1 /**********************************************************************
2 unicode.c - Oniguruma (regular expression library)
3 **********************************************************************/
4 /*-
5 * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 */
29
30 #include "regint.h"
31
32 struct PoolPropertyNameCtype {
33 short int name;
34 short int ctype;
35 };
36
37 #define ONIGENC_IS_UNICODE_ISO_8859_1_CTYPE(code,ctype) \
38 ((EncUNICODE_ISO_8859_1_CtypeTable[code] & CTYPE_TO_BIT(ctype)) != 0)
39
40 static const unsigned short EncUNICODE_ISO_8859_1_CtypeTable[256] = {
41 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
42 0x4008, 0x428c, 0x4289, 0x4288, 0x4288, 0x4288, 0x4008, 0x4008,
43 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
44 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
45 0x4284, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
46 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
47 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0,
48 0x78b0, 0x78b0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
49 0x41a0, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x74a2,
50 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2,
51 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2,
52 0x74a2, 0x74a2, 0x74a2, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x51a0,
53 0x41a0, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x70e2,
54 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2,
55 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2,
56 0x70e2, 0x70e2, 0x70e2, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x4008,
57 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0288, 0x0008, 0x0008,
58 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008,
59 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008,
60 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008,
61 0x0284, 0x01a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0,
62 0x00a0, 0x00a0, 0x30e2, 0x01a0, 0x00a0, 0x00a8, 0x00a0, 0x00a0,
63 0x00a0, 0x00a0, 0x10a0, 0x10a0, 0x00a0, 0x30e2, 0x00a0, 0x01a0,
64 0x00a0, 0x10a0, 0x30e2, 0x01a0, 0x10a0, 0x10a0, 0x10a0, 0x01a0,
65 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2,
66 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2,
67 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x00a0,
68 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x30e2,
69 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2,
70 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2,
71 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x00a0,
72 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2
73 };
74
75 #include "st.h"
76
77 #include "unicode_fold_data.c"
78
79 extern int
onigenc_unicode_mbc_case_fold(OnigEncoding enc,OnigCaseFoldType flag ARG_UNUSED,const UChar ** pp,const UChar * end,UChar * fold)80 onigenc_unicode_mbc_case_fold(OnigEncoding enc,
