Allow for more flexibility in the \p{In...} names, now

case doesn't matter, and any space or dash can be
matched by any space, dash, underbar, or empty.
(may be going too far on leniency)

p4raw-id: //depot/perl@12264

1 files changed, 416 insertions, 136 deletions

diff --git a/lib/unicore/In.pl b/lib/unicore/In.pl
index a6c24199a7..c11445c26b 100644
--- a/lib/unicore/In.pl
+++ b/lib/unicore/In.pl
@@ -2,140 +2,420 @@
 # This file is built by mktables.PL from e.g. Unicode.txt.
 # Any changes made here will be lost!
 %utf8::In = (
-'Latin'                                  =>   0,
-'Greek'                                  =>   1,
-'Cyrillic'                               =>   2,
-'Armenian'                               =>   3,
-'Hebrew'                                 =>   4,
-'Arabic'                                 =>   5,
-'Syriac'                                 =>   6,
-'Thaana'                                 =>   7,
-'Devanagari'                             =>   8,
-'Bengali'                                =>   9,
-'Gurmukhi'                               =>  10,
-'Gujarati'                               =>  11,
-'Oriya'                                  =>  12,
-'Tamil'                                  =>  13,
-'Telugu'                                 =>  14,
-'Kannada'                                =>  15,
-'Malayalam'                              =>  16,
-'Sinhala'                                =>  17,
-'Thai'                                   =>  18,
-'Lao'                                    =>  19,
-'Tibetan'                                =>  20,
-'Myanmar'                                =>  21,
-'Georgian'                               =>  22,
-'Hangul'                                 =>  23,
-'Ethiopic'                               =>  24,
-'Cherokee'                               =>  25,
-'CanadianAboriginal'                     =>  26,
-'Ogham'                                  =>  27,
-'Runic'                                  =>  28,
-'Khmer'                                  =>  29,
-'Mongolian'                              =>  30,
-'Hiragana'                               =>  31,
-'Katakana'                               =>  32,
-'Bopomofo'                               =>  33,
-'Han'                                    =>  34,
-'Yi'                                     =>  35,
-'OldItalic'                              =>  36,
-'Gothic'                                 =>  37,
-'Deseret'                                =>  38,
-'Inherited'                              =>  39,
-'BasicLatin'                             =>  40,
-'Latin1Supplement'                       =>  41,
-'LatinExtendedA'                         =>  42,
-'LatinExtendedB'                         =>  43,
-'IPAExtensions'                          =>  44,
-'SpacingModifierLetters'                 =>  45,
-'CombiningDiacriticalMarks'              =>  46,
-'GreekBlock'                             =>  47,
-'CyrillicBlock'                          =>  48,
-'ArmenianBlock'                          =>  49,
-'HebrewBlock'                            =>  50,
-'ArabicBlock'                            =>  51,
-'SyriacBlock'                            =>  52,
-'ThaanaBlock'                            =>  53,
-'DevanagariBlock'                        =>  54,
-'BengaliBlock'                           =>  55,
-'GurmukhiBlock'                          =>  56,
-'GujaratiBlock'                          =>  57,
-'OriyaBlock'                             =>  58,
-'TamilBlock'                             =>  59,
-'TeluguBlock'                            =>  60,
-'KannadaBlock'                           =>  61,
-'MalayalamBlock'                         =>  62,
