Upgrade to Unicode 4.0.0. (Well, upgrade at least
[p5sagit/p5-mst-13.2.git] / lib / unicore / Properties
1 ##
2 ## This file created by ./mktables
3 ## List of built-in \p{...}/\P{...} properties.
4 ##
5 ## '*' means name may be 'fuzzy'
6 ##
7
8   \p{ASCII}                                  [[:ASCII:]]
9   \p{Alnum}                                  [[:Alnum:]]
10 * \p{Alphabetic}                             [\p{L}\p{OtherAlphabetic}]
11   \p{Alpha}                                  [[:Alpha:]]
12   \p{Any}                                    Alias for \p{Any} ([\x{0000}-\x{10FFFF}])
13   \p{Any}                                    [\x{0000}-\x{10FFFF}]
14 * \p{Arabic}                                 Script 'ARABIC'
15 * \p{Armenian}                               Script 'ARMENIAN'
16 * \p{AsciiHexDigit}                          Extended property 'ASCII_Hex_Digit'
17   \p{Assigned}                               All assigned code points
18 * \p{Bengali}                                Script 'BENGALI'
19   \p{BidiAL}                                 Bi-directional category 'AL'
20   \p{BidiAN}                                 Bi-directional category 'AN'
21   \p{BidiBN}                                 Bi-directional category 'BN'
22   \p{BidiB}                                  Bi-directional category 'B'
23   \p{BidiCS}                                 Bi-directional category 'CS'
24 * \p{BidiControl}                            Extended property 'Bidi_Control'
25   \p{BidiEN}                                 Bi-directional category 'EN'
26   \p{BidiES}                                 Bi-directional category 'ES'
27   \p{BidiET}                                 Bi-directional category 'ET'
28   \p{BidiLRE}                                Bi-directional category 'LRE'
29   \p{BidiLRO}                                Bi-directional category 'LRO'
30   \p{BidiL}                                  Bi-directional category 'L'
31   \p{BidiNSM}                                Bi-directional category 'NSM'
32   \p{BidiON}                                 Bi-directional category 'ON'
33   \p{BidiPDF}                                Bi-directional category 'PDF'
34   \p{BidiRLE}                                Bi-directional category 'RLE'
35   \p{BidiRLO}                                Bi-directional category 'RLO'
36   \p{BidiR}                                  Bi-directional category 'R'
37   \p{BidiS}                                  Bi-directional category 'S'
38   \p{BidiWS}                                 Bi-directional category 'WS'
39   \p{Blank}                                  [[:Blank:]]
40 * \p{Bopomofo}                               Script 'BOPOMOFO'
41 * \p{Braille}                                Script 'BRAILLE'
42 * \p{Buhid}                                  Script 'BUHID'
43 * \p{CanadianAboriginal}                     Script 'CANADIAN_ABORIGINAL'
44   \p{Canon}                                  Decomposes to multiple characters
45 * \p{Cc}                                     Alias for \p{Cc} (General Category 'Cc')
46   \p{Cc}                                     General Category 'Cc'
47 * \p{Cf}                                     Alias for \p{Cf} (General Category 'Cf')
48   \p{Cf}                                     General Category 'Cf'
49 * \p{Cherokee}                               Script 'CHEROKEE'
50   \p{Cntrl}                                  [[:Cntrl:]]
51 * \p{Cn}                                     Alias for \p{Cn} (General Category 'Cn' [not functional in Perl])
52   \p{Cn}                                     General Category 'Cn' [not functional in Perl]
53 * \p{Common}                                 Pseudo-Script of codepoints not in other Unicode scripts
54   \p{Compat}                                 Compatible with a more-basic character
55 * \p{Co}                                     Alias for \p{Co} (General Category 'Co')
56   \p{Co}                                     General Category 'Co'
57 * \p{Cs}                                     Alias for \p{Cs} (General Category 'Cs')
58   \p{Cs}                                     General Category 'Cs'
59 * \p{Cypriot}                                Script 'CYPRIOT'
60 * \p{Cyrillic}                               Script 'CYRILLIC'
61 * \p{C}                                      Alias for \p{C} (Major Category 'C')
