v / vlib / encoding / utf8 / utf8_tables.v
1394 lines · 1367 sloc · 29.79 KB · acf6b344f733169ec6ecc9881f8a8c2c795b9883
Raw
1// Copyright (c) 2019-2024 Alexander Medvednikov. All rights reserved.
2// Use of this source code is governed by an MIT license that can be found in the LICENSE file.
3module utf8
4
5// for unicode type fast lookup
6const p_c = 1 // a control character.
7
8const p_p = 2 // a punctuation character.
9
10const p_n = 4 // a numeral.
11
12const p_s = 8 // a symbolic character.
13
14const p_z = 16 // a spacing character.
15
16const p_lu = 32 // an up_prer-case letter.
17
18const p_ll = 64 // a lower-case letter.
19
20const p_pr = 128 // a printable character according to Go's definition.
21
22const p_g = p_pr | p_z // a graphical character according to the Unicode definition.
23
24const p_lo = p_lu | p_ll // a letter that is neither up_prer nor lower case.
25
26const p_l_mask = p_lo
27
28const props = [
29 p_c
30 //'\x00'
31 p_c
32 //'\x01'
33 p_c
34 //'\x02'
35 p_c
36 //'\x03'
37 p_c
38 //'\x04'
39 p_c
40 //'\x05'
41 p_c
42 //'\x06'
43 p_c
44 //'\a'
45 p_c
46 //'\b'
47 p_c
48 //'\t'
49 p_c
50 //'\n'
51 p_c
52 //'\v'
53 p_c
54 //'\f'
55 p_c
56 //'\r'
57 p_c
58 //'\x0e'
59 p_c
60 //'\x0f'
61 p_c
62 //'\x10'
63 p_c
64 //'\x11'
65 p_c
66 //'\x12'
67 p_c
68 //'\x13'
69 p_c
70 //'\x14'
71 p_c
72 //'\x15'
73 p_c
74 //'\x16'
75 p_c
76 //'\x17'
77 p_c
78 //'\x18'
79 p_c
80 //'\x19'
81 p_c
82 //'\x1a'
83 p_c
84 //'\x1b'
85 p_c
86 //'\x1c'
87 p_c
88 //'\x1d'
89 p_c
90 //'\x1e'
91 p_c
92 //'\x1f'
93 p_z | p_pr
94 //' '
95 p_p | p_pr
96 //'!'
97 p_p | p_pr
98 //'"'
99 p_p | p_pr
100 //'#'
101 p_s | p_pr
102 //'$'
103 p_p | p_pr
104 //'%'
105 p_p | p_pr
106 //'&'
107 p_p | p_pr
108 //'\''
109 p_p | p_pr
110 //'('
111 p_p | p_pr
112 //')'
113 p_p | p_pr
114 //'*'
115 p_s | p_pr
116 //'+'
117 p_p | p_pr
118 //','
119 p_p | p_pr
120 //'-'
121 p_p | p_pr
122 //'.'
123 p_p | p_pr
124 //'/'
125 p_n | p_pr
126 //'0'
127 p_n | p_pr
128 //'1'
129 p_n | p_pr
130 //'2'
131 p_n | p_pr
132 //'3'
133 p_n | p_pr
134 //'4'
135 p_n | p_pr
136 //'5'
137 p_n | p_pr
138 //'6'
139 p_n | p_pr
140 //'7'
141 p_n | p_pr
142 //'8'
143 p_n | p_pr
144 //'9'
145 p_p | p_pr
146 //':'
147 p_p | p_pr
148 //';'
149 p_s | p_pr
150 //'<'
151 p_s | p_pr
152 //'='
153 p_s | p_pr
154 //'>'
155 p_p | p_pr
156 //'?'
157 p_p | p_pr
158 //'@'
159 p_lu | p_pr
160 //'A'
161 p_lu | p_pr
162 //'B'
163 p_lu | p_pr
164 //'C'
165 p_lu | p_pr
166 //'D'
167 p_lu | p_pr
168 //'E'
169 p_lu | p_pr
170 //'F'
171 p_lu | p_pr
172 //'G'
173 p_lu | p_pr
174 //'H'
175 p_lu | p_pr
176 //'I'
177 p_lu | p_pr
178 //'J'
179 p_lu | p_pr
180 //'K'
181 p_lu | p_pr
182 //'L'
183 p_lu | p_pr
184 //'M'
185 p_lu | p_pr
186 //'N'
187 p_lu | p_pr
188 //'O'
189 p_lu | p_pr
190 //'P'
191 p_lu | p_pr
192 //'Q'
193 p_lu | p_pr
194 //'R'
195 p_lu | p_pr
196 //'S'
197 p_lu | p_pr
198 //'T'
199 p_lu | p_pr
200 //'U'
201 p_lu | p_pr
202 //'V'
203 p_lu | p_pr
204 //'W'
205 p_lu | p_pr
206 //'X'
207 p_lu | p_pr
208 //'Y'
209 p_lu | p_pr
210 //'Z'
211 p_p | p_pr
212 //'['
213 p_p | p_pr
214 //'\\'
215 p_p | p_pr
216 //']'
217 p_s | p_pr
218 //'^'
219 p_p | p_pr
220 //'_'
221 p_s | p_pr
222 //'`'
223 p_ll | p_pr
224 //'a'
225 p_ll | p_pr
226 //'b'
227 p_ll | p_pr
228 //'c'
229 p_ll | p_pr
230 //'d'
231 p_ll | p_pr
232 //'e'
233 p_ll | p_pr
234 //'f'
235 p_ll | p_pr
236 //'g'
237 p_ll | p_pr
238 //'h'
239 p_ll | p_pr
240 //'i'
241 p_ll | p_pr
242 //'j'
243 p_ll | p_pr
244 //'k'
245 p_ll | p_pr
246 //'l'
247 p_ll | p_pr
248 //'m'
249 p_ll | p_pr
250 //'n'
251 p_ll | p_pr
252 //'o'
253 p_ll | p_pr
254 //'p'
255 p_ll | p_pr
256 //'q'
257 p_ll | p_pr
258 //'r'
259 p_ll | p_pr
260 //'s'
261 p_ll | p_pr
262 //'t'
263 p_ll | p_pr
264 //'u'
265 p_ll | p_pr
