.txt.
# Any changes made here will be lost!
%utf8::In =
(
'Alphabetic'                                  => '164',
'Alphabetic Presentation Forms'               => '129',
'Any'                                         => '171',
'ARABIC'                                      => '16',
'Arabic Block'                                => '62',
'Arabic Presentation Forms-A'                 => '130',
'Arabic Presentation Forms-B'                 => '134',
'ARMENIAN'                                    => '14',
'Armenian Block'                              => '60',
'Arrows'                                      => '96',
'ASCII_Hex_Digit'                             => '152',
'Assigned'                                    => '163',
'Basic Latin'                                 => '51',
'BENGALI'                                     => '20',
'Bengali Block'                               => '66',
'Bidi_Control'                                => '159',
'Block Elements'                              => '103',
'BOPOMOFO'                                    => '45',
'Bopomofo Block'                              => '114',
'Bopomofo Extended'                           => '117',
'Box Drawing'                                 => '102',
'Braille Patterns'                            => '107',
'Byzantine Musical Symbols'                   => '140',
'CANADIAN-ABORIGINAL'                         => '37',
'CHEROKEE'                                    => '36',
'Cherokee Block'                              => '82',
'CJK Compatibility'                           => '119',
'CJK Compatibility Forms'                     => '132',
'CJK Compatibility Ideographs'                => '128',
'CJK Compatibility Ideographs Supplement'     => '144',
'CJK Ideograph'                               => '1',
'CJK Ideograph Extension A'                   => '0',
'CJK Ideograph Extension B'                   => '7',
'CJK Radicals Supplement'                     => '108',
'CJK Symbols and Punctuation'                 => '111',
'CJK Unified Ideographs'                      => '121',
'CJK Unified Ideographs Extension A'          => '120',
'CJK Unified Ideographs Extension B'          => '143',
'Combining Diacritical Marks'                 => '57',
'Combining Half Marks'                        => '131',
'Combining Marks for Symbols'                 => '93',
'Common'                                      => '50',
'Control Pictures'                            => '99',
'Currency Symbols'                            => '92',
'CYRILLIC'                                    => '13',
'Cyrillic Block'                              => '59',
'Dash'                                        => '151',
'DESERET'                                     => '49',
'Deseret Block'                               => '139',
'DEVANAGARI'                                  => '19',
'Devanagari Block'                            => '65',
'Diacritic'                                   => '154',
'Dingbats'                                    => '106',
'Enclosed Alphanumerics'                      => '101',
'Enclosed CJK Letters and Months'             => '118',
'ETHIOPIC'                                    => '35',
'Ethiopic Block'                              => '81',
'Extender'                                    => '155',
'General Punctuation'                         => '90',
'Geometric Shapes'                            => '104',
'GEORGIAN'                                    => '33',
'Georgian Block'                              => '79',
'GOTHIC'                                      => '48',
'Gothic Block'                                => '138',
'GREEK'                                       => '11',
'Greek Block'                                 => '58',
'Greek Extended'                              => '89',
'GUJARATI'                                    => '22',
'Gujarati Block'                              => '68',
'GURMUKHI'                                    => '21',
'Gurmukhi Block'                              => '67',
'Halfwidth and Fullwidth Forms'               => '136',
'HAN'                                         => '42',
'HANGUL'                                      => '34',
'Hangul Compatibility Jamo'                   => '115',