62   \p{C}                                      Major Category 'C'
63   \p{DCcircle}                               Compatible with 'circle'
64   \p{DCcompat}                               Compatible with 'compat'
65   \p{DCfinal}                                Compatible with 'final'
66   \p{DCfont}                                 Compatible with 'font'
67   \p{DCfraction}                             Compatible with 'fraction'
68   \p{DCinitial}                              Compatible with 'initial'
69   \p{DCisolated}                             Compatible with 'isolated'
70   \p{DCmedial}                               Compatible with 'medial'
71   \p{DCnarrow}                               Compatible with 'narrow'
72   \p{DCnoBreak}                              Compatible with 'noBreak'
73   \p{DCsmall}                                Compatible with 'small'
74   \p{DCsquare}                               Compatible with 'square'
75   \p{DCsub}                                  Compatible with 'sub'
76   \p{DCsuper}                                Compatible with 'super'
77   \p{DCvertical}                             Compatible with 'vertical'
78   \p{DCwide}                                 Compatible with 'wide'
79 * \p{Dash}                                   Extended property 'Dash'
80 * \p{Deprecated}                             Extended property 'Deprecated'
81 * \p{Deseret}                                Script 'DESERET'
82 * \p{Devanagari}                             Script 'DEVANAGARI'
83 * \p{Diacritic}                              Extended property 'Diacritic'
84   \p{Digit}                                  [[:Digit:]]
85 * \p{Ethiopic}                               Script 'ETHIOPIC'
86 * \p{Extender}                               Extended property 'Extender'
87 * \p{Georgian}                               Script 'GEORGIAN'
88 * \p{Gothic}                                 Script 'GOTHIC'
89 * \p{GraphemeLink}                           Extended property 'Grapheme_Link'
90   \p{Graph}                                  [[:Graph:]]
91 * \p{Greek}                                  Script 'GREEK'
92 * \p{Gujarati}                               Script 'GUJARATI'
93 * \p{Gurmukhi}                               Script 'GURMUKHI'
94 * \p{Hangul}                                 Script 'HANGUL'
95 * \p{Hanunoo}                                Script 'HANUNOO'
96 * \p{Han}                                    Script 'HAN'
97 * \p{Hebrew}                                 Script 'HEBREW'
98 * \p{HexDigit}                               Extended property 'Hex_Digit'
99 * \p{Hiragana}                               Script 'HIRAGANA'
100 * \p{Hyphen}                                 Extended property 'Hyphen'
101 * \p{IdContinue}                             [\p{ID_Start}\p{Mn}\p{Mc}\p{Nd}\p{Pc}]
102 * \p{IdStart}                                [\p{Ll}\p{Lu}\p{Lt}\p{Lm}\p{Lo}\p{Nl}]
103 * \p{Ideographic}                            Extended property 'Ideographic'
104 * \p{IdsBinaryOperator}                      Extended property 'IDS_Binary_Operator'
105 * \p{IdsTrinaryOperator}                     Extended property 'IDS_Trinary_Operator'
106 * \p{InAegeanNumbers}                        Block 'Aegean Numbers'
107 * \p{InAlphabeticPresentationForms}          Block 'Alphabetic Presentation Forms'
108 * \p{InArabicPresentationFormsA}             Block 'Arabic Presentation Forms-A'
109 * \p{InArabicPresentationFormsB}             Block 'Arabic Presentation Forms-B'
110 * \p{InArabic}                               Block 'Arabic'
111 * \p{InArmenian}                             Block 'Armenian'
112 * \p{InArrows}                               Block 'Arrows'
113 * \p{InBasicLatin}                           Block 'Basic Latin'
114 * \p{InBengali}                              Block 'Bengali'
115 * \p{InBlockElements}                        Block 'Block Elements'
116 * \p{InBopomofoExtended}                     Block 'Bopomofo Extended'
117 * \p{InBopomofo}                             Block 'Bopomofo'
118 * \p{InBoxDrawing}                           Block 'Box Drawing'
119 * \p{InBraillePatterns}                      Block 'Braille Patterns'
120 * \p{InBuhid}                                Block 'Buhid'
121 * \p{InByzantineMusicalSymbols}              Block 'Byzantine Musical Symbols'
122 * \p{InCherokee}                             Block 'Cherokee'