266 //'v'
267 p_ll | p_pr
268 //'w'
269 p_ll | p_pr
270 //'x'
271 p_ll | p_pr
272 //'y'
273 p_ll | p_pr
274 //'z'
275 p_p | p_pr
276 //'{'
277 p_s | p_pr
278 //'|'
279 p_p | p_pr
280 //'}'
281 p_s | p_pr
282 //'~'
283 p_c
284 //'\u007f'
285 p_c
286 //'\u0080'
287 p_c
288 //'\u0081'
289 p_c
290 //'\u0082'
291 p_c
292 //'\u0083'
293 p_c
294 //'\u0084'
295 p_c
296 //'\u0085'
297 p_c
298 //'\u0086'
299 p_c
300 //'\u0087'
301 p_c
302 //'\u0088'
303 p_c
304 //'\u0089'
305 p_c
306 //'\u008a'
307 p_c
308 //'\u008b'
309 p_c
310 //'\u008c'
311 p_c
312 //'\u008d'
313 p_c
314 //'\u008e'
315 p_c
316 //'\u008f'
317 p_c
318 //'\u0090'
319 p_c
320 //'\u0091'
321 p_c
322 //'\u0092'
323 p_c
324 //'\u0093'
325 p_c
326 //'\u0094'
327 p_c
328 //'\u0095'
329 p_c
330 //'\u0096'
331 p_c
332 //'\u0097'
333 p_c
334 //'\u0098'
335 p_c
336 //'\u0099'
337 p_c
338 //'\u009a'
339 p_c
340 //'\u009b'
341 p_c
342 //'\u009c'
343 p_c
344 //'\u009d'
345 p_c
346 //'\u009e'
347 p_c
348 //'\u009f'
349 p_z
350 //'\u00a0'
351 p_p | p_pr
352 //'¡'
353 p_s | p_pr
354 //'¢'
355 p_s | p_pr
356 //'£'
357 p_s | p_pr
358 //'¤'
359 p_s | p_pr
360 //'¥'
361 p_s | p_pr
362 //'¦'
363 p_p | p_pr
364 //'§'
365 p_s | p_pr
366 //'¨'
367 p_s | p_pr
368 //'©'
369 p_lo | p_pr
370 //'ª'
371 p_p | p_pr
372 //'«'
373 p_s | p_pr
374 //'¬'
375 0
376 //'\u00ad'
377 p_s | p_pr
378 //'®'
379 p_s | p_pr
380 //'¯'
381 p_s | p_pr
382 //'°'
383 p_s | p_pr
384 //'±'
385 p_n | p_pr
386 //'²'
387 p_n | p_pr
388 //'³'
389 p_s | p_pr
390 //'´'
391 p_ll | p_pr
392 //'µ'
393 p_p | p_pr
394 //'¶'
395 p_p | p_pr
396 //'·'
397 p_s | p_pr
398 //'¸'
399 p_n | p_pr
400 //'¹'
401 p_lo | p_pr
402 //'º'
403 p_p | p_pr
404 //'»'
405 p_n | p_pr
406 //'¼'
407 p_n | p_pr
408 //'½'
409 p_n | p_pr
410 //'¾'
411 p_p | p_pr
412 //'¿'
413 p_lu | p_pr
414 //'À'
415 p_lu | p_pr
416 //'Á'
417 p_lu | p_pr
418 //'Â'
419 p_lu | p_pr
420 //'Ã'
421 p_lu | p_pr
422 //'Ä'
423 p_lu | p_pr
424 //'Å'
425 p_lu | p_pr
426 //'Æ'
427 p_lu | p_pr
428 //'Ç'
429 p_lu | p_pr
430 //'È'
431 p_lu | p_pr
432 //'É'
433 p_lu | p_pr
434 //'Ê'
435 p_lu | p_pr
436 //'Ë'
437 p_lu | p_pr
438 //'Ì'
439 p_lu | p_pr
440 //'Í'
441 p_lu | p_pr
442 //'Î'
443 p_lu | p_pr
444 //'Ï'
445 p_lu | p_pr
446 //'Ð'
447 p_lu | p_pr
448 //'Ñ'
449 p_lu | p_pr
450 //'Ò'
451 p_lu | p_pr
452 //'Ó'
453 p_lu | p_pr
454 //'Ô'
455 p_lu | p_pr
456 //'Õ'
457 p_lu | p_pr
458 //'Ö'
459 p_s | p_pr
460 //'×'
461 p_lu | p_pr
462 //'Ø'
463 p_lu | p_pr
464 //'Ù'
465 p_lu | p_pr
466 //'Ú'
467 p_lu | p_pr
468 //'Û'
469 p_lu | p_pr
470 //'Ü'
471 p_lu | p_pr
472 //'Ý'
473 p_lu | p_pr
474 //'Þ'
475 p_ll | p_pr
476 //'ß'
477 p_ll | p_pr
478 //'à'
479 p_ll | p_pr
480 //'á'
481 p_ll | p_pr
482 //'â'
483 p_ll | p_pr
484 //'ã'
485 p_ll | p_pr
486 //'ä'
487 p_ll | p_pr
488 //'å'
489 p_ll | p_pr
490 //'æ'
491 p_ll | p_pr
492 //'ç'
493 p_ll | p_pr
494 //'è'
495 p_ll | p_pr
496 //'é'
497 p_ll | p_pr
498 //'ê'
499 p_ll | p_pr
500 //'ë'
501 p_ll | p_pr
502 //'ì'
503 p_ll | p_pr
504 //'í'
505 p_ll | p_pr
506 //'î'
507 p_ll | p_pr
508 //'ï'
509 p_ll | p_pr
510 //'ð'
511 p_ll | p_pr
512 //'ñ'
513 p_ll | p_pr
514 //'ò'
515 p_ll | p_pr
516 //'ó'
517 p_ll | p_pr
518 //'ô'
519 p_ll | p_pr
520 //'õ'
521 p_ll | p_pr
522 //'ö'
523 p_s | p_pr
524 //'÷'
525 p_ll | p_pr
526 //'ø'
527 p_ll | p_pr
528 //'ù'
529 p_ll | p_pr
530 //'ú'
531 p_ll | p_pr
532 //'û'
533 p_ll | p_pr
534 //'ü'
535 p_ll | p_pr
536 //'ý'
537 p_ll | p_pr
538 //'þ'
539 p_ll | p_pr
540 // 'ÿ'
541]!
542
543// These tables are based on Go lang's tables: https://cs.opensource.google/go/go/+/refs/tags/go1.17.1:src/unicode/tables.go.
544// There is no need to investigate unicodes' type like letter yourself.
545const max_latin_1 = rune(0x00ff)
546
547// Represents all unicode in unicode category L.
