summaryrefslogtreecommitdiff
path: root/Master/perltl/lib/unicore/Canonical.pl
diff options
context:
space:
mode:
Diffstat (limited to 'Master/perltl/lib/unicore/Canonical.pl')
-rw-r--r--Master/perltl/lib/unicore/Canonical.pl816
1 files changed, 816 insertions, 0 deletions
diff --git a/Master/perltl/lib/unicore/Canonical.pl b/Master/perltl/lib/unicore/Canonical.pl
new file mode 100644
index 00000000000..947fd2b9d5c
--- /dev/null
+++ b/Master/perltl/lib/unicore/Canonical.pl
@@ -0,0 +1,816 @@
+# !!!!!!! DO NOT EDIT THIS FILE !!!!!!!
+# This file is built by ./mktables from e.g. UnicodeData.txt.
+# Any changes made here will be lost!
+
+##
+## Data in this file used by ../utf8_heavy.pl
+##
+
+## Mapping from lc(canonical name) to filename in ./lib
+%utf8::Canonical = (
+ alphabetic => 'Alphabet',
+
+ # InAlphabeticPresentationForms
+ alphabeticpresentationforms => 'InAlphab',
+
+ arabic => 'Arabic',
+
+ # InArabicPresentationFormsA
+ arabicpresentationformsa => 'InArabi3',
+
+ # InArabicPresentationFormsB
+ arabicpresentationformsb => 'InArabi2',
+
+ armenian => 'Armenian',
+
+ # InArrows
+ arrows => 'InArrows',
+
+ # AsciiHexDigit
+ asciihexdigit => 'AsciiHex',
+
+ # InBasicLatin
+ basiclatin => 'InBasicL',
+
+ bengali => 'Bengali',
+
+ # BidiControl
+ bidicontrol => 'BidiCont',
+
+ # InBlockElements
+ blockelements => 'InBlockE',
+
+ bopomofo => 'Bopomofo',
+
+ # InBopomofoExtended
+ bopomofoextended => 'InBopom2',
+
+ # InBoxDrawing
+ boxdrawing => 'InBoxDra',
+
+ # InBraillePatterns
+ braillepatterns => 'InBraill',
+
+ buhid => 'Buhid',
+
+ # InByzantineMusicalSymbols
+ byzantinemusicalsymbols => 'InByzant',
+
+ # CanadianAboriginal
+ canadianaboriginal => 'Canadian',
+
+ cherokee => 'Cherokee',
+
+ # InCjkCompatibility
+ cjkcompatibility => 'InCjkCom',
+
+ # InCjkCompatibilityForms
+ cjkcompatibilityforms => 'InCjkCo2',
+
+ # InCjkCompatibilityIdeographs
+ cjkcompatibilityideographs => 'InCjkCo3',
+
+ # InCjkCompatibilityIdeographsSupplement
+ cjkcompatibilityideographssupplement => 'InCjkCo4',
+
+ # InCjkRadicalsSupplement
+ cjkradicalssupplement => 'InCjkRad',
+
+ # InCjkSymbolsAndPunctuation
+ cjksymbolsandpunctuation => 'InCjkSym',
+
+ # InCjkUnifiedIdeographs
+ cjkunifiedideographs => 'InCjkUni',
+
+ # InCjkUnifiedIdeographsExtensionA
+ cjkunifiedideographsextensiona => 'InCjkUn3',
+
+ # InCjkUnifiedIdeographsExtensionB
+ cjkunifiedideographsextensionb => 'InCjkUn2',
+
+ # ClosePunctuation
+ closepunctuation => 'Pe',
+
+ # InCombiningDiacriticalMarks
+ combiningdiacriticalmarks => 'InCombi2',
+
+ # InCombiningDiacriticalMarksForSymbols
+ combiningdiacriticalmarksforsymbols => 'InCombi3',
+
+ # InCombiningHalfMarks
+ combininghalfmarks => 'InCombin',
+
