Commit | Line | Data |
99598c8c |
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:]] |
99598c8c |
9 | \p{Alnum} [[:Alnum:]] |
10 | * \p{Alphabetic} [\p{L}\p{OtherAlphabetic}] |
11 | \p{Alpha} [[:Alpha:]] |
5beb625e |
12 | \p{Any} Alias for \p{Any} ([\x{0000}-\x{10FFFF}]) |
13 | \p{Any} [\x{0000}-\x{10FFFF}] |
99598c8c |
14 | * \p{Arabic} Script 'ARABIC' |
15 | * \p{Armenian} Script 'ARMENIAN' |
16 | * \p{AsciiHexDigit} Extended property 'ASCII_Hex_Digit' |
5beb625e |
17 | \p{Assigned} All assigned code points |
99598c8c |
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{CanadianAboriginal} Script 'CANADIAN-ABORIGINAL' |
42 | \p{Canon} Decomposes to multiple characters |
5beb625e |
43 | * \p{Cc} Alias for \p{Cc} (General Category 'Cc') |
99598c8c |
44 | \p{Cc} General Category 'Cc' |
5beb625e |
45 | * \p{Cf} Alias for \p{Cf} (General Category 'Cf') |
99598c8c |
46 | \p{Cf} General Category 'Cf' |
47 | * \p{Cherokee} Script 'CHEROKEE' |
99598c8c |
48 | \p{Cntrl} [[:Cntrl:]] |
5beb625e |
49 | * \p{Cn} Alias for \p{Cn} (General Category 'Cn' [not functional in Perl]) |
99598c8c |
50 | \p{Cn} General Category 'Cn' [not functional in Perl] |
51 | * \p{Common} Pseudo-Script of codepoints not in other Unicode scripts |
52 | \p{Compat} Compatible with a more-basic character |
5beb625e |
53 | * \p{Co} Alias for \p{Co} (General Category 'Co') |
99598c8c |
54 | \p{Co} General Category 'Co' |
5beb625e |
55 | * \p{Cs} Alias for \p{Cs} (General Category 'Cs') |
99598c8c |
56 | \p{Cs} General Category 'Cs' |
99598c8c |
57 | * \p{Cyrillic} Script 'CYRILLIC' |
5beb625e |
58 | * \p{C} Alias for \p{C} (Major Category 'C') |
99598c8c |
59 | \p{C} Major Category 'C' |
60 | \p{DCcircle} Compatible with 'circle' |
61 | \p{DCcompat} Compatible with 'compat' |
62 | \p{DCfinal} Compatible with 'final' |
63 | \p{DCfont} Compatible with 'font' |
64 | \p{DCfraction} Compatible with 'fraction' |
65 | \p{DCinitial} Compatible with 'initial' |
66 | \p{DCisolated} Compatible with 'isolated' |
67 | \p{DCmedial} Compatible with 'medial' |
68 | \p{DCnarrow} Compatible with 'narrow' |
69 | \p{DCnoBreak} Compatible with 'noBreak' |
70 | \p{DCsmall} Compatible with 'small' |
71 | \p{DCsquare} Compatible with 'square' |
72 | \p{DCsub} Compatible with 'sub' |
73 | \p{DCsuper} Compatible with 'super' |
74 | \p{DCvertical} Compatible with 'vertical' |
75 | \p{DCwide} Compatible with 'wide' |
99598c8c |
76 | * \p{Dash} Extended property 'Dash' |
99598c8c |
77 | * \p{Deseret} Script 'DESERET' |
78 | * \p{Devanagari} Script 'DEVANAGARI' |
79 | * \p{Diacritic} Extended property 'Diacritic' |
80 | \p{Digit} [[:Digit:]] |
99598c8c |
81 | * \p{Ethiopic} Script 'ETHIOPIC' |
82 | * \p{Extender} Extended property 'Extender' |
99598c8c |
83 | * \p{Georgian} Script 'GEORGIAN' |
84 | * \p{Gothic} Script 'GOTHIC' |
85 | \p{Graph} [[:Graph:]] |
86 | * \p{Greek} Script 'GREEK' |