81 OnigCaseFoldType flag ARG_UNUSED, const UChar** pp, const UChar* end,
82 UChar* fold)
83 {
84 const struct ByUnfoldKey* buk;
85
86 OnigCodePoint code;
87 int i, len, rlen;
88 const UChar *p = *pp;
89
90 code = ONIGENC_MBC_TO_CODE(enc, p, end);
91 len = enclen(enc, p);
92 *pp += len;
93
94 #ifdef USE_UNICODE_CASE_FOLD_TURKISH_AZERI
95 if ((flag & ONIGENC_CASE_FOLD_TURKISH_AZERI) != 0) {
96 if (code == 0x0130) {
97 return ONIGENC_CODE_TO_MBC(enc, 0x0069, fold);
98 }
99 #if 0
100 if (code == 0x0049) {
101 return ONIGENC_CODE_TO_MBC(enc, 0x0131, fold);
102 }
103 #endif
104 }
105 #endif
106
107 buk = onigenc_unicode_unfold_key(code);
108 if (buk != 0) {
109 if (buk->fold_len == 1) {
110 return ONIGENC_CODE_TO_MBC(enc, *FOLDS1_FOLD(buk->index), fold);
111 }
112 else {
113 OnigCodePoint* addr;
114
115 FOLDS_FOLD_ADDR_BUK(buk, addr);
116 rlen = 0;
117 for (i = 0; i < buk->fold_len; i++) {
118 OnigCodePoint c = addr[i];
119 len = ONIGENC_CODE_TO_MBC(enc, c, fold);
120 fold += len;
121 rlen += len;
122 }
123 return rlen;
124 }
125 }
126
127 for (i = 0; i < len; i++) {
128 *fold++ = *p++;
129 }
130 return len;
131 }
132
133 static int
apply_case_fold1(int from,int to,OnigApplyAllCaseFoldFunc f,void * arg)134 apply_case_fold1(int from, int to, OnigApplyAllCaseFoldFunc f, void* arg)
135 {
136 int i, j, k, n, r;
137
138 for (i = from; i < to; ) {
139 OnigCodePoint fold = *FOLDS1_FOLD(i);
140 n = FOLDS1_UNFOLDS_NUM(i);
141 for (j = 0; j < n; j++) {
142 OnigCodePoint unfold = FOLDS1_UNFOLDS(i)[j];
143
144 r = (*f)(fold, &unfold, 1, arg);
145 if (r != 0) return r;
146 r = (*f)(unfold, &fold, 1, arg);
147 if (r != 0) return r;
148
149 for (k = 0; k < j; k++) {
150 OnigCodePoint unfold2 = FOLDS1_UNFOLDS(i)[k];
151 r = (*f)(unfold, &unfold2, 1, arg);
152 if (r != 0) return r;
153 r = (*f)(unfold2, &unfold, 1, arg);
154 if (r != 0) return r;
155 }
156 }
157
158 i = FOLDS1_NEXT_INDEX(i);
159 }
160
161 return 0;
162 }
163
164 static int
apply_case_fold2(int from,int to,OnigApplyAllCaseFoldFunc f,void * arg)165 apply_case_fold2(int from, int to, OnigApplyAllCaseFoldFunc f, void* arg)
166 {
167 int i, j, k, n, r;
168
169 for (i = from; i < to; ) {
170 OnigCodePoint* fold = FOLDS2_FOLD(i);
171 n = FOLDS2_UNFOLDS_NUM(i);
172 for (j = 0; j < n; j++) {
173 OnigCodePoint unfold = FOLDS2_UNFOLDS(i)[j];
174
175 r = (*f)(unfold, fold, 2, arg);
176 if (r != 0) return r;
177
178 for (k = 0; k < j; k++) {
179 OnigCodePoint unfold2 = FOLDS2_UNFOLDS(i)[k];
180 r = (*f)(unfold, &unfold2, 1, arg);
181 if (r != 0) return r;
182 r = (*f)(unfold2, &unfold, 1, arg);
183 if (r != 0) return r;
184 }
185 }
186
187 i = FOLDS2_NEXT_INDEX(i);
188 }
189
190 return 0;
191 }
192
193 static int
apply_case_fold3(int from,int to,OnigApplyAllCaseFoldFunc f,void * arg)194 apply_case_fold3(int from, int to, OnigApplyAllCaseFoldFunc f, void* arg)
195 {
196 int i, j, k, n, r;
197
198 for (i = from; i < to; ) {
199 OnigCodePoint* fold = FOLDS3_FOLD(i);