'Hangul Jamo'                                 => '80',
'Hangul Syllable'                             => '2',
'Hangul Syllables'                            => '124',
'HEBREW'                                      => '15',
'Hebrew Block'                                => '61',
'Hex_Digit'                                   => '153',
'High Private Use Surrogates'                 => '126',
'High Surrogates'                             => '125',
'HIRAGANA'                                    => '43',
'Hiragana Block'                              => '112',
'Hyphen'                                      => '150',
'ID_Continue'                                 => '170',
'ID_Start'                                    => '169',
'Ideographic'                                 => '161',
'Ideographic Description Characters'          => '110',
'INHERITED'                                   => '12',
'IPA Extensions'                              => '55',
'Join_Control'                                => '158',
'Kanbun'                                      => '116',
'Kangxi Radicals'                             => '109',
'KANNADA'                                     => '26',
'Kannada Block'                               => '72',
'KATAKANA'                                    => '44',
'Katakana Block'                              => '113',
'KHMER'                                       => '40',
'Khmer Block'                                 => '86',
'Lampersand'                                  => '168',
'LAO'                                         => '30',
'Lao Block'                                   => '76',
'LATIN'                                       => '10',
'Latin Extended Additional'                   => '88',
'Latin Extended-A'                            => '53',
'Latin Extended-B'                            => '54',
'Latin-1 Supplement'                          => '52',
'Letterlike Symbols'                          => '94',
'Low Surrogate'                               => '5',
'Low Surrogates'                              => '127',
'Lowercase'                                   => '165',
'MALAYALAM'                                   => '27',
'Malayalam Block'                             => '73',
'Math'                                        => '167',
'Mathematical Alphanumeric Symbols'           => '142',
'Mathematical Operators'                      => '97',
'Miscellaneous Symbols'                       => '105',
'Miscellaneous Technical'                     => '98',
'MONGOLIAN'                                   => '41',
'Mongolian Block'                             => '87',
'Musical Symbols'                             => '141',
'MYANMAR'                                     => '32',
'Myanmar Block'                               => '78',
'Non Private Use High Surrogate'              => '3',
'Noncharacter_Code_Point'                     => '162',
'Number Forms'                                => '95',
'OGHAM'                                       => '38',
'Ogham Block'                                 => '84',
'Old Italic Block'                            => '137',
'OLD-ITALIC'                                  => '47',
'Optical Character Recognition'               => '100',
'ORIYA'                                       => '23',
'Oriya Block'                                 => '69',
'Other_Alphabetic'                            => '157',
'Other_Lowercase'                             => '156',
'Other_Math'                                  => '149',
'Other_Uppercase'                             => '160',
'Plane 15 Private Use'                        => '8',
'Plane 16 Private Use'                        => '9',
'Private Use'                                 => '6',
'Private Use High Surrogate'                  => '4',
'Quotation_Mark'                              => '148',
'RUNIC'                                       => '39',
'Runic Block'                                 => '85',
'SINHALA'                                     => '28',
'Sinhala Block'                               => '74',
'Small Form Variants'                         => '133',
'Spacing Modifier Letters'                    => '56',