87 | * \p{Gujarati} Script 'GUJARATI' |
88 | * \p{Gurmukhi} Script 'GURMUKHI' |
89 | * \p{Hangul} Script 'HANGUL' |
90 | * \p{Han} Script 'HAN' |
91 | * \p{Hebrew} Script 'HEBREW' |
92 | * \p{HexDigit} Extended property 'Hex_Digit' |
93 | * \p{Hiragana} Script 'HIRAGANA' |
94 | * \p{Hyphen} Extended property 'Hyphen' |
95 | * \p{IdContinue} [\p{ID_Start}\p{Mn}\p{Mc}\p{Nd}\p{Pc}] |
96 | * \p{IdStart} [\p{Ll}\p{Lu}\p{Lt}\p{Lm}\p{Lo}\p{Nl}] |
97 | * \p{Ideographic} Extended property 'Ideographic' |
98 | * \p{InAlphabeticPresentationForms} Block 'Alphabetic Presentation Forms' |
99 | * \p{InArabicPresentationFormsA} Block 'Arabic Presentation Forms-A' |
100 | * \p{InArabicPresentationFormsB} Block 'Arabic Presentation Forms-B' |
101 | * \p{InArabic} Block 'Arabic' |
102 | * \p{InArmenian} Block 'Armenian' |
103 | * \p{InArrows} Block 'Arrows' |
104 | * \p{InBasicLatin} Block 'Basic Latin' |
105 | * \p{InBengali} Block 'Bengali' |
106 | * \p{InBlockElements} Block 'Block Elements' |
107 | * \p{InBopomofoExtended} Block 'Bopomofo Extended' |
108 | * \p{InBopomofo} Block 'Bopomofo' |
109 | * \p{InBoxDrawing} Block 'Box Drawing' |
110 | * \p{InBraillePatterns} Block 'Braille Patterns' |
111 | * \p{InByzantineMusicalSymbols} Block 'Byzantine Musical Symbols' |
112 | * \p{InCherokee} Block 'Cherokee' |
113 | * \p{InCjkCompatibilityForms} Block 'CJK Compatibility Forms' |
114 | * \p{InCjkCompatibilityIdeographsSupplement} Block 'CJK Compatibility Ideographs Supplement' |
115 | * \p{InCjkCompatibilityIdeographs} Block 'CJK Compatibility Ideographs' |
116 | * \p{InCjkCompatibility} Block 'CJK Compatibility' |
117 | * \p{InCjkRadicalsSupplement} Block 'CJK Radicals Supplement' |
118 | * \p{InCjkSymbolsAndPunctuation} Block 'CJK Symbols and Punctuation' |
119 | * \p{InCjkUnifiedIdeographsExtensionA} Block 'CJK Unified Ideographs Extension A' |
120 | * \p{InCjkUnifiedIdeographsExtensionB} Block 'CJK Unified Ideographs Extension B' |
121 | * \p{InCjkUnifiedIdeographs} Block 'CJK Unified Ideographs' |
122 | * \p{InCombiningDiacriticalMarks} Block 'Combining Diacritical Marks' |
123 | * \p{InCombiningHalfMarks} Block 'Combining Half Marks' |
124 | * \p{InCombiningMarksForSymbols} Block 'Combining Marks for Symbols' |
125 | * \p{InControlPictures} Block 'Control Pictures' |
126 | * \p{InCurrencySymbols} Block 'Currency Symbols' |
127 | * \p{InCyrillic} Block 'Cyrillic' |
128 | * \p{InDeseret} Block 'Deseret' |
129 | * \p{InDevanagari} Block 'Devanagari' |
130 | * \p{InDingbats} Block 'Dingbats' |
131 | * \p{InEnclosedAlphanumerics} Block 'Enclosed Alphanumerics' |
132 | * \p{InEnclosedCjkLettersAndMonths} Block 'Enclosed CJK Letters and Months' |
133 | * \p{InEthiopic} Block 'Ethiopic' |
134 | * \p{InGeneralPunctuation} Block 'General Punctuation' |
135 | * \p{InGeometricShapes} Block 'Geometric Shapes' |
136 | * \p{InGeorgian} Block 'Georgian' |
137 | * \p{InGothic} Block 'Gothic' |