200 n = FOLDS3_UNFOLDS_NUM(i);
201 for (j = 0; j < n; j++) {
202 OnigCodePoint unfold = FOLDS3_UNFOLDS(i)[j];
203
204 r = (*f)(unfold, fold, 3, arg);
205 if (r != 0) return r;
206
207 for (k = 0; k < j; k++) {
208 OnigCodePoint unfold2 = FOLDS3_UNFOLDS(i)[k];
209 r = (*f)(unfold, &unfold2, 1, arg);
210 if (r != 0) return r;
211 r = (*f)(unfold2, &unfold, 1, arg);
212 if (r != 0) return r;
213 }
214 }
215
216 i = FOLDS3_NEXT_INDEX(i);
217 }
218
219 return 0;
220 }
221
222 extern int
onigenc_unicode_apply_all_case_fold(OnigCaseFoldType flag,OnigApplyAllCaseFoldFunc f,void * arg)223 onigenc_unicode_apply_all_case_fold(OnigCaseFoldType flag,
224 OnigApplyAllCaseFoldFunc f, void* arg)
225 {
226 int r;
227
228 r = apply_case_fold1(0, FOLDS1_NORMAL_END_INDEX, f, arg);
229 if (r != 0) return r;
230
231 #ifdef USE_UNICODE_CASE_FOLD_TURKISH_AZERI
232 if ((flag & ONIGENC_CASE_FOLD_TURKISH_AZERI) != 0) {
233 code = 0x0131;
234 r = (*f)(0x0049, &code, 1, arg);
235 if (r != 0) return r;
236 code = 0x0049;
237 r = (*f)(0x0131, &code, 1, arg);
238 if (r != 0) return r;
239
240 code = 0x0130;
241 r = (*f)(0x0069, &code, 1, arg);
242 if (r != 0) return r;
243 code = 0x0069;
244 r = (*f)(0x0130, &code, 1, arg);
245 if (r != 0) return r;
246 }
247 else {
248 #endif
249 r = apply_case_fold1(FOLDS1_NORMAL_END_INDEX, FOLDS1_END_INDEX, f, arg);
250 if (r != 0) return r;
251 #ifdef USE_UNICODE_CASE_FOLD_TURKISH_AZERI
252 }
253 #endif
254
255 if ((flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) == 0)
256 return 0;
257
258 r = apply_case_fold2(0, FOLDS2_NORMAL_END_INDEX, f, arg);
259 if (r != 0) return r;
260
261 #ifdef USE_UNICODE_CASE_FOLD_TURKISH_AZERI
262 if ((flag & ONIGENC_CASE_FOLD_TURKISH_AZERI) == 0) {
263 #endif
264 r = apply_case_fold2(FOLDS2_NORMAL_END_INDEX, FOLDS2_END_INDEX, f, arg);
265 if (r != 0) return r;
266 #ifdef USE_UNICODE_CASE_FOLD_TURKISH_AZERI
267 }
268 #endif
269
270 r = apply_case_fold3(0, FOLDS3_NORMAL_END_INDEX, f, arg);
271 if (r != 0) return r;
272
273 return 0;
274 }
275
276 extern int
onigenc_unicode_get_case_fold_codes_by_str(OnigEncoding enc,OnigCaseFoldType flag,const OnigUChar * p,const OnigUChar * end,OnigCaseFoldCodeItem items[])277 onigenc_unicode_get_case_fold_codes_by_str(OnigEncoding enc,
278 OnigCaseFoldType flag, const OnigUChar* p, const OnigUChar* end,
279 OnigCaseFoldCodeItem items[])
280 {
281 int n, m, i, j, k, len;
282 OnigCodePoint code, codes[3];
283 const struct ByUnfoldKey* buk;
284
285 n = 0;
286
287 code = ONIGENC_MBC_TO_CODE(enc, p, end);
288 len = enclen(enc, p);
289
290 #ifdef USE_UNICODE_CASE_FOLD_TURKISH_AZERI
291 if ((flag & ONIGENC_CASE_FOLD_TURKISH_AZERI) != 0) {
292 if (code == 0x0049) {
293 items[0].byte_len = len;
294 items[0].code_len = 1;
295 items[0].code[0] = 0x0131;
296 return 1;
297 }
298 else if (code == 0x0130) {
299 items[0].byte_len = len;
300 items[0].code_len = 1;
301 items[0].code[0] = 0x0069;
302 return 1;
303 }
304 else if (code == 0x0131) {
