]> git.proxmox.com Git - mirror_edk2.git/blob - MdeModulePkg/Universal/RegularExpressionDxe/Oniguruma/regparse.c
projects / mirror_edk2.git / blob
? search:


01ac2b3136d4364f0c067f48904847f344f981de
[mirror_edk2.git] / MdeModulePkg / Universal / RegularExpressionDxe / Oniguruma / regparse.c
1 /**********************************************************************
2 regparse.c - Oniguruma (regular expression library)
3 **********************************************************************/
4 /*-
5 * Copyright (c) 2002-2008 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
6 * All rights reserved.
7 *
8 * (C) Copyright 2015 Hewlett Packard Enterprise Development LP<BR>
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 */
31
32 #include "regparse.h"
33 #include "st.h"
34
35 #define WARN_BUFSIZE 256
36
37 #define CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS
38
39
40 OnigSyntaxType OnigSyntaxRuby = {
41 (( SYN_GNU_REGEX_OP | ONIG_SYN_OP_QMARK_NON_GREEDY |
42 ONIG_SYN_OP_ESC_OCTAL3 | ONIG_SYN_OP_ESC_X_HEX2 |
43 ONIG_SYN_OP_ESC_X_BRACE_HEX8 | ONIG_SYN_OP_ESC_CONTROL_CHARS |
44 ONIG_SYN_OP_ESC_C_CONTROL )
45 & ~ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END )
46 , ( ONIG_SYN_OP2_QMARK_GROUP_EFFECT |
47 ONIG_SYN_OP2_OPTION_RUBY |
48 ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP | ONIG_SYN_OP2_ESC_K_NAMED_BACKREF |
49 ONIG_SYN_OP2_ESC_G_SUBEXP_CALL |
50 ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY |
51 ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT |
52 ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT |
53 ONIG_SYN_OP2_CCLASS_SET_OP | ONIG_SYN_OP2_ESC_CAPITAL_C_BAR_CONTROL |
54 ONIG_SYN_OP2_ESC_CAPITAL_M_BAR_META | ONIG_SYN_OP2_ESC_V_VTAB |
55 ONIG_SYN_OP2_ESC_H_XDIGIT )
56 , ( SYN_GNU_REGEX_BV |
57 ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV |
58 ONIG_SYN_DIFFERENT_LEN_ALT_LOOK_BEHIND |
59 ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP |
60 ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME |
61 ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY |
62 ONIG_SYN_WARN_CC_OP_NOT_ESCAPED |
63 ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT )
64 , ONIG_OPTION_NONE
65 ,
66 {
67 (OnigCodePoint )'\\' /* esc */
68 , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anychar '.' */
69 , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anytime '*' */
70 , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* zero or one time '?' */
71 , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* one or more time '+' */
72 , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anychar anytime */
73 }
74 };
75
76 OnigSyntaxType* OnigDefaultSyntax = ONIG_SYNTAX_RUBY;
77
78 extern void onig_null_warn(const char* s ARG_UNUSED) { }
79
80 #ifdef DEFAULT_WARN_FUNCTION
81 static OnigWarnFunc onig_warn = (OnigWarnFunc )DEFAULT_WARN_FUNCTION;
82 #else
83 static OnigWarnFunc onig_warn = onig_null_warn;
84 #endif
85
86 #ifdef DEFAULT_VERB_WARN_FUNCTION
87 static OnigWarnFunc onig_verb_warn = (OnigWarnFunc )DEFAULT_VERB_WARN_FUNCTION;
88 #else
89 static OnigWarnFunc onig_verb_warn = onig_null_warn;
90 #endif
91
92 extern void onig_set_warn_func(OnigWarnFunc f)
93 {
94 onig_warn = f;
95 }
96
97 extern void onig_set_verb_warn_func(OnigWarnFunc f)
98 {
99 onig_verb_warn = f;
100 }
101
102 static void
103 bbuf_free(BBuf* bbuf)
104 {
105 if (IS_NOT_NULL(bbuf)) {
106 if (IS_NOT_NULL(bbuf->p)) xfree(bbuf->p);
107 xfree(bbuf);
108 }
109 }
110
111 static int
112 bbuf_clone(BBuf** rto, BBuf* from)
113 {
114 int r;
115 BBuf *to;
116
117 *rto = to = (BBuf* )xmalloc(sizeof(BBuf));
118 CHECK_NULL_RETURN_MEMERR(to);
119 r = BBUF_INIT(to, from->alloc);
120 if (r != 0) return r;
121 to->used = from->used;
122 xmemcpy(to->p, from->p, from->used);
123 return 0;
124 }
125
126 #define BACKREF_REL_TO_ABS(rel_no, env) \
127 ((env)->num_mem + 1 + (rel_no))
128
129 #define ONOFF(v,f,negative) (negative) ? ((v) &= ~(f)) : ((v) |= (f))
130
131 #define MBCODE_START_POS(enc) \
132 (OnigCodePoint )(ONIGENC_MBC_MINLEN(enc) > 1 ? 0 : 0x80)
133
134 #define SET_ALL_MULTI_BYTE_RANGE(enc, pbuf) \
135 add_code_range_to_buf(pbuf, MBCODE_START_POS(enc), ~((OnigCodePoint )0))
136
137 #define ADD_ALL_MULTI_BYTE_RANGE(enc, mbuf) do {\
138 if (! ONIGENC_IS_SINGLEBYTE(enc)) {\
139 r = SET_ALL_MULTI_BYTE_RANGE(enc, &(mbuf));\
140 if (r) return r;\
141 }\
142 } while (0)
143
144
145 #define BITSET_IS_EMPTY(bs,empty) do {\
146 int i;\
147 empty = 1;\
148 for (i = 0; i < (int )BITSET_SIZE; i++) {\
149 if ((bs)[i] != 0) {\
150 empty = 0; break;\
151 }\
152 }\
153 } while (0)
154
155 static void
156 bitset_set_range(BitSetRef bs, int from, int to)
157 {
158 int i;
159 for (i = from; i <= to && i < SINGLE_BYTE_SIZE; i++) {
160 BITSET_SET_BIT(bs, i);
161 }
162 }
163
164 #if 0
165 static void
166 bitset_set_all(BitSetRef bs)
167 {
168 int i;
169 for (i = 0; i < BITSET_SIZE; i++) { bs[i] = ~((Bits )0); }
170 }
171 #endif
172
173 static void
174 bitset_invert(BitSetRef bs)
175 {
176 int i;
177 for (i = 0; i < (int )BITSET_SIZE; i++) { bs[i] = ~(bs[i]); }
178 }
179
180 static void
181 bitset_invert_to(BitSetRef from, BitSetRef to)
182 {
183 int i;
184 for (i = 0; i < (int )BITSET_SIZE; i++) { to[i] = ~(from[i]); }
185 }
186
187 static void
188 bitset_and(BitSetRef dest, BitSetRef bs)
189 {
190 int i;
191 for (i = 0; i < (int )BITSET_SIZE; i++) { dest[i] &= bs[i]; }
192 }
193
194 static void
195 bitset_or(BitSetRef dest, BitSetRef bs)
196 {
197 int i;
198 for (i = 0; i < (int )BITSET_SIZE; i++) { dest[i] |= bs[i]; }
199 }
200
201 static void
202 bitset_copy(BitSetRef dest, BitSetRef bs)
203 {
204 int i;
205 for (i = 0; i < (int )BITSET_SIZE; i++) { dest[i] = bs[i]; }
206 }
207
208 extern int
209 onig_strncmp(const UChar* s1, const UChar* s2, int n)
210 {
211 int x;
212
213 while (n-- > 0) {
214 x = *s2++ - *s1++;
215 if (x) return x;
216 }
217 return 0;
218 }
219
220 extern void
221 onig_strcpy(UChar* dest, const UChar* src, const UChar* end)
222 {
223 int len = (int)(end - src);
224 if (len > 0) {
225 xmemcpy(dest, src, len);
226 dest[len] = (UChar )0;
227 }
228 }
229
230 #ifdef USE_NAMED_GROUP
231 static UChar*
232 strdup_with_null(OnigEncoding enc, UChar* s, UChar* end)
233 {
234 int slen, term_len, i;
235 UChar *r;
236
237 slen = (int)(end - s);
238 term_len = ONIGENC_MBC_MINLEN(enc);
239
240 r = (UChar* )xmalloc(slen + term_len);
241 CHECK_NULL_RETURN(r);
242 xmemcpy(r, s, slen);
243
244 for (i = 0; i < term_len; i++)
245 r[slen + i] = (UChar )0;
246
247 return r;
248 }
249 #endif
250
251 /* scan pattern methods */
252 #define PEND_VALUE 0
253
254 #define PFETCH_READY UChar* pfetch_prev
255 #define PEND (p < end ? 0 : 1)
256 #define PUNFETCH p = pfetch_prev
257 #define PINC do { \
258 pfetch_prev = p; \
259 p += ONIGENC_MBC_ENC_LEN(enc, p); \
260 } while (0)
261 #define PFETCH(c) do { \
262 c = ONIGENC_MBC_TO_CODE(enc, p, end); \
263 pfetch_prev = p; \
264 p += ONIGENC_MBC_ENC_LEN(enc, p); \
265 } while (0)
266
267 #define PINC_S do { \
268 p += ONIGENC_MBC_ENC_LEN(enc, p); \
269 } while (0)
270 #define PFETCH_S(c) do { \
271 c = ONIGENC_MBC_TO_CODE(enc, p, end); \
272 p += ONIGENC_MBC_ENC_LEN(enc, p); \
273 } while (0)
274
275 #define PPEEK (p < end ? ONIGENC_MBC_TO_CODE(enc, p, end) : PEND_VALUE)
276 #define PPEEK_IS(c) (PPEEK == (OnigCodePoint )c)
277
278 static UChar*
279 strcat_capa(UChar* dest, UChar* dest_end, const UChar* src, const UChar* src_end,
280 int capa, int oldCapa)
281 {
282 UChar* r;
283
284 if (dest)
285 r = (UChar* )xrealloc(dest, capa + 1, oldCapa);
286 else
287 r = (UChar* )xmalloc(capa + 1);
288
289 CHECK_NULL_RETURN(r);
290 onig_strcpy(r + (dest_end - dest), src, src_end);
291 return r;
292 }
293
294 /* dest on static area */
295 static UChar*
296 strcat_capa_from_static(UChar* dest, UChar* dest_end,
297 const UChar* src, const UChar* src_end, int capa)
298 {
299 UChar* r;
300
301 r = (UChar* )xmalloc(capa + 1);
302 CHECK_NULL_RETURN(r);
303 onig_strcpy(r, dest, dest_end);
304 onig_strcpy(r + (dest_end - dest), src, src_end);
305 return r;
306 }
307
308
309 #ifdef USE_ST_LIBRARY
310
311 typedef struct {
312 UChar* s;
313 UChar* end;
314 } st_str_end_key;
315
316 static int
317 str_end_cmp(st_str_end_key* x, st_str_end_key* y)
318 {
319 UChar *p, *q;
320 int c;
321
322 if ((x->end - x->s) != (y->end - y->s))
323 return 1;
324
325 p = x->s;
326 q = y->s;
327 while (p < x->end) {
328 c = (int )*p - (int )*q;
329 if (c != 0) return c;
330
331 p++; q++;
332 }
333
334 return 0;
335 }
336
337 static int
338 str_end_hash(st_str_end_key* x)
339 {
340 UChar *p;
341 int val = 0;
342
343 p = x->s;
344 while (p < x->end) {
345 val = val * 997 + (int )*p++;
346 }
347
348 return val + (val >> 5);
349 }
350
351 extern hash_table_type*
352 onig_st_init_strend_table_with_size(int size)
353 {
354 static struct st_hash_type hashType = {
355 str_end_cmp,
356 str_end_hash,
357 };
358
359 return (hash_table_type* )
360 onig_st_init_table_with_size(&hashType, size);
361 }
362
363 extern int
364 onig_st_lookup_strend(hash_table_type* table, const UChar* str_key,
365 const UChar* end_key, hash_data_type *value)
366 {
367 st_str_end_key key;
368
369 key.s = (UChar* )str_key;
370 key.end = (UChar* )end_key;
371
372 return onig_st_lookup(table, (st_data_t )(&key), value);
373 }
374
375 extern int
376 onig_st_insert_strend(hash_table_type* table, const UChar* str_key,
377 const UChar* end_key, hash_data_type value)
378 {
379 st_str_end_key* key;
380 int result;
381
382 key = (st_str_end_key* )xmalloc(sizeof(st_str_end_key));
383 key->s = (UChar* )str_key;
384 key->end = (UChar* )end_key;
385 result = onig_st_insert(table, (st_data_t )key, value);
386 if (result) {
387 xfree(key);
388 }
389 return result;
390 }
391
392 #endif /* USE_ST_LIBRARY */
393
394
395 #ifdef USE_NAMED_GROUP
396
397 #define INIT_NAME_BACKREFS_ALLOC_NUM 8
398
399 typedef struct {
400 UChar* name;
401 int name_len; /* byte length */
402 int back_num; /* number of backrefs */
403 int back_alloc;
404 int back_ref1;
405 int* back_refs;
406 } NameEntry;
407
408 #ifdef USE_ST_LIBRARY
409
410 typedef st_table NameTable;
411 typedef st_data_t HashDataType; /* 1.6 st.h doesn't define st_data_t type */
412
413 #define NAMEBUF_SIZE 24
414 #define NAMEBUF_SIZE_1 25
415
416 #ifdef ONIG_DEBUG
417 static int
418 i_print_name_entry(UChar* key, NameEntry* e, void* arg)
419 {
420 int i;
421 FILE* fp = (FILE* )arg;
422
423 fprintf(fp, "%s: ", e->name);
424 if (e->back_num == 0)
425 fputs("-", fp);
426 else if (e->back_num == 1)
427 fprintf(fp, "%d", e->back_ref1);
428 else {
429 for (i = 0; i < e->back_num; i++) {
430 if (i > 0) fprintf(fp, ", ");
431 fprintf(fp, "%d", e->back_refs[i]);
432 }
433 }
434 fputs("\n", fp);
435 return ST_CONTINUE;
436 }
437
438 extern int
439 onig_print_names(FILE* fp, regex_t* reg)
440 {
441 NameTable* t = (NameTable* )reg->name_table;
442
443 if (IS_NOT_NULL(t)) {
444 fprintf(fp, "name table\n");
445 onig_st_foreach(t, i_print_name_entry, (HashDataType )fp);
446 fputs("\n", fp);
447 }
448 return 0;
449 }
450 #endif /* ONIG_DEBUG */
451
452 static int
453 i_free_name_entry(UChar* key, NameEntry* e, void* arg ARG_UNUSED)
454 {
455 xfree(e->name);
456 if (IS_NOT_NULL(e->back_refs)) xfree(e->back_refs);
457 xfree(key);
458 xfree(e);
459 return ST_DELETE;
460 }
461
462 static int
463 names_clear(regex_t* reg)
464 {
465 NameTable* t = (NameTable* )reg->name_table;
466
467 if (IS_NOT_NULL(t)) {
468 onig_st_foreach(t, i_free_name_entry, 0);
469 }
470 return 0;
471 }
472
473 extern int
474 onig_names_free(regex_t* reg)
475 {
476 int r;
477 NameTable* t;
478
479 r = names_clear(reg);
480 if (r) return r;
481
482 t = (NameTable* )reg->name_table;
483 if (IS_NOT_NULL(t)) onig_st_free_table(t);
484 reg->name_table = (void* )NULL;
485 return 0;
486 }
487
488 static NameEntry*
489 name_find(regex_t* reg, const UChar* name, const UChar* name_end)
490 {
491 NameEntry* e;
492 NameTable* t = (NameTable* )reg->name_table;
493
494 e = (NameEntry* )NULL;
495 if (IS_NOT_NULL(t)) {
496 onig_st_lookup_strend(t, name, name_end, (HashDataType* )((void* )(&e)));
497 }
498 return e;
499 }
500
501 typedef struct {
502 int (*func)(const UChar*, const UChar*,int,int*,regex_t*,void*);
503 regex_t* reg;
504 void* arg;
505 int ret;
506 OnigEncoding enc;
507 } INamesArg;
508
509 static int
510 i_names(UChar* key ARG_UNUSED, NameEntry* e, INamesArg* arg)
511 {
512 int r = (*(arg->func))(e->name,
513 e->name + e->name_len,
514 e->back_num,
515 (e->back_num > 1 ? e->back_refs : &(e->back_ref1)),
516 arg->reg, arg->arg);
517 if (r != 0) {
518 arg->ret = r;
519 return ST_STOP;
520 }
521 return ST_CONTINUE;
522 }
523
524 extern int
525 onig_foreach_name(regex_t* reg,
526 int (*func)(const UChar*, const UChar*,int,int*,regex_t*,void*), void* arg)
527 {
528 INamesArg narg;
529 NameTable* t = (NameTable* )reg->name_table;
530
531 narg.ret = 0;
532 if (IS_NOT_NULL(t)) {
533 narg.func = func;
534 narg.reg = reg;
535 narg.arg = arg;
536 narg.enc = reg->enc; /* should be pattern encoding. */
537 onig_st_foreach(t, i_names, (HashDataType )&narg);
538 }
539 return narg.ret;
540 }
541
542 static int
543 i_renumber_name(UChar* key ARG_UNUSED, NameEntry* e, GroupNumRemap* map)
544 {
545 int i;
546
547 if (e->back_num > 1) {
548 for (i = 0; i < e->back_num; i++) {
549 e->back_refs[i] = map[e->back_refs[i]].new_val;
550 }
551 }
552 else if (e->back_num == 1) {
553 e->back_ref1 = map[e->back_ref1].new_val;
554 }
555
556 return ST_CONTINUE;
557 }
558
559 extern int
560 onig_renumber_name_table(regex_t* reg, GroupNumRemap* map)
561 {
562 NameTable* t = (NameTable* )reg->name_table;
563
564 if (IS_NOT_NULL(t)) {
565 onig_st_foreach(t, i_renumber_name, (HashDataType )map);
566 }
567 return 0;
568 }
569
570
571 extern int
572 onig_number_of_names(regex_t* reg)
573 {
574 NameTable* t = (NameTable* )reg->name_table;
575
576 if (IS_NOT_NULL(t))
577 return t->num_entries;
578 else
579 return 0;
580 }
581
582 #else /* USE_ST_LIBRARY */
583
584 #define INIT_NAMES_ALLOC_NUM 8
585
586 typedef struct {
587 NameEntry* e;
588 int num;
589 int alloc;
590 } NameTable;
591
592 #ifdef ONIG_DEBUG
593 extern int
594 onig_print_names(FILE* fp, regex_t* reg)
595 {
596 int i, j;
597 NameEntry* e;
598 NameTable* t = (NameTable* )reg->name_table;
599
600 if (IS_NOT_NULL(t) && t->num > 0) {
601 fprintf(fp, "name table\n");
602 for (i = 0; i < t->num; i++) {
603 e = &(t->e[i]);
604 fprintf(fp, "%s: ", e->name);
605 if (e->back_num == 0) {
606 fputs("-", fp);
607 }
608 else if (e->back_num == 1) {
