Commit | Line | Data |
d73e5302 |
1 | # !!!!!!! DO NOT EDIT THIS FILE !!!!!!! |
99598c8c |
2 | # This file is built by ./mktables from e.g. Unicode.txt. |
d73e5302 |
3 | # Any changes made here will be lost! |
cf25bb62 |
4 | |
5 | ## |
6 | ## Data in this file used by ../utf8_heavy.pl |
7 | ## |
8 | |
9 | ## Mapping from name to filename in ./Is |
10 | %utf8::Is = ( |
11 | 'ASCII' => 'ASCII', |
12 | 'Alnum' => 'Alnum', |
13 | 'Alpha' => 'Alpha', |
14 | 'BidiAL' => 'BidiAL', |
15 | 'BidiAN' => 'BidiAN', |
16 | 'BidiB' => 'BidiB', |
17 | 'BidiBN' => 'BidiBN', |
18 | 'BidiCS' => 'BidiCS', |
19 | 'BidiEN' => 'BidiEN', |
20 | 'BidiES' => 'BidiES', |
21 | 'BidiET' => 'BidiET', |
22 | 'BidiL' => 'BidiL', |
23 | 'BidiLRE' => 'BidiLRE', |
24 | 'BidiLRO' => 'BidiLRO', |
25 | 'BidiNSM' => 'BidiNSM', |
26 | 'BidiON' => 'BidiON', |
27 | 'BidiPDF' => 'BidiPDF', |
28 | 'BidiR' => 'BidiR', |
29 | 'BidiRLE' => 'BidiRLE', |
30 | 'BidiRLO' => 'BidiRLO', |
31 | 'BidiS' => 'BidiS', |
32 | 'BidiWS' => 'BidiWS', |
33 | 'Blank' => 'Blank', |
34 | 'C' => 'C', |
35 | 'Canon' => 'Canon', |
36 | 'Cc' => 'Cc', |
37 | 'Cf' => 'Cf', |
38 | 'Cn' => 'Cn', |
39 | 'Cntrl' => 'Cntrl', |
40 | 'Co' => 'Co', |
41 | 'Compat' => 'Compat', |
42 | 'Cs' => 'Cs', |
43 | 'DCcircle' => 'DCcircle', |
44 | 'DCcompat' => 'DCcompat', |
45 | 'DCfinal' => 'DCfinal', |
46 | 'DCfont' => 'DCfont', |
47 | 'DCfraction' => 'DCfracti', |
48 | 'DCinitial' => 'DCinitia', |
49 | 'DCisolated' => 'DCisolat', |
50 | 'DCmedial' => 'DCmedial', |
51 | 'DCnarrow' => 'DCnarrow', |
52 | 'DCnoBreak' => 'DCnoBrea', |
53 | 'DCsmall' => 'DCsmall', |
54 | 'DCsquare' => 'DCsquare', |
55 | 'DCsub' => 'DCsub', |
56 | 'DCsuper' => 'DCsuper', |
57 | 'DCvertical' => 'DCvertic', |
58 | 'DCwide' => 'DCwide', |
59 | 'Digit' => 'Digit', |
60 | 'Graph' => 'Graph', |
99598c8c |
61 | 'L' => 'L', |
62 | 'L&' => 'L_', |
cf25bb62 |
63 | 'LbrkAI' => 'LbrkAI', |
64 | 'LbrkAL' => 'LbrkAL', |
65 | 'LbrkB2' => 'LbrkB2', |
66 | 'LbrkBA' => 'LbrkBA', |
67 | 'LbrkBB' => 'LbrkBB', |
68 | 'LbrkBK' => 'LbrkBK', |
69 | 'LbrkCB' => 'LbrkCB', |
70 | 'LbrkCL' => 'LbrkCL', |
71 | 'LbrkCM' => 'LbrkCM', |
72 | 'LbrkCR' => 'LbrkCR', |
73 | 'LbrkEX' => 'LbrkEX', |
74 | 'LbrkGL' => 'LbrkGL', |
75 | 'LbrkHY' => 'LbrkHY', |
76 | 'LbrkID' => 'LbrkID', |
77 | 'LbrkIN' => 'LbrkIN', |
78 | 'LbrkIS' => 'LbrkIS', |
79 | 'LbrkLF' => 'LbrkLF', |
80 | 'LbrkNS' => 'LbrkNS', |
81 | 'LbrkNU' => 'LbrkNU', |
82 | 'LbrkOP' => 'LbrkOP', |
83 | 'LbrkPO' => 'LbrkPO', |
84 | 'LbrkPR' => 'LbrkPR', |
85 | 'LbrkQU' => 'LbrkQU', |
86 | 'LbrkSA' => 'LbrkSA', |
87 | 'LbrkSG' => 'LbrkSG', |
88 | 'LbrkSP' => 'LbrkSP', |
89 | 'LbrkSY' => 'LbrkSY', |
90 | 'LbrkXX' => 'LbrkXX', |
91 | 'LbrkZW' => 'LbrkZW', |