123 * \p{InCjkCompatibilityForms}                Block 'CJK Compatibility Forms'
124 * \p{InCjkCompatibilityIdeographsSupplement} Block 'CJK Compatibility Ideographs Supplement'
125 * \p{InCjkCompatibilityIdeographs}           Block 'CJK Compatibility Ideographs'
126 * \p{InCjkCompatibility}                     Block 'CJK Compatibility'
127 * \p{InCjkRadicalsSupplement}                Block 'CJK Radicals Supplement'
128 * \p{InCjkSymbolsAndPunctuation}             Block 'CJK Symbols and Punctuation'
129 * \p{InCjkUnifiedIdeographsExtensionA}       Block 'CJK Unified Ideographs Extension A'
130 * \p{InCjkUnifiedIdeographsExtensionB}       Block 'CJK Unified Ideographs Extension B'
131 * \p{InCjkUnifiedIdeographs}                 Block 'CJK Unified Ideographs'
132 * \p{InCombiningDiacriticalMarksForSymbols}  Block 'Combining Diacritical Marks for Symbols'
133 * \p{InCombiningDiacriticalMarks}            Block 'Combining Diacritical Marks'
134 * \p{InCombiningHalfMarks}                   Block 'Combining Half Marks'
135 * \p{InControlPictures}                      Block 'Control Pictures'
136 * \p{InCurrencySymbols}                      Block 'Currency Symbols'
137 * \p{InCypriotSyllabary}                     Block 'Cypriot Syllabary'
138 * \p{InCyrillicSupplementary}                Block 'Cyrillic Supplementary'
139 * \p{InCyrillic}                             Block 'Cyrillic'
140 * \p{InDeseret}                              Block 'Deseret'
141 * \p{InDevanagari}                           Block 'Devanagari'
142 * \p{InDingbats}                             Block 'Dingbats'
143 * \p{InEnclosedAlphanumerics}                Block 'Enclosed Alphanumerics'
144 * \p{InEnclosedCjkLettersAndMonths}          Block 'Enclosed CJK Letters and Months'
145 * \p{InEthiopic}                             Block 'Ethiopic'
146 * \p{InGeneralPunctuation}                   Block 'General Punctuation'
147 * \p{InGeometricShapes}                      Block 'Geometric Shapes'
148 * \p{InGeorgian}                             Block 'Georgian'
149 * \p{InGothic}                               Block 'Gothic'
150 * \p{InGreekAndCoptic}                       Block 'Greek and Coptic'
151 * \p{InGreekExtended}                        Block 'Greek Extended'
152 * \p{InGujarati}                             Block 'Gujarati'
153 * \p{InGurmukhi}                             Block 'Gurmukhi'
154 * \p{InHalfwidthAndFullwidthForms}           Block 'Halfwidth and Fullwidth Forms'
155 * \p{InHangulCompatibilityJamo}              Block 'Hangul Compatibility Jamo'
156 * \p{InHangulJamo}                           Block 'Hangul Jamo'
157 * \p{InHangulSyllables}                      Block 'Hangul Syllables'
158 * \p{InHanunoo}                              Block 'Hanunoo'
159 * \p{InHebrew}                               Block 'Hebrew'
160 * \p{InHighPrivateUseSurrogates}             Block 'High Private Use Surrogates'
161 * \p{InHighSurrogates}                       Block 'High Surrogates'
162 * \p{InHiragana}                             Block 'Hiragana'
163 * \p{InIdeographicDescriptionCharacters}     Block 'Ideographic Description Characters'
164 * \p{InIpaExtensions}                        Block 'IPA Extensions'
165 * \p{InKanbun}                               Block 'Kanbun'
166 * \p{InKangxiRadicals}                       Block 'Kangxi Radicals'
167 * \p{InKannada}                              Block 'Kannada'
168 * \p{InKatakanaPhoneticExtensions}           Block 'Katakana Phonetic Extensions'
169 * \p{InKatakana}                             Block 'Katakana'
170 * \p{InKhmerSymbols}                         Block 'Khmer Symbols'
171 * \p{InKhmer}                                Block 'Khmer'
172 * \p{InLao}                                  Block 'Lao'
173 * \p{InLatin1Supplement}                     Block 'Latin-1 Supplement'
174 * \p{InLatinExtendedAdditional}              Block 'Latin Extended Additional'
175 * \p{InLatinExtendedA}                       Block 'Latin Extended-A'
176 * \p{InLatinExtendedB}                       Block 'Latin Extended-B'
177 * \p{InLetterlikeSymbols}                    Block 'Letterlike Symbols'
178 * \p{InLimbu}                                Block 'Limbu'
179 * \p{InLinearBIdeograms}                     Block 'Linear B Ideograms'
180 * \p{InLinearBSyllabary}                     Block 'Linear B Syllabary'