548const letter_table = RangeTable{
549 r16: [
550 Range16{0x0041, 0x005a, 1},
551 Range16{0x0061, 0x007a, 1},
552 Range16{0x00aa, 0x00b5, 11},
553 Range16{0x00ba, 0x00c0, 6},
554 Range16{0x00c1, 0x00d6, 1},
555 Range16{0x00d8, 0x00f6, 1},
556 Range16{0x00f8, 0x02c1, 1},
557 Range16{0x02c6, 0x02d1, 1},
558 Range16{0x02e0, 0x02e4, 1},
559 Range16{0x02ec, 0x02ee, 2},
560 Range16{0x0370, 0x0374, 1},
561 Range16{0x0376, 0x0377, 1},
562 Range16{0x037a, 0x037d, 1},
563 Range16{0x037f, 0x0386, 7},
564 Range16{0x0388, 0x038a, 1},
565 Range16{0x038c, 0x038e, 2},
566 Range16{0x038f, 0x03a1, 1},
567 Range16{0x03a3, 0x03f5, 1},
568 Range16{0x03f7, 0x0481, 1},
569 Range16{0x048a, 0x052f, 1},
570 Range16{0x0531, 0x0556, 1},
571 Range16{0x0559, 0x0560, 7},
572 Range16{0x0561, 0x0588, 1},
573 Range16{0x05d0, 0x05ea, 1},
574 Range16{0x05ef, 0x05f2, 1},
575 Range16{0x0620, 0x064a, 1},
576 Range16{0x066e, 0x066f, 1},
577 Range16{0x0671, 0x06d3, 1},
578 Range16{0x06d5, 0x06e5, 16},
579 Range16{0x06e6, 0x06ee, 8},
580 Range16{0x06ef, 0x06fa, 11},
581 Range16{0x06fb, 0x06fc, 1},
582 Range16{0x06ff, 0x0710, 17},
583 Range16{0x0712, 0x072f, 1},
584 Range16{0x074d, 0x07a5, 1},
585 Range16{0x07b1, 0x07ca, 25},
586 Range16{0x07cb, 0x07ea, 1},
587 Range16{0x07f4, 0x07f5, 1},
588 Range16{0x07fa, 0x0800, 6},
589 Range16{0x0801, 0x0815, 1},
590 Range16{0x081a, 0x0824, 10},
591 Range16{0x0828, 0x0840, 24},
592 Range16{0x0841, 0x0858, 1},
593 Range16{0x0860, 0x086a, 1},
594 Range16{0x08a0, 0x08b4, 1},
595 Range16{0x08b6, 0x08c7, 1},
596 Range16{0x0904, 0x0939, 1},
597 Range16{0x093d, 0x0950, 19},
598 Range16{0x0958, 0x0961, 1},
599 Range16{0x0971, 0x0980, 1},
600 Range16{0x0985, 0x098c, 1},
601 Range16{0x098f, 0x0990, 1},
602 Range16{0x0993, 0x09a8, 1},
603 Range16{0x09aa, 0x09b0, 1},
604 Range16{0x09b2, 0x09b6, 4},
605 Range16{0x09b7, 0x09b9, 1},
606 Range16{0x09bd, 0x09ce, 17},
607 Range16{0x09dc, 0x09dd, 1},
608 Range16{0x09df, 0x09e1, 1},
609 Range16{0x09f0, 0x09f1, 1},
610 Range16{0x09fc, 0x0a05, 9},
611 Range16{0x0a06, 0x0a0a, 1},
612 Range16{0x0a0f, 0x0a10, 1},
613 Range16{0x0a13, 0x0a28, 1},
614 Range16{0x0a2a, 0x0a30, 1},
615 Range16{0x0a32, 0x0a33, 1},
616 Range16{0x0a35, 0x0a36, 1},
617 Range16{0x0a38, 0x0a39, 1},
618 Range16{0x0a59, 0x0a5c, 1},
619 Range16{0x0a5e, 0x0a72, 20},
620 Range16{0x0a73, 0x0a74, 1},
621 Range16{0x0a85, 0x0a8d, 1},
622 Range16{0x0a8f, 0x0a91, 1},
623 Range16{0x0a93, 0x0aa8, 1},
624 Range16{0x0aaa, 0x0ab0, 1},
625 Range16{0x0ab2, 0x0ab3, 1},
626 Range16{0x0ab5, 0x0ab9, 1},
627 Range16{0x0abd, 0x0ad0, 19},
628 Range16{0x0ae0, 0x0ae1, 1},
629 Range16{0x0af9, 0x0b05, 12},
630 Range16{0x0b06, 0x0b0c, 1},
631 Range16{0x0b0f, 0x0b10, 1},
632 Range16{0x0b13, 0x0b28, 1},
633 Range16{0x0b2a, 0x0b30, 1},
634 Range16{0x0b32, 0x0b33, 1},
635 Range16{0x0b35, 0x0b39, 1},
636 Range16{0x0b3d, 0x0b5c, 31},
637 Range16{0x0b5d, 0x0b5f, 2},
638 Range16{0x0b60, 0x0b61, 1},
639 Range16{0x0b71, 0x0b83, 18},
640 Range16{0x0b85, 0x0b8a, 1},
641 Range16{0x0b8e, 0x0b90, 1},
642 Range16{0x0b92, 0x0b95, 1},
643 Range16{0x0b99, 0x0b9a, 1},
644 Range16{0x0b9c, 0x0b9e, 2},
645 Range16{0x0b9f, 0x0ba3, 4},
646 Range16{0x0ba4, 0x0ba8, 4},
647 Range16{0x0ba9, 0x0baa, 1},
648 Range16{0x0bae, 0x0bb9, 1},
649 Range16{0x0bd0, 0x0c05, 53},
650 Range16{0x0c06, 0x0c0c, 1},
651 Range16{0x0c0e, 0x0c10, 1},
652 Range16{0x0c12, 0x0c28, 1},
653 Range16{0x0c2a, 0x0c39, 1},
654 Range16{0x0c3d, 0x0c58, 27},
655 Range16{0x0c59, 0x0c5a, 1},
656 Range16{0x0c60, 0x0c61, 1},
657 Range16{0x0c80, 0x0c85, 5},
658 Range16{0x0c86, 0x0c8c, 1},
659 Range16{0x0c8e, 0x0c90, 1},
660 Range16{0x0c92, 0x0ca8, 1},
661 Range16{0x0caa, 0x0cb3, 1},
662 Range16{0x0cb5, 0x0cb9, 1},
663 Range16{0x0cbd, 0x0cde, 33},
664 Range16{0x0ce0, 0x0ce1, 1},
665 Range16{0x0cf1, 0x0cf2, 1},
666 Range16{0x0d04, 0x0d0c, 1},
667 Range16{0x0d0e, 0x0d10, 1},
668 Range16{0x0d12, 0x0d3a, 1},
669 Range16{0x0d3d, 0x0d4e, 17},
670 Range16{0x0d54, 0x0d56, 1},
671 Range16{0x0d5f, 0x0d61, 1},
672 Range16{0x0d7a, 0x0d7f, 1},
673 Range16{0x0d85, 0x0d96, 1},
674 Range16{0x0d9a, 0x0db1, 1},
675 Range16{0x0db3, 0x0dbb, 1},
676 Range16{0x0dbd, 0x0dc0, 3},
677 Range16{0x0dc1, 0x0dc6, 1},
678 Range16{0x0e01, 0x0e30, 1},
679 Range16{0x0e32, 0x0e33, 1},
680 Range16{0x0e40, 0x0e46, 1},
681 Range16{0x0e81, 0x0e82, 1},
682 Range16{0x0e84, 0x0e86, 2},
683 Range16{0x0e87, 0x0e8a, 1},
684 Range16{0x0e8c, 0x0ea3, 1},
685 Range16{0x0ea5, 0x0ea7, 2},
686 Range16{0x0ea8, 0x0eb0, 1},