305 items[0].byte_len = len;
306 items[0].code_len = 1;
307 items[0].code[0] = 0x0049;
308 return 1;
309 }
310 else if (code == 0x0069) {
311 items[0].byte_len = len;
312 items[0].code_len = 1;
313 items[0].code[0] = 0x0130;
314 return 1;
315 }
316 }
317 #endif
318
319 buk = onigenc_unicode_unfold_key(code);
320 if (buk != 0) {
321 if (buk->fold_len == 1) {
322 int un;
323 items[0].byte_len = len;
324 items[0].code_len = 1;
325 items[0].code[0] = *FOLDS1_FOLD(buk->index);
326 n++;
327
328 un = FOLDS1_UNFOLDS_NUM(buk->index);
329 for (i = 0; i < un; i++) {
330 OnigCodePoint unfold = FOLDS1_UNFOLDS(buk->index)[i];
331 if (unfold != code) {
332 items[n].byte_len = len;
333 items[n].code_len = 1;
334 items[n].code[0] = unfold;
335 n++;
336 }
337 }
338 code = items[0].code[0]; /* for multi-code to unfold search. */
339 }
340 else if ((flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
341 OnigCodePoint cs[3][4];
342 int fn, ncs[3];
343
344 if (buk->fold_len == 2) {
345 m = FOLDS2_UNFOLDS_NUM(buk->index);
346 for (i = 0; i < m; i++) {
347 OnigCodePoint unfold = FOLDS2_UNFOLDS(buk->index)[i];
348 if (unfold == code) continue;
349
350 items[n].byte_len = len;
351 items[n].code_len = 1;
352 items[n].code[0] = unfold;
353 n++;
354 }
355
356 for (fn = 0; fn < 2; fn++) {
357 int index;
358 cs[fn][0] = FOLDS2_FOLD(buk->index)[fn];
359 index = onigenc_unicode_fold1_key(&cs[fn][0]);
360 if (index >= 0) {
361 int m = FOLDS1_UNFOLDS_NUM(index);
362 for (i = 0; i < m; i++) {
363 cs[fn][i+1] = FOLDS1_UNFOLDS(index)[i];
364 }
365 ncs[fn] = m + 1;
366 }
367 else
368 ncs[fn] = 1;
369 }
370
371 for (i = 0; i < ncs[0]; i++) {
372 for (j = 0; j < ncs[1]; j++) {
373 items[n].byte_len = len;
374 items[n].code_len = 2;
375 items[n].code[0] = cs[0][i];
376 items[n].code[1] = cs[1][j];
377 n++;
378 }
379 }
380 }
381 else { /* fold_len == 3 */
382 m = FOLDS3_UNFOLDS_NUM(buk->index);
383 for (i = 0; i < m; i++) {
384 OnigCodePoint unfold = FOLDS3_UNFOLDS(buk->index)[i];
385 if (unfold == code) continue;
386
387 items[n].byte_len = len;
388 items[n].code_len = 1;
389 items[n].code[0] = unfold;
390 n++;
391 }
392
393 for (fn = 0; fn < 3; fn++) {
394 int index;
395 cs[fn][0] = FOLDS3_FOLD(buk->index)[fn];
396 index = onigenc_unicode_fold1_key(&cs[fn][0]);
397 if (index >= 0) {
398 int m = FOLDS1_UNFOLDS_NUM(index);
399 for (i = 0; i < m; i++) {
400 cs[fn][i+1] = FOLDS1_UNFOLDS(index)[i];
401 }
402 ncs[fn] = m + 1;
403 }
404 else
405 ncs[fn] = 1;
406 }
407
408 for (i = 0; i < ncs[0]; i++) {
409 for (j = 0; j < ncs[1]; j++) {
410 for (k = 0; k < ncs[2]; k++) {
411 items[n].byte_len = len;
412 items[n].code_len = 3;
413 items[n].code[0] = cs[0][i];
414 items[n].code[1] = cs[1][j];
415 items[n].code[2] = cs[2][k];
416 n++;
417 }
418 }
419 }
420 }
421
422 /* multi char folded code is not head of another folded multi char */
423 return n;
424 }
425 }
426 else {
427 int index = onigenc_unicode_fold1_key(&code);