92 | 'Ll' => 'Ll', |
93 | 'Lm' => 'Lm', |
94 | 'Lo' => 'Lo', |
95 | 'Lower' => 'Lower', |
96 | 'Lt' => 'Lt', |
97 | 'Lu' => 'Lu', |
98 | 'M' => 'M', |
99 | 'Mc' => 'Mc', |
100 | 'Me' => 'Me', |
101 | 'Mirrored' => 'Mirrored', |
102 | 'Mn' => 'Mn', |
103 | 'N' => 'N', |
104 | 'Nd' => 'Nd', |
105 | 'Nl' => 'Nl', |
106 | 'No' => 'No', |
107 | 'P' => 'P', |
108 | 'Pc' => 'Pc', |
109 | 'Pd' => 'Pd', |
110 | 'Pe' => 'Pe', |
111 | 'Pf' => 'Pf', |
112 | 'Pi' => 'Pi', |
113 | 'Po' => 'Po', |
114 | 'Print' => 'Print', |
115 | 'Ps' => 'Ps', |
116 | 'Punct' => 'Punct', |
117 | 'S' => 'S', |
118 | 'Sc' => 'Sc', |
119 | 'Sk' => 'Sk', |
120 | 'Sm' => 'Sm', |
121 | 'So' => 'So', |
122 | 'Space' => 'Space', |
123 | 'SpacePerl' => 'SpacePer', |
124 | 'Title' => 'Title', |
125 | 'Upper' => 'Upper', |
126 | 'Word' => 'Word', |
127 | 'XDigit' => 'XDigit', |
128 | 'Z' => 'Z', |
129 | 'Zl' => 'Zl', |
130 | 'Zp' => 'Zp', |
131 | 'Zs' => 'Zs', |
99598c8c |
132 | '_CanonDCIJ' => '_CanonDC', |
133 | '_CaseIgnorable' => '_CaseIgn', |
134 | '_CombAbove' => '_CombAbo', |
d73e5302 |
135 | ); |
cf25bb62 |
136 | |
137 | ## Mappings from regex to filename in ./Is/ |
138 | %utf8::IsPat = ( |
139 | 'al' => { |
cf25bb62 |
140 | 'All' => 'Any', |
99598c8c |
141 | 'Alphabetic' => 'Alphabet', |
cf25bb62 |
142 | }, |
143 | 'an' => { |
144 | 'Any' => 'Any', |
145 | }, |
146 | 'ar' => { |
147 | 'Armenian' => 'Armenian', |
148 | 'Arabic' => 'Arabic', |
149 | }, |
150 | 'as' => { |
99598c8c |
151 | 'Ascii(?:[-_]|\s+)?Hex(?:[-_]|\s+)?Digit' => 'AsciiHex', |
cf25bb62 |
152 | 'Assigned' => 'Assigned', |
153 | }, |
154 | 'be' => { |
155 | 'Bengali' => 'Bengali', |
156 | }, |
157 | 'bi' => { |
158 | 'Bidi(?:[-_]|\s+)?Control' => 'BidiCont', |
159 | }, |
160 | 'bo' => { |
161 | 'Bopomofo' => 'Bopomofo', |
162 | }, |
163 | 'ca' => { |
164 | 'Canadian(?:[-_]|\s+)?Aboriginal' => 'Canadian', |
165 | }, |
166 | 'ch' => { |
167 | 'Cherokee' => 'Cherokee', |
168 | }, |
169 | 'cl' => { |
d73e5302 |
170 | 'Close(?:[-_]|\s+)?Punctuation' => 'Pe', |
cf25bb62 |
171 | }, |
172 | 'co' => { |
d73e5302 |
173 | 'Control' => 'Cc', |
cf25bb62 |
174 | 'Common' => 'Common', |
99598c8c |
175 | 'Connector(?:[-_]|\s+)?Punctuation' => 'Pc', |
cf25bb62 |
176 | }, |
177 | 'cu' => { |
d73e5302 |
178 | 'Currency(?:[-_]|\s+)?Symbol' => 'Sc', |
cf25bb62 |
179 | }, |
180 | 'cy' => { |
181 | 'Cyrillic' => 'Cyrillic', |
182 | }, |
183 | 'da' => { |
d73e5302 |
184 | 'Dash(?:[-_]|\s+)?Punctuation' => 'Pd', |
cf25bb62 |
185 | 'Dash' => 'Dash', |
186 | }, |
187 | 'de' => { |
99598c8c |
188 | 'Deseret' => 'Deseret', |
cf25bb62 |
189 | 'Devanagari' => 'Devanaga', |
e150c829 |
190 | 'Decimal(?:[-_]|\s+)?Number' => 'Nd', |
cf25bb62 |
191 | }, |
192 | 'di' => { |