687 Range16{0x0eb2, 0x0eb3, 1},
688 Range16{0x0ebd, 0x0ec0, 3},
689 Range16{0x0ec1, 0x0ec4, 1},
690 Range16{0x0ec6, 0x0edc, 22},
691 Range16{0x0edd, 0x0edf, 1},
692 Range16{0x0f00, 0x0f40, 64},
693 Range16{0x0f41, 0x0f47, 1},
694 Range16{0x0f49, 0x0f6c, 1},
695 Range16{0x0f88, 0x0f8c, 1},
696 Range16{0x1000, 0x102a, 1},
697 Range16{0x103f, 0x1050, 17},
698 Range16{0x1051, 0x1055, 1},
699 Range16{0x105a, 0x105d, 1},
700 Range16{0x1061, 0x1065, 4},
701 Range16{0x1066, 0x106e, 8},
702 Range16{0x106f, 0x1070, 1},
703 Range16{0x1075, 0x1081, 1},
704 Range16{0x108e, 0x10a0, 18},
705 Range16{0x10a1, 0x10c5, 1},
706 Range16{0x10c7, 0x10cd, 6},
707 Range16{0x10d0, 0x10fa, 1},
708 Range16{0x10fc, 0x1248, 1},
709 Range16{0x124a, 0x124d, 1},
710 Range16{0x1250, 0x1256, 1},
711 Range16{0x1258, 0x125a, 2},
712 Range16{0x125b, 0x125d, 1},
713 Range16{0x1260, 0x1288, 1},
714 Range16{0x128a, 0x128d, 1},
715 Range16{0x1290, 0x12b0, 1},
716 Range16{0x12b2, 0x12b5, 1},
717 Range16{0x12b8, 0x12be, 1},
718 Range16{0x12c0, 0x12c2, 2},
719 Range16{0x12c3, 0x12c5, 1},
720 Range16{0x12c8, 0x12d6, 1},
721 Range16{0x12d8, 0x1310, 1},
722 Range16{0x1312, 0x1315, 1},
723 Range16{0x1318, 0x135a, 1},
724 Range16{0x1380, 0x138f, 1},
725 Range16{0x13a0, 0x13f5, 1},
726 Range16{0x13f8, 0x13fd, 1},
727 Range16{0x1401, 0x166c, 1},
728 Range16{0x166f, 0x167f, 1},
729 Range16{0x1681, 0x169a, 1},
730 Range16{0x16a0, 0x16ea, 1},
731 Range16{0x16f1, 0x16f8, 1},
732 Range16{0x1700, 0x170c, 1},
733 Range16{0x170e, 0x1711, 1},
734 Range16{0x1720, 0x1731, 1},
735 Range16{0x1740, 0x1751, 1},
736 Range16{0x1760, 0x176c, 1},
737 Range16{0x176e, 0x1770, 1},
738 Range16{0x1780, 0x17b3, 1},
739 Range16{0x17d7, 0x17dc, 5},
740 Range16{0x1820, 0x1878, 1},
741 Range16{0x1880, 0x1884, 1},
742 Range16{0x1887, 0x18a8, 1},
743 Range16{0x18aa, 0x18b0, 6},
744 Range16{0x18b1, 0x18f5, 1},
745 Range16{0x1900, 0x191e, 1},
746 Range16{0x1950, 0x196d, 1},
747 Range16{0x1970, 0x1974, 1},
748 Range16{0x1980, 0x19ab, 1},
749 Range16{0x19b0, 0x19c9, 1},
750 Range16{0x1a00, 0x1a16, 1},
751 Range16{0x1a20, 0x1a54, 1},
752 Range16{0x1aa7, 0x1b05, 94},
753 Range16{0x1b06, 0x1b33, 1},
754 Range16{0x1b45, 0x1b4b, 1},
755 Range16{0x1b83, 0x1ba0, 1},
756 Range16{0x1bae, 0x1baf, 1},
757 Range16{0x1bba, 0x1be5, 1},
758 Range16{0x1c00, 0x1c23, 1},
759 Range16{0x1c4d, 0x1c4f, 1},
760 Range16{0x1c5a, 0x1c7d, 1},
761 Range16{0x1c80, 0x1c88, 1},
762 Range16{0x1c90, 0x1cba, 1},
763 Range16{0x1cbd, 0x1cbf, 1},
764 Range16{0x1ce9, 0x1cec, 1},
765 Range16{0x1cee, 0x1cf3, 1},
766 Range16{0x1cf5, 0x1cf6, 1},
767 Range16{0x1cfa, 0x1d00, 6},
768 Range16{0x1d01, 0x1dbf, 1},
769 Range16{0x1e00, 0x1f15, 1},
770 Range16{0x1f18, 0x1f1d, 1},
771 Range16{0x1f20, 0x1f45, 1},
772 Range16{0x1f48, 0x1f4d, 1},
773 Range16{0x1f50, 0x1f57, 1},
774 Range16{0x1f59, 0x1f5f, 2},
775 Range16{0x1f60, 0x1f7d, 1},
776 Range16{0x1f80, 0x1fb4, 1},
777 Range16{0x1fb6, 0x1fbc, 1},
778 Range16{0x1fbe, 0x1fc2, 4},
779 Range16{0x1fc3, 0x1fc4, 1},
780 Range16{0x1fc6, 0x1fcc, 1},
781 Range16{0x1fd0, 0x1fd3, 1},
782 Range16{0x1fd6, 0x1fdb, 1},
783 Range16{0x1fe0, 0x1fec, 1},
784 Range16{0x1ff2, 0x1ff4, 1},
785 Range16{0x1ff6, 0x1ffc, 1},
786 Range16{0x2071, 0x207f, 14},
787 Range16{0x2090, 0x209c, 1},
788 Range16{0x2102, 0x2107, 5},
789 Range16{0x210a, 0x2113, 1},
790 Range16{0x2115, 0x2119, 4},
791 Range16{0x211a, 0x211d, 1},
792 Range16{0x2124, 0x212a, 2},
793 Range16{0x212b, 0x212d, 1},
794 Range16{0x212f, 0x2139, 1},
795 Range16{0x213c, 0x213f, 1},
796 Range16{0x2145, 0x2149, 1},
797 Range16{0x214e, 0x2183, 53},
798 Range16{0x2184, 0x2c00, 2684},
799 Range16{0x2c01, 0x2c2e, 1},
800 Range16{0x2c30, 0x2c5e, 1},
801 Range16{0x2c60, 0x2ce4, 1},
802 Range16{0x2ceb, 0x2cee, 1},
803 Range16{0x2cf2, 0x2cf3, 1},
804 Range16{0x2d00, 0x2d25, 1},
805 Range16{0x2d27, 0x2d2d, 6},
806 Range16{0x2d30, 0x2d67, 1},
807 Range16{0x2d6f, 0x2d80, 17},
808 Range16{0x2d81, 0x2d96, 1},
809 Range16{0x2da0, 0x2da6, 1},
810 Range16{0x2da8, 0x2dae, 1},
811 Range16{0x2db0, 0x2db6, 1},
812 Range16{0x2db8, 0x2dbe, 1},
813 Range16{0x2dc0, 0x2dc6, 1},
814 Range16{0x2dc8, 0x2dce, 1},
815 Range16{0x2dd0, 0x2dd6, 1},
816 Range16{0x2dd8, 0x2dde, 1},
817 Range16{0x2e2f, 0x3005, 470},
818 Range16{0x3006, 0x3031, 43},
819 Range16{0x3032, 0x3035, 1},
820 Range16{0x303b, 0x303c, 1},
821 Range16{0x3041, 0x3096, 1},
822 Range16{0x309d, 0x309f, 1},
823 Range16{0x30a1, 0x30fa, 1},
824 Range16{0x30fc, 0x30ff, 1},
