Commit | Line | Data |
5beb625e |
1 | # !!!!!!! DO NOT EDIT THIS FILE !!!!!!! |
eb1102fc |
2 | # This file is built by ./mktables from e.g. UnicodeData.txt. |
5beb625e |
3 | # Any changes made here will be lost! |
4 | |
5 | ## |
6 | ## Data in this file used by ../utf8_heavy.pl |
7 | ## |
8 | |
9 | ## Mapping from lc(canonical name) to filename in ./lib |
10 | %utf8::Canonical = ( |
11 | alphabetic => 'Alphabet', |
12 | |
13 | # InAlphabeticPresentationForms |
14 | alphabeticpresentationforms => 'InAlphab', |
15 | |
16 | arabic => 'Arabic', |
17 | |
18 | # InArabicPresentationFormsA |
19 | arabicpresentationformsa => 'InArabi3', |
20 | |
21 | # InArabicPresentationFormsB |
22 | arabicpresentationformsb => 'InArabi2', |
23 | |
24 | armenian => 'Armenian', |
25 | |
26 | # InArrows |
27 | arrows => 'InArrows', |
28 | |
29 | # AsciiHexDigit |
30 | asciihexdigit => 'AsciiHex', |
31 | |
32 | # InBasicLatin |
33 | basiclatin => 'InBasicL', |
34 | |
35 | bengali => 'Bengali', |
36 | |
37 | # BidiControl |
38 | bidicontrol => 'BidiCont', |
39 | |
40 | # InBlockElements |
41 | blockelements => 'InBlockE', |
42 | |
43 | bopomofo => 'Bopomofo', |
44 | |
45 | # InBopomofoExtended |
46 | bopomofoextended => 'InBopom2', |
47 | |
48 | # InBoxDrawing |
49 | boxdrawing => 'InBoxDra', |
50 | |
51 | # InBraillePatterns |
52 | braillepatterns => 'InBraill', |
53 | |
54 | # InByzantineMusicalSymbols |
55 | byzantinemusicalsymbols => 'InByzant', |
56 | |
57 | # CanadianAboriginal |
58 | canadianaboriginal => 'Canadian', |
59 | |
60 | cherokee => 'Cherokee', |
61 | |
62 | # InCjkCompatibility |
63 | cjkcompatibility => 'InCjkCom', |
64 | |
65 | # InCjkCompatibilityForms |
66 | cjkcompatibilityforms => 'InCjkCo2', |
67 | |
68 | # InCjkCompatibilityIdeographs |
69 | cjkcompatibilityideographs => 'InCjkCo3', |
70 | |
71 | # InCjkCompatibilityIdeographsSupplement |
72 | cjkcompatibilityideographssupplement => 'InCjkCo4', |
73 | |
74 | # InCjkRadicalsSupplement |
75 | cjkradicalssupplement => 'InCjkRad', |
76 | |
77 | # InCjkSymbolsAndPunctuation |
78 | cjksymbolsandpunctuation => 'InCjkSym', |
79 | |
80 | # InCjkUnifiedIdeographs |
81 | cjkunifiedideographs => 'InCjkUni', |
82 | |
83 | # InCjkUnifiedIdeographsExtensionA |
84 | cjkunifiedideographsextensiona => 'InCjkUn3', |
85 | |
86 | # InCjkUnifiedIdeographsExtensionB |
87 | cjkunifiedideographsextensionb => 'InCjkUn2', |
88 | |
89 | # ClosePunctuation |
90 | closepunctuation => 'Pe', |
91 | |
92 | # InCombiningDiacriticalMarks |
93 | combiningdiacriticalmarks => 'InCombi3', |
94 | |
95 | # InCombiningHalfMarks |
96 | combininghalfmarks => 'InCombin', |
97 | |
98 | # InCombiningMarksForSymbols |
99 | combiningmarksforsymbols => 'InCombi2', |
100 | |
101 | common => 'Common', |
102 | |
103 | # ConnectorPunctuation |
104 | connectorpunctuation => 'Pc', |
105 | |
106 | control => 'Cc', |
107 | |
108 | # InControlPictures |