+ common => 'Common',
+
+ # ConnectorPunctuation
+ connectorpunctuation => 'Pc',
+
+ control => 'Cc',
+
+ # InControlPictures
+ controlpictures => 'InContro',
+
+ # CurrencySymbol
+ currencysymbol => 'Sc',
+
+ # InCurrencySymbols
+ currencysymbols => 'InCurren',
+
+ cyrillic => 'Cyrillic',
+
+ # InCyrillicSupplementary
+ cyrillicsupplementary => 'InCyril2',
+
+ dash => 'Dash',
+
+ # DashPunctuation
+ dashpunctuation => 'Pd',
+
+ # DecimalNumber
+ decimalnumber => 'Nd',
+
+ deprecated => 'Deprecat',
+ deseret => 'Deseret',
+ devanagari => 'Devanaga',
+ diacritic => 'Diacriti',
+
+ # InDingbats
+ dingbats => 'InDingba',
+
+ # InEnclosedAlphanumerics
+ enclosedalphanumerics => 'InEnclos',
+
+ # InEnclosedCjkLettersAndMonths
+ enclosedcjklettersandmonths => 'InEnclo2',
+
+ # EnclosingMark
+ enclosingmark => 'Me',
+
+ ethiopic => 'Ethiopic',
+ extender => 'Extender',
+
+ # FinalPunctuation
+ finalpunctuation => 'Pf',
+
+ format => 'Cf',
+
+ # InGeneralPunctuation
+ generalpunctuation => 'InGenera',
+
+ # InGeometricShapes
+ geometricshapes => 'InGeomet',
+
+ georgian => 'Georgian',
+ gothic => 'Gothic',
+
+ # GraphemeLink
+ graphemelink => 'Grapheme',
+
+ greek => 'Greek',
+
+ # InGreekAndCoptic
+ greekandcoptic => 'InGreekA',
+
+ # InGreekExtended
+ greekextended => 'InGreekE',
+
+ gujarati => 'Gujarati',
+ gurmukhi => 'Gurmukhi',
+
+ # InHalfwidthAndFullwidthForms
+ halfwidthandfullwidthforms => 'InHalfwi',
+
+ han => 'Han',
+ hangul => 'Hangul',
+
+ # InHangulCompatibilityJamo
+ hangulcompatibilityjamo => 'InHangu3',
+
+ # InHangulJamo
+ hanguljamo => 'InHangul',
+
+ # InHangulSyllables
+ hangulsyllables => 'InHangu2',
+
+ hanunoo => 'Hanunoo',
+ hebrew => 'Hebrew',
+
+ # HexDigit
+ hexdigit => 'HexDigit',
+
+ # InHighPrivateUseSurrogates
+ highprivateusesurrogates => 'InHighPr',
+
+ # InHighSurrogates
+ highsurrogates => 'InHighSu',
+
+ hiragana => 'Hiragana',
+ hyphen => 'Hyphen',
+
+ # IdContinue
+ idcontinue => 'IdContin',
+
+ ideographic => 'Ideograp',
+
+ # InIdeographicDescriptionCharacters
+ ideographicdescriptioncharacters => 'InIdeogr',
+
+ # IdsBinaryOperator
+ idsbinaryoperator => 'IdsBinar',
+
+ # IdStart
+ idstart => 'IdStart',
+
+ # IdsTrinaryOperator
+ idstrinaryoperator => 'IdsTrina',
+
+ # InAlphabeticPresentationForms
+ inalphabeticpresentationforms => 'InAlphab',
+
+ # InArabic
+ inarabic => 'InArabic',
+
+ # InArabicPresentationFormsA
+ inarabicpresentationformsa => 'InArabi3',
+
+ # InArabicPresentationFormsB
+ inarabicpresentationformsb => 'InArabi2',
+
+ # InArmenian
+ inarmenian => 'InArmeni',
+
+ # InArrows
+ inarrows => 'InArrows',
+
+ # InBasicLatin
+ inbasiclatin => 'InBasicL',
+
+ # InBengali
+ inbengali => 'InBengal',
+
+ # InBlockElements