'Specials'                                    => '135',
'Superscripts and Subscripts'                 => '91',
'SYRIAC'                                      => '17',
'Syriac Block'                                => '63',
'Tags'                                        => '145',
'TAMIL'                                       => '24',
'Tamil Block'                                 => '70',
'TELUGU'                                      => '25',
'Telugu Block'                                => '71',
'Terminal_Punctuation'                        => '147',
'THAANA'                                      => '18',
'Thaana Block'                                => '64',
'THAI'                                        => '29',
'Thai Block'                                  => '75',
'TIBETAN'                                     => '31',
'Tibetan Block'                               => '77',
'Unified Canadian Aboriginal Syllabics'       => '83',
'Uppercase'                                   => '166',
'White_space'                                 => '146',
'YI'                                          => '46',
'Yi Radicals'                                 => '123',
'Yi Syllables'                                => '122',
);
%utf8::InPat =
(
'al' => {
	'Alphabetic' => '164',
	'Alphabetic(?:[-_]|\s+)?Presentation(?:[-_]|\s+)?Forms' => '129',
},
'an' => {
	'Any' => '171',
},
'ar' => {
	'ARABIC' => '16',
	'Arabic(?:[-_]|\s+)?Block' => '62',
	'Arabic(?:[-_]|\s+)?Presentation(?:[-_]|\s+)?Forms(?:[-_]|\s+)?A' => '130',
	'Arabic(?:[-_]|\s+)?Presentation(?:[-_]|\s+)?Forms(?:[-_]|\s+)?B' => '134',
	'ARMENIAN' => '14',
	'Armenian(?:[-_]|\s+)?Block' => '60',
	'Arrows' => '96',
},
'as' => {
	'ASCII(?:[-_]|\s+)?Hex(?:[-_]|\s+)?Digit' => '152',
	'Assigned' => '163',
},
'ba' => {
	'Basic(?:[-_]|\s+)?Latin' => '51',
},
'be' => {
	'BENGALI' => '20',
	'Bengali(?:[-_]|\s+)?Block' => '66',
},
'bi' => {
	'Bidi(?:[-_]|\s+)?Control' => '159',
},
'bl' => {
	'Block(?:[-_]|\s+)?Elements' => '103',
},
'bo' => {
	'BOPOMOFO' => '45',
	'Bopomofo(?:[-_]|\s+)?Block' => '114',
	'Bopomofo(?:[-_]|\s+)?Extended' => '117',
	'Box(?:[-_]|\s+)?Drawing' => '102',
},
'br' => {
	'Braille(?:[-_]|\s+)?Patterns' => '107',
},
'by' => {
	'Byzantine(?:[-_]|\s+)?Musical(?:[-_]|\s+)?Symbols' => '140',
},
'ca' => {
	'CANADIAN(?:[-_]|\s+)?ABORIGINAL' => '37',
},
'ch' => {
	'CHEROKEE' => '36',
	'Cherokee(?:[-_]|\s+)?Block' => '82',
},
'cj' => {
	'CJK(?:[-_]|\s+)?Compatibility' => '119',
	'CJK(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Forms' => '132',
	'CJK(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Ideographs' => '128',
	'CJK(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Ideographs(?:[-_]|\s+)?Supplement' => '144',
	'CJK(?:[-_]|\s+)?Ideograph' => '1',
	'CJK(?:[-_]|\s+)?Ideograph(?:[-_]|\s+)?Extension(?:[-_]|\s+)?A' => '0',
	'CJK(?:[-_]|\s+)?Ideograph(?:[-_]|\s+)?Extension(?:[-_]|\s+)?B' => '7',
	'CJK(?:[-_]|\s+)?Radicals(?:[-_]|\s+)?Supplement' => '108',
	'CJK(?:[-_]|\s+)?Symbols(?:[-_]|\s+)?and(?:[-_]|\s+)?Punctuation' => '111',
	'CJK(?:[-_]|\s+)?Unified(?:[-_]|\s+)?Ideographs' => '121',
	'CJK(?:[-_]|\s+)?Unified(?:[-_]|\s+)?Ideographs(?:[-_]|\s+)?Extension(?:[-_]|\s+)?A' => '120',
	'CJK(?:[-_]|\s+)?Unified(?:[-_]|\s+)?Ideographs(?:[-_]|\s+)?Extension(?:[-_]|\s+)?B' => '143',
},
'co' => {
	'Combining(?:[-_]|\s+)?Diacritical(?:[-_]|\s+)?Marks' => '57',
	'Combining(?:[-_]|\s+)?Half(?:[-_]|\s+)?Marks' => '131',
	'Combining(?:[-_]|\s+)?Marks(?:[-_]|\s+)?for(?:[-_]|\s+)?Symbols' => '93',
	'Common' => '50',
	'Control(?:[-_]|\s+)?Pictures' => '99',
},
'cu' => {
	'Currency(?:[-_]|\s+)?Symbols' => '92',
},
'cy' => {
	'CYRILLIC' => '13',
	'Cyrillic(?:[-_]|\s+)?Block' => '59',
},
'da' => {
	'Dash' => '151',
},
'de' => {
	'DESERET' => '49',
	'Deseret(?:[-_]|\s+)?Block' => '139',
	'DEVANAGARI' => '19',
	'Devanagari(?:[-_]|\s+)?Block' => '65',
},
'di' => {
	'Diacritic' => '154',
	'Dingbats' => '106',
},
'en' => {
	'Enclosed(?:[-_]|\s+)?Alphanumerics' => '101',
	'Enclosed(?:[-_]|\s+)?CJK(?:[-_]|\s+)?Letters(?:[-_]|\s+)?and(?:[-_]|\s+)?Months' => '118',
},
'et' => {
	'ETHIOPIC' => '35',
	'Ethiopic(?:[-_]|\s+)?Block' => '81',
},
'ex' => {
	'Extender' => '155',
},
'ge' => {
	'General(?:[-_]|\s+)?Punctuation' => '90',