193 | 'Diacritic' => 'Diacriti', |
194 | }, |
195 | 'en' => { |
d73e5302 |
196 | 'Enclosing(?:[-_]|\s+)?Mark' => 'Me', |
cf25bb62 |
197 | }, |
198 | 'et' => { |
199 | 'Ethiopic' => 'Ethiopic', |
200 | }, |
201 | 'ex' => { |
202 | 'Extender' => 'Extender', |
203 | }, |
204 | 'fi' => { |
d73e5302 |
205 | 'Final(?:[-_]|\s+)?Punctuation' => 'Pf', |
cf25bb62 |
206 | }, |
207 | 'fo' => { |
d73e5302 |
208 | 'Format' => 'Cf', |
cf25bb62 |
209 | }, |
210 | 'ge' => { |
211 | 'Georgian' => 'Georgian', |
212 | }, |
213 | 'go' => { |
214 | 'Gothic' => 'Gothic', |
215 | }, |
216 | 'gr' => { |
217 | 'Greek' => 'Greek', |
218 | }, |
219 | 'gu' => { |
cf25bb62 |
220 | 'Gujarati' => 'Gujarati', |
99598c8c |
221 | 'Gurmukhi' => 'Gurmukhi', |
cf25bb62 |
222 | }, |
223 | 'ha' => { |
cf25bb62 |
224 | 'Hangul' => 'Hangul', |
99598c8c |
225 | 'Han' => 'Han', |
cf25bb62 |
226 | }, |
227 | 'he' => { |
228 | 'Hebrew' => 'Hebrew', |
229 | 'Hex(?:[-_]|\s+)?Digit' => 'HexDigit', |
230 | }, |
231 | 'hi' => { |
232 | 'Hiragana' => 'Hiragana', |
233 | }, |
234 | 'hy' => { |
235 | 'Hyphen' => 'Hyphen', |
236 | }, |
237 | 'id' => { |
238 | 'Ideographic' => 'Ideograp', |
99598c8c |
239 | 'Id(?:[-_]|\s+)?Continue' => 'IdContin', |
240 | 'Id(?:[-_]|\s+)?Start' => 'IdStart', |
cf25bb62 |
241 | }, |
242 | 'in' => { |
243 | 'Inherited' => 'Inherite', |
d73e5302 |
244 | 'Initial(?:[-_]|\s+)?Punctuation' => 'Pi', |
cf25bb62 |
245 | }, |
246 | 'jo' => { |
247 | 'Join(?:[-_]|\s+)?Control' => 'JoinCont', |
248 | }, |
249 | 'ka' => { |
cf25bb62 |
250 | 'Kannada' => 'Kannada', |
99598c8c |
251 | 'Katakana' => 'Katakana', |
cf25bb62 |
252 | }, |
253 | 'kh' => { |
254 | 'Khmer' => 'Khmer', |
255 | }, |
256 | 'la' => { |
cf25bb62 |
257 | 'Lao' => 'Lao', |
99598c8c |
258 | 'Latin' => 'Latin', |
cf25bb62 |
259 | }, |
260 | 'le' => { |
d73e5302 |
261 | 'Letter(?:[-_]|\s+)?Number' => 'Nl', |
99598c8c |
262 | 'Letter' => 'L', |
cf25bb62 |
263 | }, |
264 | 'li' => { |
d73e5302 |
265 | 'Line(?:[-_]|\s+)?Separator' => 'Zl', |
cf25bb62 |
266 | }, |
267 | 'lo' => { |
cf25bb62 |
268 | 'Lowercase' => 'Lowercas', |
99598c8c |
269 | 'Lowercase(?:[-_]|\s+)?Letter' => 'Ll', |
cf25bb62 |
270 | }, |
271 | 'ma' => { |
cf25bb62 |
272 | 'Math' => 'Math', |
273 | 'Malayalam' => 'Malayala', |
99598c8c |
274 | 'Mark' => 'M', |
275 | 'Math(?:[-_]|\s+)?Symbol' => 'Sm', |
cf25bb62 |
276 | }, |
277 | 'mo' => { |
d73e5302 |
278 | 'Modifier(?:[-_]|\s+)?Symbol' => 'Sk', |
cf25bb62 |
279 | 'Mongolian' => 'Mongolia', |
99598c8c |
280 | 'Modifier(?:[-_]|\s+)?Letter' => 'Lm', |
cf25bb62 |
281 | }, |
282 | 'my' => { |
283 | 'Myanmar' => 'Myanmar', |
284 | }, |
285 | 'no' => { |
cf25bb62 |
286 | 'Noncharacter(?:[-_]|\s+)?Code(?:[-_]|\s+)?Point' => 'Nonchara', |