428 if (index >= 0) {
429 int m = FOLDS1_UNFOLDS_NUM(index);
430 for (i = 0; i < m; i++) {
431 items[n].byte_len = len;
432 items[n].code_len = 1;
433 items[n].code[0] = FOLDS1_UNFOLDS(index)[i];
434 n++;
435 }
436 }
437 }
438
439 if ((flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) == 0)
440 return n;
441
442 p += len;
443 if (p < end) {
444 int clen;
445 int index;
446
447 codes[0] = code;
448 code = ONIGENC_MBC_TO_CODE(enc, p, end);
449
450 buk = onigenc_unicode_unfold_key(code);
451 if (buk != 0 && buk->fold_len == 1) {
452 codes[1] = *FOLDS1_FOLD(buk->index);
453 }
454 else
455 codes[1] = code;
456
457 clen = enclen(enc, p);
458 len += clen;
459
460 index = onigenc_unicode_fold2_key(codes);
461 if (index >= 0) {
462 m = FOLDS2_UNFOLDS_NUM(index);
463 for (i = 0; i < m; i++) {
464 items[n].byte_len = len;
465 items[n].code_len = 1;
466 items[n].code[0] = FOLDS2_UNFOLDS(index)[i];
467 n++;
468 }
469 }
470
471 p += clen;
472 if (p < end) {
473 code = ONIGENC_MBC_TO_CODE(enc, p, end);
474 buk = onigenc_unicode_unfold_key(code);
475 if (buk != 0 && buk->fold_len == 1) {
476 codes[2] = *FOLDS1_FOLD(buk->index);
477 }
478 else
479 codes[2] = code;
480
481 clen = enclen(enc, p);
482 len += clen;
483
484 index = onigenc_unicode_fold3_key(codes);
485 if (index >= 0) {
486 m = FOLDS3_UNFOLDS_NUM(index);
487 for (i = 0; i < m; i++) {
488 items[n].byte_len = len;
489 items[n].code_len = 1;
490 items[n].code[0] = FOLDS3_UNFOLDS(index)[i];
491 n++;
492 }
493 }
494 }
495 }
496
497 return n;
498 }
499
500 #ifdef USE_UNICODE_PROPERTIES
501 #include "unicode_property_data.c"
502 #else
503 #include "unicode_property_data_posix.c"
504 #endif
505
506
507 #ifdef USE_UNICODE_EXTENDED_GRAPHEME_CLUSTER
508
509 enum EGCB_BREAK_TYPE {
510 EGCB_NOT_BREAK = 0,
511 EGCB_BREAK = 1,
512 EGCB_BREAK_UNDEF_GB11 = 2,
513 EGCB_BREAK_UNDEF_RI_RI = 3
514 };
515
516 enum EGCB_TYPE {
517 EGCB_Other = 0,
518 EGCB_CR = 1,
519 EGCB_LF = 2,
520 EGCB_Control = 3,
521 EGCB_Extend = 4,
522 EGCB_Prepend = 5,
523 EGCB_Regional_Indicator = 6,
524 EGCB_SpacingMark = 7,
525 EGCB_ZWJ = 8,
526 #if 0
527 /* obsoleted */
528 EGCB_E_Base = 9,
529 EGCB_E_Base_GAZ = 10,
530 EGCB_E_Modifier = 11,
531 EGCB_Glue_After_Zwj = 12,
532 #endif
533 EGCB_L = 13,
534 EGCB_LV = 14,
535 EGCB_LVT = 15,
536 EGCB_T = 16,
537 EGCB_V = 17
538 };
539
540 typedef struct {
541 OnigCodePoint start;
542 OnigCodePoint end;
543 enum EGCB_TYPE type;
544 } EGCB_RANGE_TYPE;
545
546 #include "unicode_egcb_data.c"
547
548 static enum EGCB_TYPE
egcb_get_type(OnigCodePoint code)549 egcb_get_type(OnigCodePoint code)
550 {
551 OnigCodePoint low, high, x;
552 enum EGCB_TYPE type;
553
554 for (low = 0, high = (OnigCodePoint )EGCB_RANGE_NUM; low < high; ) {
555 x = (low + high) >> 1;
556 if (code > EGCB_RANGES[x].end)
557 low = x + 1;
558 else
559 high = x;
560 }
561
562 type = (low < (OnigCodePoint )EGCB_RANGE_NUM &&
563 code >= EGCB_RANGES[low].start) ?