609 fprintf(fp, "%d", e->back_ref1);
610 }
611 else {
612 for (j = 0; j < e->back_num; j++) {
613 if (j > 0) fprintf(fp, ", ");
614 fprintf(fp, "%d", e->back_refs[j]);
615 }
616 }
617 fputs("\n", fp);
618 }
619 fputs("\n", fp);
620 }
621 return 0;
622 }
623 #endif
624
625 static int
626 names_clear(regex_t* reg)
627 {
628 int i;
629 NameEntry* e;
630 NameTable* t = (NameTable* )reg->name_table;
631
632 if (IS_NOT_NULL(t)) {
633 for (i = 0; i < t->num; i++) {
634 e = &(t->e[i]);
635 if (IS_NOT_NULL(e->name)) {
636 xfree(e->name);
637 e->name = NULL;
638 e->name_len = 0;
639 e->back_num = 0;
640 e->back_alloc = 0;
641 if (IS_NOT_NULL(e->back_refs)) xfree(e->back_refs);
642 e->back_refs = (int* )NULL;
643 }
644 }
645 if (IS_NOT_NULL(t->e)) {
646 xfree(t->e);
647 t->e = NULL;
648 }
649 t->num = 0;
650 }
651 return 0;
652 }
653
654 extern int
655 onig_names_free(regex_t* reg)
656 {
657 int r;
658 NameTable* t;
659
660 r = names_clear(reg);
661 if (r) return r;
662
663 t = (NameTable* )reg->name_table;
664 if (IS_NOT_NULL(t)) xfree(t);
665 reg->name_table = NULL;
666 return 0;
667 }
668
669 static NameEntry*
670 name_find(regex_t* reg, UChar* name, UChar* name_end)
671 {
672 int i, len;
673 NameEntry* e;
674 NameTable* t = (NameTable* )reg->name_table;
675
676 if (IS_NOT_NULL(t)) {
677 len = name_end - name;
678 for (i = 0; i < t->num; i++) {
679 e = &(t->e[i]);
680 if (len == e->name_len && onig_strncmp(name, e->name, len) == 0)
681 return e;
682 }
683 }
684 return (NameEntry* )NULL;
685 }
686
687 extern int
688 onig_foreach_name(regex_t* reg,
689 int (*func)(const UChar*, const UChar*,int,int*,regex_t*,void*), void* arg)
690 {
691 int i, r;
692 NameEntry* e;
693 NameTable* t = (NameTable* )reg->name_table;
694
695 if (IS_NOT_NULL(t)) {
696 for (i = 0; i < t->num; i++) {
697 e = &(t->e[i]);
698 r = (*func)(e->name, e->name + e->name_len, e->back_num,
699 (e->back_num > 1 ? e->back_refs : &(e->back_ref1)),
700 reg, arg);
701 if (r != 0) return r;
702 }
703 }
704 return 0;
705 }
706
707 extern int
708 onig_number_of_names(regex_t* reg)
709 {
710 NameTable* t = (NameTable* )reg->name_table;
711
712 if (IS_NOT_NULL(t))
713 return t->num;
714 else
715 return 0;
716 }
717
718 #endif /* else USE_ST_LIBRARY */
719
720 static int
721 name_add(regex_t* reg, UChar* name, UChar* name_end, int backref, ScanEnv* env)
722 {
723 int alloc;
724 NameEntry* e;
725 NameTable* t = (NameTable* )reg->name_table;
726
727 if (name_end - name <= 0)
728 return ONIGERR_EMPTY_GROUP_NAME;
729
730 e = name_find(reg, name, name_end);
731 if (IS_NULL(e)) {
732 #ifdef USE_ST_LIBRARY
733 if (IS_NULL(t)) {
734 t = onig_st_init_strend_table_with_size(5);
735 reg->name_table = (void* )t;
736 }
737 e = (NameEntry* )xmalloc(sizeof(NameEntry));
738 CHECK_NULL_RETURN_MEMERR(e);
739
740 e->name = strdup_with_null(reg->enc, name, name_end);
741 if (IS_NULL(e->name)) {
742 xfree(e); return ONIGERR_MEMORY;
743 }
744 onig_st_insert_strend(t, e->name, (e->name + (name_end - name)),
745 (HashDataType )e);
746
747 e->name_len = (int)(name_end - name);
748 e->back_num = 0;
749 e->back_alloc = 0;
750 e->back_refs = (int* )NULL;
751
752 #else
753
754 if (IS_NULL(t)) {
755 alloc = INIT_NAMES_ALLOC_NUM;
756 t = (NameTable* )xmalloc(sizeof(NameTable));
757 CHECK_NULL_RETURN_MEMERR(t);
758 t->e = NULL;
759 t->alloc = 0;
760 t->num = 0;
761
762 t->e = (NameEntry* )xmalloc(sizeof(NameEntry) * alloc);
763 if (IS_NULL(t->e)) {
764 xfree(t);
765 return ONIGERR_MEMORY;
766 }
767 t->alloc = alloc;
768 reg->name_table = t;
769 goto clear;
770 }
771 else if (t->num == t->alloc) {
772 int i;
773
774 alloc = t->alloc * 2;
775 t->e = (NameEntry* )xrealloc(t->e, sizeof(NameEntry) * alloc);
776 CHECK_NULL_RETURN_MEMERR(t->e);
777 t->alloc = alloc;
778
779 clear:
780 for (i = t->num; i < t->alloc; i++) {
781 t->e[i].name = NULL;
782 t->e[i].name_len = 0;
783 t->e[i].back_num = 0;
784 t->e[i].back_alloc = 0;
785 t->e[i].back_refs = (int* )NULL;
786 }
787 }
788 e = &(t->e[t->num]);
789 t->num++;
790 e->name = strdup_with_null(reg->enc, name, name_end);
791 if (IS_NULL(e->name)) return ONIGERR_MEMORY;
792 e->name_len = name_end - name;
793 #endif
794 }
795
796 if (e->back_num >= 1 &&
797 ! IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME)) {
798 onig_scan_env_set_error_string(env, ONIGERR_MULTIPLEX_DEFINED_NAME,
799 name, name_end);
800 return ONIGERR_MULTIPLEX_DEFINED_NAME;
801 }
802
803 e->back_num++;
804 if (e->back_num == 1) {
805 e->back_ref1 = backref;
806 }
807 else {
808 if (e->back_num == 2) {
809 alloc = INIT_NAME_BACKREFS_ALLOC_NUM;
810 e->back_refs = (int* )xmalloc(sizeof(int) * alloc);
811 CHECK_NULL_RETURN_MEMERR(e->back_refs);
812 e->back_alloc = alloc;
813 e->back_refs[0] = e->back_ref1;
814 e->back_refs[1] = backref;
815 }
816 else {
817 if (e->back_num > e->back_alloc) {
818 alloc = e->back_alloc * 2;
819 e->back_refs = (int* )xrealloc(e->back_refs, sizeof(int) * alloc, sizeof(int) * e->back_alloc);
820 CHECK_NULL_RETURN_MEMERR(e->back_refs);
821 e->back_alloc = alloc;
822 }
823 e->back_refs[e->back_num - 1] = backref;
824 }
825 }
826
827 return 0;
828 }
829
830 extern int
831 onig_name_to_group_numbers(regex_t* reg, const UChar* name,
832 const UChar* name_end, int** nums)
833 {
834 NameEntry* e = name_find(reg, name, name_end);
835
836 if (IS_NULL(e)) return ONIGERR_UNDEFINED_NAME_REFERENCE;
837
838 switch (e->back_num) {
839 case 0:
840 break;
841 case 1:
842 *nums = &(e->back_ref1);
843 break;
844 default:
845 *nums = e->back_refs;
846 break;
847 }
848 return e->back_num;
849 }
850
851 extern int
852 onig_name_to_backref_number(regex_t* reg, const UChar* name,
853 const UChar* name_end, OnigRegion *region)
854 {
855 int i, n, *nums;
856
857 n = onig_name_to_group_numbers(reg, name, name_end, &nums);
858 if (n < 0)
859 return n;
860 else if (n == 0)
861 return ONIGERR_PARSER_BUG;
862 else if (n == 1)
863 return nums[0];
864 else {
865 if (IS_NOT_NULL(region)) {
866 for (i = n - 1; i >= 0; i--) {
867 if (region->beg[nums[i]] != ONIG_REGION_NOTPOS)
868 return nums[i];
869 }
870 }
871 return nums[n - 1];
872 }
873 }
874
875 #else /* USE_NAMED_GROUP */
876
877 extern int
878 onig_name_to_group_numbers(regex_t* reg, const UChar* name,
879 const UChar* name_end, int** nums)
880 {
881 return ONIG_NO_SUPPORT_CONFIG;
882 }
883
884 extern int
885 onig_name_to_backref_number(regex_t* reg, const UChar* name,
886 const UChar* name_end, OnigRegion* region)
887 {
888 return ONIG_NO_SUPPORT_CONFIG;
889 }
890
891 extern int
892 onig_foreach_name(regex_t* reg,
893 int (*func)(const UChar*, const UChar*,int,int*,regex_t*,void*), void* arg)
894 {
895 return ONIG_NO_SUPPORT_CONFIG;
896 }
897
898 extern int
899 onig_number_of_names(regex_t* reg)
900 {
901 return 0;
902 }
903 #endif /* else USE_NAMED_GROUP */
904
905 extern int
906 onig_noname_group_capture_is_active(regex_t* reg)
907 {
908 if (ONIG_IS_OPTION_ON(reg->options, ONIG_OPTION_DONT_CAPTURE_GROUP))
909 return 0;
910
911 #ifdef USE_NAMED_GROUP
912 if (onig_number_of_names(reg) > 0 &&
913 IS_SYNTAX_BV(reg->syntax, ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP) &&
914 !ONIG_IS_OPTION_ON(reg->options, ONIG_OPTION_CAPTURE_GROUP)) {
915 return 0;
916 }
917 #endif
918
919 return 1;
920 }
921
922
923 #define INIT_SCANENV_MEMNODES_ALLOC_SIZE 16
924
925 static void
926 scan_env_clear(ScanEnv* env)
927 {
928 int i;
929
930 BIT_STATUS_CLEAR(env->capture_history);
931 BIT_STATUS_CLEAR(env->bt_mem_start);
932 BIT_STATUS_CLEAR(env->bt_mem_end);
933 BIT_STATUS_CLEAR(env->backrefed_mem);
934 env->error = (UChar* )NULL;
935 env->error_end = (UChar* )NULL;
936 env->num_call = 0;
937 env->num_mem = 0;
938 #ifdef USE_NAMED_GROUP
939 env->num_named = 0;
940 #endif
941 env->mem_alloc = 0;
942 env->mem_nodes_dynamic = (Node** )NULL;
943
944 for (i = 0; i < SCANENV_MEMNODES_SIZE; i++)
945 env->mem_nodes_static[i] = NULL_NODE;
946
947 #ifdef USE_COMBINATION_EXPLOSION_CHECK
948 env->num_comb_exp_check = 0;
949 env->comb_exp_max_regnum = 0;
950 env->curr_max_regnum = 0;
951 env->has_recursion = 0;
952 #endif
953 }
954
955 static int
956 scan_env_add_mem_entry(ScanEnv* env)
957 {
958 int i, need, alloc;
959 Node** p;
960
961 need = env->num_mem + 1;
962 if (need >= SCANENV_MEMNODES_SIZE) {
963 if (env->mem_alloc <= need) {
964 if (IS_NULL(env->mem_nodes_dynamic)) {
965 alloc = INIT_SCANENV_MEMNODES_ALLOC_SIZE;
966 p = (Node** )xmalloc(sizeof(Node*) * alloc);
967 xmemcpy(p, env->mem_nodes_static,
968 sizeof(Node*) * SCANENV_MEMNODES_SIZE);
969 }
970 else {
971 alloc = env->mem_alloc * 2;
972 p = (Node** )xrealloc(env->mem_nodes_dynamic, sizeof(Node*) * alloc, sizeof(Node*) * env->mem_alloc);
973 }
974 CHECK_NULL_RETURN_MEMERR(p);
975
976 for (i = env->num_mem + 1; i < alloc; i++)
977 p[i] = NULL_NODE;
978
979 env->mem_nodes_dynamic = p;
980 env->mem_alloc = alloc;
981 }
982 }
983
984 env->num_mem++;
985 return env->num_mem;
986 }
987
988 static int
989 scan_env_set_mem_node(ScanEnv* env, int num, Node* node)
990 {
991 if (env->num_mem >= num)
992 SCANENV_MEM_NODES(env)[num] = node;
993 else
994 return ONIGERR_PARSER_BUG;
995 return 0;
996 }
997
998
999 #ifdef USE_PARSE_TREE_NODE_RECYCLE
1000 typedef struct _FreeNode {
1001 struct _FreeNode* next;
1002 } FreeNode;
1003
1004 static FreeNode* FreeNodeList = (FreeNode* )NULL;
1005 #endif
1006
1007 extern void
1008 onig_node_free(Node* node)
1009 {
1010 start:
1011 if (IS_NULL(node)) return ;
1012
1013 switch (NTYPE(node)) {
1014 case NT_STR:
1015 if (NSTR(node)->capa != 0 &&
1016 IS_NOT_NULL(NSTR(node)->s) && NSTR(node)->s != NSTR(node)->buf) {
1017 xfree(NSTR(node)->s);
1018 }
1019 break;
1020
1021 case NT_LIST:
1022 case NT_ALT:
1023 onig_node_free(NCAR(node));
1024 {
1025 Node* next_node = NCDR(node);
1026
1027 #ifdef USE_PARSE_TREE_NODE_RECYCLE
1028 {
1029 FreeNode* n = (FreeNode* )node;
1030
1031 THREAD_ATOMIC_START;
1032 n->next = FreeNodeList;
1033 FreeNodeList = n;
1034 THREAD_ATOMIC_END;
1035 }
1036 #else
1037 xfree(node);
1038 #endif
1039 node = next_node;
1040 goto start;
1041 }
1042 break;
1043
1044 case NT_CCLASS:
1045 {
1046 CClassNode* cc = NCCLASS(node);
1047
1048 if (IS_NCCLASS_SHARE(cc)) return ;
1049 if (cc->mbuf)
1050 bbuf_free(cc->mbuf);
1051 }
1052 break;
1053
1054 case NT_QTFR:
1055 if (NQTFR(node)->target)
1056 onig_node_free(NQTFR(node)->target);
1057 break;
1058
1059 case NT_ENCLOSE:
1060 if (NENCLOSE(node)->target)
1061 onig_node_free(NENCLOSE(node)->target);
1062 break;
1063
1064 case NT_BREF:
1065 if (IS_NOT_NULL(NBREF(node)->back_dynamic))
1066 xfree(NBREF(node)->back_dynamic);
1067 break;
1068
1069 case NT_ANCHOR:
1070 if (NANCHOR(node)->target)
1071 onig_node_free(NANCHOR(node)->target);
1072 break;
1073 }
1074
1075 #ifdef USE_PARSE_TREE_NODE_RECYCLE
1076 {
1077 FreeNode* n = (FreeNode* )node;
1078
1079 THREAD_ATOMIC_START;
1080 n->next = FreeNodeList;
1081 FreeNodeList = n;
1082 THREAD_ATOMIC_END;
1083 }
1084 #else
1085 xfree(node);
1086 #endif
1087 }
1088
1089 #ifdef USE_PARSE_TREE_NODE_RECYCLE
1090 extern int
1091 onig_free_node_list(void)
1092 {
1093 FreeNode* n;
1094
1095 /* THREAD_ATOMIC_START; */
1096 while (IS_NOT_NULL(FreeNodeList)) {
1097 n = FreeNodeList;
1098 FreeNodeList = FreeNodeList->next;
1099 xfree(n);
1100 }
1101 /* THREAD_ATOMIC_END; */
1102 return 0;
1103 }
1104 #endif
1105
1106 static Node*
1107 node_new(void)
1108 {
1109 Node* node;
1110
1111 #ifdef USE_PARSE_TREE_NODE_RECYCLE
1112 THREAD_ATOMIC_START;
1113 if (IS_NOT_NULL(FreeNodeList)) {
1114 node = (Node* )FreeNodeList;
1115 FreeNodeList = FreeNodeList->next;
1116 THREAD_ATOMIC_END;
1117 return node;
1118 }
1119 THREAD_ATOMIC_END;
1120 #endif
1121
1122 node = (Node* )xmalloc(sizeof(Node));
1123 /* xmemset(node, 0, sizeof(Node)); */
1124 return node;
1125 }
1126
1127
1128 static void
1129 initialize_cclass(CClassNode* cc)
1130 {
1131 BITSET_CLEAR(cc->bs);
1132 /* cc->base.flags = 0; */
1133 cc->flags = 0;
1134 cc->mbuf = NULL;
1135 }
1136
1137 static Node*
1138 node_new_cclass(void)
1139 {
1140 Node* node = node_new();
1141 CHECK_NULL_RETURN(node);
1142
1143 SET_NTYPE(node, NT_CCLASS);
1144 initialize_cclass(NCCLASS(node));
1145 return node;
1146 }
1147
1148 static Node*
1149 node_new_cclass_by_codepoint_range(int not, OnigCodePoint sb_out,
1150 const OnigCodePoint ranges[])
1151 {
1152 int n, i;
1153 CClassNode* cc;
1154 OnigCodePoint j;
1155
1156 Node* node = node_new_cclass();
1157 CHECK_NULL_RETURN(node);
1158
1159 cc = NCCLASS(node);
1160 if (not != 0) NCCLASS_SET_NOT(cc);
1161
1162 BITSET_CLEAR(cc->bs);
1163 if (sb_out > 0 && IS_NOT_NULL(ranges)) {
1164 n = ONIGENC_CODE_RANGE_NUM(ranges);
1165 for (i = 0; i < n; i++) {
1166 for (j = ONIGENC_CODE_RANGE_FROM(ranges, i);
1167 j <= (OnigCodePoint )ONIGENC_CODE_RANGE_TO(ranges, i); j++) {
1168 if (j >= sb_out) goto sb_end;
1169
1170 BITSET_SET_BIT(cc->bs, j);
1171 }
1172 }
1173 }
1174
1175 sb_end:
1176 if (IS_NULL(ranges)) {
1177 is_null:
1178 cc->mbuf = NULL;
1179 }
1180 else {
1181 BBuf* bbuf;
1182
1183 n = ONIGENC_CODE_RANGE_NUM(ranges);
1184 if (n == 0) goto is_null;
1185
1186 bbuf = (BBuf* )xmalloc(sizeof(BBuf));
1187 CHECK_NULL_RETURN(bbuf);
1188 bbuf->alloc = n + 1;
1189 bbuf->used = n + 1;
1190 bbuf->p = (UChar* )((void* )ranges);
1191
1192 cc->mbuf = bbuf;
1193 }
1194
1195 return node;
1196 }
1197
1198 static Node*
1199 node_new_ctype(int type, int not)
1200 {
1201 Node* node = node_new();
1202 CHECK_NULL_RETURN(node);
1203
1204 SET_NTYPE(node, NT_CTYPE);
1205 NCTYPE(node)->ctype = type;
1206 NCTYPE(node)->not = not;
1207 return node;
1208 }
1209
1210 static Node*
1211 node_new_anychar(void)
1212 {
1213 Node* node = node_new();
1214 CHECK_NULL_RETURN(node);
1215
1216 SET_NTYPE(node, NT_CANY);
1217 return node;
1218 }
1219
1220 static Node*
1221 node_new_list(Node* left, Node* right)
1222 {
1223 Node* node = node_new();
1224 CHECK_NULL_RETURN(node);
1225
1226 SET_NTYPE(node, NT_LIST);
1227 NCAR(node) = left;
1228 NCDR(node) = right;
1229 return node;
1230 }
1231
1232 extern Node*
1233 onig_node_new_list(Node* left, Node* right)
1234 {
1235 return node_new_list(left, right);
1236 }
1237
1238 extern Node*
1239 onig_node_list_add(Node* list, Node* x)
1240 {
1241 Node *n;
1242
1243 n = onig_node_new_list(x, NULL);
1244 if (IS_NULL(n)) return NULL_NODE;
1245
1246 if (IS_NOT_NULL(list)) {
1247 while (IS_NOT_NULL(NCDR(list)))
1248 list = NCDR(list);
1249
1250 NCDR(list) = n;
1251 }
1252
1253 return n;
1254 }
1255
1256 extern Node*
1257 onig_node_new_alt(Node* left, Node* right)
1258 {
1259 Node* node = node_new();