181 * \p{InLowSurrogates}                        Block 'Low Surrogates'
182 * \p{InMalayalam}                            Block 'Malayalam'
183 * \p{InMathematicalAlphanumericSymbols}      Block 'Mathematical Alphanumeric Symbols'
184 * \p{InMathematicalOperators}                Block 'Mathematical Operators'
185 * \p{InMiscellaneousMathematicalSymbolsA}    Block 'Miscellaneous Mathematical Symbols-A'
186 * \p{InMiscellaneousMathematicalSymbolsB}    Block 'Miscellaneous Mathematical Symbols-B'
187 * \p{InMiscellaneousSymbolsAndArrows}        Block 'Miscellaneous Symbols and Arrows'
188 * \p{InMiscellaneousSymbols}                 Block 'Miscellaneous Symbols'
189 * \p{InMiscellaneousTechnical}               Block 'Miscellaneous Technical'
190 * \p{InMongolian}                            Block 'Mongolian'
191 * \p{InMusicalSymbols}                       Block 'Musical Symbols'
192 * \p{InMyanmar}                              Block 'Myanmar'
193 * \p{InNumberForms}                          Block 'Number Forms'
194 * \p{InOgham}                                Block 'Ogham'
195 * \p{InOldItalic}                            Block 'Old Italic'
196 * \p{InOpticalCharacterRecognition}          Block 'Optical Character Recognition'
197 * \p{InOriya}                                Block 'Oriya'
198 * \p{InOsmanya}                              Block 'Osmanya'
199 * \p{InPhoneticExtensions}                   Block 'Phonetic Extensions'
200 * \p{InPrivateUseArea}                       Block 'Private Use Area'
201 * \p{InRunic}                                Block 'Runic'
202 * \p{InShavian}                              Block 'Shavian'
203 * \p{InSinhala}                              Block 'Sinhala'
204 * \p{InSmallFormVariants}                    Block 'Small Form Variants'
205 * \p{InSpacingModifierLetters}               Block 'Spacing Modifier Letters'
206 * \p{InSpecials}                             Block 'Specials'
207 * \p{InSuperscriptsAndSubscripts}            Block 'Superscripts and Subscripts'
208 * \p{InSupplementalArrowsA}                  Block 'Supplemental Arrows-A'
209 * \p{InSupplementalArrowsB}                  Block 'Supplemental Arrows-B'
210 * \p{InSupplementalMathematicalOperators}    Block 'Supplemental Mathematical Operators'
211 * \p{InSupplementaryPrivateUseAreaA}         Block 'Supplementary Private Use Area-A'
212 * \p{InSupplementaryPrivateUseAreaB}         Block 'Supplementary Private Use Area-B'
213 * \p{InSyriac}                               Block 'Syriac'
214 * \p{InTagalog}                              Block 'Tagalog'
215 * \p{InTagbanwa}                             Block 'Tagbanwa'
216 * \p{InTags}                                 Block 'Tags'
217 * \p{InTaiLe}                                Block 'Tai Le'
218 * \p{InTaiXuanJingSymbols}                   Block 'Tai Xuan Jing Symbols'
219 * \p{InTamil}                                Block 'Tamil'
220 * \p{InTelugu}                               Block 'Telugu'
221 * \p{InThaana}                               Block 'Thaana'
222 * \p{InThai}                                 Block 'Thai'
223 * \p{InTibetan}                              Block 'Tibetan'
224 * \p{InUgaritic}                             Block 'Ugaritic'
225 * \p{InUnifiedCanadianAboriginalSyllabics}   Block 'Unified Canadian Aboriginal Syllabics'
226 * \p{InVariationSelectorsSupplement}         Block 'Variation Selectors Supplement'
227 * \p{InVariationSelectors}                   Block 'Variation Selectors'
228 * \p{InYiRadicals}                           Block 'Yi Radicals'
229 * \p{InYiSyllables}                          Block 'Yi Syllables'
230 * \p{InYijingHexagramSymbols}                Block 'Yijing Hexagram Symbols'
231 * \p{Inherited}                              Script 'INHERITED'
232 * \p{JoinControl}                            Extended property 'Join_Control'
233 * \p{Kannada}                                Script 'KANNADA'
234 * \p{Katakana}                               Script 'KATAKANA'
235 * \p{Khmer}                                  Script 'KHMER'
236   \p{L&}                                     [\p{Ll}\p{Lu}\p{Lt}]
237 * \p{Lao}                                    Script 'LAO'
238 * \p{Latin}                                  Script 'LATIN'
239 * \p{Limbu}                                  Script 'LIMBU'
240 * \p{LinearB}                                Script 'LINEAR_B'