	'Geometric(?:[-_]|\s+)?Shapes' => '104',
	'GEORGIAN' => '33',
	'Georgian(?:[-_]|\s+)?Block' => '79',
},
'go' => {
	'GOTHIC' => '48',
	'Gothic(?:[-_]|\s+)?Block' => '138',
},
'gr' => {
	'GREEK' => '11',
	'Greek(?:[-_]|\s+)?Block' => '58',
	'Greek(?:[-_]|\s+)?Extended' => '89',
},
'gu' => {
	'GUJARATI' => '22',
	'Gujarati(?:[-_]|\s+)?Block' => '68',
	'GURMUKHI' => '21',
	'Gurmukhi(?:[-_]|\s+)?Block' => '67',
},
'ha' => {
	'Halfwidth(?:[-_]|\s+)?and(?:[-_]|\s+)?Fullwidth(?:[-_]|\s+)?Forms' => '136',
	'HAN' => '42',
	'HANGUL' => '34',
	'Hangul(?:[-_]|\s+)?Compatibility(?:[-_]|\s+)?Jamo' => '115',
	'Hangul(?:[-_]|\s+)?Jamo' => '80',
	'Hangul(?:[-_]|\s+)?Syllable' => '2',
	'Hangul(?:[-_]|\s+)?Syllables' => '124',
},
'he' => {
	'HEBREW' => '15',
	'Hebrew(?:[-_]|\s+)?Block' => '61',
	'Hex(?:[-_]|\s+)?Digit' => '153',
},
'hi' => {
	'High(?:[-_]|\s+)?Private(?:[-_]|\s+)?Use(?:[-_]|\s+)?Surrogates' => '126',
	'High(?:[-_]|\s+)?Surrogates' => '125',
	'HIRAGANA' => '43',
	'Hiragana(?:[-_]|\s+)?Block' => '112',
},
'hy' => {
	'Hyphen' => '150',
},
'id' => {
	'ID(?:[-_]|\s+)?Continue' => '170',
	'ID(?:[-_]|\s+)?Start' => '169',
	'Ideographic' => '161',
	'Ideographic(?:[-_]|\s+)?Description(?:[-_]|\s+)?Characters' => '110',
},
'in' => {
	'INHERITED' => '12',
},
'ip' => {
	'IPA(?:[-_]|\s+)?Extensions' => '55',
},
'jo' => {
	'Join(?:[-_]|\s+)?Control' => '158',
},
'ka' => {
	'Kanbun' => '116',
	'Kangxi(?:[-_]|\s+)?Radicals' => '109',
	'KANNADA' => '26',
	'Kannada(?:[-_]|\s+)?Block' => '72',
	'KATAKANA' => '44',
	'Katakana(?:[-_]|\s+)?Block' => '113',
},
'kh' => {
	'KHMER' => '40',
	'Khmer(?:[-_]|\s+)?Block' => '86',
},
'la' => {
	'Lampersand' => '168',
	'LAO' => '30',
	'Lao(?:[-_]|\s+)?Block' => '76',
	'LATIN' => '10',
	'Latin(?:[-_]|\s+)?Extended(?:[-_]|\s+)?Additional' => '88',
	'Latin(?:[-_]|\s+)?Extended(?:[-_]|\s+)?A' => '53',
	'Latin(?:[-_]|\s+)?Extended(?:[-_]|\s+)?B' => '54',
	'Latin(?:[-_]|\s+)?1(?:[-_]|\s+)?Supplement' => '52',
},
'le' => {
	'Letterlike(?:[-_]|\s+)?Symbols' => '94',
},
'lo' => {
	'Low(?:[-_]|\s+)?Surrogate' => '5',
	'Low(?:[-_]|\s+)?Surrogates' => '127',
	'Lowercase' => '165',
},
'ma' => {
	'MALAYALAM' => '27',
	'Malayalam(?:[-_]|\s+)?Block' => '73',
	'Math' => '167',
	'Mathematical(?:[-_]|\s+)?Alphanumeric(?:[-_]|\s+)?Symbols' => '142',
	'Mathematical(?:[-_]|\s+)?Operators' => '97',
},
'mi' => {
	'Miscellaneous(?:[-_]|\s+)?Symbols' => '105',
	'Miscellaneous(?:[-_]|\s+)?Technical' => '98',
},
'mo' => {
	'MONGOLIAN' => '41',
	'Mongolian(?:[-_]|\s+)?Block' => '87',
},
'mu' => {
	'Musical(?:[-_]|\s+)?Symbols' => '141',
},
'my' => {
	'MYANMAR' => '32',
	'Myanmar(?:[-_]|\s+)?Block' => '78',
},
'no' => {
	'Non(?:[-_]|\s+)?Private(?:[-_]|\s+)?Use(?:[-_]|\s+)?High(?:[-_]|\s+)?Surrogate' => '3',
	'Noncharacter(?:[-_]|\s+)?Code(?:[-_]|\s+)?Point' => '162',
},
'nu' => {
	'Number(?:[-_]|\s+)?Forms' => '95',
},
'og' => {
	'OGHAM' => '38',
	'Ogham(?:[-_]|\s+)?Block' => '84',
},
'ol' => {
	'Old(?:[-_]|\s+)?Italic(?:[-_]|\s+)?Block' => '137',
	'OLD(?:[-_]|\s+)?ITALIC' => '47',
},
'op' => {
	'Optical(?:[-_]|\s+)?Character(?:[-_]|\s+)?Recognition' => '100',
},
'or' => {
	'ORIYA' => '23',
	'Oriya(?:[-_]|\s+)?Block' => '69',
},
'ot' => {
	'Other(?:[-_]|\s+)?Alphabetic' => '157',
	'Other(?:[-_]|\s+)?Lowercase' => '156',
	'Other(?:[-_]|\s+)?Math' => '149',
	'Other(?:[-_]|\s+)?Uppercase' => '160',
},
'pl' => {
	'Plane(?:[-_]|\s+)?15(?:[-_]|\s+)?Private(?:[-_]|\s+)?Use' => '8',
	'Plane(?:[-_]|\s+)?16(?:[-_]|\s+)?Private(?:[-_]|\s+)?Use' => '9',
},
'pr' => {
	'Private(?:[-_]|\s+)?Use' => '6',
	'Private(?:[-_]|\s+)?Use(?:[-_]|\s+)?High(?:[-_]|\s+)?Surrogate' => '4',
},
'qu' => {
	'Quotation(?:[-_]|\s+)?Mark' => '148',
},
'ru' => {
	'RUNIC' => '39',
	'Runic(?:[-_]|\s+)?Block' => '85',
},
'si' => {
	'SINHALA' => '28',
	'Sinhala(?:[-_]|\s+)?Block' => '74',
},
'sm' => {
	'Small(?:[-_]|\s+)?Form(?:[-_]|\s+)?Variants' => '133',
},
'sp' => {
	'Spacing(?:[-_]|\s+)?Modifier(?:[-_]|\s+)?Letters' => '56',
	'Specials' => '135',
},
'su' => {
	'Superscripts(?:[-_]|\s+)?and(?:[-_]|\s+)?Subscripts' => '91',
},
'sy' => {
	'SYRIAC' => '17',
	'Syriac(?:[-_]|\s+)?Block' => '63',
},
'ta' => {
	'Tags' => '145',
	'TAMIL' => '24',