+ inblockelements => 'InBlockE',
+
+ # InBopomofo
+ inbopomofo => 'InBopomo',
+
+ # InBopomofoExtended
+ inbopomofoextended => 'InBopom2',
+
+ # InBoxDrawing
+ inboxdrawing => 'InBoxDra',
+
+ # InBraillePatterns
+ inbraillepatterns => 'InBraill',
+
+ # InBuhid
+ inbuhid => 'InBuhid',
+
+ # InByzantineMusicalSymbols
+ inbyzantinemusicalsymbols => 'InByzant',
+
+ # InCherokee
+ incherokee => 'InCherok',
+
+ # InCjkCompatibility
+ incjkcompatibility => 'InCjkCom',
+
+ # InCjkCompatibilityForms
+ incjkcompatibilityforms => 'InCjkCo2',
+
+ # InCjkCompatibilityIdeographs
+ incjkcompatibilityideographs => 'InCjkCo3',
+
+ # InCjkCompatibilityIdeographsSupplement
+ incjkcompatibilityideographssupplement => 'InCjkCo4',
+
+ # InCjkRadicalsSupplement
+ incjkradicalssupplement => 'InCjkRad',
+
+ # InCjkSymbolsAndPunctuation
+ incjksymbolsandpunctuation => 'InCjkSym',
+
+ # InCjkUnifiedIdeographs
+ incjkunifiedideographs => 'InCjkUni',
+
+ # InCjkUnifiedIdeographsExtensionA
+ incjkunifiedideographsextensiona => 'InCjkUn3',
+
+ # InCjkUnifiedIdeographsExtensionB
+ incjkunifiedideographsextensionb => 'InCjkUn2',
+
+ # InCombiningDiacriticalMarks
+ incombiningdiacriticalmarks => 'InCombi2',
+
+ # InCombiningDiacriticalMarksForSymbols
+ incombiningdiacriticalmarksforsymbols => 'InCombi3',
+
+ # InCombiningHalfMarks
+ incombininghalfmarks => 'InCombin',
+
+ # InControlPictures
+ incontrolpictures => 'InContro',
+
+ # InCurrencySymbols
+ incurrencysymbols => 'InCurren',
+
+ # InCyrillic
+ incyrillic => 'InCyrill',
+
+ # InCyrillicSupplementary
+ incyrillicsupplementary => 'InCyril2',
+
+ # InDeseret
+ indeseret => 'InDesere',
+
+ # InDevanagari
+ indevanagari => 'InDevana',
+
+ # InDingbats
+ indingbats => 'InDingba',
+
+ # InEnclosedAlphanumerics
+ inenclosedalphanumerics => 'InEnclos',
+
+ # InEnclosedCjkLettersAndMonths
+ inenclosedcjklettersandmonths => 'InEnclo2',
+
+ # InEthiopic
+ inethiopic => 'InEthiop',
+
+ # InGeneralPunctuation
+ ingeneralpunctuation => 'InGenera',
+
+ # InGeometricShapes
+ ingeometricshapes => 'InGeomet',
+
+ # InGeorgian
+ ingeorgian => 'InGeorgi',
+
+ # InGothic
+ ingothic => 'InGothic',
+
+ # InGreekAndCoptic
+ ingreekandcoptic => 'InGreekA',
+
+ # InGreekExtended
+ ingreekextended => 'InGreekE',
+
+ # InGujarati
+ ingujarati => 'InGujara',
+
+ # InGurmukhi
+ ingurmukhi => 'InGurmuk',
+
+ # InHalfwidthAndFullwidthForms
+ inhalfwidthandfullwidthforms => 'InHalfwi',
+
+ # InHangulCompatibilityJamo
+ inhangulcompatibilityjamo => 'InHangu3',
+
+ # InHangulJamo
+ inhanguljamo => 'InHangul',
+
+ # InHangulSyllables
+ inhangulsyllables => 'InHangu2',
+
+ # InHanunoo
+ inhanunoo => 'InHanuno',
+
+ # InHebrew
+ inhebrew => 'InHebrew',
+