241 * \p{Ll}                                     Alias for \p{Ll} (General Category 'Ll')
242   \p{Ll}                                     General Category 'Ll'
243 * \p{Lm}                                     Alias for \p{Lm} (General Category 'Lm')
244   \p{Lm}                                     General Category 'Lm'
245 * \p{LogicalOrderException}                  Extended property 'Logical_Order_Exception'
246 * \p{Lowercase}                              [\p{Ll}\p{OtherLowercase}]
247   \p{Lower}                                  [[:Lower:]]
248 * \p{Lo}                                     Alias for \p{Lo} (General Category 'Lo')
249   \p{Lo}                                     General Category 'Lo'
250 * \p{Lt}                                     Alias for \p{Lt} (General Category 'Lt')
251   \p{Lt}                                     General Category 'Lt'
252 * \p{Lu}                                     Alias for \p{Lu} (General Category 'Lu')
253   \p{Lu}                                     General Category 'Lu'
254 * \p{L}                                      Alias for \p{L} (Major Category 'L')
255   \p{L}                                      Major Category 'L'
256 * \p{Malayalam}                              Script 'MALAYALAM'
257 * \p{Math}                                   [\p{Sm}\p{OtherMath}]
258 * \p{Mc}                                     Alias for \p{Mc} (General Category 'Mc')
259   \p{Mc}                                     General Category 'Mc'
260 * \p{Me}                                     Alias for \p{Me} (General Category 'Me')
261   \p{Me}                                     General Category 'Me'
262   \p{Mirrored}                               Mirrored in bidirectional text
263 * \p{Mn}                                     Alias for \p{Mn} (General Category 'Mn')
264   \p{Mn}                                     General Category 'Mn'
265 * \p{Mongolian}                              Script 'MONGOLIAN'
266 * \p{Myanmar}                                Script 'MYANMAR'
267 * \p{M}                                      Alias for \p{M} (Major Category 'M')
268   \p{M}                                      Major Category 'M'
269 * \p{Nd}                                     Alias for \p{Nd} (General Category 'Nd')
270   \p{Nd}                                     General Category 'Nd'
271 * \p{Nl}                                     Alias for \p{Nl} (General Category 'Nl')
272   \p{Nl}                                     General Category 'Nl'
273 * \p{NoncharacterCodePoint}                  Extended property 'Noncharacter_Code_Point'
274 * \p{No}                                     Alias for \p{No} (General Category 'No')
275   \p{No}                                     General Category 'No'
276 * \p{N}                                      Alias for \p{N} (Major Category 'N')
277   \p{N}                                      Major Category 'N'
278 * \p{Ogham}                                  Script 'OGHAM'
279 * \p{OldItalic}                              Script 'OLD_ITALIC'
280 * \p{Oriya}                                  Script 'ORIYA'
281 * \p{Osmanya}                                Script 'OSMANYA'
282 * \p{OtherAlphabetic}                        Extended property 'Other_Alphabetic'
283 * \p{OtherDefaultIgnorableCodePoint}         Extended property 'Other_Default_Ignorable_Code_Point'
284 * \p{OtherGraphemeExtend}                    Extended property 'Other_Grapheme_Extend'
285 * \p{OtherIdStart}                           Extended property 'Other_ID_Start'
286 * \p{OtherLowercase}                         Extended property 'Other_Lowercase'
287 * \p{OtherMath}                              Extended property 'Other_Math'
288 * \p{OtherUppercase}                         Extended property 'Other_Uppercase'
289 * \p{Pc}                                     Alias for \p{Pc} (General Category 'Pc')
290   \p{Pc}                                     General Category 'Pc'
291 * \p{Pd}                                     Alias for \p{Pd} (General Category 'Pd')
292   \p{Pd}                                     General Category 'Pd'
293 * \p{Pe}                                     Alias for \p{Pe} (General Category 'Pe')
294   \p{Pe}                                     General Category 'Pe'
295 * \p{Pf}                                     Alias for \p{Pf} (General Category 'Pf')