	'Tamil(?:[-_]|\s+)?Block' => '70',
},
'te' => {
	'TELUGU' => '25',
	'Telugu(?:[-_]|\s+)?Block' => '71',
	'Terminal(?:[-_]|\s+)?Punctuation' => '147',
},
'th' => {
	'THAANA' => '18',
	'Thaana(?:[-_]|\s+)?Block' => '64',
	'THAI' => '29',
	'Thai(?:[-_]|\s+)?Block' => '75',
},
'ti' => {
	'TIBETAN' => '31',
	'Tibetan(?:[-_]|\s+)?Block' => '77',
},
'un' => {
	'Unified(?:[-_]|\s+)?Canadian(?:[-_]|\s+)?Aboriginal(?:[-_]|\s+)?Syllabics' => '83',
},
'up' => {
	'Uppercase' => '166',
},
'wh' => {
	'White(?:[-_]|\s+)?space' => '146',
},
'yi' => {
	'YI' => '46',
	'Yi(?:[-_]|\s+)?Radicals' => '123',
	'Yi(?:[-_]|\s+)?Syllables' => '122',
},
);

%utf8::InScript =
(
  10 => 'LATIN',
  11 => 'GREEK',
  12 => 'INHERITED',
  13 => 'CYRILLIC',
  14 => 'ARMENIAN',
  15 => 'HEBREW',
  16 => 'ARABIC',
  17 => 'SYRIAC',
  18 => 'THAANA',
  19 => 'DEVANAGARI',
  20 => 'BENGALI',
  21 => 'GURMUKHI',
  22 => 'GUJARATI',
  23 => 'ORIYA',
  24 => 'TAMIL',
  25 => 'TELUGU',
  26 => 'KANNADA',
  27 => 'MALAYALAM',
  28 => 'SINHALA',
  29 => 'THAI',
  30 => 'LAO',
  31 => 'TIBETAN',
  32 => 'MYANMAR',
  33 => 'GEORGIAN',
  34 => 'HANGUL',
  35 => 'ETHIOPIC',
  36 => 'CHEROKEE',
  37 => 'CANADIAN-ABORIGINAL',
  38 => 'OGHAM',
  39 => 'RUNIC',
  40 => 'KHMER',
  41 => 'MONGOLIAN',
  42 => 'HAN',
  43 => 'HIRAGANA',
  44 => 'KATAKANA',
  45 => 'BOPOMOFO',
  46 => 'YI',
  47 => 'OLD-ITALIC',
  48 => 'GOTHIC',
  49 => 'DESERET',
);

%utf8::InBlock =
(
  51 => 'Basic Latin',
  52 => 'Latin-1 Supplement',
  53 => 'Latin Extended-A',
  54 => 'Latin Extended-B',
  55 => 'IPA Extensions',
  56 => 'Spacing Modifier Letters',
  57 => 'Combining Diacritical Marks',
  58 => 'Greek',
  59 => 'Cyrillic',
  60 => 'Armenian',
  61 => 'Hebrew',
  62 => 'Arabic',
  63 => 'Syriac',
  64 => 'Thaana',
  65 => 'Devanagari',
  66 => 'Bengali',
  67 => 'Gurmukhi',
  68 => 'Gujarati',
  69 => 'Oriya',
  70 => 'Tamil',
  71 => 'Telugu',
  72 => 'Kannada',
  73 => 'Malayalam',
  74 => 'Sinhala',
  75 => 'Thai',
  76 => 'Lao',
  77 => 'Tibetan',
  78 => 'Myanmar',
  79 => 'Georgian',
  80 => 'Hangul Jamo',
  81 => 'Ethiopic',
  82 => 'Cherokee',
  83 => 'Unified Canadian Aboriginal Syllabics',
  84 => 'Ogham',
  85 => 'Runic',
  86 => 'Khmer',
  87 => 'Mongolian',
  88 => 'Latin Extended Additional',
  89 => 'Greek Extended',
  90 => 'General Punctuation',
  91 => 'Superscripts and Subscripts',
  92 => 'Currency Symbols',
  93 => 'Combining Marks for Symbols',
  94 => 'Letterlike Symbols',
  95 => 'Number Forms',
  96 => 'Arrows',
  97 => 'Mathematical Operators',
  98 => 'Miscellaneous Technical',
  99 => 'Control Pictures',
 100 => 'Optical Character Recognition',
 101 => 'Enclosed Alphanumerics',
 102 => 'Box Drawing',
 103 => 'Block Elements',
 104 => 'Geometric Shapes',
 105 => 'Miscellaneous Symbols',
 106 => 'Dingbats',
 107 => 'Braille Patterns',
 108 => 'CJK Radicals Supplement',
 109 => 'Kangxi Radicals',
 110 => 'Ideographic Description Characters',
 111 => 'CJK Symbols and Punctuation',
 112 => 'Hiragana',
 113 => 'Katakana',
 114 => 'Bopomofo',
 115 => 'Hangul Compatibility Jamo',
 116 => 'Kanbun',
 117 => 'Bopomofo Extended',
 118 => 'Enclosed CJK Letters and Months',
 119 => 'CJK Compatibility',
 120 => 'CJK Unified Ideographs Extension A',
 121 => 'CJK Unified Ideographs',
 122 => 'Yi Syllables',
 123 => 'Yi Radicals',
 124 => 'Hangul Syllables',
 125 => 'High Surrogates',
 126 => 'High Private Use Surrogates',
 127 => 'Low Surrogates',
 128 => 'CJK Compatibility Ideographs',
 129 => 'Alphabetic Presentation Forms',
 130 => 'Arabic Presentation Forms-A',
 131 => 'Combining Half Marks',
 132 => 'CJK Compatibility Forms',
 133 => 'Small Form Variants',
 134 => 'Arabic Presentation Forms-B',
 135 => 'Specials',
 136 => 'Halfwidth and Fullwidth Forms',
 137 => 'Old Italic',
 138 => 'Gothic',
 139 => 'Deseret',
 140 => 'Byzantine Musical Symbols',
 141 => 'Musical Symbols',
 142 => 'Mathematical Alphanumeric Symbols',
 143 => 'CJK Unified Ideographs Extension B',
 144 => 'CJK Compatibility Ideographs Supplement',
 145 => 'Tags',
);
Commit	Line	Data
9fdf68be	1	# !!!!!!! DO NOT EDIT THIS FILE !!!!!!!
d73e5302	2	# This file is built by mktables from e.g. Unicode.txt.