+ inherited => 'Inherite',
+
+ # InHighPrivateUseSurrogates
+ inhighprivateusesurrogates => 'InHighPr',
+
+ # InHighSurrogates
+ inhighsurrogates => 'InHighSu',
+
+ # InHiragana
+ inhiragana => 'InHiraga',
+
+ # InIdeographicDescriptionCharacters
+ inideographicdescriptioncharacters => 'InIdeogr',
+
+ # InIpaExtensions
+ inipaextensions => 'InIpaExt',
+
+ # InitialPunctuation
+ initialpunctuation => 'Pi',
+
+ # InKanbun
+ inkanbun => 'InKanbun',
+
+ # InKangxiRadicals
+ inkangxiradicals => 'InKangxi',
+
+ # InKannada
+ inkannada => 'InKannad',
+
+ # InKatakana
+ inkatakana => 'InKataka',
+
+ # InKatakanaPhoneticExtensions
+ inkatakanaphoneticextensions => 'InKatak2',
+
+ # InKhmer
+ inkhmer => 'InKhmer',
+
+ # InLao
+ inlao => 'InLao',
+
+ # InLatin1Supplement
+ inlatin1supplement => 'InLatin1',
+
+ # InLatinExtendedA
+ inlatinextendeda => 'InLatin2',
+
+ # InLatinExtendedAdditional
+ inlatinextendedadditional => 'InLatin3',
+
+ # InLatinExtendedB
+ inlatinextendedb => 'InLatinE',
+
+ # InLetterlikeSymbols
+ inletterlikesymbols => 'InLetter',
+
+ # InLowSurrogates
+ inlowsurrogates => 'InLowSur',
+
+ # InMalayalam
+ inmalayalam => 'InMalaya',
+
+ # InMathematicalAlphanumericSymbols
+ inmathematicalalphanumericsymbols => 'InMathe2',
+
+ # InMathematicalOperators
+ inmathematicaloperators => 'InMathem',
+
+ # InMiscellaneousMathematicalSymbolsA
+ inmiscellaneousmathematicalsymbolsa => 'InMisce3',
+
+ # InMiscellaneousMathematicalSymbolsB
+ inmiscellaneousmathematicalsymbolsb => 'InMisce4',
+
+ # InMiscellaneousSymbols
+ inmiscellaneoussymbols => 'InMiscel',
+
+ # InMiscellaneousTechnical
+ inmiscellaneoustechnical => 'InMisce2',
+
+ # InMongolian
+ inmongolian => 'InMongol',
+
+ # InMusicalSymbols
+ inmusicalsymbols => 'InMusica',
+
+ # InMyanmar
+ inmyanmar => 'InMyanma',
+
+ # InNumberForms
+ innumberforms => 'InNumber',
+
+ # InOgham
+ inogham => 'InOgham',
+
+ # InOldItalic
+ inolditalic => 'InOldIta',
+
+ # InOpticalCharacterRecognition
+ inopticalcharacterrecognition => 'InOptica',
+
+ # InOriya
+ inoriya => 'InOriya',
+
+ # InPrivateUseArea
+ inprivateusearea => 'InPrivat',
+
+ # InRunic
+ inrunic => 'InRunic',
+
+ # InSinhala
+ insinhala => 'InSinhal',
+
+ # InSmallFormVariants
+ insmallformvariants => 'InSmallF',
+
+ # InSpacingModifierLetters
+ inspacingmodifierletters => 'InSpacin',
+
+ # InSpecials
+ inspecials => 'InSpecia',
+
+ # InSuperscriptsAndSubscripts
+ insuperscriptsandsubscripts => 'InSupers',
+
+ # InSupplementalArrowsA
+ insupplementalarrowsa => 'InSuppl2',
+
+ # InSupplementalArrowsB
+ insupplementalarrowsb => 'InSupple',
+
+ # InSupplementalMathematicalOperators
+ insupplementalmathematicaloperators => 'InSuppl5',
+
+ # InSupplementaryPrivateUseAreaA