1260 CHECK_NULL_RETURN(node);
1261
1262 SET_NTYPE(node, NT_ALT);
1263 NCAR(node) = left;
1264 NCDR(node) = right;
1265 return node;
1266 }
1267
1268 extern Node*
1269 onig_node_new_anchor(int type)
1270 {
1271 Node* node = node_new();
1272 CHECK_NULL_RETURN(node);
1273
1274 SET_NTYPE(node, NT_ANCHOR);
1275 NANCHOR(node)->type = type;
1276 NANCHOR(node)->target = NULL;
1277 NANCHOR(node)->char_len = -1;
1278 return node;
1279 }
1280
1281 static Node*
1282 node_new_backref(int back_num, int* backrefs, int by_name,
1283 #ifdef USE_BACKREF_WITH_LEVEL
1284 int exist_level, int nest_level,
1285 #endif
1286 ScanEnv* env)
1287 {
1288 int i;
1289 Node* node = node_new();
1290
1291 CHECK_NULL_RETURN(node);
1292
1293 SET_NTYPE(node, NT_BREF);
1294 NBREF(node)->state = 0;
1295 NBREF(node)->back_num = back_num;
1296 NBREF(node)->back_dynamic = (int* )NULL;
1297 if (by_name != 0)
1298 NBREF(node)->state |= NST_NAME_REF;
1299
1300 #ifdef USE_BACKREF_WITH_LEVEL
1301 if (exist_level != 0) {
1302 NBREF(node)->state |= NST_NEST_LEVEL;
1303 NBREF(node)->nest_level = nest_level;
1304 }
1305 #endif
1306
1307 for (i = 0; i < back_num; i++) {
1308 if (backrefs[i] <= env->num_mem &&
1309 IS_NULL(SCANENV_MEM_NODES(env)[backrefs[i]])) {
1310 NBREF(node)->state |= NST_RECURSION; /* /...(\1).../ */
1311 break;
1312 }
1313 }
1314
1315 if (back_num <= NODE_BACKREFS_SIZE) {
1316 for (i = 0; i < back_num; i++)
1317 NBREF(node)->back_static[i] = backrefs[i];
1318 }
1319 else {
1320 int* p = (int* )xmalloc(sizeof(int) * back_num);
1321 if (IS_NULL(p)) {
1322 onig_node_free(node);
1323 return NULL;
1324 }
1325 NBREF(node)->back_dynamic = p;
1326 for (i = 0; i < back_num; i++)
1327 p[i] = backrefs[i];
1328 }
1329 return node;
1330 }
1331
1332 #ifdef USE_SUBEXP_CALL
1333 static Node*
1334 node_new_call(UChar* name, UChar* name_end, int gnum)
1335 {
1336 Node* node = node_new();
1337 CHECK_NULL_RETURN(node);
1338
1339 SET_NTYPE(node, NT_CALL);
1340 NCALL(node)->state = 0;
1341 NCALL(node)->target = NULL_NODE;
1342 NCALL(node)->name = name;
1343 NCALL(node)->name_end = name_end;
1344 NCALL(node)->group_num = gnum; /* call by number if gnum != 0 */
1345 return node;
1346 }
1347 #endif
1348
1349 static Node*
1350 node_new_quantifier(int lower, int upper, int by_number)
1351 {
1352 Node* node = node_new();
1353 CHECK_NULL_RETURN(node);
1354
1355 SET_NTYPE(node, NT_QTFR);
1356 NQTFR(node)->state = 0;
1357 NQTFR(node)->target = NULL;
1358 NQTFR(node)->lower = lower;
1359 NQTFR(node)->upper = upper;
1360 NQTFR(node)->greedy = 1;
1361 NQTFR(node)->target_empty_info = NQ_TARGET_ISNOT_EMPTY;
1362 NQTFR(node)->head_exact = NULL_NODE;
1363 NQTFR(node)->next_head_exact = NULL_NODE;
1364 NQTFR(node)->is_refered = 0;
1365 if (by_number != 0)
1366 NQTFR(node)->state |= NST_BY_NUMBER;
1367
1368 #ifdef USE_COMBINATION_EXPLOSION_CHECK
1369 NQTFR(node)->comb_exp_check_num = 0;
1370 #endif
1371
1372 return node;
1373 }
1374
1375 static Node*
1376 node_new_enclose(int type)
1377 {
1378 Node* node = node_new();
1379 CHECK_NULL_RETURN(node);
1380
1381 SET_NTYPE(node, NT_ENCLOSE);
1382 NENCLOSE(node)->type = type;
1383 NENCLOSE(node)->state = 0;
1384 NENCLOSE(node)->regnum = 0;
1385 NENCLOSE(node)->option = 0;
1386 NENCLOSE(node)->target = NULL;
1387 NENCLOSE(node)->call_addr = -1;
1388 NENCLOSE(node)->opt_count = 0;
1389 return node;
1390 }
1391
1392 extern Node*
1393 onig_node_new_enclose(int type)
1394 {
1395 return node_new_enclose(type);
1396 }
1397
1398 static Node*
1399 node_new_enclose_memory(OnigOptionType option, int is_named)
1400 {
1401 Node* node = node_new_enclose(ENCLOSE_MEMORY);
1402 CHECK_NULL_RETURN(node);
1403 if (is_named != 0)
1404 SET_ENCLOSE_STATUS(node, NST_NAMED_GROUP);
1405
1406 #ifdef USE_SUBEXP_CALL
1407 NENCLOSE(node)->option = option;
1408 #endif
1409 return node;
1410 }
1411
1412 static Node*
1413 node_new_option(OnigOptionType option)
1414 {
1415 Node* node = node_new_enclose(ENCLOSE_OPTION);
1416 CHECK_NULL_RETURN(node);
1417 NENCLOSE(node)->option = option;
1418 return node;
1419 }
1420
1421 extern int
1422 onig_node_str_cat(Node* node, const UChar* s, const UChar* end)
1423 {
1424 int addlen = (int)(end - s);
1425
1426 if (addlen > 0) {
1427 int len = (int)(NSTR(node)->end - NSTR(node)->s);
1428
1429 if (NSTR(node)->capa > 0 || (len + addlen > NODE_STR_BUF_SIZE - 1)) {
1430 UChar* p;
1431 int capa = len + addlen + NODE_STR_MARGIN;
1432
1433 if (capa <= NSTR(node)->capa) {
1434 onig_strcpy(NSTR(node)->s + len, s, end);
1435 }
1436 else {
1437 if (NSTR(node)->s == NSTR(node)->buf)
1438 p = strcat_capa_from_static(NSTR(node)->s, NSTR(node)->end,
1439 s, end, capa);
1440 else
1441 p = strcat_capa(NSTR(node)->s, NSTR(node)->end, s, end, capa, NSTR(node)->capa);
1442
1443 CHECK_NULL_RETURN_MEMERR(p);
1444 NSTR(node)->s = p;
1445 NSTR(node)->capa = capa;
1446 }
1447 }
1448 else {
1449 onig_strcpy(NSTR(node)->s + len, s, end);
1450 }
1451 NSTR(node)->end = NSTR(node)->s + len + addlen;
1452 }
1453
1454 return 0;
1455 }
1456
1457 extern int
1458 onig_node_str_set(Node* node, const UChar* s, const UChar* end)
1459 {
1460 onig_node_str_clear(node);
1461 return onig_node_str_cat(node, s, end);
1462 }
1463
1464 static int
1465 node_str_cat_char(Node* node, UChar c)
1466 {
1467 UChar s[1];
1468
1469 s[0] = c;
1470 return onig_node_str_cat(node, s, s + 1);
1471 }
1472
1473 extern void
1474 onig_node_conv_to_str_node(Node* node, int flag)
1475 {
1476 SET_NTYPE(node, NT_STR);
1477 NSTR(node)->flag = flag;
1478 NSTR(node)->capa = 0;
1479 NSTR(node)->s = NSTR(node)->buf;
1480 NSTR(node)->end = NSTR(node)->buf;
1481 }
1482
1483 extern void
1484 onig_node_str_clear(Node* node)
1485 {
1486 if (NSTR(node)->capa != 0 &&
1487 IS_NOT_NULL(NSTR(node)->s) && NSTR(node)->s != NSTR(node)->buf) {
1488 xfree(NSTR(node)->s);
1489 }
1490
1491 NSTR(node)->capa = 0;
1492 NSTR(node)->flag = 0;
1493 NSTR(node)->s = NSTR(node)->buf;
1494 NSTR(node)->end = NSTR(node)->buf;
1495 }
1496
1497 static Node*
1498 node_new_str(const UChar* s, const UChar* end)
1499 {
1500 Node* node = node_new();
1501 CHECK_NULL_RETURN(node);
1502
1503 SET_NTYPE(node, NT_STR);
1504 NSTR(node)->capa = 0;
1505 NSTR(node)->flag = 0;
1506 NSTR(node)->s = NSTR(node)->buf;
1507 NSTR(node)->end = NSTR(node)->buf;
1508 if (onig_node_str_cat(node, s, end)) {
1509 onig_node_free(node);
1510 return NULL;
1511 }
1512 return node;
1513 }
1514
1515 extern Node*
1516 onig_node_new_str(const UChar* s, const UChar* end)
1517 {
1518 return node_new_str(s, end);
1519 }
1520
1521 static Node*
1522 node_new_str_raw(UChar* s, UChar* end)
1523 {
1524 Node* node = node_new_str(s, end);
1525 NSTRING_SET_RAW(node);
1526 return node;
1527 }
1528
1529 static Node*
1530 node_new_empty(void)
1531 {
1532 return node_new_str(NULL, NULL);
1533 }
1534
1535 static Node*
1536 node_new_str_raw_char(UChar c)
1537 {
1538 UChar p[1];
1539
1540 p[0] = c;
1541 return node_new_str_raw(p, p + 1);
1542 }
1543
1544 static Node*
1545 str_node_split_last_char(StrNode* sn, OnigEncoding enc)
1546 {
1547 const UChar *p;
1548 Node* n = NULL_NODE;
1549
1550 if (sn->end > sn->s) {
1551 p = onigenc_get_prev_char_head(enc, sn->s, sn->end);
1552 if (p && p > sn->s) { /* can be splitted. */
1553 n = node_new_str(p, sn->end);
1554 if ((sn->flag & NSTR_RAW) != 0)
1555 NSTRING_SET_RAW(n);
1556 sn->end = (UChar* )p;
1557 }
1558 }
1559 return n;
1560 }
1561
1562 static int
1563 str_node_can_be_split(StrNode* sn, OnigEncoding enc)
1564 {
1565 if (sn->end > sn->s) {
1566 return ((enclen(enc, sn->s) < sn->end - sn->s) ? 1 : 0);
1567 }
1568 return 0;
1569 }
1570
1571 #ifdef USE_PAD_TO_SHORT_BYTE_CHAR
1572 static int
1573 node_str_head_pad(StrNode* sn, int num, UChar val)
1574 {
1575 UChar buf[NODE_STR_BUF_SIZE];
1576 int i, len;
1577
1578 len = sn->end - sn->s;
1579 onig_strcpy(buf, sn->s, sn->end);
1580 onig_strcpy(&(sn->s[num]), buf, buf + len);
1581 sn->end += num;
1582
1583 for (i = 0; i < num; i++) {
1584 sn->s[i] = val;
1585 }
1586 }
1587 #endif
1588
1589 extern int
1590 onig_scan_unsigned_number(UChar** src, const UChar* end, OnigEncoding enc)
1591 {
1592 unsigned int num, val;
1593 OnigCodePoint c;
1594 UChar* p = *src;
1595 PFETCH_READY;
1596
1597 num = 0;
1598 while (!PEND) {
1599 PFETCH(c);
1600 if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
1601 val = (unsigned int )DIGITVAL(c);
1602 if ((INT_MAX_LIMIT - val) / 10UL < num)
1603 return -1; /* overflow */
1604
1605 num = num * 10 + val;
1606 }
1607 else {
1608 PUNFETCH;
1609 break;
1610 }
1611 }
1612 *src = p;
1613 return num;
1614 }
1615
1616 static int
1617 scan_unsigned_hexadecimal_number(UChar** src, UChar* end, int maxlen,
1618 OnigEncoding enc)
1619 {
1620 OnigCodePoint c;
1621 unsigned int num, val;
1622 UChar* p = *src;
1623 PFETCH_READY;
1624
1625 num = 0;
1626 while (!PEND && maxlen-- != 0) {
1627 PFETCH(c);
1628 if (ONIGENC_IS_CODE_XDIGIT(enc, c)) {
1629 val = (unsigned int )XDIGITVAL(enc,c);
1630 if ((INT_MAX_LIMIT - val) / 16UL < num)
1631 return -1; /* overflow */
1632
1633 num = (num << 4) + XDIGITVAL(enc,c);
1634 }
1635 else {
1636 PUNFETCH;
1637 break;
1638 }
1639 }
1640 *src = p;
1641 return num;
1642 }
1643
1644 static int
1645 scan_unsigned_octal_number(UChar** src, UChar* end, int maxlen,
1646 OnigEncoding enc)
1647 {
1648 OnigCodePoint c;
1649 unsigned int num, val;
1650 UChar* p = *src;
1651 PFETCH_READY;
1652
1653 num = 0;
1654 while (!PEND && maxlen-- != 0) {
1655 PFETCH(c);
1656 if (ONIGENC_IS_CODE_DIGIT(enc, c) && c < '8') {
1657 val = ODIGITVAL(c);
1658 if ((INT_MAX_LIMIT - val) / 8UL < num)
1659 return -1; /* overflow */
1660
1661 num = (num << 3) + val;
1662 }
1663 else {
1664 PUNFETCH;
1665 break;
1666 }
1667 }
1668 *src = p;
1669 return num;
1670 }
1671
1672
1673 #define BBUF_WRITE_CODE_POINT(bbuf,pos,code) \
1674 BBUF_WRITE(bbuf, pos, &(code), SIZE_CODE_POINT)
1675
1676 /* data format:
1677 [n][from-1][to-1][from-2][to-2] ... [from-n][to-n]
1678 (all data size is OnigCodePoint)
1679 */
1680 static int
1681 new_code_range(BBuf** pbuf)
1682 {
1683 #define INIT_MULTI_BYTE_RANGE_SIZE (SIZE_CODE_POINT * 5)
1684 int r;
1685 OnigCodePoint n;
1686 BBuf* bbuf;
1687
1688 bbuf = *pbuf = (BBuf* )xmalloc(sizeof(BBuf));
1689 CHECK_NULL_RETURN_MEMERR(*pbuf);
1690 r = BBUF_INIT(*pbuf, INIT_MULTI_BYTE_RANGE_SIZE);
1691 if (r) return r;
1692
1693 n = 0;
1694 BBUF_WRITE_CODE_POINT(bbuf, 0, n);
1695 return 0;
1696 }
1697
1698 static int
1699 add_code_range_to_buf(BBuf** pbuf, OnigCodePoint from, OnigCodePoint to)
1700 {
1701 int r, inc_n, pos;
1702 int low, high, bound, x;
1703 OnigCodePoint n, *data;
1704 BBuf* bbuf;
1705
1706 if (from > to) {
1707 n = from; from = to; to = n;
1708 }
1709
1710 if (IS_NULL(*pbuf)) {
1711 r = new_code_range(pbuf);
1712 if (r) return r;
1713 bbuf = *pbuf;
1714 n = 0;
1715 }
1716 else {
1717 bbuf = *pbuf;
1718 GET_CODE_POINT(n, bbuf->p);
1719 }
1720 data = (OnigCodePoint* )(bbuf->p);
1721 data++;
1722
1723 for (low = 0, bound = n; low < bound; ) {
1724 x = (low + bound) >> 1;
1725 if (from > data[x*2 + 1])
1726 low = x + 1;
1727 else
1728 bound = x;
1729 }
1730
1731 for (high = low, bound = n; high < bound; ) {
1732 x = (high + bound) >> 1;
1733 if (to >= data[x*2] - 1)
1734 high = x + 1;
1735 else
1736 bound = x;
1737 }
1738
1739 inc_n = low + 1 - high;
1740 if (n + inc_n > ONIG_MAX_MULTI_BYTE_RANGES_NUM)
1741 return ONIGERR_TOO_MANY_MULTI_BYTE_RANGES;
1742
1743 if (inc_n != 1) {
1744 if (from > data[low*2])
1745 from = data[low*2];
1746 if (to < data[(high - 1)*2 + 1])
1747 to = data[(high - 1)*2 + 1];
1748 }
1749
1750 if (inc_n != 0 && (OnigCodePoint )high < n) {
1751 int from_pos = SIZE_CODE_POINT * (1 + high * 2);
1752 int to_pos = SIZE_CODE_POINT * (1 + (low + 1) * 2);
1753 int size = (n - high) * 2 * SIZE_CODE_POINT;
1754
1755 if (inc_n > 0) {
1756 BBUF_MOVE_RIGHT(bbuf, from_pos, to_pos, size);
1757 }
1758 else {
1759 BBUF_MOVE_LEFT_REDUCE(bbuf, from_pos, to_pos);
1760 }
1761 }
1762
1763 pos = SIZE_CODE_POINT * (1 + low * 2);
1764 BBUF_ENSURE_SIZE(bbuf, pos + SIZE_CODE_POINT * 2);
1765 BBUF_WRITE_CODE_POINT(bbuf, pos, from);
1766 BBUF_WRITE_CODE_POINT(bbuf, pos + SIZE_CODE_POINT, to);
1767 n += inc_n;
1768 BBUF_WRITE_CODE_POINT(bbuf, 0, n);
1769
1770 return 0;
1771 }
1772
1773 static int
1774 add_code_range(BBuf** pbuf, ScanEnv* env, OnigCodePoint from, OnigCodePoint to)
1775 {
1776 if (from > to) {
1777 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC))
1778 return 0;
1779 else
1780 return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS;
1781 }
1782
1783 return add_code_range_to_buf(pbuf, from, to);
1784 }
1785
1786 static int
1787 not_code_range_buf(OnigEncoding enc, BBuf* bbuf, BBuf** pbuf)
1788 {
1789 int r, i, n;
1790 OnigCodePoint pre, from, *data, to = 0;
1791
1792 *pbuf = (BBuf* )NULL;
1793 if (IS_NULL(bbuf)) {
1794 set_all:
1795 return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);
1796 }
1797
1798 data = (OnigCodePoint* )(bbuf->p);
1799 GET_CODE_POINT(n, data);
1800 data++;
1801 if (n <= 0) goto set_all;
1802
1803 r = 0;
1804 pre = MBCODE_START_POS(enc);
1805 for (i = 0; i < n; i++) {
1806 from = data[i*2];
1807 to = data[i*2+1];
1808 if (pre <= from - 1) {
1809 r = add_code_range_to_buf(pbuf, pre, from - 1);
1810 if (r != 0) return r;
1811 }
1812 if (to == ~((OnigCodePoint )0)) break;
1813 pre = to + 1;
1814 }
1815 if (to < ~((OnigCodePoint )0)) {
1816 r = add_code_range_to_buf(pbuf, to + 1, ~((OnigCodePoint )0));
1817 }
1818 return r;
1819 }
1820
1821 #define SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2) do {\
1822 BBuf *tbuf; \
1823 int tnot; \
1824 tnot = not1; not1 = not2; not2 = tnot; \
1825 tbuf = bbuf1; bbuf1 = bbuf2; bbuf2 = tbuf; \
1826 } while (0)
1827
1828 static int
1829 or_code_range_buf(OnigEncoding enc, BBuf* bbuf1, int not1,
1830 BBuf* bbuf2, int not2, BBuf** pbuf)
1831 {
1832 int r;
1833 OnigCodePoint i, n1, *data1;
1834 OnigCodePoint from, to;
1835
1836 *pbuf = (BBuf* )NULL;
1837 if (IS_NULL(bbuf1) && IS_NULL(bbuf2)) {
1838 if (not1 != 0 || not2 != 0)
1839 return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);
1840 return 0;
1841 }
1842
1843 r = 0;
1844 if (IS_NULL(bbuf2))
1845 SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);
1846
1847 if (IS_NULL(bbuf1)) {
1848 if (not1 != 0) {
1849 return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);
1850 }
1851 else {
1852 if (not2 == 0) {
1853 return bbuf_clone(pbuf, bbuf2);
1854 }
1855 else {
1856 return not_code_range_buf(enc, bbuf2, pbuf);
1857 }
1858 }
1859 }
1860
1861 if (not1 != 0)
1862 SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);