9fdf68be	3	# Any changes made here will be lost!
d73e5302	4	%utf8::In =
d73e5302	5	(
67765ba6	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',
71d929cb	136	'Old Italic Block' => '137',
67765ba6	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',
d9efae67	178	);
d73e5302	179	%utf8::InPat =
d73e5302	180	(
c8b5a1e3	181	'al' => {
d73e5302	182	'Alphabetic' => '164',
d73e5302	183	'Alphabetic(?:[-_]\|\s+)?Presentation(?:[-_]\|\s+)?Forms' => '129',
1ac13f9a	184	},
c8b5a1e3	185	'an' => {
d73e5302	186	'Any' => '171',
d9efae67	187	},
c8b5a1e3	188	'ar' => {
d73e5302	189	'ARABIC' => '16',
d73e5302	190	'Arabic(?:[-_]\|\s+)?Block' => '62',
	191	'Arabic(?:[-_]\|\s+)?Presentation(?:[-_]\|\s+)?Forms(?:[-_]\|\s+)?A' => '130',
	192	'Arabic(?:[-_]\|\s+)?Presentation(?:[-_]\|\s+)?Forms(?:[-_]\|\s+)?B' => '134',
67765ba6	193	'ARMENIAN' => '14',
d73e5302	194	'Armenian(?:[-_]\|\s+)?Block' => '60',
d73e5302	195	'Arrows' => '96',
d9efae67	196	},
c8b5a1e3	197	'as' => {
d73e5302	198	'ASCII(?:[-_]\|\s+)?Hex(?:[-_]\|\s+)?Digit' => '152',
d73e5302	199	'Assigned' => '163',
1ac13f9a	200	},
c8b5a1e3	201	'ba' => {
d73e5302	202	'Basic(?:[-_]\|\s+)?Latin' => '51',
d9efae67	203	},
c8b5a1e3	204	'be' => {
d73e5302	205	'BENGALI' => '20',
d73e5302	206	'Bengali(?:[-_]\|\s+)?Block' => '66',
d9efae67	207	},
c8b5a1e3	208	'bi' => {
d73e5302	209	'Bidi(?:[-_]\|\s+)?Control' => '159',
1ac13f9a	210	},
c8b5a1e3	211	'bl' => {
d73e5302	212	'Block(?:[-_]\|\s+)?Elements' => '103',
d9efae67	213	},
c8b5a1e3	214	'bo' => {
d73e5302	215	'BOPOMOFO' => '45',
	216	'Bopomofo(?:[-_]\|\s+)?Block' => '114',
	217	'Bopomofo(?:[-_]\|\s+)?Extended' => '117',
	218	'Box(?:[-_]\|\s+)?Drawing' => '102',
d9efae67	219	},
c8b5a1e3	220	'br' => {
d73e5302	221	'Braille(?:[-_]\|\s+)?Patterns' => '107',
d9efae67	222	},
c8b5a1e3	223	'by' => {
d73e5302	224	'Byzantine(?:[-_]\|\s+)?Musical(?:[-_]\|\s+)?Symbols' => '140',
d9efae67	225	},
c8b5a1e3	226	'ca' => {
d73e5302	227	'CANADIAN(?:[-_]\|\s+)?ABORIGINAL' => '37',
d9efae67	228	},
c8b5a1e3	229	'ch' => {
d73e5302	230	'CHEROKEE' => '36',
d73e5302	231	'Cherokee(?:[-_]\|\s+)?Block' => '82',
d9efae67	232	},
c8b5a1e3	233	'cj' => {
d73e5302	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',
d9efae67	246	},
c8b5a1e3	247	'co' => {
d73e5302	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',
d9efae67	253	},
c8b5a1e3	254	'cu' => {
d73e5302	255	'Currency(?:[-_]\|\s+)?Symbols' => '92',
d9efae67	256	},
c8b5a1e3	257	'cy' => {
d73e5302	258	'CYRILLIC' => '13',
d73e5302	259	'Cyrillic(?:[-_]\|\s+)?Block' => '59',
d9efae67	260	},
c8b5a1e3	261	'da' => {
d73e5302	262	'Dash' => '151',
1ac13f9a	263	},
c8b5a1e3	264	'de' => {
d73e5302	265	'DESERET' => '49',
d73e5302	266	'Deseret(?:[-_]\|\s+)?Block' => '139',
67765ba6	267	'DEVANAGARI' => '19',
d73e5302	268	'Devanagari(?:[-_]\|\s+)?Block' => '65',
d9efae67	269	},
c8b5a1e3	270	'di' => {
d73e5302	271	'Diacritic' => '154',
d73e5302	272	'Dingbats' => '106',
d9efae67	273	},
c8b5a1e3	274	'en' => {
d73e5302	275	'Enclosed(?:[-_]\|\s+)?Alphanumerics' => '101',
d73e5302	276	'Enclosed(?:[-_]\|\s+)?CJK(?:[-_]\|\s+)?Letters(?:[-_]\|\s+)?and(?:[-_]\|\s+)?Months' => '118',
d9efae67	277	},
c8b5a1e3	278	'et' => {
d73e5302	279	'ETHIOPIC' => '35',
d73e5302	280	'Ethiopic(?:[-_]\|\s+)?Block' => '81',
d9efae67	281	},
c8b5a1e3	282	'ex' => {
d73e5302	283	'Extender' => '155',
1ac13f9a	284	},
c8b5a1e3	285	'ge' => {
d73e5302	286	'General(?:[-_]\|\s+)?Punctuation' => '90',