-'SinhalaBlock'                           =>  63,
-'ThaiBlock'                              =>  64,
-'LaoBlock'                               =>  65,
-'TibetanBlock'                           =>  66,
-'MyanmarBlock'                           =>  67,
-'GeorgianBlock'                          =>  68,
-'HangulJamo'                             =>  69,
-'EthiopicBlock'                          =>  70,
-'CherokeeBlock'                          =>  71,
-'UnifiedCanadianAboriginalSyllabics'     =>  72,
-'OghamBlock'                             =>  73,
-'RunicBlock'                             =>  74,
-'KhmerBlock'                             =>  75,
-'MongolianBlock'                         =>  76,
-'LatinExtendedAdditional'                =>  77,
-'GreekExtended'                          =>  78,
-'GeneralPunctuation'                     =>  79,
-'SuperscriptsandSubscripts'              =>  80,
-'CurrencySymbols'                        =>  81,
-'CombiningMarksforSymbols'               =>  82,
-'LetterlikeSymbols'                      =>  83,
-'NumberForms'                            =>  84,
-'Arrows'                                 =>  85,
-'MathematicalOperators'                  =>  86,
-'MiscellaneousTechnical'                 =>  87,
-'ControlPictures'                        =>  88,
-'OpticalCharacterRecognition'            =>  89,
-'EnclosedAlphanumerics'                  =>  90,
-'BoxDrawing'                             =>  91,
-'BlockElements'                          =>  92,
-'GeometricShapes'                        =>  93,
-'MiscellaneousSymbols'                   =>  94,
-'Dingbats'                               =>  95,
-'BraillePatterns'                        =>  96,
-'CJKRadicalsSupplement'                  =>  97,
-'KangxiRadicals'                         =>  98,
-'IdeographicDescriptionCharacters'       =>  99,
-'CJKSymbolsandPunctuation'               => 100,
-'HiraganaBlock'                          => 101,
-'KatakanaBlock'                          => 102,
-'BopomofoBlock'                          => 103,
-'HangulCompatibilityJamo'                => 104,
-'Kanbun'                                 => 105,
-'BopomofoExtended'                       => 106,
-'EnclosedCJKLettersandMonths'            => 107,
-'CJKCompatibility'                       => 108,
-'CJKUnifiedIdeographsExtensionA'         => 109,
-'CJKUnifiedIdeographs'                   => 110,
-'YiSyllables'                            => 111,
-'YiRadicals'                             => 112,
-'HangulSyllables'                        => 113,
-'HighSurrogates'                         => 114,
-'HighPrivateUseSurrogates'               => 115,
-'LowSurrogates'                          => 116,
-'PrivateUse'                             => 117,
-'CJKCompatibilityIdeographs'             => 118,
-'AlphabeticPresentationForms'            => 119,
-'ArabicPresentationFormsA'               => 120,
-'CombiningHalfMarks'                     => 121,
-'CJKCompatibilityForms'                  => 122,
-'SmallFormVariants'                      => 123,
-'ArabicPresentationFormsB'               => 124,
-'Specials'                               => 125,
-'HalfwidthandFullwidthForms'             => 126,
-'OldItalicBlock'                         => 127,
-'GothicBlock'                            => 128,
-'DeseretBlock'                           => 129,
-'ByzantineMusicalSymbols'                => 130,
-'MusicalSymbols'                         => 131,
-'MathematicalAlphanumericSymbols'        => 132,
-'CJKUnifiedIdeographsExtensionB'         => 133,
-'CJKCompatibilityIdeographsSupplement'   => 134,
-'Tags'                                   => 135,
+'LATIN'                                       =>   0,