1863
1864 data1 = (OnigCodePoint* )(bbuf1->p);
1865 GET_CODE_POINT(n1, data1);
1866 data1++;
1867
1868 if (not2 == 0 && not1 == 0) { /* 1 OR 2 */
1869 r = bbuf_clone(pbuf, bbuf2);
1870 }
1871 else if (not1 == 0) { /* 1 OR (not 2) */
1872 r = not_code_range_buf(enc, bbuf2, pbuf);
1873 }
1874 if (r != 0) return r;
1875
1876 for (i = 0; i < n1; i++) {
1877 from = data1[i*2];
1878 to = data1[i*2+1];
1879 r = add_code_range_to_buf(pbuf, from, to);
1880 if (r != 0) return r;
1881 }
1882 return 0;
1883 }
1884
1885 static int
1886 and_code_range1(BBuf** pbuf, OnigCodePoint from1, OnigCodePoint to1,
1887 OnigCodePoint* data, int n)
1888 {
1889 int i, r;
1890 OnigCodePoint from2, to2;
1891
1892 for (i = 0; i < n; i++) {
1893 from2 = data[i*2];
1894 to2 = data[i*2+1];
1895 if (from2 < from1) {
1896 if (to2 < from1) continue;
1897 else {
1898 from1 = to2 + 1;
1899 }
1900 }
1901 else if (from2 <= to1) {
1902 if (to2 < to1) {
1903 if (from1 <= from2 - 1) {
1904 r = add_code_range_to_buf(pbuf, from1, from2-1);
1905 if (r != 0) return r;
1906 }
1907 from1 = to2 + 1;
1908 }
1909 else {
1910 to1 = from2 - 1;
1911 }
1912 }
1913 else {
1914 from1 = from2;
1915 }
1916 if (from1 > to1) break;
1917 }
1918 if (from1 <= to1) {
1919 r = add_code_range_to_buf(pbuf, from1, to1);
1920 if (r != 0) return r;
1921 }
1922 return 0;
1923 }
1924
1925 static int
1926 and_code_range_buf(BBuf* bbuf1, int not1, BBuf* bbuf2, int not2, BBuf** pbuf)
1927 {
1928 int r;
1929 OnigCodePoint i, j, n1, n2, *data1, *data2;
1930 OnigCodePoint from, to, from1, to1, from2, to2;
1931
1932 *pbuf = (BBuf* )NULL;
1933 if (IS_NULL(bbuf1)) {
1934 if (not1 != 0 && IS_NOT_NULL(bbuf2)) /* not1 != 0 -> not2 == 0 */
1935 return bbuf_clone(pbuf, bbuf2);
1936 return 0;
1937 }
1938 else if (IS_NULL(bbuf2)) {
1939 if (not2 != 0)
1940 return bbuf_clone(pbuf, bbuf1);
1941 return 0;
1942 }
1943
1944 if (not1 != 0)
1945 SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);
1946
1947 data1 = (OnigCodePoint* )(bbuf1->p);
1948 data2 = (OnigCodePoint* )(bbuf2->p);
1949 GET_CODE_POINT(n1, data1);
1950 GET_CODE_POINT(n2, data2);
1951 data1++;
1952 data2++;
1953
1954 if (not2 == 0 && not1 == 0) { /* 1 AND 2 */
1955 for (i = 0; i < n1; i++) {
1956 from1 = data1[i*2];
1957 to1 = data1[i*2+1];
1958 for (j = 0; j < n2; j++) {
1959 from2 = data2[j*2];
1960 to2 = data2[j*2+1];
1961 if (from2 > to1) break;
1962 if (to2 < from1) continue;
1963 from = MAX(from1, from2);
1964 to = MIN(to1, to2);
1965 r = add_code_range_to_buf(pbuf, from, to);
1966 if (r != 0) return r;
1967 }
1968 }
1969 }
1970 else if (not1 == 0) { /* 1 AND (not 2) */
1971 for (i = 0; i < n1; i++) {
1972 from1 = data1[i*2];
1973 to1 = data1[i*2+1];
1974 r = and_code_range1(pbuf, from1, to1, data2, n2);
1975 if (r != 0) return r;
1976 }
1977 }
1978
1979 return 0;
1980 }
1981
1982 static int
1983 and_cclass(CClassNode* dest, CClassNode* cc, OnigEncoding enc)
1984 {
1985 int r, not1, not2;
1986 BBuf *buf1, *buf2, *pbuf;
1987 BitSetRef bsr1, bsr2;
1988 BitSet bs1, bs2;
1989
1990 not1 = IS_NCCLASS_NOT(dest);
1991 bsr1 = dest->bs;
1992 buf1 = dest->mbuf;
1993 not2 = IS_NCCLASS_NOT(cc);
1994 bsr2 = cc->bs;
1995 buf2 = cc->mbuf;
1996
1997 if (not1 != 0) {
1998 bitset_invert_to(bsr1, bs1);
1999 bsr1 = bs1;
2000 }
2001 if (not2 != 0) {
2002 bitset_invert_to(bsr2, bs2);
2003 bsr2 = bs2;
2004 }
2005 bitset_and(bsr1, bsr2);
2006 if (bsr1 != dest->bs) {
2007 bitset_copy(dest->bs, bsr1);
2008 bsr1 = dest->bs;
2009 }
2010 if (not1 != 0) {
2011 bitset_invert(dest->bs);
2012 }
2013
2014 if (! ONIGENC_IS_SINGLEBYTE(enc)) {
2015 if (not1 != 0 && not2 != 0) {
2016 r = or_code_range_buf(enc, buf1, 0, buf2, 0, &pbuf);
2017 }
2018 else {
2019 r = and_code_range_buf(buf1, not1, buf2, not2, &pbuf);
2020 if (r == 0 && not1 != 0) {
2021 BBuf *tbuf;
2022 r = not_code_range_buf(enc, pbuf, &tbuf);
2023 if (r != 0) {
2024 bbuf_free(pbuf);
2025 return r;
2026 }
2027 bbuf_free(pbuf);
2028 pbuf = tbuf;
2029 }
2030 }
2031 if (r != 0) return r;
2032
2033 dest->mbuf = pbuf;
2034 bbuf_free(buf1);
2035 return r;
2036 }
2037 return 0;
2038 }
2039
2040 static int
2041 or_cclass(CClassNode* dest, CClassNode* cc, OnigEncoding enc)
2042 {
2043 int r, not1, not2;
2044 BBuf *buf1, *buf2, *pbuf;
2045 BitSetRef bsr1, bsr2;
2046 BitSet bs1, bs2;
2047
2048 not1 = IS_NCCLASS_NOT(dest);
2049 bsr1 = dest->bs;
2050 buf1 = dest->mbuf;
2051 not2 = IS_NCCLASS_NOT(cc);
2052 bsr2 = cc->bs;
2053 buf2 = cc->mbuf;
2054
2055 if (not1 != 0) {
2056 bitset_invert_to(bsr1, bs1);
2057 bsr1 = bs1;
2058 }
2059 if (not2 != 0) {
2060 bitset_invert_to(bsr2, bs2);
2061 bsr2 = bs2;
2062 }
2063 bitset_or(bsr1, bsr2);
2064 if (bsr1 != dest->bs) {
2065 bitset_copy(dest->bs, bsr1);
2066 bsr1 = dest->bs;
2067 }
2068 if (not1 != 0) {
2069 bitset_invert(dest->bs);
2070 }
2071
2072 if (! ONIGENC_IS_SINGLEBYTE(enc)) {
2073 if (not1 != 0 && not2 != 0) {
2074 r = and_code_range_buf(buf1, 0, buf2, 0, &pbuf);
2075 }
2076 else {
2077 r = or_code_range_buf(enc, buf1, not1, buf2, not2, &pbuf);
2078 if (r == 0 && not1 != 0) {
2079 BBuf *tbuf;
2080 r = not_code_range_buf(enc, pbuf, &tbuf);
2081 if (r != 0) {
2082 bbuf_free(pbuf);
2083 return r;
2084 }
2085 bbuf_free(pbuf);
2086 pbuf = tbuf;
2087 }
2088 }
2089 if (r != 0) return r;
2090
2091 dest->mbuf = pbuf;
2092 bbuf_free(buf1);
2093 return r;
2094 }
2095 else
2096 return 0;
2097 }
2098
2099 static int
2100 conv_backslash_value(int c, ScanEnv* env)
2101 {
2102 if (IS_SYNTAX_OP(env->syntax, ONIG_SYN_OP_ESC_CONTROL_CHARS)) {
2103 switch (c) {
2104 case 'n': return '\n';
2105 case 't': return '\t';
2106 case 'r': return '\r';
2107 case 'f': return '\f';
2108 case 'a': return '\007';
2109 case 'b': return '\010';
2110 case 'e': return '\033';
2111 case 'v':
2112 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ESC_V_VTAB))
2113 return '\v';
2114 break;
2115
2116 default:
2117 break;
2118 }
2119 }
2120 return c;
2121 }
2122
2123 static int
2124 is_invalid_quantifier_target(Node* node)
2125 {
2126 switch (NTYPE(node)) {
2127 case NT_ANCHOR:
2128 return 1;
2129 break;
2130
2131 case NT_ENCLOSE:
2132 /* allow enclosed elements */
2133 /* return is_invalid_quantifier_target(NENCLOSE(node)->target); */
2134 break;
2135
2136 case NT_LIST:
2137 do {
2138 if (! is_invalid_quantifier_target(NCAR(node))) return 0;
2139 } while (IS_NOT_NULL(node = NCDR(node)));
2140 return 0;
2141 break;
2142
2143 case NT_ALT:
2144 do {
2145 if (is_invalid_quantifier_target(NCAR(node))) return 1;
2146 } while (IS_NOT_NULL(node = NCDR(node)));
2147 break;
2148
2149 default:
2150 break;
2151 }
2152 return 0;
2153 }
2154
2155 /* ?:0, *:1, +:2, ??:3, *?:4, +?:5 */
2156 static int
2157 popular_quantifier_num(QtfrNode* q)
2158 {
2159 if (q->greedy) {
2160 if (q->lower == 0) {
2161 if (q->upper == 1) return 0;
2162 else if (IS_REPEAT_INFINITE(q->upper)) return 1;
2163 }
2164 else if (q->lower == 1) {
2165 if (IS_REPEAT_INFINITE(q->upper)) return 2;
2166 }
2167 }
2168 else {
2169 if (q->lower == 0) {
2170 if (q->upper == 1) return 3;
2171 else if (IS_REPEAT_INFINITE(q->upper)) return 4;
2172 }
2173 else if (q->lower == 1) {
2174 if (IS_REPEAT_INFINITE(q->upper)) return 5;
2175 }
2176 }
2177 return -1;
2178 }
2179
2180
2181 enum ReduceType {
2182 RQ_ASIS = 0, /* as is */
2183 RQ_DEL = 1, /* delete parent */
2184 RQ_A, /* to '*' */
2185 RQ_AQ, /* to '*?' */
2186 RQ_QQ, /* to '??' */
2187 RQ_P_QQ, /* to '+)??' */
2188 RQ_PQ_Q /* to '+?)?' */
2189 };
2190
2191 static enum ReduceType ReduceTypeTable[6][6] = {
2192 {RQ_DEL, RQ_A, RQ_A, RQ_QQ, RQ_AQ, RQ_ASIS}, /* '?' */
2193 {RQ_DEL, RQ_DEL, RQ_DEL, RQ_P_QQ, RQ_P_QQ, RQ_DEL}, /* '*' */
2194 {RQ_A, RQ_A, RQ_DEL, RQ_ASIS, RQ_P_QQ, RQ_DEL}, /* '+' */
2195 {RQ_DEL, RQ_AQ, RQ_AQ, RQ_DEL, RQ_AQ, RQ_AQ}, /* '??' */
2196 {RQ_DEL, RQ_DEL, RQ_DEL, RQ_DEL, RQ_DEL, RQ_DEL}, /* '*?' */
2197 {RQ_ASIS, RQ_PQ_Q, RQ_DEL, RQ_AQ, RQ_AQ, RQ_DEL} /* '+?' */
2198 };
2199
2200 extern void
2201 onig_reduce_nested_quantifier(Node* pnode, Node* cnode)
2202 {
2203 int pnum, cnum;
2204 QtfrNode *p, *c;
2205
2206 p = NQTFR(pnode);
2207 c = NQTFR(cnode);
2208 pnum = popular_quantifier_num(p);
2209 cnum = popular_quantifier_num(c);
2210 if (pnum < 0 || cnum < 0) return ;
2211
2212 switch(ReduceTypeTable[cnum][pnum]) {
2213 case RQ_DEL:
2214 *pnode = *cnode;
2215 break;
2216 case RQ_A:
2217 p->target = c->target;
2218 p->lower = 0; p->upper = REPEAT_INFINITE; p->greedy = 1;
2219 break;
2220 case RQ_AQ:
2221 p->target = c->target;
2222 p->lower = 0; p->upper = REPEAT_INFINITE; p->greedy = 0;
2223 break;
2224 case RQ_QQ:
2225 p->target = c->target;
2226 p->lower = 0; p->upper = 1; p->greedy = 0;
2227 break;
2228 case RQ_P_QQ:
2229 p->target = cnode;
2230 p->lower = 0; p->upper = 1; p->greedy = 0;
2231 c->lower = 1; c->upper = REPEAT_INFINITE; c->greedy = 1;
2232 return ;
2233 break;
2234 case RQ_PQ_Q:
2235 p->target = cnode;
2236 p->lower = 0; p->upper = 1; p->greedy = 1;
2237 c->lower = 1; c->upper = REPEAT_INFINITE; c->greedy = 0;
2238 return ;
2239 break;
2240 case RQ_ASIS:
2241 p->target = cnode;
2242 return ;
2243 break;
2244 }
2245
2246 c->target = NULL_NODE;
2247 onig_node_free(cnode);
2248 }
2249
2250
2251 enum TokenSyms {
2252 TK_EOT = 0, /* end of token */
2253 TK_RAW_BYTE = 1,
2254 TK_CHAR,
2255 TK_STRING,
2256 TK_CODE_POINT,
2257 TK_ANYCHAR,
2258 TK_CHAR_TYPE,
2259 TK_BACKREF,
2260 TK_CALL,
2261 TK_ANCHOR,
2262 TK_OP_REPEAT,
2263 TK_INTERVAL,
2264 TK_ANYCHAR_ANYTIME, /* SQL '%' == .* */
2265 TK_ALT,
2266 TK_SUBEXP_OPEN,
2267 TK_SUBEXP_CLOSE,
2268 TK_CC_OPEN,
2269 TK_QUOTE_OPEN,
2270 TK_CHAR_PROPERTY, /* \p{...}, \P{...} */
2271 /* in cc */
2272 TK_CC_CLOSE,
2273 TK_CC_RANGE,
2274 TK_POSIX_BRACKET_OPEN,
2275 TK_CC_AND, /* && */
2276 TK_CC_CC_OPEN /* [ */
2277 };
2278
2279 typedef struct {
2280 enum TokenSyms type;
2281 int escaped;
2282 int base; /* is number: 8, 16 (used in [....]) */
2283 UChar* backp;
2284 union {
2285 UChar* s;
2286 int c;
2287 OnigCodePoint code;
2288 int anchor;
2289 int subtype;
2290 struct {
2291 int lower;
2292 int upper;
2293 int greedy;
2294 int possessive;
2295 } repeat;
2296 struct {
2297 int num;
2298 int ref1;
2299 int* refs;
2300 int by_name;
2301 #ifdef USE_BACKREF_WITH_LEVEL
2302 int exist_level;
2303 int level; /* \k<name+n> */
2304 #endif
2305 } backref;
2306 struct {
2307 UChar* name;
2308 UChar* name_end;
2309 int gnum;
2310 } call;
2311 struct {
2312 int ctype;
2313 int not;
2314 } prop;
2315 } u;
2316 } OnigToken;
2317
2318
2319 static int
2320 fetch_range_quantifier(UChar** src, UChar* end, OnigToken* tok, ScanEnv* env)
2321 {
2322 int low, up, syn_allow, non_low = 0;
2323 int r = 0;
2324 OnigCodePoint c;
2325 OnigEncoding enc = env->enc;
2326 UChar* p = *src;
2327 PFETCH_READY;
2328
2329 syn_allow = IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_INVALID_INTERVAL);
2330
2331 if (PEND) {
2332 if (syn_allow)
2333 return 1; /* "....{" : OK! */
2334 else
2335 return ONIGERR_END_PATTERN_AT_LEFT_BRACE; /* "....{" syntax error */
2336 }
2337
2338 if (! syn_allow) {
2339 c = PPEEK;
2340 if (c == ')' || c == '(' || c == '|') {
2341 return ONIGERR_END_PATTERN_AT_LEFT_BRACE;
2342 }
2343 }
2344
2345 low = onig_scan_unsigned_number(&p, end, env->enc);
2346 if (low < 0) return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
2347 if (low > ONIG_MAX_REPEAT_NUM)
2348 return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
2349
2350 if (p == *src) { /* can't read low */
2351 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV)) {
2352 /* allow {,n} as {0,n} */
2353 low = 0;
2354 non_low = 1;
2355 }
2356 else
2357 goto invalid;
2358 }
2359
2360 if (PEND) goto invalid;
2361 PFETCH(c);
2362 if (c == ',') {
2363 UChar* prev = p;
2364 up = onig_scan_unsigned_number(&p, end, env->enc);
2365 if (up < 0) return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
2366 if (up > ONIG_MAX_REPEAT_NUM)
2367 return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
2368
2369 if (p == prev) {
2370 if (non_low != 0)
2371 goto invalid;
2372 up = REPEAT_INFINITE; /* {n,} : {n,infinite} */
2373 }
2374 }
2375 else {
2376 if (non_low != 0)
2377 goto invalid;
2378
2379 PUNFETCH;
2380 up = low; /* {n} : exact n times */
2381 r = 2; /* fixed */
2382 }
2383
2384 if (PEND) goto invalid;
2385 PFETCH(c);
2386 if (IS_SYNTAX_OP(env->syntax, ONIG_SYN_OP_ESC_BRACE_INTERVAL)) {
2387 if (c != MC_ESC(env->syntax)) goto invalid;
2388 PFETCH(c);
2389 }
2390 if (c != '}') goto invalid;
2391
2392 if (!IS_REPEAT_INFINITE(up) && low > up) {
2393 return ONIGERR_UPPER_SMALLER_THAN_LOWER_IN_REPEAT_RANGE;
2394 }
2395
2396 tok->type = TK_INTERVAL;
2397 tok->u.repeat.lower = low;
2398 tok->u.repeat.upper = up;
2399 *src = p;
2400 return r; /* 0: normal {n,m}, 2: fixed {n} */
2401
2402 invalid:
2403 if (syn_allow)
2404 return 1; /* OK */
2405 else
2406 return ONIGERR_INVALID_REPEAT_RANGE_PATTERN;
2407 }
2408
2409 /* \M-, \C-, \c, or \... */
2410 static int
2411 fetch_escaped_value(UChar** src, UChar* end, ScanEnv* env)
2412 {
2413 int v;
2414 OnigCodePoint c;
2415 OnigEncoding enc = env->enc;
2416 UChar* p = *src;
2417
2418 if (PEND) return ONIGERR_END_PATTERN_AT_ESCAPE;
2419
2420 PFETCH_S(c);
2421 switch (c) {
2422 case 'M':
2423 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ESC_CAPITAL_M_BAR_META)) {
2424 if (PEND) return ONIGERR_END_PATTERN_AT_META;
2425 PFETCH_S(c);
2426 if (c != '-') return ONIGERR_META_CODE_SYNTAX;
2427 if (PEND) return ONIGERR_END_PATTERN_AT_META;
2428 PFETCH_S(c);
2429 if (c == MC_ESC(env->syntax)) {
2430 v = fetch_escaped_value(&p, end, env);
2431 if (v < 0) return v;
2432 c = (OnigCodePoint )v;
2433 }
2434 c = ((c & 0xff) | 0x80);
2435 }
2436 else
2437 goto backslash;
2438 break;
2439
2440 case 'C':
2441 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ESC_CAPITAL_C_BAR_CONTROL)) {
2442 if (PEND) return ONIGERR_END_PATTERN_AT_CONTROL;
2443 PFETCH_S(c);
2444 if (c != '-') return ONIGERR_CONTROL_CODE_SYNTAX;
2445 goto control;
2446 }
2447 else
2448 goto backslash;
2449
2450 case 'c':
2451 if (IS_SYNTAX_OP(env->syntax, ONIG_SYN_OP_ESC_C_CONTROL)) {
2452 control:
2453 if (PEND) return ONIGERR_END_PATTERN_AT_CONTROL;
2454 PFETCH_S(c);
2455 if (c == '?') {
2456 c = 0177;
2457 }
2458 else {
2459 if (c == MC_ESC(env->syntax)) {
2460 v = fetch_escaped_value(&p, end, env);
2461 if (v < 0) return v;
2462 c = (OnigCodePoint )v;
2463 }
2464 c &= 0x9f;
2465 }
2466 break;
2467 }
2468 /* fall through */
2469
2470 default:
2471 {
2472 backslash:
2473 c = conv_backslash_value(c, env);
2474 }
2475 break;
2476 }
2477
2478 *src = p;