296   \p{Pf}                                     General Category 'Pf'
297 * \p{Pi}                                     Alias for \p{Pi} (General Category 'Pi')
298   \p{Pi}                                     General Category 'Pi'
299 * \p{Po}                                     Alias for \p{Po} (General Category 'Po')
300   \p{Po}                                     General Category 'Po'
301   \p{Print}                                  [[:Print:]]
302 * \p{Ps}                                     Alias for \p{Ps} (General Category 'Ps')
303   \p{Ps}                                     General Category 'Ps'
304   \p{Punct}                                  [[:Punct:]]
305 * \p{P}                                      Alias for \p{P} (Major Category 'P')
306   \p{P}                                      Major Category 'P'
307 * \p{QuotationMark}                          Extended property 'Quotation_Mark'
308 * \p{Radical}                                Extended property 'Radical'
309 * \p{Runic}                                  Script 'RUNIC'
310 * \p{Sc}                                     Alias for \p{Sc} (General Category 'Sc')
311   \p{Sc}                                     General Category 'Sc'
312 * \p{Shavian}                                Script 'SHAVIAN'
313 * \p{Sinhala}                                Script 'SINHALA'
314 * \p{Sk}                                     Alias for \p{Sk} (General Category 'Sk')
315   \p{Sk}                                     General Category 'Sk'
316 * \p{Sm}                                     Alias for \p{Sm} (General Category 'Sm')
317   \p{Sm}                                     General Category 'Sm'
318 * \p{SoftDotted}                             Extended property 'Soft_Dotted'
319 * \p{So}                                     Alias for \p{So} (General Category 'So')
320   \p{So}                                     General Category 'So'
321   \p{SpacePerl}                              \s
322   \p{Space}                                  [[:Space:]]
323 * \p{Syriac}                                 Script 'SYRIAC'
324 * \p{S}                                      Alias for \p{S} (Major Category 'S')
325   \p{S}                                      Major Category 'S'
326 * \p{Tagalog}                                Script 'TAGALOG'
327 * \p{Tagbanwa}                               Script 'TAGBANWA'
328 * \p{TaiLe}                                  Script 'TAI_LE'
329 * \p{Tamil}                                  Script 'TAMIL'
330 * \p{Telugu}                                 Script 'TELUGU'
331 * \p{TerminalPunctuation}                    Extended property 'Terminal_Punctuation'
332 * \p{Thaana}                                 Script 'THAANA'
333 * \p{Thai}                                   Script 'THAI'
334 * \p{Tibetan}                                Script 'TIBETAN'
335   \p{Title}                                  [[:Title:]]
336 * \p{Ugaritic}                               Script 'UGARITIC'
337 * \p{UnifiedIdeograph}                       Extended property 'Unified_Ideograph'
338 * \p{Uppercase}                              [\p{Lu}\p{Other_Uppercase}]
339   \p{Upper}                                  [[:Upper:]]
340 * \p{WhiteSpace}                             Extended property 'White_Space'
341   \p{Word}                                   [[:Word:]]
342   \p{XDigit}                                 [[:XDigit:]]
343 * \p{Yi}                                     Script 'YI'
344 * \p{Zl}                                     Alias for \p{Zl} (General Category 'Zl')
345   \p{Zl}                                     General Category 'Zl'
346 * \p{Zp}                                     Alias for \p{Zp} (General Category 'Zp')
347   \p{Zp}                                     General Category 'Zp'
348 * \p{Zs}                                     Alias for \p{Zs} (General Category 'Zs')
349   \p{Zs}                                     General Category 'Zs'
350 * \p{Z}                                      Alias for \p{Z} (Major Category 'Z')
351   \p{Z}                                      Major Category 'Z'
352   \p{_CanonDCIJ}                             (for internal casefolding use)
353   \p{_CaseIgnorable}                         (for internal casefolding use)
354   \p{_CombAbove}                             (for internal casefolding use)