564 EGCB_RANGES[low].type : EGCB_Other;
565
566 return type;
567 }
568
569 #define IS_CONTROL_CR_LF(code) ((code) <= EGCB_Control && (code) >= EGCB_CR)
570 #define IS_HANGUL(code) ((code) >= EGCB_L)
571
572 /* GB1 and GB2 are outside of this function. */
573 static enum EGCB_BREAK_TYPE
unicode_egcb_is_break_2code(OnigCodePoint from_code,OnigCodePoint to_code)574 unicode_egcb_is_break_2code(OnigCodePoint from_code, OnigCodePoint to_code)
575 {
576 enum EGCB_TYPE from;
577 enum EGCB_TYPE to;
578
579 from = egcb_get_type(from_code);
580 to = egcb_get_type(to_code);
581
582 /* short cut */
583 if (from == 0 && to == 0) goto GB999;
584
585 /* GB3 */
586 if (from == EGCB_CR && to == EGCB_LF) return EGCB_NOT_BREAK;
587 /* GB4 */
588 if (IS_CONTROL_CR_LF(from)) return EGCB_BREAK;
589 /* GB5 */
590 if (IS_CONTROL_CR_LF(to)) return EGCB_BREAK;
591
592 if (IS_HANGUL(from) && IS_HANGUL(to)) {
593 /* GB6 */
594 if (from == EGCB_L && to != EGCB_T) return EGCB_NOT_BREAK;
595 /* GB7 */
596 if ((from == EGCB_LV || from == EGCB_V)
597 && (to == EGCB_V || to == EGCB_T)) return EGCB_NOT_BREAK;
598
599 /* GB8 */
600 if ((to == EGCB_T) && (from == EGCB_LVT || from == EGCB_T))
601 return EGCB_NOT_BREAK;
602
603 goto GB999;
604 }
605
606 /* GB9 */
607 if (to == EGCB_Extend || to == EGCB_ZWJ) return EGCB_NOT_BREAK;
608
609 /* GB9a */
610 if (to == EGCB_SpacingMark) return EGCB_NOT_BREAK;
611 /* GB9b */
612 if (from == EGCB_Prepend) return EGCB_NOT_BREAK;
613
614 /* GB10 removed */
615
616 /* GB11 */
617 if (from == EGCB_ZWJ) {
618 if (onigenc_unicode_is_code_ctype(to_code, PROP_INDEX_EXTENDEDPICTOGRAPHIC))
619 return EGCB_BREAK_UNDEF_GB11;
620
621 goto GB999;
622 }
623
624 /* GB12, GB13 */
625 if (from == EGCB_Regional_Indicator && to == EGCB_Regional_Indicator) {
626 return EGCB_BREAK_UNDEF_RI_RI;
627 }
628
629 GB999:
630 return EGCB_BREAK;
631 }
632
633 #endif /* USE_UNICODE_EXTENDED_GRAPHEME_CLUSTER */
634
635 extern int
onigenc_egcb_is_break_position(OnigEncoding enc,UChar * p,UChar * prev,const UChar * start,const UChar * end)636 onigenc_egcb_is_break_position(OnigEncoding enc, UChar* p, UChar* prev,
637 const UChar* start, const UChar* end)
638 {
639 OnigCodePoint from;
640 OnigCodePoint to;
641 #ifdef USE_UNICODE_EXTENDED_GRAPHEME_CLUSTER
642 enum EGCB_BREAK_TYPE btype;
643 enum EGCB_TYPE type;
644 #endif
645
646 /* GB1 and GB2 */
647 if (p == start) return 1;
648 if (p == end) return 1;
649
650 if (IS_NULL(prev)) {
651 prev = onigenc_get_prev_char_head(enc, start, p);
652 if (IS_NULL(prev)) return 1;
653 }
654
655 from = ONIGENC_MBC_TO_CODE(enc, prev, end);
656 to = ONIGENC_MBC_TO_CODE(enc, p, end);
657
658 #ifdef USE_UNICODE_EXTENDED_GRAPHEME_CLUSTER
659 if (! ONIGENC_IS_UNICODE_ENCODING(enc)) {
660 return from != 0x000d || to != 0x000a;
661 }
662
663 btype = unicode_egcb_is_break_2code(from, to);