2479 return c;
2480 }
2481
2482 static int fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env);
2483
2484 static OnigCodePoint
2485 get_name_end_code_point(OnigCodePoint start)
2486 {
2487 switch (start) {
2488 case '<': return (OnigCodePoint )'>'; break;
2489 case '\'': return (OnigCodePoint )'\''; break;
2490 default:
2491 break;
2492 }
2493
2494 return (OnigCodePoint )0;
2495 }
2496
2497 #ifdef USE_NAMED_GROUP
2498 #ifdef USE_BACKREF_WITH_LEVEL
2499 /*
2500 \k<name+n>, \k<name-n>
2501 \k<num+n>, \k<num-n>
2502 \k<-num+n>, \k<-num-n>
2503 */
2504 static int
2505 fetch_name_with_level(OnigCodePoint start_code, UChar** src, UChar* end,
2506 UChar** rname_end, ScanEnv* env,
2507 int* rback_num, int* rlevel)
2508 {
2509 int r, sign, is_num, exist_level;
2510 OnigCodePoint end_code;
2511 OnigCodePoint c = 0;
2512 OnigEncoding enc = env->enc;
2513 UChar *name_end;
2514 UChar *pnum_head;
2515 UChar *p = *src;
2516 PFETCH_READY;
2517
2518 *rback_num = 0;
2519 is_num = exist_level = 0;
2520 sign = 1;
2521 pnum_head = *src;
2522
2523 end_code = get_name_end_code_point(start_code);
2524
2525 name_end = end;
2526 r = 0;
2527 if (PEND) {
2528 return ONIGERR_EMPTY_GROUP_NAME;
2529 }
2530 else {
2531 PFETCH(c);
2532 if (c == end_code)
2533 return ONIGERR_EMPTY_GROUP_NAME;
2534
2535 if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
2536 is_num = 1;
2537 }
2538 else if (c == '-') {
2539 is_num = 2;
2540 sign = -1;
2541 pnum_head = p;
2542 }
2543 else if (!ONIGENC_IS_CODE_WORD(enc, c)) {
2544 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2545 }
2546 }
2547
2548 while (!PEND) {
2549 name_end = p;
2550 PFETCH(c);
2551 if (c == end_code || c == ')' || c == '+' || c == '-') {
2552 if (is_num == 2) r = ONIGERR_INVALID_GROUP_NAME;
2553 break;
2554 }
2555
2556 if (is_num != 0) {
2557 if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
2558 is_num = 1;
2559 }
2560 else {
2561 r = ONIGERR_INVALID_GROUP_NAME;
2562 is_num = 0;
2563 }
2564 }
2565 else if (!ONIGENC_IS_CODE_WORD(enc, c)) {
2566 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2567 }
2568 }
2569
2570 if (r == 0 && c != end_code) {
2571 if (c == '+' || c == '-') {
2572 int level;
2573 int flag = (c == '-' ? -1 : 1);
2574
2575 PFETCH(c);
2576 if (! ONIGENC_IS_CODE_DIGIT(enc, c)) goto err;
2577 PUNFETCH;
2578 level = onig_scan_unsigned_number(&p, end, enc);
2579 if (level < 0) return ONIGERR_TOO_BIG_NUMBER;
2580 *rlevel = (level * flag);
2581 exist_level = 1;
2582
2583 PFETCH(c);
2584 if (c == end_code)
2585 goto end;
2586 }
2587
2588 err:
2589 r = ONIGERR_INVALID_GROUP_NAME;
2590 name_end = end;
2591 }
2592
2593 end:
2594 if (r == 0) {
2595 if (is_num != 0) {
2596 *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc);
2597 if (*rback_num < 0) return ONIGERR_TOO_BIG_NUMBER;
2598 else if (*rback_num == 0) goto err;
2599
2600 *rback_num *= sign;
2601 }
2602
2603 *rname_end = name_end;
2604 *src = p;
2605 return (exist_level ? 1 : 0);
2606 }
2607 else {
2608 onig_scan_env_set_error_string(env, r, *src, name_end);
2609 return r;
2610 }
2611 }
2612 #endif /* USE_BACKREF_WITH_LEVEL */
2613
2614 /*
2615 def: 0 -> define name (don't allow number name)
2616 1 -> reference name (allow number name)
2617 */
2618 static int
2619 fetch_name(OnigCodePoint start_code, UChar** src, UChar* end,
2620 UChar** rname_end, ScanEnv* env, int* rback_num, int ref)
2621 {
2622 int r, is_num, sign;
2623 OnigCodePoint end_code;
2624 OnigCodePoint c = 0;
2625 OnigEncoding enc = env->enc;
2626 UChar *name_end;
2627 UChar *pnum_head;
2628 UChar *p = *src;
2629
2630 *rback_num = 0;
2631
2632 end_code = get_name_end_code_point(start_code);
2633
2634 name_end = end;
2635 pnum_head = *src;
2636 r = 0;
2637 is_num = 0;
2638 sign = 1;
2639 if (PEND) {
2640 return ONIGERR_EMPTY_GROUP_NAME;
2641 }
2642 else {
2643 PFETCH_S(c);
2644 if (c == end_code)
2645 return ONIGERR_EMPTY_GROUP_NAME;
2646
2647 if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
2648 if (ref == 1)
2649 is_num = 1;
2650 else {
2651 r = ONIGERR_INVALID_GROUP_NAME;
2652 is_num = 0;
2653 }
2654 }
2655 else if (c == '-') {
2656 if (ref == 1) {
2657 is_num = 2;
2658 sign = -1;
2659 pnum_head = p;
2660 }
2661 else {
2662 r = ONIGERR_INVALID_GROUP_NAME;
2663 is_num = 0;
2664 }
2665 }
2666 else if (!ONIGENC_IS_CODE_WORD(enc, c)) {
2667 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2668 }
2669 }
2670
2671 if (r == 0) {
2672 while (!PEND) {
2673 name_end = p;
2674 PFETCH_S(c);
2675 if (c == end_code || c == ')') {
2676 if (is_num == 2) r = ONIGERR_INVALID_GROUP_NAME;
2677 break;
2678 }
2679
2680 if (is_num != 0) {
2681 if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
2682 is_num = 1;
2683 }
2684 else {
2685 if (!ONIGENC_IS_CODE_WORD(enc, c))
2686 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2687 else
2688 r = ONIGERR_INVALID_GROUP_NAME;
2689 is_num = 0;
2690 }
2691 }
2692 else {
2693 if (!ONIGENC_IS_CODE_WORD(enc, c)) {
2694 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2695 }
2696 }
2697 }
2698
2699 if (c != end_code) {
2700 r = ONIGERR_INVALID_GROUP_NAME;
2701 name_end = end;
2702 }
2703
2704 if (is_num != 0) {
2705 *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc);
2706 if (*rback_num < 0) return ONIGERR_TOO_BIG_NUMBER;
2707 else if (*rback_num == 0) {
2708 r = ONIGERR_INVALID_GROUP_NAME;
2709 goto err;
2710 }
2711
2712 *rback_num *= sign;
2713 }
2714
2715 *rname_end = name_end;
2716 *src = p;
2717 return 0;
2718 }
2719 else {
2720 while (!PEND) {
2721 name_end = p;
2722 PFETCH_S(c);
2723 if (c == end_code || c == ')')
2724 break;
2725 }
2726 if (PEND)
2727 name_end = end;
2728
2729 err:
2730 onig_scan_env_set_error_string(env, r, *src, name_end);
2731 return r;
2732 }
2733 }
2734 #else
2735 static int
2736 fetch_name(OnigCodePoint start_code, UChar** src, UChar* end,
2737 UChar** rname_end, ScanEnv* env, int* rback_num, int ref)
2738 {
2739 int r, is_num, sign;
2740 OnigCodePoint end_code;
2741 OnigCodePoint c = 0;
2742 UChar *name_end;
2743 OnigEncoding enc = env->enc;
2744 UChar *pnum_head;
2745 UChar *p = *src;
2746 PFETCH_READY;
2747
2748 *rback_num = 0;
2749
2750 end_code = get_name_end_code_point(start_code);
2751
2752 *rname_end = name_end = end;
2753 r = 0;
2754 pnum_head = *src;
2755 is_num = 0;
2756 sign = 1;
2757
2758 if (PEND) {
2759 return ONIGERR_EMPTY_GROUP_NAME;
2760 }
2761 else {
2762 PFETCH(c);
2763 if (c == end_code)
2764 return ONIGERR_EMPTY_GROUP_NAME;
2765
2766 if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
2767 is_num = 1;
2768 }
2769 else if (c == '-') {
2770 is_num = 2;
2771 sign = -1;
2772 pnum_head = p;
2773 }
2774 else {
2775 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2776 }
2777 }
2778
2779 while (!PEND) {
2780 name_end = p;
2781
2782 PFETCH(c);
2783 if (c == end_code || c == ')') break;
2784 if (! ONIGENC_IS_CODE_DIGIT(enc, c))
2785 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2786 }
2787 if (r == 0 && c != end_code) {
2788 r = ONIGERR_INVALID_GROUP_NAME;
2789 name_end = end;
2790 }
2791
2792 if (r == 0) {
2793 *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc);
2794 if (*rback_num < 0) return ONIGERR_TOO_BIG_NUMBER;
2795 else if (*rback_num == 0) {
2796 r = ONIGERR_INVALID_GROUP_NAME;
2797 goto err;
2798 }
2799 *rback_num *= sign;
2800
2801 *rname_end = name_end;
2802 *src = p;
2803 return 0;
2804 }
2805 else {
2806 err:
2807 onig_scan_env_set_error_string(env, r, *src, name_end);
2808 return r;
2809 }
2810 }
2811 #endif /* USE_NAMED_GROUP */
2812
2813 static void
2814 CC_ESC_WARN(ScanEnv* env, UChar *c)
2815 {
2816 if (onig_warn == onig_null_warn) return ;
2817
2818 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_CC_OP_NOT_ESCAPED) &&
2819 IS_SYNTAX_BV(env->syntax, ONIG_SYN_BACKSLASH_ESCAPE_IN_CC)) {
2820 UChar buf[WARN_BUFSIZE];
2821 onig_snprintf_with_pattern(buf, WARN_BUFSIZE, env->enc,
2822 env->pattern, env->pattern_end,
2823 (UChar* )"character class has '%s' without escape", c);
2824 (*onig_warn)((char* )buf);
2825 }
2826 }
2827
2828 static void
2829 CLOSE_BRACKET_WITHOUT_ESC_WARN(ScanEnv* env, UChar* c)
2830 {
2831 if (onig_warn == onig_null_warn) return ;
2832
2833 if (IS_SYNTAX_BV((env)->syntax, ONIG_SYN_WARN_CC_OP_NOT_ESCAPED)) {
2834 UChar buf[WARN_BUFSIZE];
2835 onig_snprintf_with_pattern(buf, WARN_BUFSIZE, (env)->enc,
2836 (env)->pattern, (env)->pattern_end,
2837 (UChar* )"regular expression has '%s' without escape", c);
2838 (*onig_warn)((char* )buf);
2839 }
2840 }
2841
2842 static UChar*
2843 find_str_position(OnigCodePoint s[], int n, UChar* from, UChar* to,
2844 UChar **next, OnigEncoding enc)
2845 {
2846 int i;
2847 OnigCodePoint x;
2848 UChar *q;
2849 UChar *p = from;
2850
2851 while (p < to) {
2852 x = ONIGENC_MBC_TO_CODE(enc, p, to);
2853 q = p + enclen(enc, p);
2854 if (x == s[0]) {
2855 for (i = 1; i < n && q < to; i++) {
2856 x = ONIGENC_MBC_TO_CODE(enc, q, to);
2857 if (x != s[i]) break;
2858 q += enclen(enc, q);
2859 }
2860 if (i >= n) {
2861 if (IS_NOT_NULL(next))
2862 *next = q;
2863 return p;
2864 }
2865 }
2866 p = q;
2867 }
2868 return NULL_UCHARP;
2869 }
2870
2871 static int
2872 str_exist_check_with_esc(OnigCodePoint s[], int n, UChar* from, UChar* to,
2873 OnigCodePoint bad, OnigEncoding enc, OnigSyntaxType* syn)
2874 {
2875 int i, in_esc;
2876 OnigCodePoint x;
2877 UChar *q;
2878 UChar *p = from;
2879
2880 in_esc = 0;
2881 while (p < to) {
2882 if (in_esc) {
2883 in_esc = 0;
2884 p += enclen(enc, p);
2885 }
2886 else {
2887 x = ONIGENC_MBC_TO_CODE(enc, p, to);
2888 q = p + enclen(enc, p);
2889 if (x == s[0]) {
2890 for (i = 1; i < n && q < to; i++) {
2891 x = ONIGENC_MBC_TO_CODE(enc, q, to);
2892 if (x != s[i]) break;
2893 q += enclen(enc, q);
2894 }
2895 if (i >= n) return 1;
2896 p += enclen(enc, p);
2897 }
2898 else {
2899 x = ONIGENC_MBC_TO_CODE(enc, p, to);
2900 if (x == bad) return 0;
2901 else if (x == MC_ESC(syn)) in_esc = 1;
2902 p = q;
2903 }
2904 }
2905 }
2906 return 0;
2907 }
2908
2909 static int
2910 fetch_token_in_cc(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env)
2911 {
2912 int num;
2913 OnigCodePoint c, c2;
2914 OnigSyntaxType* syn = env->syntax;
2915 OnigEncoding enc = env->enc;
2916 UChar* prev;
2917 UChar* p = *src;
2918 PFETCH_READY;
2919
2920 if (PEND) {
2921 tok->type = TK_EOT;
2922 return tok->type;
2923 }
2924
2925 PFETCH(c);
2926 tok->type = TK_CHAR;
2927 tok->base = 0;
2928 tok->u.c = c;
2929 tok->escaped = 0;
2930
2931 if (c == ']') {
2932 tok->type = TK_CC_CLOSE;
2933 }
2934 else if (c == '-') {
2935 tok->type = TK_CC_RANGE;
2936 }
2937 else if (c == MC_ESC(syn)) {
2938 if (! IS_SYNTAX_BV(syn, ONIG_SYN_BACKSLASH_ESCAPE_IN_CC))
2939 goto end;
2940
2941 if (PEND) return ONIGERR_END_PATTERN_AT_ESCAPE;
2942
2943 PFETCH(c);
2944 tok->escaped = 1;
2945 tok->u.c = c;
2946 switch (c) {
2947 case 'w':
2948 tok->type = TK_CHAR_TYPE;
2949 tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
2950 tok->u.prop.not = 0;
2951 break;
2952 case 'W':
2953 tok->type = TK_CHAR_TYPE;
2954 tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
2955 tok->u.prop.not = 1;
2956 break;
2957 case 'd':
2958 tok->type = TK_CHAR_TYPE;
2959 tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
2960 tok->u.prop.not = 0;
2961 break;
2962 case 'D':
2963 tok->type = TK_CHAR_TYPE;
2964 tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
2965 tok->u.prop.not = 1;
2966 break;
2967 case 's':
2968 tok->type = TK_CHAR_TYPE;
2969 tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
2970 tok->u.prop.not = 0;
2971 break;
2972 case 'S':
2973 tok->type = TK_CHAR_TYPE;
2974 tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
2975 tok->u.prop.not = 1;
2976 break;
2977 case 'h':
2978 if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT)) break;
2979 tok->type = TK_CHAR_TYPE;
2980 tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;
2981 tok->u.prop.not = 0;
2982 break;
2983 case 'H':
2984 if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT)) break;
2985 tok->type = TK_CHAR_TYPE;
2986 tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;
2987 tok->u.prop.not = 1;
2988 break;
2989
2990 case 'p':
2991 case 'P':
2992 c2 = PPEEK;
2993 if (c2 == '{' &&
2994 IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY)) {
2995 PINC;
2996 tok->type = TK_CHAR_PROPERTY;
2997 tok->u.prop.not = (c == 'P' ? 1 : 0);
2998
2999 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT)) {
3000 PFETCH(c2);
3001 if (c2 == '^') {
3002 tok->u.prop.not = (tok->u.prop.not == 0 ? 1 : 0);
3003 }
3004 else
3005 PUNFETCH;
3006 }
3007 }
3008 break;
3009
3010 case 'x':
3011 if (PEND) break;
3012
3013 prev = p;
3014 if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_BRACE_HEX8)) {
3015 PINC;
3016 num = scan_unsigned_hexadecimal_number(&p, end, 8, enc);
3017 if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;
3018 if (!PEND) {
3019 c2 = PPEEK;
3020 if (ONIGENC_IS_CODE_XDIGIT(enc, c2))
3021 return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE;
3022 }
3023
3024 if (p > prev + enclen(enc, prev) && !PEND && (PPEEK_IS('}'))) {
3025 PINC;
3026 tok->type = TK_CODE_POINT;
3027 tok->base = 16;
3028 tok->u.code = (OnigCodePoint )num;
3029 }
3030 else {
3031 /* can't read nothing or invalid format */
3032 p = prev;
3033 }
3034 }
3035 else if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_HEX2)) {
3036 num = scan_unsigned_hexadecimal_number(&p, end, 2, enc);
3037 if (num < 0) return ONIGERR_TOO_BIG_NUMBER;
3038 if (p == prev) { /* can't read nothing. */
3039 num = 0; /* but, it's not error */
3040 }
3041 tok->type = TK_RAW_BYTE;
3042 tok->base = 16;
3043 tok->u.c = num;
3044 }
3045 break;
3046
3047 case 'u':
3048 if (PEND) break;
3049
3050 prev = p;
3051 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_U_HEX4)) {
3052 num = scan_unsigned_hexadecimal_number(&p, end, 4, enc);
3053 if (num < 0) return ONIGERR_TOO_BIG_NUMBER;
3054 if (p == prev) { /* can't read nothing. */
3055 num = 0; /* but, it's not error */
3056 }
3057 tok->type = TK_CODE_POINT;
3058 tok->base = 16;
3059 tok->u.code = (OnigCodePoint )num;
3060 }
3061 break;
3062
3063 case '0':
3064 case '1': case '2': case '3': case '4': case '5': case '6': case '7':
3065 if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_OCTAL3)) {
3066 PUNFETCH;
3067 prev = p;
3068 num = scan_unsigned_octal_number(&p, end, 3, enc);
3069 if (num < 0) return ONIGERR_TOO_BIG_NUMBER;
3070 if (p == prev) { /* can't read nothing. */
3071 num = 0; /* but, it's not error */
3072 }
3073 tok->type = TK_RAW_BYTE;
3074 tok->base = 8;