+'GREEK'                                       =>   1,
+'CYRILLIC'                                    =>   2,
+'ARMENIAN'                                    =>   3,
+'HEBREW'                                      =>   4,
+'ARABIC'                                      =>   5,
+'SYRIAC'                                      =>   6,
+'THAANA'                                      =>   7,
+'DEVANAGARI'                                  =>   8,
+'BENGALI'                                     =>   9,
+'GURMUKHI'                                    =>  10,
+'GUJARATI'                                    =>  11,
+'ORIYA'                                       =>  12,
+'TAMIL'                                       =>  13,
+'TELUGU'                                      =>  14,
+'KANNADA'                                     =>  15,
+'MALAYALAM'                                   =>  16,
+'SINHALA'                                     =>  17,
+'THAI'                                        =>  18,
+'LAO'                                         =>  19,
+'TIBETAN'                                     =>  20,
+'MYANMAR'                                     =>  21,
+'GEORGIAN'                                    =>  22,
+'HANGUL'                                      =>  23,
+'ETHIOPIC'                                    =>  24,
+'CHEROKEE'                                    =>  25,
+'CANADIAN-ABORIGINAL'                         =>  26,
+'OGHAM'                                       =>  27,
+'RUNIC'                                       =>  28,
+'KHMER'                                       =>  29,
+'MONGOLIAN'                                   =>  30,
+'HIRAGANA'                                    =>  31,
+'KATAKANA'                                    =>  32,
+'BOPOMOFO'                                    =>  33,
+'HAN'                                         =>  34,
+'YI'                                          =>  35,
+'OLD-ITALIC'                                  =>  36,
+'GOTHIC'                                      =>  37,
+'DESERET'                                     =>  38,
+'INHERITED'                                   =>  39,
+'Basic Latin'                                 =>  40,
+'Latin-1 Supplement'                          =>  41,
+'Latin Extended-A'                            =>  42,
+'Latin Extended-B'                            =>  43,
+'IPA Extensions'                              =>  44,
+'Spacing Modifier Letters'                    =>  45,
+'Combining Diacritical Marks'                 =>  46,
+'Greek Block'                                 =>  47,
+'Cyrillic Block'                              =>  48,
+'Armenian Block'                              =>  49,
+'Hebrew Block'                                =>  50,
+'Arabic Block'                                =>  51,
+'Syriac Block'                                =>  52,
+'Thaana Block'                                =>  53,
+'Devanagari Block'                            =>  54,
+'Bengali Block'                               =>  55,
+'Gurmukhi Block'                              =>  56,
+'Gujarati Block'                              =>  57,
+'Oriya Block'                                 =>  58,
+'Tamil Block'                                 =>  59,
+'Telugu Block'                                =>  60,
+'Kannada Block'                               =>  61,
+'Malayalam Block'                             =>  62,
+'Sinhala Block'                               =>  63,
+'Thai Block'                                  =>  64,
+'Lao Block'                                   =>  65,
+'Tibetan Block'                               =>  66,
+'Myanmar Block'                               =>  67,
+'Georgian Block'                              =>  68,