138 | * \p{InGreekExtended} Block 'Greek Extended' |
139 | * \p{InGreek} Block 'Greek' |
140 | * \p{InGujarati} Block 'Gujarati' |
141 | * \p{InGurmukhi} Block 'Gurmukhi' |
142 | * \p{InHalfwidthAndFullwidthForms} Block 'Halfwidth and Fullwidth Forms' |
143 | * \p{InHangulCompatibilityJamo} Block 'Hangul Compatibility Jamo' |
144 | * \p{InHangulJamo} Block 'Hangul Jamo' |
145 | * \p{InHangulSyllables} Block 'Hangul Syllables' |
146 | * \p{InHebrew} Block 'Hebrew' |
147 | * \p{InHighPrivateUseSurrogates} Block 'High Private Use Surrogates' |
148 | * \p{InHighSurrogates} Block 'High Surrogates' |
149 | * \p{InHiragana} Block 'Hiragana' |
150 | * \p{InIdeographicDescriptionCharacters} Block 'Ideographic Description Characters' |
151 | * \p{InIpaExtensions} Block 'IPA Extensions' |
152 | * \p{InKanbun} Block 'Kanbun' |
153 | * \p{InKangxiRadicals} Block 'Kangxi Radicals' |
154 | * \p{InKannada} Block 'Kannada' |
155 | * \p{InKatakana} Block 'Katakana' |
156 | * \p{InKhmer} Block 'Khmer' |
157 | * \p{InLao} Block 'Lao' |
158 | * \p{InLatin1Supplement} Block 'Latin-1 Supplement' |
159 | * \p{InLatinExtendedAdditional} Block 'Latin Extended Additional' |
160 | * \p{InLatinExtendedA} Block 'Latin Extended-A' |
161 | * \p{InLatinExtendedB} Block 'Latin Extended-B' |
162 | * \p{InLetterlikeSymbols} Block 'Letterlike Symbols' |
163 | * \p{InLowSurrogates} Block 'Low Surrogates' |
164 | * \p{InMalayalam} Block 'Malayalam' |
165 | * \p{InMathematicalAlphanumericSymbols} Block 'Mathematical Alphanumeric Symbols' |
166 | * \p{InMathematicalOperators} Block 'Mathematical Operators' |
167 | * \p{InMiscellaneousSymbols} Block 'Miscellaneous Symbols' |
168 | * \p{InMiscellaneousTechnical} Block 'Miscellaneous Technical' |
169 | * \p{InMongolian} Block 'Mongolian' |
170 | * \p{InMusicalSymbols} Block 'Musical Symbols' |
171 | * \p{InMyanmar} Block 'Myanmar' |
172 | * \p{InNumberForms} Block 'Number Forms' |
173 | * \p{InOgham} Block 'Ogham' |
174 | * \p{InOldItalic} Block 'Old Italic' |
175 | * \p{InOpticalCharacterRecognition} Block 'Optical Character Recognition' |
176 | * \p{InOriya} Block 'Oriya' |
177 | * \p{InPrivateUse} Block 'Private Use' |
178 | * \p{InRunic} Block 'Runic' |
179 | * \p{InSinhala} Block 'Sinhala' |
180 | * \p{InSmallFormVariants} Block 'Small Form Variants' |
181 | * \p{InSpacingModifierLetters} Block 'Spacing Modifier Letters' |
182 | * \p{InSpecials} Block 'Specials' |
183 | * \p{InSuperscriptsAndSubscripts} Block 'Superscripts and Subscripts' |
184 | * \p{InSyriac} Block 'Syriac' |
185 | * \p{InTags} Block 'Tags' |
186 | * \p{InTamil} Block 'Tamil' |
187 | * \p{InTelugu} Block 'Telugu' |
188 | * \p{InThaana} Block 'Thaana' |
189 | * \p{InThai} Block 'Thai' |
190 | * \p{InTibetan} Block 'Tibetan' |
191 | * \p{InUnifiedCanadianAboriginalSyllabics} Block 'Unified Canadian Aboriginal Syllabics' |
192 | * \p{InYiRadicals} Block 'Yi Radicals' |