825 Range16{0x3105, 0x312f, 1},
826 Range16{0x3131, 0x318e, 1},
827 Range16{0x31a0, 0x31bf, 1},
828 Range16{0x31f0, 0x31ff, 1},
829 Range16{0x3400, 0x4dbf, 1},
830 Range16{0x4e00, 0x9ffc, 1},
831 Range16{0xa000, 0xa48c, 1},
832 Range16{0xa4d0, 0xa4fd, 1},
833 Range16{0xa500, 0xa60c, 1},
834 Range16{0xa610, 0xa61f, 1},
835 Range16{0xa62a, 0xa62b, 1},
836 Range16{0xa640, 0xa66e, 1},
837 Range16{0xa67f, 0xa69d, 1},
838 Range16{0xa6a0, 0xa6e5, 1},
839 Range16{0xa717, 0xa71f, 1},
840 Range16{0xa722, 0xa788, 1},
841 Range16{0xa78b, 0xa7bf, 1},
842 Range16{0xa7c2, 0xa7ca, 1},
843 Range16{0xa7f5, 0xa801, 1},
844 Range16{0xa803, 0xa805, 1},
845 Range16{0xa807, 0xa80a, 1},
846 Range16{0xa80c, 0xa822, 1},
847 Range16{0xa840, 0xa873, 1},
848 Range16{0xa882, 0xa8b3, 1},
849 Range16{0xa8f2, 0xa8f7, 1},
850 Range16{0xa8fb, 0xa8fd, 2},
851 Range16{0xa8fe, 0xa90a, 12},
852 Range16{0xa90b, 0xa925, 1},
853 Range16{0xa930, 0xa946, 1},
854 Range16{0xa960, 0xa97c, 1},
855 Range16{0xa984, 0xa9b2, 1},
856 Range16{0xa9cf, 0xa9e0, 17},
857 Range16{0xa9e1, 0xa9e4, 1},
858 Range16{0xa9e6, 0xa9ef, 1},
859 Range16{0xa9fa, 0xa9fe, 1},
860 Range16{0xaa00, 0xaa28, 1},
861 Range16{0xaa40, 0xaa42, 1},
862 Range16{0xaa44, 0xaa4b, 1},
863 Range16{0xaa60, 0xaa76, 1},
864 Range16{0xaa7a, 0xaa7e, 4},
865 Range16{0xaa7f, 0xaaaf, 1},
866 Range16{0xaab1, 0xaab5, 4},
867 Range16{0xaab6, 0xaab9, 3},
868 Range16{0xaaba, 0xaabd, 1},
869 Range16{0xaac0, 0xaac2, 2},
870 Range16{0xaadb, 0xaadd, 1},
871 Range16{0xaae0, 0xaaea, 1},
872 Range16{0xaaf2, 0xaaf4, 1},
873 Range16{0xab01, 0xab06, 1},
874 Range16{0xab09, 0xab0e, 1},
875 Range16{0xab11, 0xab16, 1},
876 Range16{0xab20, 0xab26, 1},
877 Range16{0xab28, 0xab2e, 1},
878 Range16{0xab30, 0xab5a, 1},
879 Range16{0xab5c, 0xab69, 1},
880 Range16{0xab70, 0xabe2, 1},
881 Range16{0xac00, 0xd7a3, 1},
882 Range16{0xd7b0, 0xd7c6, 1},
883 Range16{0xd7cb, 0xd7fb, 1},
884 Range16{0xf900, 0xfa6d, 1},
885 Range16{0xfa70, 0xfad9, 1},
886 Range16{0xfb00, 0xfb06, 1},
887 Range16{0xfb13, 0xfb17, 1},
888 Range16{0xfb1d, 0xfb1f, 2},
889 Range16{0xfb20, 0xfb28, 1},
890 Range16{0xfb2a, 0xfb36, 1},
891 Range16{0xfb38, 0xfb3c, 1},
892 Range16{0xfb3e, 0xfb40, 2},
893 Range16{0xfb41, 0xfb43, 2},
894 Range16{0xfb44, 0xfb46, 2},
895 Range16{0xfb47, 0xfbb1, 1},
896 Range16{0xfbd3, 0xfd3d, 1},
897 Range16{0xfd50, 0xfd8f, 1},
898 Range16{0xfd92, 0xfdc7, 1},
899 Range16{0xfdf0, 0xfdfb, 1},
900 Range16{0xfe70, 0xfe74, 1},
901 Range16{0xfe76, 0xfefc, 1},
902 Range16{0xff21, 0xff3a, 1},
903 Range16{0xff41, 0xff5a, 1},
904 Range16{0xff66, 0xffbe, 1},
905 Range16{0xffc2, 0xffc7, 1},
906 Range16{0xffca, 0xffcf, 1},
907 Range16{0xffd2, 0xffd7, 1},
908 Range16{0xffda, 0xffdc, 1},
909 ]
910 r32: [
911 Range32{0x10000, 0x1000b, 1},
912 Range32{0x1000d, 0x10026, 1},
913 Range32{0x10028, 0x1003a, 1},
914 Range32{0x1003c, 0x1003d, 1},
915 Range32{0x1003f, 0x1004d, 1},
916 Range32{0x10050, 0x1005d, 1},
917 Range32{0x10080, 0x100fa, 1},
918 Range32{0x10280, 0x1029c, 1},
919 Range32{0x102a0, 0x102d0, 1},
920 Range32{0x10300, 0x1031f, 1},
921 Range32{0x1032d, 0x10340, 1},
922 Range32{0x10342, 0x10349, 1},
923 Range32{0x10350, 0x10375, 1},
924 Range32{0x10380, 0x1039d, 1},
925 Range32{0x103a0, 0x103c3, 1},
926 Range32{0x103c8, 0x103cf, 1},
927 Range32{0x10400, 0x1049d, 1},
928 Range32{0x104b0, 0x104d3, 1},
929 Range32{0x104d8, 0x104fb, 1},
930 Range32{0x10500, 0x10527, 1},
931 Range32{0x10530, 0x10563, 1},
932 Range32{0x10600, 0x10736, 1},
933 Range32{0x10740, 0x10755, 1},
934 Range32{0x10760, 0x10767, 1},
935 Range32{0x10800, 0x10805, 1},
936 Range32{0x10808, 0x1080a, 2},
937 Range32{0x1080b, 0x10835, 1},
938 Range32{0x10837, 0x10838, 1},
939 Range32{0x1083c, 0x1083f, 3},
940 Range32{0x10840, 0x10855, 1},
941 Range32{0x10860, 0x10876, 1},
942 Range32{0x10880, 0x1089e, 1},
943 Range32{0x108e0, 0x108f2, 1},
944 Range32{0x108f4, 0x108f5, 1},
945 Range32{0x10900, 0x10915, 1},
946 Range32{0x10920, 0x10939, 1},
947 Range32{0x10980, 0x109b7, 1},
948 Range32{0x109be, 0x109bf, 1},
949 Range32{0x10a00, 0x10a10, 16},
950 Range32{0x10a11, 0x10a13, 1},
951 Range32{0x10a15, 0x10a17, 1},
952 Range32{0x10a19, 0x10a35, 1},
953 Range32{0x10a60, 0x10a7c, 1},
954 Range32{0x10a80, 0x10a9c, 1},
955 Range32{0x10ac0, 0x10ac7, 1},
956 Range32{0x10ac9, 0x10ae4, 1},
957 Range32{0x10b00, 0x10b35, 1},
958 Range32{0x10b40, 0x10b55, 1},
959 Range32{0x10b60, 0x10b72, 1},
960 Range32{0x10b80, 0x10b91, 1},
961 Range32{0x10c00, 0x10c48, 1},