+'Hangul Jamo'                                 =>  69,
+'Ethiopic Block'                              =>  70,
+'Cherokee Block'                              =>  71,
+'Unified Canadian Aboriginal Syllabics'       =>  72,
+'Ogham Block'                                 =>  73,
+'Runic Block'                                 =>  74,
+'Khmer Block'                                 =>  75,
+'Mongolian Block'                             =>  76,
+'Latin Extended Additional'                   =>  77,
+'Greek Extended'                              =>  78,
+'General Punctuation'                         =>  79,
+'Superscripts and Subscripts'                 =>  80,
+'Currency Symbols'                            =>  81,
+'Combining Marks for Symbols'                 =>  82,
+'Letterlike Symbols'                          =>  83,
+'Number Forms'                                =>  84,
+'Arrows'                                      =>  85,
+'Mathematical Operators'                      =>  86,
+'Miscellaneous Technical'                     =>  87,
+'Control Pictures'                            =>  88,
+'Optical Character Recognition'               =>  89,
+'Enclosed Alphanumerics'                      =>  90,
+'Box Drawing'                                 =>  91,
+'Block Elements'                              =>  92,
+'Geometric Shapes'                            =>  93,
+'Miscellaneous Symbols'                       =>  94,
+'Dingbats'                                    =>  95,
+'Braille Patterns'                            =>  96,
+'CJK Radicals Supplement'                     =>  97,
+'Kangxi Radicals'                             =>  98,
+'Ideographic Description Characters'          =>  99,
+'CJK Symbols and Punctuation'                 => 100,
+'Hiragana Block'                              => 101,
+'Katakana Block'                              => 102,
+'Bopomofo Block'                              => 103,
+'Hangul Compatibility Jamo'                   => 104,
+'Kanbun'                                      => 105,
+'Bopomofo Extended'                           => 106,
+'Enclosed CJK Letters and Months'             => 107,
+'CJK Compatibility'                           => 108,
+'CJK Unified Ideographs Extension A'          => 109,
+'CJK Unified Ideographs'                      => 110,
+'Yi Syllables'                                => 111,
+'Yi Radicals'                                 => 112,
+'Hangul Syllables'                            => 113,
+'High Surrogates'                             => 114,
+'High Private Use Surrogates'                 => 115,
+'Low Surrogates'                              => 116,
+'Private Use'                                 => 117,
+'CJK Compatibility Ideographs'                => 118,
+'Alphabetic Presentation Forms'               => 119,
+'Arabic Presentation Forms-A'                 => 120,
+'Combining Half Marks'                        => 121,
+'CJK Compatibility Forms'                     => 122,
+'Small Form Variants'                         => 123,
+'Arabic Presentation Forms-B'                 => 124,
+'Specials'                                    => 125,
+'Halfwidth and Fullwidth Forms'               => 126,
+'Old Italic'                                  => 127,
+'Gothic Block'                                => 128,
+'Deseret Block'                               => 129,
+'Byzantine Musical Symbols'                   => 130,
+'Musical Symbols'                             => 131,
+'Mathematical Alphanumeric Symbols'           => 132,
+'CJK Unified Ideographs Extension B'          => 133,
+'CJK Compatibility Ideographs Supplement'     => 134,
+'Tags'                                        => 135,