193 | * \p{InYiSyllables} Block 'Yi Syllables' |
194 | * \p{Inherited} Script 'INHERITED' |
99598c8c |
195 | * \p{JoinControl} Extended property 'Join_Control' |
196 | * \p{Kannada} Script 'KANNADA' |
197 | * \p{Katakana} Script 'KATAKANA' |
198 | * \p{Khmer} Script 'KHMER' |
199 | \p{L&} [\p{Ll}\p{Lu}\p{Lt}] |
200 | * \p{Lao} Script 'LAO' |
201 | * \p{Latin} Script 'LATIN' |
5beb625e |
202 | * \p{Ll} Alias for \p{Ll} (General Category 'Ll') |
99598c8c |
203 | \p{Ll} General Category 'Ll' |
5beb625e |
204 | * \p{Lm} Alias for \p{Lm} (General Category 'Lm') |
99598c8c |
205 | \p{Lm} General Category 'Lm' |
99598c8c |
206 | * \p{Lowercase} [\p{Ll}\p{OtherLowercase}] |
207 | \p{Lower} [[:Lower:]] |
5beb625e |
208 | * \p{Lo} Alias for \p{Lo} (General Category 'Lo') |
99598c8c |
209 | \p{Lo} General Category 'Lo' |
5beb625e |
210 | * \p{Lt} Alias for \p{Lt} (General Category 'Lt') |
99598c8c |
211 | \p{Lt} General Category 'Lt' |
5beb625e |
212 | * \p{Lu} Alias for \p{Lu} (General Category 'Lu') |
99598c8c |
213 | \p{Lu} General Category 'Lu' |
5beb625e |
214 | * \p{L} Alias for \p{L} (Major Category 'L') |
99598c8c |
215 | \p{L} Major Category 'L' |
216 | * \p{Malayalam} Script 'MALAYALAM' |
99598c8c |
217 | * \p{Math} [\p{Sm}\p{OtherMath}] |
5beb625e |
218 | * \p{Mc} Alias for \p{Mc} (General Category 'Mc') |
99598c8c |
219 | \p{Mc} General Category 'Mc' |
5beb625e |
220 | * \p{Me} Alias for \p{Me} (General Category 'Me') |
99598c8c |
221 | \p{Me} General Category 'Me' |
222 | \p{Mirrored} Mirrored in bidirectional text |
5beb625e |
223 | * \p{Mn} Alias for \p{Mn} (General Category 'Mn') |
99598c8c |
224 | \p{Mn} General Category 'Mn' |
99598c8c |
225 | * \p{Mongolian} Script 'MONGOLIAN' |
226 | * \p{Myanmar} Script 'MYANMAR' |
5beb625e |
227 | * \p{M} Alias for \p{M} (Major Category 'M') |
99598c8c |
228 | \p{M} Major Category 'M' |
5beb625e |
229 | * \p{Nd} Alias for \p{Nd} (General Category 'Nd') |
99598c8c |
230 | \p{Nd} General Category 'Nd' |
5beb625e |
231 | * \p{Nl} Alias for \p{Nl} (General Category 'Nl') |
99598c8c |
232 | \p{Nl} General Category 'Nl' |
99598c8c |
233 | * \p{NoncharacterCodePoint} Extended property 'Noncharacter_Code_Point' |
5beb625e |
234 | * \p{No} Alias for \p{No} (General Category 'No') |
99598c8c |
235 | \p{No} General Category 'No' |
5beb625e |
236 | * \p{N} Alias for \p{N} (Major Category 'N') |
99598c8c |
237 | \p{N} Major Category 'N' |
238 | * \p{Ogham} Script 'OGHAM' |
239 | * \p{OldItalic} Script 'OLD-ITALIC' |
99598c8c |
240 | * \p{Oriya} Script 'ORIYA' |
241 | * \p{OtherAlphabetic} Extended property 'Other_Alphabetic' |
99598c8c |
242 | * \p{OtherLowercase} Extended property 'Other_Lowercase' |
243 | * \p{OtherMath} Extended property 'Other_Math' |
99598c8c |
244 | * \p{OtherUppercase} Extended property 'Other_Uppercase' |
5beb625e |
245 | * \p{Pc} Alias for \p{Pc} (General Category 'Pc') |
99598c8c |