109 | controlpictures => 'InContro', |
110 | |
111 | # CurrencySymbol |
112 | currencysymbol => 'Sc', |
113 | |
114 | # InCurrencySymbols |
115 | currencysymbols => 'InCurren', |
116 | |
117 | cyrillic => 'Cyrillic', |
118 | dash => 'Dash', |
119 | |
120 | # DashPunctuation |
121 | dashpunctuation => 'Pd', |
122 | |
123 | # DecimalNumber |
124 | decimalnumber => 'Nd', |
125 | |
126 | deseret => 'Deseret', |
127 | devanagari => 'Devanaga', |
128 | diacritic => 'Diacriti', |
129 | |
130 | # InDingbats |
131 | dingbats => 'InDingba', |
132 | |
133 | # InEnclosedAlphanumerics |
134 | enclosedalphanumerics => 'InEnclos', |
135 | |
136 | # InEnclosedCjkLettersAndMonths |
137 | enclosedcjklettersandmonths => 'InEnclo2', |
138 | |
139 | # EnclosingMark |
140 | enclosingmark => 'Me', |
141 | |
142 | ethiopic => 'Ethiopic', |
143 | extender => 'Extender', |
144 | |
145 | # FinalPunctuation |
146 | finalpunctuation => 'Pf', |
147 | |
148 | format => 'Cf', |
149 | |
150 | # InGeneralPunctuation |
151 | generalpunctuation => 'InGenera', |
152 | |
153 | # InGeometricShapes |
154 | geometricshapes => 'InGeomet', |
155 | |
156 | georgian => 'Georgian', |
157 | gothic => 'Gothic', |
158 | greek => 'Greek', |
159 | |
160 | # InGreekExtended |
161 | greekextended => 'InGreekE', |
162 | |
163 | gujarati => 'Gujarati', |
164 | gurmukhi => 'Gurmukhi', |
165 | |
166 | # InHalfwidthAndFullwidthForms |
167 | halfwidthandfullwidthforms => 'InHalfwi', |
168 | |
169 | han => 'Han', |
170 | hangul => 'Hangul', |
171 | |
172 | # InHangulCompatibilityJamo |
173 | hangulcompatibilityjamo => 'InHangu3', |
174 | |
175 | # InHangulJamo |
176 | hanguljamo => 'InHangul', |
177 | |
178 | # InHangulSyllables |
179 | hangulsyllables => 'InHangu2', |
180 | |
181 | hebrew => 'Hebrew', |
182 | |
183 | # HexDigit |
184 | hexdigit => 'HexDigit', |
185 | |
186 | # InHighPrivateUseSurrogates |
187 | highprivateusesurrogates => 'InHighPr', |
188 | |
189 | # InHighSurrogates |
190 | highsurrogates => 'InHighSu', |
191 | |
192 | hiragana => 'Hiragana', |
193 | hyphen => 'Hyphen', |
194 | |
195 | # IdContinue |
196 | idcontinue => 'IdContin', |
197 | |
198 | ideographic => 'Ideograp', |
199 | |
200 | # InIdeographicDescriptionCharacters |
201 | ideographicdescriptioncharacters => 'InIdeogr', |
202 | |
203 | # IdStart |
204 | idstart => 'IdStart', |
205 | |
206 | # InAlphabeticPresentationForms |
207 | inalphabeticpresentationforms => 'InAlphab', |
208 | |
209 | # InArabic |
210 | inarabic => 'InArabic', |
211 | |
212 | # InArabicPresentationFormsA |
213 | inarabicpresentationformsa => 'InArabi3', |
214 | |
215 | # InArabicPresentationFormsB |
216 | inarabicpresentationformsb => 'InArabi2', |
217 | |
218 | # InArmenian |
219 | inarmenian => 'InArmeni', |
220 | |
221 | # InArrows |
222 | inarrows => 'InArrows', |
223 | |
224 | # InBasicLatin |
225 | inbasiclatin => 'InBasicL', |
226 | |