+);
+%utf8::InPat = (
+'alp' => {
+	'Alphabetic[- _]?Presentation[- _]?Forms' => 'Alphabetic Presentation Forms',
+},
+'ara' => {
+	'ARABIC' => 'ARABIC',
+	'Arabic[- _]?Block' => 'Arabic Block',
+	'Arabic[- _]?Presentation[- _]?Forms[- _]?A' => 'Arabic Presentation Forms-A',
+	'Arabic[- _]?Presentation[- _]?Forms[- _]?B' => 'Arabic Presentation Forms-B',
+},
+'arm' => {
+	'ARMENIAN' => 'ARMENIAN',
+	'Armenian[- _]?Block' => 'Armenian Block',
+},
+'arr' => {
+	'Arrows' => 'Arrows',
+},
+'bas' => {
+	'Basic[- _]?Latin' => 'Basic Latin',
+},
+'ben' => {
+	'BENGALI' => 'BENGALI',
+	'Bengali[- _]?Block' => 'Bengali Block',
+},
+'blo' => {
+	'Block[- _]?Elements' => 'Block Elements',
+},
+'bop' => {
+	'BOPOMOFO' => 'BOPOMOFO',
+	'Bopomofo[- _]?Block' => 'Bopomofo Block',
+	'Bopomofo[- _]?Extended' => 'Bopomofo Extended',
+},
+'box' => {
+	'Box[- _]?Drawing' => 'Box Drawing',
+},
+'bra' => {
+	'Braille[- _]?Patterns' => 'Braille Patterns',
+},
+'byz' => {
+	'Byzantine[- _]?Musical[- _]?Symbols' => 'Byzantine Musical Symbols',
+},
+'can' => {
+	'CANADIAN[- _]?ABORIGINAL' => 'CANADIAN-ABORIGINAL',
+},
+'che' => {
+	'CHEROKEE' => 'CHEROKEE',
+	'Cherokee[- _]?Block' => 'Cherokee Block',
+},
+'cjk' => {
+	'CJK[- _]?Radicals[- _]?Supplement' => 'CJK Radicals Supplement',
+	'CJK[- _]?Symbols[- _]?and[- _]?Punctuation' => 'CJK Symbols and Punctuation',
+	'CJK[- _]?Compatibility' => 'CJK Compatibility',
+	'CJK[- _]?Unified[- _]?Ideographs[- _]?Extension[- _]?A' => 'CJK Unified Ideographs Extension A',
+	'CJK[- _]?Unified[- _]?Ideographs' => 'CJK Unified Ideographs',
+	'CJK[- _]?Compatibility[- _]?Ideographs' => 'CJK Compatibility Ideographs',
+	'CJK[- _]?Compatibility[- _]?Forms' => 'CJK Compatibility Forms',
+	'CJK[- _]?Unified[- _]?Ideographs[- _]?Extension[- _]?B' => 'CJK Unified Ideographs Extension B',
+	'CJK[- _]?Compatibility[- _]?Ideographs[- _]?Supplement' => 'CJK Compatibility Ideographs Supplement',
+},
+'com' => {
+	'Combining[- _]?Diacritical[- _]?Marks' => 'Combining Diacritical Marks',
+	'Combining[- _]?Marks[- _]?for[- _]?Symbols' => 'Combining Marks for Symbols',
+	'Combining[- _]?Half[- _]?Marks' => 'Combining Half Marks',
+},
+'con' => {
+	'Control[- _]?Pictures' => 'Control Pictures',
+},
+'cur' => {
+	'Currency[- _]?Symbols' => 'Currency Symbols',
+},
+'cyr' => {
+	'CYRILLIC' => 'CYRILLIC',
+	'Cyrillic[- _]?Block' => 'Cyrillic Block',
+},
+'des' => {
+	'DESERET' => 'DESERET',
+	'Deseret[- _]?Block' => 'Deseret Block',
+},
+'dev' => {
+	'DEVANAGARI' => 'DEVANAGARI',
+	'Devanagari[- _]?Block' => 'Devanagari Block',
+},
+'din' => {
+	'Dingbats' => 'Dingbats',
+},
+'enc' => {
+	'Enclosed[- _]?Alphanumerics' => 'Enclosed Alphanumerics',
+	'Enclosed[- _]?CJK[- _]?Letters[- _]?and[- _]?Months' => 'Enclosed CJK Letters and Months',
+},
+'eth' => {
+	'ETHIOPIC' => 'ETHIOPIC',
+	'Ethiopic[- _]?Block' => 'Ethiopic Block',
+},
+'gen' => {
+	'General[- _]?Punctuation' => 'General Punctuation',
+},
+'geo' => {
+	'GEORGIAN' => 'GEORGIAN',
+	'Georgian[- _]?Block' => 'Georgian Block',
+	'Geometric[- _]?Shapes' => 'Geometric Shapes',
+},
+'got' => {
+	'GOTHIC' => 'GOTHIC',
+	'Gothic[- _]?Block' => 'Gothic Block',
+},
+'gre' => {
+	'GREEK' => 'GREEK',
+	'Greek[- _]?Block' => 'Greek Block',
+	'Greek[- _]?Extended' => 'Greek Extended',
+},
+'guj' => {
+	'GUJARATI' => 'GUJARATI',
+	'Gujarati[- _]?Block' => 'Gujarati Block',
+},
+'gur' => {
+	'GURMUKHI' => 'GURMUKHI',
+	'Gurmukhi[- _]?Block' => 'Gurmukhi Block',
+},
+'hal' => {
+	'Halfwidth[- _]?and[- _]?Fullwidth[- _]?Forms' => 'Halfwidth and Fullwidth Forms',
+},
+'han' => {
+	'HANGUL' => 'HANGUL',
+	'HAN' => 'HAN',
+	'Hangul[- _]?Jamo' => 'Hangul Jamo',
+	'Hangul[- _]?Compatibility[- _]?Jamo' => 'Hangul Compatibility Jamo',