246 | \p{Pc} General Category 'Pc' |
5beb625e |
247 | * \p{Pd} Alias for \p{Pd} (General Category 'Pd') |
99598c8c |
248 | \p{Pd} General Category 'Pd' |
5beb625e |
249 | * \p{Pe} Alias for \p{Pe} (General Category 'Pe') |
99598c8c |
250 | \p{Pe} General Category 'Pe' |
5beb625e |
251 | * \p{Pf} Alias for \p{Pf} (General Category 'Pf') |
99598c8c |
252 | \p{Pf} General Category 'Pf' |
5beb625e |
253 | * \p{Pi} Alias for \p{Pi} (General Category 'Pi') |
99598c8c |
254 | \p{Pi} General Category 'Pi' |
5beb625e |
255 | * \p{Po} Alias for \p{Po} (General Category 'Po') |
99598c8c |
256 | \p{Po} General Category 'Po' |
257 | \p{Print} [[:Print:]] |
5beb625e |
258 | * \p{Ps} Alias for \p{Ps} (General Category 'Ps') |
99598c8c |
259 | \p{Ps} General Category 'Ps' |
99598c8c |
260 | \p{Punct} [[:Punct:]] |
5beb625e |
261 | * \p{P} Alias for \p{P} (Major Category 'P') |
99598c8c |
262 | \p{P} Major Category 'P' |
263 | * \p{QuotationMark} Extended property 'Quotation_Mark' |
264 | * \p{Runic} Script 'RUNIC' |
5beb625e |
265 | * \p{Sc} Alias for \p{Sc} (General Category 'Sc') |
99598c8c |
266 | \p{Sc} General Category 'Sc' |
99598c8c |
267 | * \p{Sinhala} Script 'SINHALA' |
5beb625e |
268 | * \p{Sk} Alias for \p{Sk} (General Category 'Sk') |
99598c8c |
269 | \p{Sk} General Category 'Sk' |
5beb625e |
270 | * \p{Sm} Alias for \p{Sm} (General Category 'Sm') |
99598c8c |
271 | \p{Sm} General Category 'Sm' |
5beb625e |
272 | * \p{So} Alias for \p{So} (General Category 'So') |
99598c8c |
273 | \p{So} General Category 'So' |
274 | \p{SpacePerl} \s |
99598c8c |
275 | \p{Space} [[:Space:]] |
99598c8c |
276 | * \p{Syriac} Script 'SYRIAC' |
5beb625e |
277 | * \p{S} Alias for \p{S} (Major Category 'S') |
99598c8c |
278 | \p{S} Major Category 'S' |
279 | * \p{Tamil} Script 'TAMIL' |
280 | * \p{Telugu} Script 'TELUGU' |
281 | * \p{TerminalPunctuation} Extended property 'Terminal_Punctuation' |
282 | * \p{Thaana} Script 'THAANA' |
283 | * \p{Thai} Script 'THAI' |
284 | * \p{Tibetan} Script 'TIBETAN' |
99598c8c |
285 | \p{Title} [[:Title:]] |
99598c8c |
286 | * \p{Uppercase} [\p{Lu}\p{Other_Uppercase}] |
287 | \p{Upper} [[:Upper:]] |
288 | * \p{WhiteSpace} Extended property 'White_space' |
289 | \p{Word} [[:Word:]] |
290 | \p{XDigit} [[:XDigit:]] |
291 | * \p{Yi} Script 'YI' |
5beb625e |
292 | * \p{Zl} Alias for \p{Zl} (General Category 'Zl') |
99598c8c |
293 | \p{Zl} General Category 'Zl' |
5beb625e |
294 | * \p{Zp} Alias for \p{Zp} (General Category 'Zp') |
99598c8c |
295 | \p{Zp} General Category 'Zp' |
5beb625e |
296 | * \p{Zs} Alias for \p{Zs} (General Category 'Zs') |
99598c8c |
297 | \p{Zs} General Category 'Zs' |
5beb625e |
298 | * \p{Z} Alias for \p{Z} (Major Category 'Z') |
99598c8c |
299 | \p{Z} Major Category 'Z' |
300 | \p{_CanonDCIJ} (for internal casefolding use) |
301 | \p{_CaseIgnorable} (for internal casefolding use) |
302 | \p{_CombAbove} (for internal casefolding use) |