227 | # InBengali |
228 | inbengali => 'InBengal', |
229 | |
230 | # InBlockElements |
231 | inblockelements => 'InBlockE', |
232 | |
233 | # InBopomofo |
234 | inbopomofo => 'InBopomo', |
235 | |
236 | # InBopomofoExtended |
237 | inbopomofoextended => 'InBopom2', |
238 | |
239 | # InBoxDrawing |
240 | inboxdrawing => 'InBoxDra', |
241 | |
242 | # InBraillePatterns |
243 | inbraillepatterns => 'InBraill', |
244 | |
245 | # InByzantineMusicalSymbols |
246 | inbyzantinemusicalsymbols => 'InByzant', |
247 | |
248 | # InCherokee |
249 | incherokee => 'InCherok', |
250 | |
251 | # InCjkCompatibility |
252 | incjkcompatibility => 'InCjkCom', |
253 | |
254 | # InCjkCompatibilityForms |
255 | incjkcompatibilityforms => 'InCjkCo2', |
256 | |
257 | # InCjkCompatibilityIdeographs |
258 | incjkcompatibilityideographs => 'InCjkCo3', |
259 | |
260 | # InCjkCompatibilityIdeographsSupplement |
261 | incjkcompatibilityideographssupplement => 'InCjkCo4', |
262 | |
263 | # InCjkRadicalsSupplement |
264 | incjkradicalssupplement => 'InCjkRad', |
265 | |
266 | # InCjkSymbolsAndPunctuation |
267 | incjksymbolsandpunctuation => 'InCjkSym', |
268 | |
269 | # InCjkUnifiedIdeographs |
270 | incjkunifiedideographs => 'InCjkUni', |
271 | |
272 | # InCjkUnifiedIdeographsExtensionA |
273 | incjkunifiedideographsextensiona => 'InCjkUn3', |
274 | |
275 | # InCjkUnifiedIdeographsExtensionB |
276 | incjkunifiedideographsextensionb => 'InCjkUn2', |
277 | |
278 | # InCombiningDiacriticalMarks |
279 | incombiningdiacriticalmarks => 'InCombi3', |
280 | |
281 | # InCombiningHalfMarks |
282 | incombininghalfmarks => 'InCombin', |
283 | |
284 | # InCombiningMarksForSymbols |
285 | incombiningmarksforsymbols => 'InCombi2', |
286 | |
287 | # InControlPictures |
288 | incontrolpictures => 'InContro', |
289 | |
290 | # InCurrencySymbols |
291 | incurrencysymbols => 'InCurren', |
292 | |
293 | # InCyrillic |
294 | incyrillic => 'InCyrill', |
295 | |
296 | # InDeseret |
297 | indeseret => 'InDesere', |
298 | |
299 | # InDevanagari |
300 | indevanagari => 'InDevana', |
301 | |
302 | # InDingbats |
303 | indingbats => 'InDingba', |
304 | |
305 | # InEnclosedAlphanumerics |
306 | inenclosedalphanumerics => 'InEnclos', |
307 | |
308 | # InEnclosedCjkLettersAndMonths |
309 | inenclosedcjklettersandmonths => 'InEnclo2', |
310 | |
311 | # InEthiopic |
312 | inethiopic => 'InEthiop', |
313 | |
314 | # InGeneralPunctuation |
315 | ingeneralpunctuation => 'InGenera', |
316 | |
317 | # InGeometricShapes |
318 | ingeometricshapes => 'InGeomet', |
319 | |
320 | # InGeorgian |
321 | ingeorgian => 'InGeorgi', |
322 | |
323 | # InGothic |
324 | ingothic => 'InGothic', |
325 | |
326 | # InGreek |
327 | ingreek => 'InGreek', |
328 | |
329 | # InGreekExtended |
330 | ingreekextended => 'InGreekE', |
331 | |
332 | # InGujarati |
333 | ingujarati => 'InGujara', |
334 | |
335 | # InGurmukhi |