3075 tok->u.c = num;
3076 }
3077 break;
3078
3079 default:
3080 PUNFETCH;
3081 num = fetch_escaped_value(&p, end, env);
3082 if (num < 0) return num;
3083 if (tok->u.c != num) {
3084 tok->u.code = (OnigCodePoint )num;
3085 tok->type = TK_CODE_POINT;
3086 }
3087 break;
3088 }
3089 }
3090 else if (c == '[') {
3091 if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_POSIX_BRACKET) && (PPEEK_IS(':'))) {
3092 OnigCodePoint send[] = { (OnigCodePoint )':', (OnigCodePoint )']' };
3093 tok->backp = p; /* point at '[' is readed */
3094 PINC;
3095 if (str_exist_check_with_esc(send, 2, p, end,
3096 (OnigCodePoint )']', enc, syn)) {
3097 tok->type = TK_POSIX_BRACKET_OPEN;
3098 }
3099 else {
3100 PUNFETCH;
3101 goto cc_in_cc;
3102 }
3103 }
3104 else {
3105 cc_in_cc:
3106 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_CCLASS_SET_OP)) {
3107 tok->type = TK_CC_CC_OPEN;
3108 }
3109 else {
3110 CC_ESC_WARN(env, (UChar* )"[");
3111 }
3112 }
3113 }
3114 else if (c == '&') {
3115 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_CCLASS_SET_OP) &&
3116 !PEND && (PPEEK_IS('&'))) {
3117 PINC;
3118 tok->type = TK_CC_AND;
3119 }
3120 }
3121
3122 end:
3123 *src = p;
3124 return tok->type;
3125 }
3126
3127 static int
3128 fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env)
3129 {
3130 int r, num;
3131 OnigCodePoint c;
3132 OnigEncoding enc = env->enc;
3133 OnigSyntaxType* syn = env->syntax;
3134 UChar* prev;
3135 UChar* p = *src;
3136 PFETCH_READY;
3137
3138 start:
3139 if (PEND) {
3140 tok->type = TK_EOT;
3141 return tok->type;
3142 }
3143
3144 tok->type = TK_STRING;
3145 tok->base = 0;
3146 tok->backp = p;
3147
3148 PFETCH(c);
3149 if (IS_MC_ESC_CODE(c, syn)) {
3150 if (PEND) return ONIGERR_END_PATTERN_AT_ESCAPE;
3151
3152 tok->backp = p;
3153 PFETCH(c);
3154
3155 tok->u.c = c;
3156 tok->escaped = 1;
3157 switch (c) {
3158 case '*':
3159 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_ASTERISK_ZERO_INF)) break;
3160 tok->type = TK_OP_REPEAT;
3161 tok->u.repeat.lower = 0;
3162 tok->u.repeat.upper = REPEAT_INFINITE;
3163 goto greedy_check;
3164 break;
3165
3166 case '+':
3167 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_PLUS_ONE_INF)) break;
3168 tok->type = TK_OP_REPEAT;
3169 tok->u.repeat.lower = 1;
3170 tok->u.repeat.upper = REPEAT_INFINITE;
3171 goto greedy_check;
3172 break;
3173
3174 case '?':
3175 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_QMARK_ZERO_ONE)) break;
3176 tok->type = TK_OP_REPEAT;
3177 tok->u.repeat.lower = 0;
3178 tok->u.repeat.upper = 1;
3179 greedy_check:
3180 if (!PEND && PPEEK_IS('?') &&
3181 IS_SYNTAX_OP(syn, ONIG_SYN_OP_QMARK_NON_GREEDY)) {
3182 PFETCH(c);
3183 tok->u.repeat.greedy = 0;
3184 tok->u.repeat.possessive = 0;
3185 }
3186 else {
3187 possessive_check:
3188 if (!PEND && PPEEK_IS('+') &&
3189 ((IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT) &&
3190 tok->type != TK_INTERVAL) ||
3191 (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_PLUS_POSSESSIVE_INTERVAL) &&
3192 tok->type == TK_INTERVAL))) {
3193 PFETCH(c);
3194 tok->u.repeat.greedy = 1;
3195 tok->u.repeat.possessive = 1;
3196 }
3197 else {
3198 tok->u.repeat.greedy = 1;
3199 tok->u.repeat.possessive = 0;
3200 }
3201 }
3202 break;
3203
3204 case '{':
3205 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_BRACE_INTERVAL)) break;
3206 r = fetch_range_quantifier(&p, end, tok, env);
3207 if (r < 0) return r; /* error */
3208 if (r == 0) goto greedy_check;
3209 else if (r == 2) { /* {n} */
3210 if (IS_SYNTAX_BV(syn, ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY))
3211 goto possessive_check;
3212
3213 goto greedy_check;
3214 }
3215 /* r == 1 : normal char */
3216 break;
3217
3218 case '|':
3219 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_VBAR_ALT)) break;
3220 tok->type = TK_ALT;
3221 break;
3222
3223 case '(':
3224 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LPAREN_SUBEXP)) break;
3225 tok->type = TK_SUBEXP_OPEN;
3226 break;
3227
3228 case ')':
3229 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LPAREN_SUBEXP)) break;
3230 tok->type = TK_SUBEXP_CLOSE;
3231 break;
3232
3233 case 'w':
3234 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_W_WORD)) break;
3235 tok->type = TK_CHAR_TYPE;
3236 tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
3237 tok->u.prop.not = 0;
3238 break;
3239
3240 case 'W':
3241 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_W_WORD)) break;
3242 tok->type = TK_CHAR_TYPE;
3243 tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
3244 tok->u.prop.not = 1;
3245 break;
3246
3247 case 'b':
3248 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_B_WORD_BOUND)) break;
3249 tok->type = TK_ANCHOR;
3250 tok->u.anchor = ANCHOR_WORD_BOUND;
3251 break;
3252
3253 case 'B':
3254 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_B_WORD_BOUND)) break;
3255 tok->type = TK_ANCHOR;
3256 tok->u.anchor = ANCHOR_NOT_WORD_BOUND;
3257 break;
3258
3259 #ifdef USE_WORD_BEGIN_END
3260 case '<':
3261 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END)) break;
3262 tok->type = TK_ANCHOR;
3263 tok->u.anchor = ANCHOR_WORD_BEGIN;
3264 break;
3265
3266 case '>':
3267 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END)) break;
3268 tok->type = TK_ANCHOR;
3269 tok->u.anchor = ANCHOR_WORD_END;
3270 break;
3271 #endif
3272
3273 case 's':
3274 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_S_WHITE_SPACE)) break;
3275 tok->type = TK_CHAR_TYPE;
3276 tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
3277 tok->u.prop.not = 0;
3278 break;
3279
3280 case 'S':
3281 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_S_WHITE_SPACE)) break;
3282 tok->type = TK_CHAR_TYPE;
3283 tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
3284 tok->u.prop.not = 1;
3285 break;
3286
3287 case 'd':
3288 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_D_DIGIT)) break;
3289 tok->type = TK_CHAR_TYPE;
3290 tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
3291 tok->u.prop.not = 0;
3292 break;
3293
3294 case 'D':
3295 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_D_DIGIT)) break;
3296 tok->type = TK_CHAR_TYPE;
3297 tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
3298 tok->u.prop.not = 1;
3299 break;
3300
3301 case 'h':
3302 if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT)) break;
3303 tok->type = TK_CHAR_TYPE;
3304 tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;
3305 tok->u.prop.not = 0;
3306 break;
3307
3308 case 'H':
3309 if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT)) break;
3310 tok->type = TK_CHAR_TYPE;
3311 tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;
3312 tok->u.prop.not = 1;
3313 break;
3314
3315 case 'A':
3316 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR)) break;
3317 begin_buf:
3318 tok->type = TK_ANCHOR;
3319 tok->u.subtype = ANCHOR_BEGIN_BUF;
3320 break;
3321
3322 case 'Z':
3323 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR)) break;
3324 tok->type = TK_ANCHOR;
3325 tok->u.subtype = ANCHOR_SEMI_END_BUF;
3326 break;
3327
3328 case 'z':
3329 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR)) break;
3330 end_buf:
3331 tok->type = TK_ANCHOR;
3332 tok->u.subtype = ANCHOR_END_BUF;
3333 break;
3334
3335 case 'G':
3336 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR)) break;
3337 tok->type = TK_ANCHOR;
3338 tok->u.subtype = ANCHOR_BEGIN_POSITION;
3339 break;
3340
3341 case '`':
3342 if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_GNU_BUF_ANCHOR)) break;
3343 goto begin_buf;
3344 break;
3345
3346 case '\'':
3347 if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_GNU_BUF_ANCHOR)) break;
3348 goto end_buf;
3349 break;
3350
3351 case 'x':
3352 if (PEND) break;
3353
3354 prev = p;
3355 if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_BRACE_HEX8)) {
3356 PINC;
3357 num = scan_unsigned_hexadecimal_number(&p, end, 8, enc);
3358 if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;
3359 if (!PEND) {
3360 if (ONIGENC_IS_CODE_XDIGIT(enc, PPEEK))
3361 return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE;
3362 }
3363
3364 if ((p > prev + enclen(enc, prev)) && !PEND && PPEEK_IS('}')) {
3365 PINC;
3366 tok->type = TK_CODE_POINT;
3367 tok->u.code = (OnigCodePoint )num;
3368 }
3369 else {
3370 /* can't read nothing or invalid format */
3371 p = prev;
3372 }
3373 }
3374 else if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_HEX2)) {
3375 num = scan_unsigned_hexadecimal_number(&p, end, 2, enc);
3376 if (num < 0) return ONIGERR_TOO_BIG_NUMBER;
3377 if (p == prev) { /* can't read nothing. */
3378 num = 0; /* but, it's not error */
3379 }
3380 tok->type = TK_RAW_BYTE;
3381 tok->base = 16;
3382 tok->u.c = num;
3383 }
3384 break;
3385
3386 case 'u':
3387 if (PEND) break;
3388
3389 prev = p;
3390 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_U_HEX4)) {
3391 num = scan_unsigned_hexadecimal_number(&p, end, 4, enc);
3392 if (num < 0) return ONIGERR_TOO_BIG_NUMBER;
3393 if (p == prev) { /* can't read nothing. */
3394 num = 0; /* but, it's not error */
3395 }
3396 tok->type = TK_CODE_POINT;
3397 tok->base = 16;
3398 tok->u.code = (OnigCodePoint )num;
3399 }
3400 break;
3401
3402 case '1': case '2': case '3': case '4':
3403 case '5': case '6': case '7': case '8': case '9':
3404 PUNFETCH;
3405 prev = p;
3406 num = onig_scan_unsigned_number(&p, end, enc);
3407 if (num < 0 || num > ONIG_MAX_BACKREF_NUM) {
3408 goto skip_backref;
3409 }
3410
3411 if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_DECIMAL_BACKREF) &&
3412 (num <= env->num_mem || num <= 9)) { /* This spec. from GNU regex */
3413 if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) {
3414 if (num > env->num_mem || IS_NULL(SCANENV_MEM_NODES(env)[num]))
3415 return ONIGERR_INVALID_BACKREF;
3416 }
3417
3418 tok->type = TK_BACKREF;
3419 tok->u.backref.num = 1;
3420 tok->u.backref.ref1 = num;
3421 tok->u.backref.by_name = 0;
3422 #ifdef USE_BACKREF_WITH_LEVEL
3423 tok->u.backref.exist_level = 0;
3424 #endif
3425 break;
3426 }
3427
3428 skip_backref:
3429 if (c == '8' || c == '9') {
3430 /* normal char */
3431 p = prev; PINC;
3432 break;
3433 }
3434
3435 p = prev;
3436 /* fall through */
3437 case '0':
3438 if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_OCTAL3)) {
3439 prev = p;
3440 num = scan_unsigned_octal_number(&p, end, (c == '0' ? 2:3), enc);
3441 if (num < 0) return ONIGERR_TOO_BIG_NUMBER;
3442 if (p == prev) { /* can't read nothing. */
3443 num = 0; /* but, it's not error */
3444 }
3445 tok->type = TK_RAW_BYTE;
3446 tok->base = 8;
3447 tok->u.c = num;
3448 }
3449 else if (c != '0') {
3450 PINC;
3451 }
3452 break;
3453
3454 #ifdef USE_NAMED_GROUP
3455 case 'k':
3456 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_K_NAMED_BACKREF)) {
3457 PFETCH(c);
3458 if (c == '<' || c == '\'') {
3459 UChar* name_end;
3460 int* backs;
3461 int back_num;
3462
3463 prev = p;
3464
3465 #ifdef USE_BACKREF_WITH_LEVEL
3466 name_end = NULL_UCHARP; /* no need. escape gcc warning. */
3467 r = fetch_name_with_level((OnigCodePoint )c, &p, end, &name_end,
3468 env, &back_num, &tok->u.backref.level);
3469 if (r == 1) tok->u.backref.exist_level = 1;
3470 else tok->u.backref.exist_level = 0;
3471 #else
3472 r = fetch_name(&p, end, &name_end, env, &back_num, 1);
3473 #endif
3474 if (r < 0) return r;
3475
3476 if (back_num != 0) {
3477 if (back_num < 0) {
3478 back_num = BACKREF_REL_TO_ABS(back_num, env);
3479 if (back_num <= 0)
3480 return ONIGERR_INVALID_BACKREF;
3481 }
3482
3483 if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) {
3484 if (back_num > env->num_mem ||
3485 IS_NULL(SCANENV_MEM_NODES(env)[back_num]))
3486 return ONIGERR_INVALID_BACKREF;
3487 }
3488 tok->type = TK_BACKREF;
3489 tok->u.backref.by_name = 0;
3490 tok->u.backref.num = 1;
3491 tok->u.backref.ref1 = back_num;
3492 }
3493 else {
3494 num = onig_name_to_group_numbers(env->reg, prev, name_end, &backs);
3495 if (num <= 0) {
3496 onig_scan_env_set_error_string(env,
3497 ONIGERR_UNDEFINED_NAME_REFERENCE, prev, name_end);
3498 return ONIGERR_UNDEFINED_NAME_REFERENCE;
3499 }
3500 if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) {
3501 int i;
3502 for (i = 0; i < num; i++) {
3503 if (backs[i] > env->num_mem ||
3504 IS_NULL(SCANENV_MEM_NODES(env)[backs[i]]))
3505 return ONIGERR_INVALID_BACKREF;
3506 }
3507 }
3508
3509 tok->type = TK_BACKREF;
3510 tok->u.backref.by_name = 1;
3511 if (num == 1) {
3512 tok->u.backref.num = 1;
3513 tok->u.backref.ref1 = backs[0];
3514 }
3515 else {
3516 tok->u.backref.num = num;
3517 tok->u.backref.refs = backs;
3518 }
3519 }
3520 }
3521 else
3522 PUNFETCH;
3523 }
3524 break;
3525 #endif
3526
3527 #ifdef USE_SUBEXP_CALL
3528 case 'g':
3529 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_G_SUBEXP_CALL)) {
3530 PFETCH(c);
3531 if (c == '<' || c == '\'') {
3532 int gnum;
3533 UChar* name_end;
3534
3535 prev = p;
3536 r = fetch_name((OnigCodePoint )c, &p, end, &name_end, env, &gnum, 1);
3537 if (r < 0) return r;
3538
3539 tok->type = TK_CALL;
3540 tok->u.call.name = prev;
3541 tok->u.call.name_end = name_end;
3542 tok->u.call.gnum = gnum;
3543 }
3544 else
3545 PUNFETCH;
3546 }
3547 break;
3548 #endif
3549
3550 case 'Q':
3551 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE)) {
3552 tok->type = TK_QUOTE_OPEN;
3553 }
3554 break;
3555
3556 case 'p':
3557 case 'P':
3558 if (PPEEK_IS('{') &&
3559 IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY)) {
3560 PINC;
3561 tok->type = TK_CHAR_PROPERTY;
3562 tok->u.prop.not = (c == 'P' ? 1 : 0);
3563
3564 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT)) {
3565 PFETCH(c);
3566 if (c == '^') {
3567 tok->u.prop.not = (tok->u.prop.not == 0 ? 1 : 0);
3568 }
3569 else
3570 PUNFETCH;
3571 }
3572 }
3573 break;
3574
3575 default:
3576 PUNFETCH;
3577 num = fetch_escaped_value(&p, end, env);
3578 if (num < 0) return num;
3579 /* set_raw: */
3580 if (tok->u.c != num) {
3581 tok->type = TK_CODE_POINT;
3582 tok->u.code = (OnigCodePoint )num;
3583 }
3584 else { /* string */
3585 p = tok->backp + enclen(enc, tok->backp);
3586 }
3587 break;
3588 }
3589 }
3590 else {
3591 tok->u.c = c;
3592 tok->escaped = 0;
3593
3594 #ifdef USE_VARIABLE_META_CHARS
3595 if ((c != ONIG_INEFFECTIVE_META_CHAR) &&
3596 IS_SYNTAX_OP(syn, ONIG_SYN_OP_VARIABLE_META_CHARACTERS)) {
3597 if (c == MC_ANYCHAR(syn))
3598 goto any_char;
3599 else if (c == MC_ANYTIME(syn))
3600 goto anytime;
3601 else if (c == MC_ZERO_OR_ONE_TIME(syn))
3602 goto zero_or_one_time;
3603 else if (c == MC_ONE_OR_MORE_TIME(syn))
3604 goto one_or_more_time;