+ insupplementaryprivateuseareaa => 'InSuppl3',
+
+ # InSupplementaryPrivateUseAreaB
+ insupplementaryprivateuseareab => 'InSuppl4',
+
+ # InSyriac
+ insyriac => 'InSyriac',
+
+ # InTagalog
+ intagalog => 'InTagalo',
+
+ # InTagbanwa
+ intagbanwa => 'InTagban',
+
+ # InTags
+ intags => 'InTags',
+
+ # InTamil
+ intamil => 'InTamil',
+
+ # InTelugu
+ intelugu => 'InTelugu',
+
+ # InThaana
+ inthaana => 'InThaana',
+
+ # InThai
+ inthai => 'InThai',
+
+ # InTibetan
+ intibetan => 'InTibeta',
+
+ # InUnifiedCanadianAboriginalSyllabics
+ inunifiedcanadianaboriginalsyllabics => 'InUnifie',
+
+ # InVariationSelectors
+ invariationselectors => 'InVariat',
+
+ # InYiRadicals
+ inyiradicals => 'InYiRadi',
+
+ # InYiSyllables
+ inyisyllables => 'InYiSyll',
+
+ # InIpaExtensions
+ ipaextensions => 'InIpaExt',
+
+ # JoinControl
+ joincontrol => 'JoinCont',
+
+ # InKanbun
+ kanbun => 'InKanbun',
+
+ # InKangxiRadicals
+ kangxiradicals => 'InKangxi',
+
+ kannada => 'Kannada',
+ katakana => 'Katakana',
+
+ # InKatakanaPhoneticExtensions
+ katakanaphoneticextensions => 'InKatak2',
+
+ khmer => 'Khmer',
+ lao => 'Lao',
+ latin => 'Latin',
+
+ # InLatin1Supplement
+ latin1supplement => 'InLatin1',
+
+ # InLatinExtendedA
+ latinextendeda => 'InLatin2',
+
+ # InLatinExtendedAdditional
+ latinextendedadditional => 'InLatin3',
+
+ # InLatinExtendedB
+ latinextendedb => 'InLatinE',
+
+ letter => 'L',
+
+ # InLetterlikeSymbols
+ letterlikesymbols => 'InLetter',
+
+ # LetterNumber
+ letternumber => 'Nl',
+
+ # LineSeparator
+ lineseparator => 'Zl',
+
+ # LogicalOrderException
+ logicalorderexception => 'LogicalO',
+
+ lowercase => 'Lowercas',
+
+ # LowercaseLetter
+ lowercaseletter => 'Ll',
+
+ # InLowSurrogates
+ lowsurrogates => 'InLowSur',
+
+ malayalam => 'Malayala',
+ mark => 'M',
+ math => 'Math',
+
+ # InMathematicalAlphanumericSymbols
+ mathematicalalphanumericsymbols => 'InMathe2',
+
+ # InMathematicalOperators
+ mathematicaloperators => 'InMathem',
+
+ # MathSymbol
+ mathsymbol => 'Sm',
+
+ # InMiscellaneousMathematicalSymbolsA
+ miscellaneousmathematicalsymbolsa => 'InMisce3',
+
+ # InMiscellaneousMathematicalSymbolsB
+ miscellaneousmathematicalsymbolsb => 'InMisce4',
+
+ # InMiscellaneousSymbols
+ miscellaneoussymbols => 'InMiscel',
+
+ # InMiscellaneousTechnical
+ miscellaneoustechnical => 'InMisce2',
+
+ # ModifierLetter
+ modifierletter => 'Lm',
+
+ # ModifierSymbol
+ modifiersymbol => 'Sk',
+
+ mongolian => 'Mongolia',
+
+ # InMusicalSymbols
+ musicalsymbols => 'InMusica',
+
+ myanmar => 'Myanmar',
+
+ # NoncharacterCodePoint
+ noncharactercodepoint => 'Nonchara',
+
+ # NonSpacingMark
+ nonspacingmark => 'Mn',
+
+ number => 'N',
+
+ # InNumberForms
+ numberforms => 'InNumber',