336 | ingurmukhi => 'InGurmuk', |
337 | |
338 | # InHalfwidthAndFullwidthForms |
339 | inhalfwidthandfullwidthforms => 'InHalfwi', |
340 | |
341 | # InHangulCompatibilityJamo |
342 | inhangulcompatibilityjamo => 'InHangu3', |
343 | |
344 | # InHangulJamo |
345 | inhanguljamo => 'InHangul', |
346 | |
347 | # InHangulSyllables |
348 | inhangulsyllables => 'InHangu2', |
349 | |
350 | # InHebrew |
351 | inhebrew => 'InHebrew', |
352 | |
353 | inherited => 'Inherite', |
354 | |
355 | # InHighPrivateUseSurrogates |
356 | inhighprivateusesurrogates => 'InHighPr', |
357 | |
358 | # InHighSurrogates |
359 | inhighsurrogates => 'InHighSu', |
360 | |
361 | # InHiragana |
362 | inhiragana => 'InHiraga', |
363 | |
364 | # InIdeographicDescriptionCharacters |
365 | inideographicdescriptioncharacters => 'InIdeogr', |
366 | |
367 | # InIpaExtensions |
368 | inipaextensions => 'InIpaExt', |
369 | |
370 | # InitialPunctuation |
371 | initialpunctuation => 'Pi', |
372 | |
373 | # InKanbun |
374 | inkanbun => 'InKanbun', |
375 | |
376 | # InKangxiRadicals |
377 | inkangxiradicals => 'InKangxi', |
378 | |
379 | # InKannada |
380 | inkannada => 'InKannad', |
381 | |
382 | # InKatakana |
383 | inkatakana => 'InKataka', |
384 | |
385 | # InKhmer |
386 | inkhmer => 'InKhmer', |
387 | |
388 | # InLao |
389 | inlao => 'InLao', |
390 | |
391 | # InLatin1Supplement |
392 | inlatin1supplement => 'InLatin1', |
393 | |
394 | # InLatinExtendedA |
395 | inlatinextendeda => 'InLatin2', |
396 | |
397 | # InLatinExtendedAdditional |
398 | inlatinextendedadditional => 'InLatin3', |
399 | |
400 | # InLatinExtendedB |
401 | inlatinextendedb => 'InLatinE', |
402 | |
403 | # InLetterlikeSymbols |
404 | inletterlikesymbols => 'InLetter', |
405 | |
406 | # InLowSurrogates |
407 | inlowsurrogates => 'InLowSur', |
408 | |
409 | # InMalayalam |
410 | inmalayalam => 'InMalaya', |
411 | |
412 | # InMathematicalAlphanumericSymbols |
413 | inmathematicalalphanumericsymbols => 'InMathe2', |
414 | |
415 | # InMathematicalOperators |
416 | inmathematicaloperators => 'InMathem', |
417 | |
418 | # InMiscellaneousSymbols |
419 | inmiscellaneoussymbols => 'InMiscel', |
420 | |
421 | # InMiscellaneousTechnical |
422 | inmiscellaneoustechnical => 'InMisce2', |
423 | |
424 | # InMongolian |
425 | inmongolian => 'InMongol', |
426 | |
427 | # InMusicalSymbols |
428 | inmusicalsymbols => 'InMusica', |
429 | |
430 | # InMyanmar |
431 | inmyanmar => 'InMyanma', |
432 | |
433 | # InNumberForms |
434 | innumberforms => 'InNumber', |
435 | |
436 | # InOgham |
437 | inogham => 'InOgham', |
438 | |
439 | # InOldItalic |
440 | inolditalic => 'InOldIta', |
441 | |
442 | # InOpticalCharacterRecognition |
443 | inopticalcharacterrecognition => 'InOptica', |
444 | |
445 | # InOriya |
446 | inoriya => 'InOriya', |
447 | |
448 | # InPrivateUse |
449 | inprivateuse => 'InPrivat', |
450 | |
451 | # InRunic |
452 | inrunic => 'InRunic', |