664 switch (btype) {
665 case EGCB_NOT_BREAK:
666 return 0;
667 break;
668 case EGCB_BREAK:
669 return 1;
670 break;
671
672 case EGCB_BREAK_UNDEF_GB11:
673 while ((prev = onigenc_get_prev_char_head(enc, start, prev)) != NULL) {
674 from = ONIGENC_MBC_TO_CODE(enc, prev, end);
675 if (onigenc_unicode_is_code_ctype(from, PROP_INDEX_EXTENDEDPICTOGRAPHIC))
676 return 0;
677
678 type = egcb_get_type(from);
679 if (type != EGCB_Extend)
680 break;
681 }
682 break;
683
684 case EGCB_BREAK_UNDEF_RI_RI:
685 {
686 int n = 0;
687 while ((prev = onigenc_get_prev_char_head(enc, start, prev)) != NULL) {
688 from = ONIGENC_MBC_TO_CODE(enc, prev, end);
689 type = egcb_get_type(from);
690 if (type != EGCB_Regional_Indicator)
691 break;
692
693 n++;
694 }
695 if ((n % 2) == 0) return 0;
696 }
697 break;
698 }
699
700 return 1;
701
702 #else
703 return from != 0x000d || to != 0x000a;
704 #endif /* USE_UNICODE_EXTENDED_GRAPHEME_CLUSTER */
705 }
706
707
708 #define USER_DEFINED_PROPERTY_MAX_NUM 20
709
710 typedef struct {
711 int ctype;
712 OnigCodePoint* ranges;
713 } UserDefinedPropertyValue;
714
715 static int UserDefinedPropertyNum;
716 static UserDefinedPropertyValue
717 UserDefinedPropertyRanges[USER_DEFINED_PROPERTY_MAX_NUM];
718 static st_table* UserDefinedPropertyTable;
719
720 extern int
onig_unicode_define_user_property(const char * name,OnigCodePoint * ranges)721 onig_unicode_define_user_property(const char* name, OnigCodePoint* ranges)
722 {
723 UserDefinedPropertyValue* e;
724 int r;
725 int i;
726 int n;
727 int len;
728 int c;
729 char* s;
730 UChar* uname;
731
732 if (UserDefinedPropertyNum >= USER_DEFINED_PROPERTY_MAX_NUM)
733 return ONIGERR_TOO_MANY_USER_DEFINED_OBJECTS;
734
735 len = (int )strlen(name);
736 if (len >= PROPERTY_NAME_MAX_SIZE)
737 return ONIGERR_TOO_LONG_PROPERTY_NAME;
738
739 s = (char* )xmalloc(len + 1);
740 if (s == 0)
741 return ONIGERR_MEMORY;
742
743 uname = (UChar* )name;
744 n = 0;
745 for (i = 0; i < len; i++) {
746 c = uname[i];
747 if (c < 0x20 || c >= 0x80) {
748 xfree(s);
749 return ONIGERR_INVALID_CHAR_PROPERTY_NAME;
750 }
751
752 if (c != ' ' && c != '-' && c != '_') {
753 s[n] = c;
754 n++;
755 }
756 }
757 s[n] = '\0';
758
759 if (UserDefinedPropertyTable == 0) {
760 UserDefinedPropertyTable = onig_st_init_strend_table_with_size(10);
761 if (IS_NULL(UserDefinedPropertyTable)) {
762 xfree(s);
763 return ONIGERR_MEMORY;
764 }
765 }
766
767 e = UserDefinedPropertyRanges + UserDefinedPropertyNum;
768 e->ctype = CODE_RANGES_NUM + UserDefinedPropertyNum;
769 e->ranges = ranges;
770 r = onig_st_insert_strend(UserDefinedPropertyTable,
771 (const UChar* )s, (const UChar* )s + n,
772 (hash_data_type )((void* )e));
773 if (r < 0) return r;
774
775 UserDefinedPropertyNum++;
776 return 0;
777 }
778
779 extern int