962 Range32{0x10c80, 0x10cb2, 1},
963 Range32{0x10cc0, 0x10cf2, 1},
964 Range32{0x10d00, 0x10d23, 1},
965 Range32{0x10e80, 0x10ea9, 1},
966 Range32{0x10eb0, 0x10eb1, 1},
967 Range32{0x10f00, 0x10f1c, 1},
968 Range32{0x10f27, 0x10f30, 9},
969 Range32{0x10f31, 0x10f45, 1},
970 Range32{0x10fb0, 0x10fc4, 1},
971 Range32{0x10fe0, 0x10ff6, 1},
972 Range32{0x11003, 0x11037, 1},
973 Range32{0x11083, 0x110af, 1},
974 Range32{0x110d0, 0x110e8, 1},
975 Range32{0x11103, 0x11126, 1},
976 Range32{0x11144, 0x11147, 3},
977 Range32{0x11150, 0x11172, 1},
978 Range32{0x11176, 0x11183, 13},
979 Range32{0x11184, 0x111b2, 1},
980 Range32{0x111c1, 0x111c4, 1},
981 Range32{0x111da, 0x111dc, 2},
982 Range32{0x11200, 0x11211, 1},
983 Range32{0x11213, 0x1122b, 1},
984 Range32{0x11280, 0x11286, 1},
985 Range32{0x11288, 0x1128a, 2},
986 Range32{0x1128b, 0x1128d, 1},
987 Range32{0x1128f, 0x1129d, 1},
988 Range32{0x1129f, 0x112a8, 1},
989 Range32{0x112b0, 0x112de, 1},
990 Range32{0x11305, 0x1130c, 1},
991 Range32{0x1130f, 0x11310, 1},
992 Range32{0x11313, 0x11328, 1},
993 Range32{0x1132a, 0x11330, 1},
994 Range32{0x11332, 0x11333, 1},
995 Range32{0x11335, 0x11339, 1},
996 Range32{0x1133d, 0x11350, 19},
997 Range32{0x1135d, 0x11361, 1},
998 Range32{0x11400, 0x11434, 1},
999 Range32{0x11447, 0x1144a, 1},
1000 Range32{0x1145f, 0x11461, 1},
1001 Range32{0x11480, 0x114af, 1},
1002 Range32{0x114c4, 0x114c5, 1},
1003 Range32{0x114c7, 0x11580, 185},
1004 Range32{0x11581, 0x115ae, 1},
1005 Range32{0x115d8, 0x115db, 1},
1006 Range32{0x11600, 0x1162f, 1},
1007 Range32{0x11644, 0x11680, 60},
1008 Range32{0x11681, 0x116aa, 1},
1009 Range32{0x116b8, 0x11700, 72},
1010 Range32{0x11701, 0x1171a, 1},
1011 Range32{0x11800, 0x1182b, 1},
1012 Range32{0x118a0, 0x118df, 1},
1013 Range32{0x118ff, 0x11906, 1},
1014 Range32{0x11909, 0x1190c, 3},
1015 Range32{0x1190d, 0x11913, 1},
1016 Range32{0x11915, 0x11916, 1},
1017 Range32{0x11918, 0x1192f, 1},
1018 Range32{0x1193f, 0x11941, 2},
1019 Range32{0x119a0, 0x119a7, 1},
1020 Range32{0x119aa, 0x119d0, 1},
1021 Range32{0x119e1, 0x119e3, 2},
1022 Range32{0x11a00, 0x11a0b, 11},
1023 Range32{0x11a0c, 0x11a32, 1},
1024 Range32{0x11a3a, 0x11a50, 22},
1025 Range32{0x11a5c, 0x11a89, 1},
1026 Range32{0x11a9d, 0x11ac0, 35},
1027 Range32{0x11ac1, 0x11af8, 1},
1028 Range32{0x11c00, 0x11c08, 1},
1029 Range32{0x11c0a, 0x11c2e, 1},
1030 Range32{0x11c40, 0x11c72, 50},
1031 Range32{0x11c73, 0x11c8f, 1},
1032 Range32{0x11d00, 0x11d06, 1},
1033 Range32{0x11d08, 0x11d09, 1},
1034 Range32{0x11d0b, 0x11d30, 1},
1035 Range32{0x11d46, 0x11d60, 26},
1036 Range32{0x11d61, 0x11d65, 1},
1037 Range32{0x11d67, 0x11d68, 1},
1038 Range32{0x11d6a, 0x11d89, 1},
1039 Range32{0x11d98, 0x11ee0, 328},
1040 Range32{0x11ee1, 0x11ef2, 1},
1041 Range32{0x11fb0, 0x12000, 80},
1042 Range32{0x12001, 0x12399, 1},
1043 Range32{0x12480, 0x12543, 1},
1044 Range32{0x13000, 0x1342e, 1},
1045 Range32{0x14400, 0x14646, 1},
1046 Range32{0x16800, 0x16a38, 1},
1047 Range32{0x16a40, 0x16a5e, 1},
1048 Range32{0x16ad0, 0x16aed, 1},
1049 Range32{0x16b00, 0x16b2f, 1},
1050 Range32{0x16b40, 0x16b43, 1},
1051 Range32{0x16b63, 0x16b77, 1},
1052 Range32{0x16b7d, 0x16b8f, 1},
1053 Range32{0x16e40, 0x16e7f, 1},
1054 Range32{0x16f00, 0x16f4a, 1},
1055 Range32{0x16f50, 0x16f93, 67},
1056 Range32{0x16f94, 0x16f9f, 1},
1057 Range32{0x16fe0, 0x16fe1, 1},
1058 Range32{0x16fe3, 0x17000, 29},
1059 Range32{0x17001, 0x187f7, 1},
1060 Range32{0x18800, 0x18cd5, 1},
1061 Range32{0x18d00, 0x18d08, 1},
1062 Range32{0x1b000, 0x1b11e, 1},
1063 Range32{0x1b150, 0x1b152, 1},
1064 Range32{0x1b164, 0x1b167, 1},
1065 Range32{0x1b170, 0x1b2fb, 1},
1066 Range32{0x1bc00, 0x1bc6a, 1},
1067 Range32{0x1bc70, 0x1bc7c, 1},
1068 Range32{0x1bc80, 0x1bc88, 1},
1069 Range32{0x1bc90, 0x1bc99, 1},
1070 Range32{0x1d400, 0x1d454, 1},
1071 Range32{0x1d456, 0x1d49c, 1},
1072 Range32{0x1d49e, 0x1d49f, 1},
1073 Range32{0x1d4a2, 0x1d4a5, 3},
1074 Range32{0x1d4a6, 0x1d4a9, 3},
1075 Range32{0x1d4aa, 0x1d4ac, 1},
1076 Range32{0x1d4ae, 0x1d4b9, 1},
1077 Range32{0x1d4bb, 0x1d4bd, 2},
1078 Range32{0x1d4be, 0x1d4c3, 1},
1079 Range32{0x1d4c5, 0x1d505, 1},
1080 Range32{0x1d507, 0x1d50a, 1},
1081 Range32{0x1d50d, 0x1d514, 1},
1082 Range32{0x1d516, 0x1d51c, 1},
1083 Range32{0x1d51e, 0x1d539, 1},
1084 Range32{0x1d53b, 0x1d53e, 1},
1085 Range32{0x1d540, 0x1d544, 1},
1086 Range32{0x1d546, 0x1d54a, 4},
1087 Range32{0x1d54b, 0x1d550, 1},
1088 Range32{0x1d552, 0x1d6a5, 1},
1089 Range32{0x1d6a8, 0x1d6c0, 1},