453 | |
454 | # InSinhala |
455 | insinhala => 'InSinhal', |
456 | |
457 | # InSmallFormVariants |
458 | insmallformvariants => 'InSmallF', |
459 | |
460 | # InSpacingModifierLetters |
461 | inspacingmodifierletters => 'InSpacin', |
462 | |
463 | # InSpecials |
464 | inspecials => 'InSpecia', |
465 | |
466 | # InSuperscriptsAndSubscripts |
467 | insuperscriptsandsubscripts => 'InSupers', |
468 | |
469 | # InSyriac |
470 | insyriac => 'InSyriac', |
471 | |
472 | # InTags |
473 | intags => 'InTags', |
474 | |
475 | # InTamil |
476 | intamil => 'InTamil', |
477 | |
478 | # InTelugu |
479 | intelugu => 'InTelugu', |
480 | |
481 | # InThaana |
482 | inthaana => 'InThaana', |
483 | |
484 | # InThai |
485 | inthai => 'InThai', |
486 | |
487 | # InTibetan |
488 | intibetan => 'InTibeta', |
489 | |
490 | # InUnifiedCanadianAboriginalSyllabics |
491 | inunifiedcanadianaboriginalsyllabics => 'InUnifie', |
492 | |
493 | # InYiRadicals |
494 | inyiradicals => 'InYiRadi', |
495 | |
496 | # InYiSyllables |
497 | inyisyllables => 'InYiSyll', |
498 | |
499 | # InIpaExtensions |
500 | ipaextensions => 'InIpaExt', |
501 | |
502 | # JoinControl |
503 | joincontrol => 'JoinCont', |
504 | |
505 | # InKanbun |
506 | kanbun => 'InKanbun', |
507 | |
508 | # InKangxiRadicals |
509 | kangxiradicals => 'InKangxi', |
510 | |
511 | kannada => 'Kannada', |
512 | katakana => 'Katakana', |
513 | khmer => 'Khmer', |
514 | lao => 'Lao', |
515 | latin => 'Latin', |
516 | |
517 | # InLatin1Supplement |
518 | latin1supplement => 'InLatin1', |
519 | |
520 | # InLatinExtendedA |
521 | latinextendeda => 'InLatin2', |
522 | |
523 | # InLatinExtendedAdditional |
524 | latinextendedadditional => 'InLatin3', |
525 | |
526 | # InLatinExtendedB |
527 | latinextendedb => 'InLatinE', |
528 | |
529 | letter => 'L', |
530 | |
531 | # InLetterlikeSymbols |
532 | letterlikesymbols => 'InLetter', |
533 | |
534 | # LetterNumber |
535 | letternumber => 'Nl', |
536 | |
537 | # LineSeparator |
538 | lineseparator => 'Zl', |
539 | |
540 | lowercase => 'Lowercas', |
541 | |
542 | # LowercaseLetter |
543 | lowercaseletter => 'Ll', |
544 | |
545 | # InLowSurrogates |
546 | lowsurrogates => 'InLowSur', |
547 | |
548 | malayalam => 'Malayala', |
549 | mark => 'M', |
550 | math => 'Math', |
551 | |
552 | # InMathematicalAlphanumericSymbols |
553 | mathematicalalphanumericsymbols => 'InMathe2', |
554 | |
555 | # InMathematicalOperators |
556 | mathematicaloperators => 'InMathem', |
557 | |
558 | # MathSymbol |
559 | mathsymbol => 'Sm', |
560 | |
561 | # InMiscellaneousSymbols |
562 | miscellaneoussymbols => 'InMiscel', |
563 | |
564 | # InMiscellaneousTechnical |
565 | miscellaneoustechnical => 'InMisce2', |
566 | |
567 | # ModifierLetter |
568 | modifierletter => 'Lm', |
569 | |
570 | # ModifierSymbol |
571 | modifiersymbol => 'Sk', |
572 | |
573 | mongolian => 'Mongolia', |
574 | |
575 | # InMusicalSymbols |