3605 else if (c == MC_ANYCHAR_ANYTIME(syn)) {
3606 tok->type = TK_ANYCHAR_ANYTIME;
3607 goto out;
3608 }
3609 }
3610 #endif
3611
3612 switch (c) {
3613 case '.':
3614 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_DOT_ANYCHAR)) break;
3615 #ifdef USE_VARIABLE_META_CHARS
3616 any_char:
3617 #endif
3618 tok->type = TK_ANYCHAR;
3619 break;
3620
3621 case '*':
3622 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ASTERISK_ZERO_INF)) break;
3623 #ifdef USE_VARIABLE_META_CHARS
3624 anytime:
3625 #endif
3626 tok->type = TK_OP_REPEAT;
3627 tok->u.repeat.lower = 0;
3628 tok->u.repeat.upper = REPEAT_INFINITE;
3629 goto greedy_check;
3630 break;
3631
3632 case '+':
3633 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_PLUS_ONE_INF)) break;
3634 #ifdef USE_VARIABLE_META_CHARS
3635 one_or_more_time:
3636 #endif
3637 tok->type = TK_OP_REPEAT;
3638 tok->u.repeat.lower = 1;
3639 tok->u.repeat.upper = REPEAT_INFINITE;
3640 goto greedy_check;
3641 break;
3642
3643 case '?':
3644 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_QMARK_ZERO_ONE)) break;
3645 #ifdef USE_VARIABLE_META_CHARS
3646 zero_or_one_time:
3647 #endif
3648 tok->type = TK_OP_REPEAT;
3649 tok->u.repeat.lower = 0;
3650 tok->u.repeat.upper = 1;
3651 goto greedy_check;
3652 break;
3653
3654 case '{':
3655 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_BRACE_INTERVAL)) break;
3656 r = fetch_range_quantifier(&p, end, tok, env);
3657 if (r < 0) return r; /* error */
3658 if (r == 0) goto greedy_check;
3659 else if (r == 2) { /* {n} */
3660 if (IS_SYNTAX_BV(syn, ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY))
3661 goto possessive_check;
3662
3663 goto greedy_check;
3664 }
3665 /* r == 1 : normal char */
3666 break;
3667
3668 case '|':
3669 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_VBAR_ALT)) break;
3670 tok->type = TK_ALT;
3671 break;
3672
3673 case '(':
3674 if (PPEEK_IS('?') &&
3675 IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_QMARK_GROUP_EFFECT)) {
3676 PINC;
3677 if (PPEEK_IS('#')) {
3678 PFETCH(c);
3679 while (1) {
3680 if (PEND) return ONIGERR_END_PATTERN_IN_GROUP;
3681 PFETCH(c);
3682 if (c == MC_ESC(syn)) {
3683 if (!PEND) PFETCH(c);
3684 }
3685 else {
3686 if (c == ')') break;
3687 }
3688 }
3689 goto start;
3690 }
3691 PUNFETCH;
3692 }
3693
3694 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LPAREN_SUBEXP)) break;
3695 tok->type = TK_SUBEXP_OPEN;
3696 break;
3697
3698 case ')':
3699 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LPAREN_SUBEXP)) break;
3700 tok->type = TK_SUBEXP_CLOSE;
3701 break;
3702
3703 case '^':
3704 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LINE_ANCHOR)) break;
3705 tok->type = TK_ANCHOR;
3706 tok->u.subtype = (IS_SINGLELINE(env->option)
3707 ? ANCHOR_BEGIN_BUF : ANCHOR_BEGIN_LINE);
3708 break;
3709
3710 case '$':
3711 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LINE_ANCHOR)) break;
3712 tok->type = TK_ANCHOR;
3713 tok->u.subtype = (IS_SINGLELINE(env->option)
3714 ? ANCHOR_SEMI_END_BUF : ANCHOR_END_LINE);
3715 break;
3716
3717 case '[':
3718 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_BRACKET_CC)) break;
3719 tok->type = TK_CC_OPEN;
3720 break;
3721
3722 case ']':
3723 if (*src > env->pattern) /* /].../ is allowed. */
3724 CLOSE_BRACKET_WITHOUT_ESC_WARN(env, (UChar* )"]");
3725 break;
3726
3727 case '#':
3728 if (IS_EXTEND(env->option)) {
3729 while (!PEND) {
3730 PFETCH(c);
3731 if (ONIGENC_IS_CODE_NEWLINE(enc, c))
3732 break;
3733 }
3734 goto start;
3735 break;
3736 }
3737 break;
3738
3739 case ' ': case '\t': case '\n': case '\r': case '\f':
3740 if (IS_EXTEND(env->option))
3741 goto start;
3742 break;
3743
3744 default:
3745 /* string */
3746 break;
3747 }
3748 }
3749
3750 #ifdef USE_VARIABLE_META_CHARS
3751 out:
3752 #endif
3753 *src = p;
3754 return tok->type;
3755 }
3756
3757 static int
3758 add_ctype_to_cc_by_range(CClassNode* cc, int ctype ARG_UNUSED, int not,
3759 OnigEncoding enc ARG_UNUSED,
3760 OnigCodePoint sb_out, const OnigCodePoint mbr[])
3761 {
3762 int i, r;
3763 OnigCodePoint j;
3764
3765 int n = ONIGENC_CODE_RANGE_NUM(mbr);
3766
3767 if (not == 0) {
3768 for (i = 0; i < n; i++) {
3769 for (j = ONIGENC_CODE_RANGE_FROM(mbr, i);
3770 j <= ONIGENC_CODE_RANGE_TO(mbr, i); j++) {
3771 if (j >= sb_out) {
3772 if (j == ONIGENC_CODE_RANGE_TO(mbr, i)) i++;
3773 else if (j > ONIGENC_CODE_RANGE_FROM(mbr, i)) {
3774 r = add_code_range_to_buf(&(cc->mbuf), j,
3775 ONIGENC_CODE_RANGE_TO(mbr, i));
3776 if (r != 0) return r;
3777 i++;
3778 }
3779
3780 goto sb_end;
3781 }
3782 BITSET_SET_BIT(cc->bs, j);
3783 }
3784 }
3785
3786 sb_end:
3787 for ( ; i < n; i++) {
3788 r = add_code_range_to_buf(&(cc->mbuf),
3789 ONIGENC_CODE_RANGE_FROM(mbr, i),
3790 ONIGENC_CODE_RANGE_TO(mbr, i));
3791 if (r != 0) return r;
3792 }
3793 }
3794 else {
3795 OnigCodePoint prev = 0;
3796
3797 for (i = 0; i < n; i++) {
3798 for (j = prev;
3799 j < ONIGENC_CODE_RANGE_FROM(mbr, i); j++) {
3800 if (j >= sb_out) {
3801 goto sb_end2;
3802 }
3803 BITSET_SET_BIT(cc->bs, j);
3804 }
3805 prev = ONIGENC_CODE_RANGE_TO(mbr, i) + 1;
3806 }
3807 for (j = prev; j < sb_out; j++) {
3808 BITSET_SET_BIT(cc->bs, j);
3809 }
3810
3811 sb_end2:
3812 prev = sb_out;
3813
3814 for (i = 0; i < n; i++) {
3815 if (prev < ONIGENC_CODE_RANGE_FROM(mbr, i)) {
3816 r = add_code_range_to_buf(&(cc->mbuf), prev,
3817 ONIGENC_CODE_RANGE_FROM(mbr, i) - 1);
3818 if (r != 0) return r;
3819 }
3820 prev = ONIGENC_CODE_RANGE_TO(mbr, i) + 1;
3821 }
3822 if (prev < 0x7fffffff) {
3823 r = add_code_range_to_buf(&(cc->mbuf), prev, 0x7fffffff);
3824 if (r != 0) return r;
3825 }
3826 }
3827
3828 return 0;
3829 }
3830
3831 static int
3832 add_ctype_to_cc(CClassNode* cc, int ctype, int not, ScanEnv* env)
3833 {
3834 int c, r;
3835 const OnigCodePoint *ranges;
3836 OnigCodePoint sb_out;
3837 OnigEncoding enc = env->enc;
3838
3839 r = ONIGENC_GET_CTYPE_CODE_RANGE(enc, ctype, &sb_out, &ranges);
3840 if (r == 0) {
3841 return add_ctype_to_cc_by_range(cc, ctype, not, env->enc, sb_out, ranges);
3842 }
3843 else if (r != ONIG_NO_SUPPORT_CONFIG) {
3844 return r;
3845 }
3846
3847 r = 0;
3848 switch (ctype) {
3849 case ONIGENC_CTYPE_ALPHA:
3850 case ONIGENC_CTYPE_BLANK:
3851 case ONIGENC_CTYPE_CNTRL:
3852 case ONIGENC_CTYPE_DIGIT:
3853 case ONIGENC_CTYPE_LOWER:
3854 case ONIGENC_CTYPE_PUNCT:
3855 case ONIGENC_CTYPE_SPACE:
3856 case ONIGENC_CTYPE_UPPER:
3857 case ONIGENC_CTYPE_XDIGIT:
3858 case ONIGENC_CTYPE_ASCII:
3859 case ONIGENC_CTYPE_ALNUM:
3860 if (not != 0) {
3861 for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
3862 if (! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
3863 BITSET_SET_BIT(cc->bs, c);
3864 }
3865 ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
3866 }
3867 else {
3868 for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
3869 if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
3870 BITSET_SET_BIT(cc->bs, c);
3871 }
3872 }
3873 break;
3874
3875 case ONIGENC_CTYPE_GRAPH:
3876 case ONIGENC_CTYPE_PRINT:
3877 if (not != 0) {
3878 for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
3879 if (! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
3880 BITSET_SET_BIT(cc->bs, c);
3881 }
3882 }
3883 else {
3884 for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
3885 if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
3886 BITSET_SET_BIT(cc->bs, c);
3887 }
3888 ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
3889 }
3890 break;
3891
3892 case ONIGENC_CTYPE_WORD:
3893 if (not == 0) {
3894 for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
3895 if (IS_CODE_SB_WORD(enc, c)) BITSET_SET_BIT(cc->bs, c);
3896 }
3897 ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
3898 }
3899 else {
3900 for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
3901 if ((ONIGENC_CODE_TO_MBCLEN(enc, c) > 0) /* check invalid code point */
3902 && ! ONIGENC_IS_CODE_WORD(enc, c))
3903 BITSET_SET_BIT(cc->bs, c);
3904 }
3905 }
3906 break;
3907
3908 default:
3909 return ONIGERR_PARSER_BUG;
3910 break;
3911 }
3912
3913 return r;
3914 }
3915
3916 static int
3917 parse_posix_bracket(CClassNode* cc, UChar** src, UChar* end, ScanEnv* env)
3918 {
3919 #define POSIX_BRACKET_CHECK_LIMIT_LENGTH 20
3920 #define POSIX_BRACKET_NAME_MIN_LEN 4
3921
3922 static PosixBracketEntryType PBS[] = {
3923 { (UChar* )"alnum", ONIGENC_CTYPE_ALNUM, 5 },
3924 { (UChar* )"alpha", ONIGENC_CTYPE_ALPHA, 5 },
3925 { (UChar* )"blank", ONIGENC_CTYPE_BLANK, 5 },
3926 { (UChar* )"cntrl", ONIGENC_CTYPE_CNTRL, 5 },
3927 { (UChar* )"digit", ONIGENC_CTYPE_DIGIT, 5 },
3928 { (UChar* )"graph", ONIGENC_CTYPE_GRAPH, 5 },
3929 { (UChar* )"lower", ONIGENC_CTYPE_LOWER, 5 },
3930 { (UChar* )"print", ONIGENC_CTYPE_PRINT, 5 },
3931 { (UChar* )"punct", ONIGENC_CTYPE_PUNCT, 5 },
3932 { (UChar* )"space", ONIGENC_CTYPE_SPACE, 5 },
3933 { (UChar* )"upper", ONIGENC_CTYPE_UPPER, 5 },
3934 { (UChar* )"xdigit", ONIGENC_CTYPE_XDIGIT, 6 },
3935 { (UChar* )"ascii", ONIGENC_CTYPE_ASCII, 5 },
3936 { (UChar* )"word", ONIGENC_CTYPE_WORD, 4 },
3937 { (UChar* )NULL, -1, 0 }
3938 };
3939
3940 PosixBracketEntryType *pb;
3941 int not, i, r;
3942 OnigCodePoint c;
3943 OnigEncoding enc = env->enc;
3944 UChar *p = *src;
3945
3946 if (PPEEK_IS('^')) {
3947 PINC_S;
3948 not = 1;
3949 }
3950 else
3951 not = 0;
3952
3953 if (onigenc_strlen(enc, p, end) < POSIX_BRACKET_NAME_MIN_LEN + 3)
3954 goto not_posix_bracket;
3955
3956 for (pb = PBS; IS_NOT_NULL(pb->name); pb++) {
3957 if (onigenc_with_ascii_strncmp(enc, p, end, pb->name, pb->len) == 0) {
3958 p = (UChar* )onigenc_step(enc, p, end, pb->len);
3959 if (onigenc_with_ascii_strncmp(enc, p, end, (UChar* )":]", 2) != 0)
3960 return ONIGERR_INVALID_POSIX_BRACKET_TYPE;
3961
3962 r = add_ctype_to_cc(cc, pb->ctype, not, env);
3963 if (r != 0) return r;
3964
3965 PINC_S; PINC_S;
3966 *src = p;
3967 return 0;
3968 }
3969 }
3970
3971 not_posix_bracket:
3972 c = 0;
3973 i = 0;
3974 while (!PEND && ((c = PPEEK) != ':') && c != ']') {
3975 PINC_S;
3976 if (++i > POSIX_BRACKET_CHECK_LIMIT_LENGTH) break;
3977 }
3978 if (c == ':' && ! PEND) {
3979 PINC_S;
3980 if (! PEND) {
3981 PFETCH_S(c);
3982 if (c == ']')
3983 return ONIGERR_INVALID_POSIX_BRACKET_TYPE;
3984 }
3985 }
3986
3987 return 1; /* 1: is not POSIX bracket, but no error. */
3988 }
3989
3990 static int
3991 fetch_char_property_to_ctype(UChar** src, UChar* end, ScanEnv* env)
3992 {
3993 int r;
3994 OnigCodePoint c;
3995 OnigEncoding enc = env->enc;
3996 UChar *prev, *start, *p = *src;
3997
3998 r = 0;
3999 start = prev = p;
4000
4001 while (!PEND) {
4002 prev = p;
4003 PFETCH_S(c);
4004 if (c == '}') {
4005 r = ONIGENC_PROPERTY_NAME_TO_CTYPE(enc, start, prev);
4006 if (r < 0) break;
4007
4008 *src = p;
4009 return r;
4010 }
4011 else if (c == '(' || c == ')' || c == '{' || c == '|') {
4012 r = ONIGERR_INVALID_CHAR_PROPERTY_NAME;
4013 break;
4014 }
4015 }
4016
4017 onig_scan_env_set_error_string(env, r, *src, prev);
4018 return r;
4019 }
4020
4021 static int
4022 parse_char_property(Node** np, OnigToken* tok, UChar** src, UChar* end,
4023 ScanEnv* env)
4024 {
4025 int r, ctype;
4026 CClassNode* cc;
4027
4028 ctype = fetch_char_property_to_ctype(src, end, env);
4029 if (ctype < 0) return ctype;
4030
4031 *np = node_new_cclass();
4032 CHECK_NULL_RETURN_MEMERR(*np);
4033 cc = NCCLASS(*np);
4034 r = add_ctype_to_cc(cc, ctype, 0, env);
4035 if (r != 0) return r;
4036 if (tok->u.prop.not != 0) NCCLASS_SET_NOT(cc);
4037
4038 return 0;
4039 }
4040
4041
4042 enum CCSTATE {
4043 CCS_VALUE,
4044 CCS_RANGE,
4045 CCS_COMPLETE,
4046 CCS_START
4047 };
4048
4049 enum CCVALTYPE {
4050 CCV_SB,
4051 CCV_CODE_POINT,
4052 CCV_CLASS
4053 };
4054
4055 static int
4056 next_state_class(CClassNode* cc, OnigCodePoint* vs, enum CCVALTYPE* type,
4057 enum CCSTATE* state, ScanEnv* env)
4058 {
4059 int r;
4060
4061 if (*state == CCS_RANGE)
4062 return ONIGERR_CHAR_CLASS_VALUE_AT_END_OF_RANGE;
4063
4064 if (*state == CCS_VALUE && *type != CCV_CLASS) {
4065 if (*type == CCV_SB)
4066 BITSET_SET_BIT(cc->bs, (int )(*vs));
4067 else if (*type == CCV_CODE_POINT) {
4068 r = add_code_range(&(cc->mbuf), env, *vs, *vs);
4069 if (r < 0) return r;
4070 }
4071 }
4072
4073 *state = CCS_VALUE;
4074 *type = CCV_CLASS;
4075 return 0;
4076 }
4077
4078 static int
4079 next_state_val(CClassNode* cc, OnigCodePoint *vs, OnigCodePoint v,
4080 int* vs_israw, int v_israw,
4081 enum CCVALTYPE intype, enum CCVALTYPE* type,
4082 enum CCSTATE* state, ScanEnv* env)
4083 {
4084 int r;
4085
4086 switch (*state) {
4087 case CCS_VALUE:
4088 if (*type == CCV_SB)
4089 BITSET_SET_BIT(cc->bs, (int )(*vs));
4090 else if (*type == CCV_CODE_POINT) {
4091 r = add_code_range(&(cc->mbuf), env, *vs, *vs);
4092 if (r < 0) return r;
4093 }
4094 break;
4095
4096 case CCS_RANGE:
4097 if (intype == *type) {
4098 if (intype == CCV_SB) {
4099 if (*vs > 0xff || v > 0xff)
4100 return ONIGERR_INVALID_CODE_POINT_VALUE;
4101
4102 if (*vs > v) {
4103 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC))
4104 goto ccs_range_end;
4105 else
4106 return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS;
4107 }
4108 bitset_set_range(cc->bs, (int )*vs, (int )v);
4109 }
4110 else {
4111 r = add_code_range(&(cc->mbuf), env, *vs, v);
4112 if (r < 0) return r;
4113 }
4114 }
4115 else {
4116 #if 0
4117 if (intype == CCV_CODE_POINT && *type == CCV_SB) {
4118 #endif
4119 if (*vs > v) {
4120 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC))
4121 goto ccs_range_end;
4122 else
4123 return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS;
4124 }
4125 bitset_set_range(cc->bs, (int )*vs, (int )(v < 0xff ? v : 0xff));
4126 r = add_code_range(&(cc->mbuf), env, (OnigCodePoint )*vs, v);
4127 if (r < 0) return r;
4128 #if 0
4129 }
4130 else
4131 return ONIGERR_MISMATCH_CODE_LENGTH_IN_CLASS_RANGE;
4132 #endif
4133 }
4134 ccs_range_end:
4135 *state = CCS_COMPLETE;
4136 break;
4137
4138 case CCS_COMPLETE:
4139 case CCS_START:
4140 *state = CCS_VALUE;
4141 break;
4142
4143 default:
4144 break;
4145 }
4146
4147 *vs_israw = v_israw;
4148 *vs = v;
4149 *type = intype;
4150 return 0;
4151 }
4152
4153 static int
4154 code_exist_check(OnigCodePoint c, UChar* from, UChar* end, int ignore_escaped,
4155 ScanEnv* env)
4156 {
4157 int in_esc;
4158 OnigCodePoint code;
4159 OnigEncoding enc = env->enc;
4160 UChar* p = from;
4161
4162 in_esc = 0;
4163 while (! PEND) {
4164 if (ignore_escaped && in_esc) {
4165 in_esc = 0;
4166 }
4167 else {
4168 PFETCH_S(code);
4169 if (code == c) return 1;
4170 if (code == MC_ESC(env->syntax)) in_esc = 1;
4171 }
4172 }
4173 return 0;
4174 }
4175
4176 static int