1090 Range32{0x1d6c2, 0x1d6da, 1},
1091 Range32{0x1d6dc, 0x1d6fa, 1},
1092 Range32{0x1d6fc, 0x1d714, 1},
1093 Range32{0x1d716, 0x1d734, 1},
1094 Range32{0x1d736, 0x1d74e, 1},
1095 Range32{0x1d750, 0x1d76e, 1},
1096 Range32{0x1d770, 0x1d788, 1},
1097 Range32{0x1d78a, 0x1d7a8, 1},
1098 Range32{0x1d7aa, 0x1d7c2, 1},
1099 Range32{0x1d7c4, 0x1d7cb, 1},
1100 Range32{0x1e100, 0x1e12c, 1},
1101 Range32{0x1e137, 0x1e13d, 1},
1102 Range32{0x1e14e, 0x1e2c0, 370},
1103 Range32{0x1e2c1, 0x1e2eb, 1},
1104 Range32{0x1e800, 0x1e8c4, 1},
1105 Range32{0x1e900, 0x1e943, 1},
1106 Range32{0x1e94b, 0x1ee00, 1205},
1107 Range32{0x1ee01, 0x1ee03, 1},
1108 Range32{0x1ee05, 0x1ee1f, 1},
1109 Range32{0x1ee21, 0x1ee22, 1},
1110 Range32{0x1ee24, 0x1ee27, 3},
1111 Range32{0x1ee29, 0x1ee32, 1},
1112 Range32{0x1ee34, 0x1ee37, 1},
1113 Range32{0x1ee39, 0x1ee3b, 2},
1114 Range32{0x1ee42, 0x1ee47, 5},
1115 Range32{0x1ee49, 0x1ee4d, 2},
1116 Range32{0x1ee4e, 0x1ee4f, 1},
1117 Range32{0x1ee51, 0x1ee52, 1},
1118 Range32{0x1ee54, 0x1ee57, 3},
1119 Range32{0x1ee59, 0x1ee61, 2},
1120 Range32{0x1ee62, 0x1ee64, 2},
1121 Range32{0x1ee67, 0x1ee6a, 1},
1122 Range32{0x1ee6c, 0x1ee72, 1},
1123 Range32{0x1ee74, 0x1ee77, 1},
1124 Range32{0x1ee79, 0x1ee7c, 1},
1125 Range32{0x1ee7e, 0x1ee80, 2},
1126 Range32{0x1ee81, 0x1ee89, 1},
1127 Range32{0x1ee8b, 0x1ee9b, 1},
1128 Range32{0x1eea1, 0x1eea3, 1},
1129 Range32{0x1eea5, 0x1eea9, 1},
1130 Range32{0x1eeab, 0x1eebb, 1},
1131 Range32{0x20000, 0x2a6dd, 1},
1132 Range32{0x2a700, 0x2b734, 1},
1133 Range32{0x2b740, 0x2b81d, 1},
1134 Range32{0x2b820, 0x2cea1, 1},
1135 Range32{0x2ceb0, 0x2ebe0, 1},
1136 Range32{0x2f800, 0x2fa1d, 1},
1137 Range32{0x30000, 0x3134a, 1},
1138 ]
1139 latin_offset: 6
1140}
1141
1142// Represents all unicodes in unicode category Z with property white space.
1143const white_space_table = RangeTable{
1144 r16: [
1145 Range16{0x0009, 0x000d, 1},
1146 Range16{0x0020, 0x0085, 101},
1147 Range16{0x00a0, 0x1680, 5600},
1148 Range16{0x2000, 0x200a, 1},
1149 Range16{0x2028, 0x2029, 1},
1150 Range16{0x202f, 0x205f, 48},
1151 Range16{0x3000, 0x3000, 1},
1152 ]
1153 r32: []
1154 latin_offset: 2
1155}
1156
1157// Represents all unicodes in unicode category N.
1158const number_table = RangeTable{
1159 r16: [
1160 Range16{0x0030, 0x0039, 1},
1161 Range16{0x00b2, 0x00b3, 1},
1162 Range16{0x00b9, 0x00bc, 3},
1163 Range16{0x00bd, 0x00be, 1},
1164 Range16{0x0660, 0x0669, 1},
1165 Range16{0x06f0, 0x06f9, 1},
1166 Range16{0x07c0, 0x07c9, 1},
1167 Range16{0x0966, 0x096f, 1},
1168 Range16{0x09e6, 0x09ef, 1},
1169 Range16{0x09f4, 0x09f9, 1},
1170 Range16{0x0a66, 0x0a6f, 1},
1171 Range16{0x0ae6, 0x0aef, 1},
1172 Range16{0x0b66, 0x0b6f, 1},
1173 Range16{0x0b72, 0x0b77, 1},
1174 Range16{0x0be6, 0x0bf2, 1},
1175 Range16{0x0c66, 0x0c6f, 1},
1176 Range16{0x0c78, 0x0c7e, 1},
1177 Range16{0x0ce6, 0x0cef, 1},
1178 Range16{0x0d58, 0x0d5e, 1},
1179 Range16{0x0d66, 0x0d78, 1},
1180 Range16{0x0de6, 0x0def, 1},
1181 Range16{0x0e50, 0x0e59, 1},
1182 Range16{0x0ed0, 0x0ed9, 1},
1183 Range16{0x0f20, 0x0f33, 1},
1184 Range16{0x1040, 0x1049, 1},
1185 Range16{0x1090, 0x1099, 1},
1186 Range16{0x1369, 0x137c, 1},
1187 Range16{0x16ee, 0x16f0, 1},
1188 Range16{0x17e0, 0x17e9, 1},
1189 Range16{0x17f0, 0x17f9, 1},
1190 Range16{0x1810, 0x1819, 1},
1191 Range16{0x1946, 0x194f, 1},
1192 Range16{0x19d0, 0x19da, 1},
1193 Range16{0x1a80, 0x1a89, 1},
1194 Range16{0x1a90, 0x1a99, 1},
1195 Range16{0x1b50, 0x1b59, 1},
1196 Range16{0x1bb0, 0x1bb9, 1},
1197 Range16{0x1c40, 0x1c49, 1},
1198 Range16{0x1c50, 0x1c59, 1},
1199 Range16{0x2070, 0x2074, 4},
1200 Range16{0x2075, 0x2079, 1},
1201 Range16{0x2080, 0x2089, 1},
1202 Range16{0x2150, 0x2182, 1},
1203 Range16{0x2185, 0x2189, 1},
1204 Range16{0x2460, 0x249b, 1},
1205 Range16{0x24ea, 0x24ff, 1},
1206 Range16{0x2776, 0x2793, 1},
1207 Range16{0x2cfd, 0x3007, 778},
1208 Range16{0x3021, 0x3029, 1},
1209 Range16{0x3038, 0x303a, 1},
1210 Range16{0x3192, 0x3195, 1},
1211 Range16{0x3220, 0x3229, 1},
1212 Range16{0x3248, 0x324f, 1},
1213 Range16{0x3251, 0x325f, 1},
1214 Range16{0x3280, 0x3289, 1},
1215 Range16{0x32b1, 0x32bf, 1},
1216 Range16{0xa620, 0xa629, 1},
1217 Range16{0xa6e6, 0xa6ef, 1},
1218 Range16{0xa830, 0xa835, 1},
1219 Range16{0xa8d0, 0xa8d9, 1},
1220 Range16{0xa900, 0xa909, 1},
1221 Range16{0xa9d0, 0xa9d9, 1},
1222 Range16{0xa9f0, 0xa9f9, 1},
1223 Range16{0xaa50, 0xaa59, 1},
1224 Range16{0xabf0, 0xabf9, 1},
1225 Range16{0xff10, 0xff19, 1},
1226 ]