d73e5302	287	'Geometric(?:[-_]\|\s+)?Shapes' => '104',
67765ba6	288	'GEORGIAN' => '33',
d73e5302	289	'Georgian(?:[-_]\|\s+)?Block' => '79',
d9efae67	290	},
c8b5a1e3	291	'go' => {
d73e5302	292	'GOTHIC' => '48',
d73e5302	293	'Gothic(?:[-_]\|\s+)?Block' => '138',
d9efae67	294	},
c8b5a1e3	295	'gr' => {
d73e5302	296	'GREEK' => '11',
	297	'Greek(?:[-_]\|\s+)?Block' => '58',
	298	'Greek(?:[-_]\|\s+)?Extended' => '89',
d9efae67	299	},
c8b5a1e3	300	'gu' => {
d73e5302	301	'GUJARATI' => '22',
d73e5302	302	'Gujarati(?:[-_]\|\s+)?Block' => '68',
67765ba6	303	'GURMUKHI' => '21',
d73e5302	304	'Gurmukhi(?:[-_]\|\s+)?Block' => '67',
d9efae67	305	},
c8b5a1e3	306	'ha' => {
67765ba6	307	'Halfwidth(?:[-_]\|\s+)?and(?:[-_]\|\s+)?Fullwidth(?:[-_]\|\s+)?Forms' => '136',
d73e5302	308	'HAN' => '42',
d73e5302	309	'HANGUL' => '34',
d73e5302	310	'Hangul(?:[-_]\|\s+)?Compatibility(?:[-_]\|\s+)?Jamo' => '115',
	311	'Hangul(?:[-_]\|\s+)?Jamo' => '80',
	312	'Hangul(?:[-_]\|\s+)?Syllable' => '2',
	313	'Hangul(?:[-_]\|\s+)?Syllables' => '124',
d9efae67	314	},
c8b5a1e3	315	'he' => {
d73e5302	316	'HEBREW' => '15',
	317	'Hebrew(?:[-_]\|\s+)?Block' => '61',
	318	'Hex(?:[-_]\|\s+)?Digit' => '153',
1ac13f9a	319	},
c8b5a1e3	320	'hi' => {
d73e5302	321	'High(?:[-_]\|\s+)?Private(?:[-_]\|\s+)?Use(?:[-_]\|\s+)?Surrogates' => '126',
d73e5302	322	'High(?:[-_]\|\s+)?Surrogates' => '125',
67765ba6	323	'HIRAGANA' => '43',
d73e5302	324	'Hiragana(?:[-_]\|\s+)?Block' => '112',
d9efae67	325	},
c8b5a1e3	326	'hy' => {
d73e5302	327	'Hyphen' => '150',
1ac13f9a	328	},
c8b5a1e3	329	'id' => {
d73e5302	330	'ID(?:[-_]\|\s+)?Continue' => '170',
	331	'ID(?:[-_]\|\s+)?Start' => '169',
	332	'Ideographic' => '161',
	333	'Ideographic(?:[-_]\|\s+)?Description(?:[-_]\|\s+)?Characters' => '110',
d9efae67	334	},
c8b5a1e3	335	'in' => {
d73e5302	336	'INHERITED' => '12',
d9efae67	337	},
c8b5a1e3	338	'ip' => {
d73e5302	339	'IPA(?:[-_]\|\s+)?Extensions' => '55',
d9efae67	340	},
c8b5a1e3	341	'jo' => {
d73e5302	342	'Join(?:[-_]\|\s+)?Control' => '158',
1ac13f9a	343	},
c8b5a1e3	344	'ka' => {
d73e5302	345	'Kanbun' => '116',
d73e5302	346	'Kangxi(?:[-_]\|\s+)?Radicals' => '109',
67765ba6	347	'KANNADA' => '26',
d73e5302	348	'Kannada(?:[-_]\|\s+)?Block' => '72',
67765ba6	349	'KATAKANA' => '44',
d73e5302	350	'Katakana(?:[-_]\|\s+)?Block' => '113',
d9efae67	351	},
c8b5a1e3	352	'kh' => {
d73e5302	353	'KHMER' => '40',
d73e5302	354	'Khmer(?:[-_]\|\s+)?Block' => '86',
d9efae67	355	},
c8b5a1e3	356	'la' => {
d73e5302	357	'Lampersand' => '168',
67765ba6	358	'LAO' => '30',
d73e5302	359	'Lao(?:[-_]\|\s+)?Block' => '76',
67765ba6	360	'LATIN' => '10',
d73e5302	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',
d9efae67	365	},
c8b5a1e3	366	'le' => {
d73e5302	367	'Letterlike(?:[-_]\|\s+)?Symbols' => '94',
d9efae67	368	},
c8b5a1e3	369	'lo' => {
d73e5302	370	'Low(?:[-_]\|\s+)?Surrogate' => '5',
	371	'Low(?:[-_]\|\s+)?Surrogates' => '127',
	372	'Lowercase' => '165',
d9efae67	373	},
c8b5a1e3	374	'ma' => {
d73e5302	375	'MALAYALAM' => '27',
	376	'Malayalam(?:[-_]\|\s+)?Block' => '73',
	377	'Math' => '167',
	378	'Mathematical(?:[-_]\|\s+)?Alphanumeric(?:[-_]\|\s+)?Symbols' => '142',
	379	'Mathematical(?:[-_]\|\s+)?Operators' => '97',
d9efae67	380	},
c8b5a1e3	381	'mi' => {
d73e5302	382	'Miscellaneous(?:[-_]\|\s+)?Symbols' => '105',
d73e5302	383	'Miscellaneous(?:[-_]\|\s+)?Technical' => '98',
d9efae67	384	},
c8b5a1e3	385	'mo' => {