576 | musicalsymbols => 'InMusica', |
577 | |
578 | myanmar => 'Myanmar', |
579 | |
580 | # NoncharacterCodePoint |
581 | noncharactercodepoint => 'Nonchara', |
582 | |
583 | # NonSpacingMark |
584 | nonspacingmark => 'Mn', |
585 | |
586 | number => 'N', |
587 | |
588 | # InNumberForms |
589 | numberforms => 'InNumber', |
590 | |
591 | ogham => 'Ogham', |
592 | |
593 | # OldItalic |
594 | olditalic => 'OldItali', |
595 | |
596 | # OpenPunctuation |
597 | openpunctuation => 'Ps', |
598 | |
599 | # InOpticalCharacterRecognition |
600 | opticalcharacterrecognition => 'InOptica', |
601 | |
602 | oriya => 'Oriya', |
603 | other => 'C', |
604 | |
605 | # OtherAlphabetic |
606 | otheralphabetic => 'OtherAlp', |
607 | |
608 | # OtherLetter |
609 | otherletter => 'Lo', |
610 | |
611 | # OtherLowercase |
612 | otherlowercase => 'OtherLow', |
613 | |
614 | # OtherMath |
615 | othermath => 'OtherMat', |
616 | |
617 | # OtherNumber |
618 | othernumber => 'No', |
619 | |
620 | # OtherPunctuation |
621 | otherpunctuation => 'Po', |
622 | |
623 | # OtherSymbol |
624 | othersymbol => 'So', |
625 | |
626 | # OtherUppercase |
627 | otheruppercase => 'OtherUpp', |
628 | |
629 | # ParagraphSeparator |
630 | paragraphseparator => 'Zp', |
631 | |
632 | # PrivateUse |
633 | privateuse => 'Co', |
634 | |
635 | punctuation => 'P', |
636 | |
637 | # QuotationMark |
638 | quotationmark => 'Quotatio', |
639 | |
640 | runic => 'Runic', |
641 | separator => 'Z', |
642 | sinhala => 'Sinhala', |
643 | |
644 | # InSmallFormVariants |
645 | smallformvariants => 'InSmallF', |
646 | |
647 | # SpaceSeparator |
648 | spaceseparator => 'Zs', |
649 | |
650 | # SpacingMark |
651 | spacingmark => 'Mc', |
652 | |
653 | # InSpacingModifierLetters |
654 | spacingmodifierletters => 'InSpacin', |
655 | |
656 | # InSpecials |
657 | specials => 'InSpecia', |
658 | |
659 | # InSuperscriptsAndSubscripts |
660 | superscriptsandsubscripts => 'InSupers', |
661 | |
662 | surrogate => 'Cs', |
663 | symbol => 'S', |
664 | syriac => 'Syriac', |
665 | |
666 | # InTags |
667 | tags => 'InTags', |
668 | |
669 | tamil => 'Tamil', |
670 | telugu => 'Telugu', |
671 | |
672 | # TerminalPunctuation |
673 | terminalpunctuation => 'Terminal', |
674 | |
675 | thaana => 'Thaana', |
676 | thai => 'Thai', |
677 | tibetan => 'Tibetan', |
678 | |
679 | # TitlecaseLetter |
680 | titlecaseletter => 'Lt', |
681 | |
682 | unassigned => 'Cn', |
683 | |
684 | # InUnifiedCanadianAboriginalSyllabics |
685 | unifiedcanadianaboriginalsyllabics => 'InUnifie', |
686 | |
687 | uppercase => 'Uppercas', |
688 | |
689 | # UppercaseLetter |
690 | uppercaseletter => 'Lu', |
691 | |
692 | # WhiteSpace |
693 | whitespace => 'WhiteSpa', |
694 | |
695 | yi => 'Yi', |
696 | |
697 | # InYiRadicals |
698 | yiradicals => 'InYiRadi', |
699 | |
700 | # InYiSyllables |
701 | yisyllables => 'InYiSyll', |
702 | |
703 | ); |
704 | 1 |