99598c8c |
287 | 'Non(?:[-_]|\s+)?Spacing(?:[-_]|\s+)?Mark' => 'Mn', |
cf25bb62 |
288 | }, |
289 | 'nu' => { |
d73e5302 |
290 | 'Number' => 'N', |
cf25bb62 |
291 | }, |
292 | 'og' => { |
293 | 'Ogham' => 'Ogham', |
294 | }, |
295 | 'ol' => { |
296 | 'Old(?:[-_]|\s+)?Italic' => 'OldItali', |
297 | }, |
298 | 'op' => { |
d73e5302 |
299 | 'Open(?:[-_]|\s+)?Punctuation' => 'Ps', |
cf25bb62 |
300 | }, |
301 | 'or' => { |
302 | 'Oriya' => 'Oriya', |
303 | }, |
304 | 'ot' => { |
d73e5302 |
305 | 'Other(?:[-_]|\s+)?Punctuation' => 'Po', |
cf25bb62 |
306 | 'Other(?:[-_]|\s+)?Uppercase' => 'OtherUpp', |
cf25bb62 |
307 | 'Other(?:[-_]|\s+)?Alphabetic' => 'OtherAlp', |
d73e5302 |
308 | 'Other(?:[-_]|\s+)?Symbol' => 'So', |
cf25bb62 |
309 | 'Other(?:[-_]|\s+)?Number' => 'No', |
310 | 'Other' => 'C', |
99598c8c |
311 | 'Other(?:[-_]|\s+)?Math' => 'OtherMat', |
312 | 'Other(?:[-_]|\s+)?Letter' => 'Lo', |
313 | 'Other(?:[-_]|\s+)?Lowercase' => 'OtherLow', |
cf25bb62 |
314 | }, |
315 | 'pa' => { |
d73e5302 |
316 | 'Paragraph(?:[-_]|\s+)?Separator' => 'Zp', |
cf25bb62 |
317 | }, |
318 | 'pr' => { |
d73e5302 |
319 | 'Private(?:[-_]|\s+)?Use' => 'Co', |
cf25bb62 |
320 | }, |
321 | 'pu' => { |
d73e5302 |
322 | 'Punctuation' => 'P', |
cf25bb62 |
323 | }, |
324 | 'qu' => { |
325 | 'Quotation(?:[-_]|\s+)?Mark' => 'Quotatio', |
326 | }, |
327 | 'ru' => { |
328 | 'Runic' => 'Runic', |
329 | }, |
330 | 'se' => { |
d73e5302 |
331 | 'Separator' => 'Z', |
cf25bb62 |
332 | }, |
333 | 'si' => { |
334 | 'Sinhala' => 'Sinhala', |
335 | }, |
336 | 'sp' => { |
d73e5302 |
337 | 'Space(?:[-_]|\s+)?Separator' => 'Zs', |
e150c829 |
338 | 'Spacing(?:[-_]|\s+)?Mark' => 'Mc', |
cf25bb62 |
339 | }, |
340 | 'su' => { |
d73e5302 |
341 | 'Surrogate' => 'Cs', |
cf25bb62 |
342 | }, |
343 | 'sy' => { |
344 | 'Syriac' => 'Syriac', |
d73e5302 |
345 | 'Symbol' => 'S', |
cf25bb62 |
346 | }, |
347 | 'ta' => { |
348 | 'Tamil' => 'Tamil', |
349 | }, |
350 | 'te' => { |
cf25bb62 |
351 | 'Telugu' => 'Telugu', |
99598c8c |
352 | 'Terminal(?:[-_]|\s+)?Punctuation' => 'Terminal', |
cf25bb62 |
353 | }, |
354 | 'th' => { |
355 | 'Thaana' => 'Thaana', |
356 | 'Thai' => 'Thai', |
357 | }, |
358 | 'ti' => { |
cf25bb62 |
359 | 'Tibetan' => 'Tibetan', |
99598c8c |
360 | 'Titlecase(?:[-_]|\s+)?Letter' => 'Lt', |
cf25bb62 |
361 | }, |
362 | 'un' => { |
e150c829 |
363 | 'Unassigned' => 'Cn', |
cf25bb62 |
364 | }, |
365 | 'up' => { |
cf25bb62 |
366 | 'Uppercase' => 'Uppercas', |
99598c8c |
367 | 'Uppercase(?:[-_]|\s+)?Letter' => 'Lu', |
cf25bb62 |
368 | }, |
369 | 'wh' => { |
99598c8c |
370 | 'White(?:[-_]|\s+)?Space' => 'WhiteSpa', |
cf25bb62 |
371 | }, |
372 | 'yi' => { |
373 | 'Yi' => 'Yi', |
374 | }, |
d73e5302 |
375 | ); |