d73e5302	386	'MONGOLIAN' => '41',
d73e5302	387	'Mongolian(?:[-_]\|\s+)?Block' => '87',
d9efae67	388	},
c8b5a1e3	389	'mu' => {
d73e5302	390	'Musical(?:[-_]\|\s+)?Symbols' => '141',
d9efae67	391	},
c8b5a1e3	392	'my' => {
d73e5302	393	'MYANMAR' => '32',
d73e5302	394	'Myanmar(?:[-_]\|\s+)?Block' => '78',
d9efae67	395	},
c8b5a1e3	396	'no' => {
d73e5302	397	'Non(?:[-_]\|\s+)?Private(?:[-_]\|\s+)?Use(?:[-_]\|\s+)?High(?:[-_]\|\s+)?Surrogate' => '3',
d73e5302	398	'Noncharacter(?:[-_]\|\s+)?Code(?:[-_]\|\s+)?Point' => '162',
1ac13f9a	399	},
c8b5a1e3	400	'nu' => {
d73e5302	401	'Number(?:[-_]\|\s+)?Forms' => '95',
d9efae67	402	},
c8b5a1e3	403	'og' => {
d73e5302	404	'OGHAM' => '38',
d73e5302	405	'Ogham(?:[-_]\|\s+)?Block' => '84',
d9efae67	406	},
c8b5a1e3	407	'ol' => {
71d929cb	408	'Old(?:[-_]\|\s+)?Italic(?:[-_]\|\s+)?Block' => '137',
67765ba6	409	'OLD(?:[-_]\|\s+)?ITALIC' => '47',
d9efae67	410	},
c8b5a1e3	411	'op' => {
d73e5302	412	'Optical(?:[-_]\|\s+)?Character(?:[-_]\|\s+)?Recognition' => '100',
d9efae67	413	},
c8b5a1e3	414	'or' => {
d73e5302	415	'ORIYA' => '23',
d73e5302	416	'Oriya(?:[-_]\|\s+)?Block' => '69',
d9efae67	417	},
c8b5a1e3	418	'ot' => {
d73e5302	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',
1ac13f9a	427	},
c8b5a1e3	428	'pr' => {
d73e5302	429	'Private(?:[-_]\|\s+)?Use' => '6',
d73e5302	430	'Private(?:[-_]\|\s+)?Use(?:[-_]\|\s+)?High(?:[-_]\|\s+)?Surrogate' => '4',
d9efae67	431	},
c8b5a1e3	432	'qu' => {
d73e5302	433	'Quotation(?:[-_]\|\s+)?Mark' => '148',
1ac13f9a	434	},
c8b5a1e3	435	'ru' => {
d73e5302	436	'RUNIC' => '39',
d73e5302	437	'Runic(?:[-_]\|\s+)?Block' => '85',
d9efae67	438	},
c8b5a1e3	439	'si' => {
d73e5302	440	'SINHALA' => '28',
d73e5302	441	'Sinhala(?:[-_]\|\s+)?Block' => '74',
d9efae67	442	},
c8b5a1e3	443	'sm' => {
d73e5302	444	'Small(?:[-_]\|\s+)?Form(?:[-_]\|\s+)?Variants' => '133',
d9efae67	445	},
c8b5a1e3	446	'sp' => {
d73e5302	447	'Spacing(?:[-_]\|\s+)?Modifier(?:[-_]\|\s+)?Letters' => '56',
d73e5302	448	'Specials' => '135',
d9efae67	449	},
c8b5a1e3	450	'su' => {
d73e5302	451	'Superscripts(?:[-_]\|\s+)?and(?:[-_]\|\s+)?Subscripts' => '91',
d9efae67	452	},
c8b5a1e3	453	'sy' => {
d73e5302	454	'SYRIAC' => '17',
d73e5302	455	'Syriac(?:[-_]\|\s+)?Block' => '63',
d9efae67	456	},
c8b5a1e3	457	'ta' => {
d73e5302	458	'Tags' => '145',
67765ba6	459	'TAMIL' => '24',
d73e5302	460	'Tamil(?:[-_]\|\s+)?Block' => '70',
d9efae67	461	},
c8b5a1e3	462	'te' => {
d73e5302	463	'TELUGU' => '25',
	464	'Telugu(?:[-_]\|\s+)?Block' => '71',
	465	'Terminal(?:[-_]\|\s+)?Punctuation' => '147',
1ac13f9a	466	},
c8b5a1e3	467	'th' => {
d73e5302	468	'THAANA' => '18',
d73e5302	469	'Thaana(?:[-_]\|\s+)?Block' => '64',
67765ba6	470	'THAI' => '29',
d73e5302	471	'Thai(?:[-_]\|\s+)?Block' => '75',
d9efae67	472	},
c8b5a1e3	473	'ti' => {
d73e5302	474	'TIBETAN' => '31',
d73e5302	475	'Tibetan(?:[-_]\|\s+)?Block' => '77',
d9efae67	476	},
c8b5a1e3	477	'un' => {
d73e5302	478	'Unified(?:[-_]\|\s+)?Canadian(?:[-_]\|\s+)?Aboriginal(?:[-_]\|\s+)?Syllabics' => '83',
d9efae67	479	},
c8b5a1e3	480	'up' => {
d73e5302	481	'Uppercase' => '166',
1ac13f9a	482	},
c8b5a1e3	483	'wh' => {
d73e5302	484	'White(?:[-_]\|\s+)?space' => '146',
1ac13f9a	485	},
d9efae67	486	'yi' => {
d73e5302	487	'YI' => '46',
	488	'Yi(?:[-_]\|\s+)?Radicals' => '123',
	489	'Yi(?:[-_]\|\s+)?Syllables' => '122',
d9efae67	490	},
9fdf68be	491	);
71d929cb	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	);