1227 r32: [
1228 Range32{0x10107, 0x10133, 1},
1229 Range32{0x10140, 0x10178, 1},
1230 Range32{0x1018a, 0x1018b, 1},
1231 Range32{0x102e1, 0x102fb, 1},
1232 Range32{0x10320, 0x10323, 1},
1233 Range32{0x10341, 0x1034a, 9},
1234 Range32{0x103d1, 0x103d5, 1},
1235 Range32{0x104a0, 0x104a9, 1},
1236 Range32{0x10858, 0x1085f, 1},
1237 Range32{0x10879, 0x1087f, 1},
1238 Range32{0x108a7, 0x108af, 1},
1239 Range32{0x108fb, 0x108ff, 1},
1240 Range32{0x10916, 0x1091b, 1},
1241 Range32{0x109bc, 0x109bd, 1},
1242 Range32{0x109c0, 0x109cf, 1},
1243 Range32{0x109d2, 0x109ff, 1},
1244 Range32{0x10a40, 0x10a48, 1},
1245 Range32{0x10a7d, 0x10a7e, 1},
1246 Range32{0x10a9d, 0x10a9f, 1},
1247 Range32{0x10aeb, 0x10aef, 1},
1248 Range32{0x10b58, 0x10b5f, 1},
1249 Range32{0x10b78, 0x10b7f, 1},
1250 Range32{0x10ba9, 0x10baf, 1},
1251 Range32{0x10cfa, 0x10cff, 1},
1252 Range32{0x10d30, 0x10d39, 1},
1253 Range32{0x10e60, 0x10e7e, 1},
1254 Range32{0x10f1d, 0x10f26, 1},
1255 Range32{0x10f51, 0x10f54, 1},
1256 Range32{0x10fc5, 0x10fcb, 1},
1257 Range32{0x11052, 0x1106f, 1},
1258 Range32{0x110f0, 0x110f9, 1},
1259 Range32{0x11136, 0x1113f, 1},
1260 Range32{0x111d0, 0x111d9, 1},
1261 Range32{0x111e1, 0x111f4, 1},
1262 Range32{0x112f0, 0x112f9, 1},
1263 Range32{0x11450, 0x11459, 1},
1264 Range32{0x114d0, 0x114d9, 1},
1265 Range32{0x11650, 0x11659, 1},
1266 Range32{0x116c0, 0x116c9, 1},
1267 Range32{0x11730, 0x1173b, 1},
1268 Range32{0x118e0, 0x118f2, 1},
1269 Range32{0x11950, 0x11959, 1},
1270 Range32{0x11c50, 0x11c6c, 1},
1271 Range32{0x11d50, 0x11d59, 1},
1272 Range32{0x11da0, 0x11da9, 1},
1273 Range32{0x11fc0, 0x11fd4, 1},
1274 Range32{0x12400, 0x1246e, 1},
1275 Range32{0x16a60, 0x16a69, 1},
1276 Range32{0x16b50, 0x16b59, 1},
1277 Range32{0x16b5b, 0x16b61, 1},
1278 Range32{0x16e80, 0x16e96, 1},
1279 Range32{0x1d2e0, 0x1d2f3, 1},
1280 Range32{0x1d360, 0x1d378, 1},
1281 Range32{0x1d7ce, 0x1d7ff, 1},
1282 Range32{0x1e140, 0x1e149, 1},
1283 Range32{0x1e2f0, 0x1e2f9, 1},
1284 Range32{0x1e8c7, 0x1e8cf, 1},
1285 Range32{0x1e950, 0x1e959, 1},
1286 Range32{0x1ec71, 0x1ecab, 1},
1287 Range32{0x1ecad, 0x1ecaf, 1},
1288 Range32{0x1ecb1, 0x1ecb4, 1},
1289 Range32{0x1ed01, 0x1ed2d, 1},
1290 Range32{0x1ed2f, 0x1ed3d, 1},
1291 Range32{0x1f100, 0x1f10c, 1},
1292 Range32{0x1fbf0, 0x1fbf9, 1},
1293 ]
1294 latin_offset: 4
1295}
1296
1297struct RangeTable {
1298pub:
1299 r16 []Range16
1300 r32 []Range32
1301 latin_offset int
1302}
1303
1304struct Range16 {
1305pub:
1306 lo u16
1307 hi u16
1308 stride u16
1309}
1310
1311struct Range32 {
1312pub:
1313 lo u32
1314 hi u32
1315 stride u32
1316}
1317
1318// tests if rune is in the given range table.
1319fn is_excluding_latin(table &RangeTable, r rune) bool {
1320 r16 := &table.r16
1321 off := table.latin_offset
1322 if r16.len > off && u32(r) < u32((*r16)[r16.len - 1].hi) {
1323 return is_16((*r16)[off..], u16(r))
1324 }
1325 r32 := &table.r32
1326 if r32.len > 0 && r >= rune((*r32)[0].lo) {
1327 return is_32(*r32, u32(r))
1328 }
1329 return false
1330}
1331
1332const linear_max = 18
1333
1334fn is_16(ranges []Range16, r u16) bool {
1335 if ranges.len <= linear_max && r <= max_latin_1 {
1336 for range in ranges {
1337 if r < range.lo {
1338 return false
1339 }
1340 if r <= range.hi {
1341 return range.stride == 1 || (r - range.lo) % range.stride == 0
1342 }
1343 }
1344 return false
1345 }
1346
1347 // binary search
1348 mut low, mut high := 0, ranges.len
1349 for low < high {
1350 medium := low + (high - low) / 2
1351 range := ranges[medium]
1352 if range.lo <= r && r <= range.hi {
1353 return range.stride == 1 || (r - range.lo) % range.stride == 0
1354 }
1355 if r < range.lo {
1356 high = medium
1357 } else {
1358 low = medium + 1
1359 }
1360 }
1361
1362 return false
1363}
1364
1365fn is_32(ranges []Range32, r u32) bool {
1366 if ranges.len <= linear_max && r <= max_latin_1 {
1367 for range in ranges {
1368 if r < range.lo {
1369 return false
1370 }
1371 if r <= range.hi {
1372 return range.stride == 1 || (r - range.lo) % range.stride == 0
1373 }
1374 }
1375 return false
1376 }
1377
1378 // binary search
1379 mut low, mut high := 0, ranges.len
1380 for low < high {
1381 medium := low + (high - low) / 2
1382 range := ranges[medium]
1383 if range.lo <= r && r <= range.hi {
1384 return range.stride == 1 || (r - range.lo) % range.stride == 0
1385 }
1386 if r < range.lo {
1387 high = medium
1388 } else {
1389 low = medium + 1
1390 }
1391 }
1392
1393 return false
1394}
1395