+	'Hangul[- _]?Syllables' => 'Hangul Syllables',
+},
+'heb' => {
+	'HEBREW' => 'HEBREW',
+	'Hebrew[- _]?Block' => 'Hebrew Block',
+},
+'hig' => {
+	'High[- _]?Surrogates' => 'High Surrogates',
+	'High[- _]?Private[- _]?Use[- _]?Surrogates' => 'High Private Use Surrogates',
+},
+'hir' => {
+	'HIRAGANA' => 'HIRAGANA',
+	'Hiragana[- _]?Block' => 'Hiragana Block',
+},
+'ide' => {
+	'Ideographic[- _]?Description[- _]?Characters' => 'Ideographic Description Characters',
+},
+'inh' => {
+	'INHERITED' => 'INHERITED',
+},
+'ipa' => {
+	'IPA[- _]?Extensions' => 'IPA Extensions',
+},
+'kan' => {
+	'KANNADA' => 'KANNADA',
+	'Kannada[- _]?Block' => 'Kannada Block',
+	'Kangxi[- _]?Radicals' => 'Kangxi Radicals',
+	'Kanbun' => 'Kanbun',
+},
+'kat' => {
+	'KATAKANA' => 'KATAKANA',
+	'Katakana[- _]?Block' => 'Katakana Block',
+},
+'khm' => {
+	'KHMER' => 'KHMER',
+	'Khmer[- _]?Block' => 'Khmer Block',
+},
+'lao' => {
+	'LAO' => 'LAO',
+	'Lao[- _]?Block' => 'Lao Block',
+},
+'lat' => {
+	'LATIN' => 'LATIN',
+	'Latin[- _]?1[- _]?Supplement' => 'Latin-1 Supplement',
+	'Latin[- _]?Extended[- _]?A' => 'Latin Extended-A',
+	'Latin[- _]?Extended[- _]?B' => 'Latin Extended-B',
+	'Latin[- _]?Extended[- _]?Additional' => 'Latin Extended Additional',
+},
+'let' => {
+	'Letterlike[- _]?Symbols' => 'Letterlike Symbols',
+},
+'low' => {
+	'Low[- _]?Surrogates' => 'Low Surrogates',
+},
+'mal' => {
+	'MALAYALAM' => 'MALAYALAM',
+	'Malayalam[- _]?Block' => 'Malayalam Block',
+},
+'mat' => {
+	'Mathematical[- _]?Operators' => 'Mathematical Operators',
+	'Mathematical[- _]?Alphanumeric[- _]?Symbols' => 'Mathematical Alphanumeric Symbols',
+},
+'mis' => {
+	'Miscellaneous[- _]?Technical' => 'Miscellaneous Technical',
+	'Miscellaneous[- _]?Symbols' => 'Miscellaneous Symbols',
+},
+'mon' => {
+	'MONGOLIAN' => 'MONGOLIAN',
+	'Mongolian[- _]?Block' => 'Mongolian Block',
+},
+'mus' => {
+	'Musical[- _]?Symbols' => 'Musical Symbols',
+},
+'mya' => {
+	'MYANMAR' => 'MYANMAR',
+	'Myanmar[- _]?Block' => 'Myanmar Block',
+},
+'num' => {
+	'Number[- _]?Forms' => 'Number Forms',
+},
+'ogh' => {
+	'OGHAM' => 'OGHAM',
+	'Ogham[- _]?Block' => 'Ogham Block',
+},
+'old' => {
+	'OLD[- _]?ITALIC' => 'OLD-ITALIC',
+	'Old[- _]?Italic' => 'Old Italic',
+},
+'opt' => {
+	'Optical[- _]?Character[- _]?Recognition' => 'Optical Character Recognition',
+},
+'ori' => {
+	'ORIYA' => 'ORIYA',
+	'Oriya[- _]?Block' => 'Oriya Block',
+},
+'pri' => {
+	'Private[- _]?Use' => 'Private Use',
+},
+'run' => {
+	'RUNIC' => 'RUNIC',
+	'Runic[- _]?Block' => 'Runic Block',
+},
+'sin' => {
+	'SINHALA' => 'SINHALA',
+	'Sinhala[- _]?Block' => 'Sinhala Block',
+},
+'sma' => {
+	'Small[- _]?Form[- _]?Variants' => 'Small Form Variants',
+},
+'spa' => {
+	'Spacing[- _]?Modifier[- _]?Letters' => 'Spacing Modifier Letters',
+},
+'spe' => {
+	'Specials' => 'Specials',
+},
+'sup' => {
+	'Superscripts[- _]?and[- _]?Subscripts' => 'Superscripts and Subscripts',
+},
+'syr' => {
+	'SYRIAC' => 'SYRIAC',
+	'Syriac[- _]?Block' => 'Syriac Block',
+},
+'tag' => {
+	'Tags' => 'Tags',
+},
+'tam' => {
+	'TAMIL' => 'TAMIL',
+	'Tamil[- _]?Block' => 'Tamil Block',
+},
+'tel' => {
+	'TELUGU' => 'TELUGU',
+	'Telugu[- _]?Block' => 'Telugu Block',
+},
+'tha' => {
+	'THAANA' => 'THAANA',
+	'THAI' => 'THAI',
+	'Thaana[- _]?Block' => 'Thaana Block',
+	'Thai[- _]?Block' => 'Thai Block',
+},
+'tib' => {
+	'TIBETAN' => 'TIBETAN',
+	'Tibetan[- _]?Block' => 'Tibetan Block',
+},
+'uni' => {
+	'Unified[- _]?Canadian[- _]?Aboriginal[- _]?Syllabics' => 'Unified Canadian Aboriginal Syllabics',
+},
+'yi' => {
+	'YI' => 'YI',
+},
+'yi ' => {
+	'Yi[- _]?Syllables' => 'Yi Syllables',
+	'Yi[- _]?Radicals' => 'Yi Radicals',
+},
 );