onigenc_unicode_is_code_ctype(OnigCodePoint code,unsigned int ctype)780 onigenc_unicode_is_code_ctype(OnigCodePoint code, unsigned int ctype)
781 {
782 if (
783 #ifdef USE_UNICODE_PROPERTIES
784 ctype <= ONIGENC_MAX_STD_CTYPE &&
785 #endif
786 code < 256) {
787 return ONIGENC_IS_UNICODE_ISO_8859_1_CTYPE(code, ctype);
788 }
789
790 if (ctype >= CODE_RANGES_NUM) {
791 int index = ctype - CODE_RANGES_NUM;
792 if (index < UserDefinedPropertyNum)
793 return onig_is_in_code_range((UChar* )UserDefinedPropertyRanges[index].ranges, code);
794 else
795 return ONIGERR_TYPE_BUG;
796 }
797
798 return onig_is_in_code_range((UChar* )CodeRanges[ctype], code);
799 }
800
801
802 extern int
onigenc_unicode_ctype_code_range(OnigCtype ctype,const OnigCodePoint * ranges[])803 onigenc_unicode_ctype_code_range(OnigCtype ctype, const OnigCodePoint* ranges[])
804 {
805 if (ctype >= CODE_RANGES_NUM) {
806 int index = ctype - CODE_RANGES_NUM;
807 if (index < UserDefinedPropertyNum) {
808 *ranges = UserDefinedPropertyRanges[index].ranges;
809 return 0;
810 }
811 else
812 return ONIGERR_TYPE_BUG;
813 }
814
815 *ranges = CodeRanges[ctype];
816 return 0;
817 }
818
819 extern int
onigenc_utf16_32_get_ctype_code_range(OnigCtype ctype,OnigCodePoint * sb_out,const OnigCodePoint * ranges[])820 onigenc_utf16_32_get_ctype_code_range(OnigCtype ctype, OnigCodePoint* sb_out,
821 const OnigCodePoint* ranges[])
822 {
823 *sb_out = 0x00;
824 return onigenc_unicode_ctype_code_range(ctype, ranges);
825 }
826
827 extern int
onigenc_unicode_property_name_to_ctype(OnigEncoding enc,UChar * name,UChar * end)828 onigenc_unicode_property_name_to_ctype(OnigEncoding enc, UChar* name, UChar* end)
829 {
830 int len;
831 UChar *p;
832 OnigCodePoint code;
833 const struct PoolPropertyNameCtype* pc;
834 char buf[PROPERTY_NAME_MAX_SIZE];
835
836 p = name;
837 len = 0;
838 while (p < end) {
839 code = ONIGENC_MBC_TO_CODE(enc, p, end);
840 if (code >= 0x80)
841 return ONIGERR_INVALID_CHAR_PROPERTY_NAME;
842
843 if (code != ' ' && code != '-' && code != '_') {
844 buf[len++] = (char )code;
845 if (len >= PROPERTY_NAME_MAX_SIZE)
846 return ONIGERR_INVALID_CHAR_PROPERTY_NAME;
847 }
848
849 p += enclen(enc, p);
850 }
851
852 buf[len] = 0;
853
854 if (UserDefinedPropertyTable != 0) {
855 UserDefinedPropertyValue* e;
856 e = (UserDefinedPropertyValue* )NULL;
857 onig_st_lookup_strend(UserDefinedPropertyTable,
858 (const UChar* )buf, (const UChar* )buf + len,
859 (hash_data_type* )((void* )(&e)));
860 if (e != 0) {
861 return e->ctype;
862 }
863 }
864
865 pc = unicode_lookup_property_name(buf, len);
866 if (pc != 0) {
867 /* fprintf(stderr, "LOOKUP: %s: %d\n", buf, pc->ctype); */
868 #ifndef USE_UNICODE_PROPERTIES
869 if (pc->ctype > ONIGENC_MAX_STD_CTYPE)
870 return ONIGERR_INVALID_CHAR_PROPERTY_NAME;
871 #endif
872
873 return (int )pc->ctype;
874 }
875
876 return ONIGERR_INVALID_CHAR_PROPERTY_NAME;
877 }
878