+
+ ogham => 'Ogham',
+
+ # OldItalic
+ olditalic => 'OldItali',
+
+ # OpenPunctuation
+ openpunctuation => 'Ps',
+
+ # InOpticalCharacterRecognition
+ opticalcharacterrecognition => 'InOptica',
+
+ oriya => 'Oriya',
+ other => 'C',
+
+ # OtherAlphabetic
+ otheralphabetic => 'OtherAlp',
+
+ # OtherDefaultIgnorableCodePoint
+ otherdefaultignorablecodepoint => 'OtherDef',
+
+ # OtherGraphemeExtend
+ othergraphemeextend => 'OtherGra',
+
+ # OtherLetter
+ otherletter => 'Lo',
+
+ # OtherLowercase
+ otherlowercase => 'OtherLow',
+
+ # OtherMath
+ othermath => 'OtherMat',
+
+ # OtherNumber
+ othernumber => 'No',
+
+ # OtherPunctuation
+ otherpunctuation => 'Po',
+
+ # OtherSymbol
+ othersymbol => 'So',
+
+ # OtherUppercase
+ otheruppercase => 'OtherUpp',
+
+ # ParagraphSeparator
+ paragraphseparator => 'Zp',
+
+ # PrivateUse
+ privateuse => 'Co',
+
+ # InPrivateUseArea
+ privateusearea => 'InPrivat',
+
+ punctuation => 'P',
+
+ # QuotationMark
+ quotationmark => 'Quotatio',
+
+ radical => 'Radical',
+ runic => 'Runic',
+ separator => 'Z',
+ sinhala => 'Sinhala',
+
+ # InSmallFormVariants
+ smallformvariants => 'InSmallF',
+
+ # SoftDotted
+ softdotted => 'SoftDott',
+
+ # SpaceSeparator
+ spaceseparator => 'Zs',
+
+ # SpacingMark
+ spacingmark => 'Mc',
+
+ # InSpacingModifierLetters
+ spacingmodifierletters => 'InSpacin',
+
+ # InSpecials
+ specials => 'InSpecia',
+
+ # InSuperscriptsAndSubscripts
+ superscriptsandsubscripts => 'InSupers',
+
+ # InSupplementalArrowsA
+ supplementalarrowsa => 'InSuppl2',
+
+ # InSupplementalArrowsB
+ supplementalarrowsb => 'InSupple',
+
+ # InSupplementalMathematicalOperators
+ supplementalmathematicaloperators => 'InSuppl5',
+
+ # InSupplementaryPrivateUseAreaA
+ supplementaryprivateuseareaa => 'InSuppl3',
+
+ # InSupplementaryPrivateUseAreaB
+ supplementaryprivateuseareab => 'InSuppl4',
+
+ surrogate => 'Cs',
+ symbol => 'S',
+ syriac => 'Syriac',
+ tagalog => 'Tagalog',
+ tagbanwa => 'Tagbanwa',
+
+ # InTags
+ tags => 'InTags',
+
+ tamil => 'Tamil',
+ telugu => 'Telugu',
+
+ # TerminalPunctuation
+ terminalpunctuation => 'Terminal',
+
+ thaana => 'Thaana',
+ thai => 'Thai',
+ tibetan => 'Tibetan',
+
+ # TitlecaseLetter
+ titlecaseletter => 'Lt',
+
+ unassigned => 'Cn',
+
+ # InUnifiedCanadianAboriginalSyllabics
+ unifiedcanadianaboriginalsyllabics => 'InUnifie',
+
+ # UnifiedIdeograph
+ unifiedideograph => 'UnifiedI',
+
+ uppercase => 'Uppercas',
+
+ # UppercaseLetter
+ uppercaseletter => 'Lu',
+
+ # InVariationSelectors
+ variationselectors => 'InVariat',
+
+ # WhiteSpace
+ whitespace => 'WhiteSpa',
+
+ yi => 'Yi',
+
+ # InYiRadicals
+ yiradicals => 'InYiRadi',
+
+ # InYiSyllables
+ yisyllables => 'InYiSyll',
+
+);
+1