4177 parse_char_class(Node** np, OnigToken* tok, UChar** src, UChar* end,
4178 ScanEnv* env)
4179 {
4180 int r, neg, len, fetched, and_start;
4181 OnigCodePoint v, vs;
4182 UChar *p;
4183 Node* node;
4184 CClassNode *cc, *prev_cc;
4185 CClassNode work_cc;
4186
4187 enum CCSTATE state;
4188 enum CCVALTYPE val_type, in_type;
4189 int val_israw, in_israw;
4190
4191 prev_cc = (CClassNode* )NULL;
4192 *np = NULL_NODE;
4193 r = fetch_token_in_cc(tok, src, end, env);
4194 if (r == TK_CHAR && tok->u.c == '^' && tok->escaped == 0) {
4195 neg = 1;
4196 r = fetch_token_in_cc(tok, src, end, env);
4197 }
4198 else {
4199 neg = 0;
4200 }
4201
4202 if (r < 0) return r;
4203 if (r == TK_CC_CLOSE) {
4204 if (! code_exist_check((OnigCodePoint )']',
4205 *src, env->pattern_end, 1, env))
4206 return ONIGERR_EMPTY_CHAR_CLASS;
4207
4208 CC_ESC_WARN(env, (UChar* )"]");
4209 r = tok->type = TK_CHAR; /* allow []...] */
4210 }
4211
4212 *np = node = node_new_cclass();
4213 CHECK_NULL_RETURN_MEMERR(node);
4214 cc = NCCLASS(node);
4215
4216 and_start = 0;
4217 state = CCS_START;
4218 p = *src;
4219 while (r != TK_CC_CLOSE) {
4220 fetched = 0;
4221 switch (r) {
4222 case TK_CHAR:
4223 len = ONIGENC_CODE_TO_MBCLEN(env->enc, tok->u.c);
4224 if (len > 1) {
4225 in_type = CCV_CODE_POINT;
4226 }
4227 else if (len < 0) {
4228 r = len;
4229 goto err;
4230 }
4231 else {
4232 sb_char:
4233 in_type = CCV_SB;
4234 }
4235 v = (OnigCodePoint )tok->u.c;
4236 in_israw = 0;
4237 goto val_entry2;
4238 break;
4239
4240 case TK_RAW_BYTE:
4241 /* tok->base != 0 : octal or hexadec. */
4242 if (! ONIGENC_IS_SINGLEBYTE(env->enc) && tok->base != 0) {
4243 UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];
4244 UChar* bufe = buf + ONIGENC_CODE_TO_MBC_MAXLEN;
4245 UChar* psave = p;
4246 int i, base = tok->base;
4247
4248 buf[0] = (UChar)tok->u.c;
4249 for (i = 1; i < ONIGENC_MBC_MAXLEN(env->enc); i++) {
4250 r = fetch_token_in_cc(tok, &p, end, env);
4251 if (r < 0) goto err;
4252 if (r != TK_RAW_BYTE || tok->base != base) {
4253 fetched = 1;
4254 break;
4255 }
4256 buf[i] = (UChar)tok->u.c;
4257 }
4258
4259 if (i < ONIGENC_MBC_MINLEN(env->enc)) {
4260 r = ONIGERR_TOO_SHORT_MULTI_BYTE_STRING;
4261 goto err;
4262 }
4263
4264 len = enclen(env->enc, buf);
4265 if (i < len) {
4266 r = ONIGERR_TOO_SHORT_MULTI_BYTE_STRING;
4267 goto err;
4268 }
4269 else if (i > len) { /* fetch back */
4270 p = psave;
4271 for (i = 1; i < len; i++) {
4272 r = fetch_token_in_cc(tok, &p, end, env);
4273 }
4274 fetched = 0;
4275 }
4276
4277 if (i == 1) {
4278 v = (OnigCodePoint )buf[0];
4279 goto raw_single;
4280 }
4281 else {
4282 v = ONIGENC_MBC_TO_CODE(env->enc, buf, bufe);
4283 in_type = CCV_CODE_POINT;
4284 }
4285 }
4286 else {
4287 v = (OnigCodePoint )tok->u.c;
4288 raw_single:
4289 in_type = CCV_SB;
4290 }
4291 in_israw = 1;
4292 goto val_entry2;
4293 break;
4294
4295 case TK_CODE_POINT:
4296 v = tok->u.code;
4297 in_israw = 1;
4298 val_entry:
4299 len = ONIGENC_CODE_TO_MBCLEN(env->enc, v);
4300 if (len < 0) {
4301 r = len;
4302 goto err;
4303 }
4304 in_type = (len == 1 ? CCV_SB : CCV_CODE_POINT);
4305 val_entry2:
4306 r = next_state_val(cc, &vs, v, &val_israw, in_israw, in_type, &val_type,
4307 &state, env);
4308 if (r != 0) goto err;
4309 break;
4310
4311 case TK_POSIX_BRACKET_OPEN:
4312 r = parse_posix_bracket(cc, &p, end, env);
4313 if (r < 0) goto err;
4314 if (r == 1) { /* is not POSIX bracket */
4315 CC_ESC_WARN(env, (UChar* )"[");
4316 p = tok->backp;
4317 v = (OnigCodePoint )tok->u.c;
4318 in_israw = 0;
4319 goto val_entry;
4320 }
4321 goto next_class;
4322 break;
4323
4324 case TK_CHAR_TYPE:
4325 r = add_ctype_to_cc(cc, tok->u.prop.ctype, tok->u.prop.not, env);
4326 if (r != 0) return r;
4327
4328 next_class:
4329 r = next_state_class(cc, &vs, &val_type, &state, env);
4330 if (r != 0) goto err;
4331 break;
4332
4333 case TK_CHAR_PROPERTY:
4334 {
4335 int ctype;
4336
4337 ctype = fetch_char_property_to_ctype(&p, end, env);
4338 if (ctype < 0) return ctype;
4339 r = add_ctype_to_cc(cc, ctype, tok->u.prop.not, env);
4340 if (r != 0) return r;
4341 goto next_class;
4342 }
4343 break;
4344
4345 case TK_CC_RANGE:
4346 if (state == CCS_VALUE) {
4347 r = fetch_token_in_cc(tok, &p, end, env);
4348 if (r < 0) goto err;
4349 fetched = 1;
4350 if (r == TK_CC_CLOSE) { /* allow [x-] */
4351 range_end_val:
4352 v = (OnigCodePoint )'-';
4353 in_israw = 0;
4354 goto val_entry;
4355 }
4356 else if (r == TK_CC_AND) {
4357 CC_ESC_WARN(env, (UChar* )"-");
4358 goto range_end_val;
4359 }
4360 state = CCS_RANGE;
4361 }
4362 else if (state == CCS_START) {
4363 /* [-xa] is allowed */
4364 v = (OnigCodePoint )tok->u.c;
4365 in_israw = 0;
4366
4367 r = fetch_token_in_cc(tok, &p, end, env);
4368 if (r < 0) goto err;
4369 fetched = 1;
4370 /* [--x] or [a&&-x] is warned. */
4371 if (r == TK_CC_RANGE || and_start != 0)
4372 CC_ESC_WARN(env, (UChar* )"-");
4373
4374 goto val_entry;
4375 }
4376 else if (state == CCS_RANGE) {
4377 CC_ESC_WARN(env, (UChar* )"-");
4378 goto sb_char; /* [!--x] is allowed */
4379 }
4380 else { /* CCS_COMPLETE */
4381 r = fetch_token_in_cc(tok, &p, end, env);
4382 if (r < 0) goto err;
4383 fetched = 1;
4384 if (r == TK_CC_CLOSE) goto range_end_val; /* allow [a-b-] */
4385 else if (r == TK_CC_AND) {
4386 CC_ESC_WARN(env, (UChar* )"-");
4387 goto range_end_val;
4388 }
4389
4390 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC)) {
4391 CC_ESC_WARN(env, (UChar* )"-");
4392 goto sb_char; /* [0-9-a] is allowed as [0-9\-a] */
4393 }
4394 r = ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS;
4395 goto err;
4396 }
4397 break;
4398
4399 case TK_CC_CC_OPEN: /* [ */
4400 {
4401 Node *anode;
4402 CClassNode* acc;
4403
4404 r = parse_char_class(&anode, tok, &p, end, env);
4405 if (r != 0) goto cc_open_err;
4406 acc = NCCLASS(anode);
4407 r = or_cclass(cc, acc, env->enc);
4408
4409 onig_node_free(anode);
4410 cc_open_err:
4411 if (r != 0) goto err;
4412 }
4413 break;
4414
4415 case TK_CC_AND: /* && */
4416 {
4417 if (state == CCS_VALUE) {
4418 r = next_state_val(cc, &vs, 0, &val_israw, 0, val_type,
4419 &val_type, &state, env);
4420 if (r != 0) goto err;
4421 }
4422 /* initialize local variables */
4423 and_start = 1;
4424 state = CCS_START;
4425
4426 if (IS_NOT_NULL(prev_cc)) {
4427 r = and_cclass(prev_cc, cc, env->enc);
4428 if (r != 0) goto err;
4429 bbuf_free(cc->mbuf);
4430 }
4431 else {
4432 prev_cc = cc;
4433 cc = &work_cc;
4434 }
4435 initialize_cclass(cc);
4436 }
4437 break;
4438
4439 case TK_EOT:
4440 r = ONIGERR_PREMATURE_END_OF_CHAR_CLASS;
4441 goto err;
4442 break;
4443 default:
4444 r = ONIGERR_PARSER_BUG;
4445 goto err;
4446 break;
4447 }
4448
4449 if (fetched)
4450 r = tok->type;
4451 else {
4452 r = fetch_token_in_cc(tok, &p, end, env);
4453 if (r < 0) goto err;
4454 }
4455 }
4456
4457 if (state == CCS_VALUE) {
4458 r = next_state_val(cc, &vs, 0, &val_israw, 0, val_type,
4459 &val_type, &state, env);
4460 if (r != 0) goto err;
4461 }
4462
4463 if (IS_NOT_NULL(prev_cc)) {
4464 r = and_cclass(prev_cc, cc, env->enc);
4465 if (r != 0) goto err;
4466 bbuf_free(cc->mbuf);
4467 cc = prev_cc;
4468 }
4469
4470 if (neg != 0)
4471 NCCLASS_SET_NOT(cc);
4472 else
4473 NCCLASS_CLEAR_NOT(cc);
4474 if (IS_NCCLASS_NOT(cc) &&
4475 IS_SYNTAX_BV(env->syntax, ONIG_SYN_NOT_NEWLINE_IN_NEGATIVE_CC)) {
4476 int is_empty;
4477
4478 is_empty = (IS_NULL(cc->mbuf) ? 1 : 0);
4479 if (is_empty != 0)
4480 BITSET_IS_EMPTY(cc->bs, is_empty);
4481
4482 if (is_empty == 0) {
4483 #define NEWLINE_CODE 0x0a
4484
4485 if (ONIGENC_IS_CODE_NEWLINE(env->enc, NEWLINE_CODE)) {
4486 if (ONIGENC_CODE_TO_MBCLEN(env->enc, NEWLINE_CODE) == 1)
4487 BITSET_SET_BIT(cc->bs, NEWLINE_CODE);
4488 else
4489 add_code_range(&(cc->mbuf), env, NEWLINE_CODE, NEWLINE_CODE);
4490 }
4491 }
4492 }
4493 *src = p;
4494 return 0;
4495
4496 err:
4497 if (cc != NCCLASS(*np))
4498 bbuf_free(cc->mbuf);
4499 onig_node_free(*np);
4500 return r;
4501 }
4502
4503 static int parse_subexp(Node** top, OnigToken* tok, int term,
4504 UChar** src, UChar* end, ScanEnv* env);
4505
4506 static int
4507 parse_enclose(Node** np, OnigToken* tok, int term, UChar** src, UChar* end,
4508 ScanEnv* env)
4509 {
4510 int r, num;
4511 Node *target;
4512 OnigOptionType option;
4513 OnigCodePoint c;
4514 OnigEncoding enc = env->enc;
4515
4516 #ifdef USE_NAMED_GROUP
4517 int list_capture;
4518 #endif
4519
4520 UChar* p = *src;
4521 PFETCH_READY;
4522
4523 *np = NULL;
4524 if (PEND) return ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS;
4525
4526 option = env->option;
4527 if (PPEEK_IS('?') &&
4528 IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_GROUP_EFFECT)) {
4529 PINC;
4530 if (PEND) return ONIGERR_END_PATTERN_IN_GROUP;
4531
4532 PFETCH(c);
4533 switch (c) {
4534 case ':': /* (?:...) grouping only */
4535 group:
4536 r = fetch_token(tok, &p, end, env);
4537 if (r < 0) return r;
4538 r = parse_subexp(np, tok, term, &p, end, env);
4539 if (r < 0) return r;
4540 *src = p;
4541 return 1; /* group */
4542 break;
4543
4544 case '=':
4545 *np = onig_node_new_anchor(ANCHOR_PREC_READ);
4546 break;
4547 case '!': /* preceding read */
4548 *np = onig_node_new_anchor(ANCHOR_PREC_READ_NOT);
4549 break;
4550 case '>': /* (?>...) stop backtrack */
4551 *np = node_new_enclose(ENCLOSE_STOP_BACKTRACK);
4552 break;
4553
4554 #ifdef USE_NAMED_GROUP
4555 case '\'':
4556 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) {
4557 goto named_group1;
4558 }
4559 else
4560 return ONIGERR_UNDEFINED_GROUP_OPTION;
4561 break;
4562 #endif
4563
4564 case '<': /* look behind (?<=...), (?<!...) */
4565 PFETCH(c);
4566 if (c == '=')
4567 *np = onig_node_new_anchor(ANCHOR_LOOK_BEHIND);
4568 else if (c == '!')
4569 *np = onig_node_new_anchor(ANCHOR_LOOK_BEHIND_NOT);
4570 #ifdef USE_NAMED_GROUP
4571 else {
4572 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) {
4573 UChar *name;
4574 UChar *name_end;
4575
4576 PUNFETCH;
4577 c = '<';
4578
4579 named_group1:
4580 list_capture = 0;
4581
4582 named_group2:
4583 name = p;
4584 r = fetch_name((OnigCodePoint )c, &p, end, &name_end, env, &num, 0);
4585 if (r < 0) return r;
4586
4587 num = scan_env_add_mem_entry(env);
4588 if (num < 0) return num;
4589 if (list_capture != 0 && num >= (int )BIT_STATUS_BITS_NUM)
4590 return ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY;
4591
4592 r = name_add(env->reg, name, name_end, num, env);
4593 if (r != 0) return r;
4594 *np = node_new_enclose_memory(env->option, 1);
4595 CHECK_NULL_RETURN_MEMERR(*np);
4596 NENCLOSE(*np)->regnum = num;
4597 if (list_capture != 0)
4598 BIT_STATUS_ON_AT_SIMPLE(env->capture_history, num);
4599 env->num_named++;
4600 }
4601 else {
4602 return ONIGERR_UNDEFINED_GROUP_OPTION;
4603 }
4604 }
4605 #else
4606 else {
4607 return ONIGERR_UNDEFINED_GROUP_OPTION;
4608 }
4609 #endif
4610 break;
4611
4612 case '@':
4613 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ATMARK_CAPTURE_HISTORY)) {
4614 #ifdef USE_NAMED_GROUP
4615 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) {
4616 PFETCH(c);
4617 if (c == '<' || c == '\'') {
4618 list_capture = 1;
4619 goto named_group2; /* (?@<name>...) */
4620 }
4621 PUNFETCH;
4622 }
4623 #endif
4624 *np = node_new_enclose_memory(env->option, 0);
4625 CHECK_NULL_RETURN_MEMERR(*np);
4626 num = scan_env_add_mem_entry(env);
4627 if (num < 0) {
4628 onig_node_free(*np);
4629 return num;
4630 }
4631 else if (num >= (int )BIT_STATUS_BITS_NUM) {
4632 onig_node_free(*np);
4633 return ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY;
4634 }
4635 NENCLOSE(*np)->regnum = num;
4636 BIT_STATUS_ON_AT_SIMPLE(env->capture_history, num);
4637 }
4638 else {
4639 return ONIGERR_UNDEFINED_GROUP_OPTION;
4640 }
4641 break;
4642
4643 #ifdef USE_POSIXLINE_OPTION
4644 case 'p':
4645 #endif
4646 case '-': case 'i': case 'm': case 's': case 'x':
4647 {
4648 int neg = 0;
4649
4650 while (1) {
4651 switch (c) {
4652 case ':':
4653 case ')':
4654 break;
4655
4656 case '-': neg = 1; break;
4657 case 'x': ONOFF(option, ONIG_OPTION_EXTEND, neg); break;
4658 case 'i': ONOFF(option, ONIG_OPTION_IGNORECASE, neg); break;
4659 case 's':
4660 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) {
4661 ONOFF(option, ONIG_OPTION_MULTILINE, neg);
4662 }
4663 else
4664 return ONIGERR_UNDEFINED_GROUP_OPTION;
4665 break;
4666
4667 case 'm':
4668 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) {
4669 ONOFF(option, ONIG_OPTION_SINGLELINE, (neg == 0 ? 1 : 0));
4670 }
4671 else if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) {
4672 ONOFF(option, ONIG_OPTION_MULTILINE, neg);
4673 }
4674 else
4675 return ONIGERR_UNDEFINED_GROUP_OPTION;
4676 break;
4677 #ifdef USE_POSIXLINE_OPTION
4678 case 'p':
4679 ONOFF(option, ONIG_OPTION_MULTILINE|ONIG_OPTION_SINGLELINE, neg);
4680 break;
4681 #endif
4682 default:
4683 return ONIGERR_UNDEFINED_GROUP_OPTION;
4684 }
4685
4686 if (c == ')') {
4687 *np = node_new_option(option);
4688 CHECK_NULL_RETURN_MEMERR(*np);
4689 *src = p;
4690 return 2; /* option only */
4691 }
4692 else if (c == ':') {
4693 OnigOptionType prev = env->option;
4694
4695 env->option = option;
4696 r = fetch_token(tok, &p, end, env);
4697 if (r < 0) return r;
4698 r = parse_subexp(&target, tok, term, &p, end, env);
4699 env->option = prev;
4700 if (r < 0) return r;
4701 *np = node_new_option(option);
4702 CHECK_NULL_RETURN_MEMERR(*np);
4703 NENCLOSE(*np)->target = target;
4704 *src = p;
4705 return 0;
4706 }
4707
4708 if (PEND) return ONIGERR_END_PATTERN_IN_GROUP;
4709 PFETCH(c);
4710 }
4711 }
4712 break;
4713
4714 default:
4715 return ONIGERR_UNDEFINED_GROUP_OPTION;
4716 }
4717 }
4718 else {
4719 if (ONIG_IS_OPTION_ON(env->option, ONIG_OPTION_DONT_CAPTURE_GROUP))
4720 goto group;
4721
4722 *np = node_new_enclose_memory(env->option, 0);
4723 CHECK_NULL_RETURN_MEMERR(*np);
4724 num = scan_env_add_mem_entry(env);
4725 if (num < 0) return num;
4726 NENCLOSE(*np)->regnum = num;
4727 }
4728
4729 CHECK_NULL_RETURN_MEMERR(*np);
4730 r = fetch_token(tok, &p, end, env);
4731 if (r < 0) return r;
4732 r = parse_subexp(&target, tok, term, &p, end, env);
4733 if (r < 0) return r;
4734
4735 if (NTYPE(*np) == NT_ANCHOR)
4736 NANCHOR(*np)->target = target;
4737 else {
4738 NENCLOSE(*np)->target = target;
4739 if (NENCLOSE(*np)->type == ENCLOSE_MEMORY) {
4740 /* Don't move this to previous of parse_subexp() */
4741 r = scan_env_set_mem_node(env, NENCLOSE(*np)->regnum, *np);
4742 if (r != 0) return r;
4743 }
4744 }
4745
4746 *src = p;
4747 return 0;
4748 }
4749
4750 static const char* PopularQStr[] = {
4751 "?", "*", "+", "??", "*?", "+?"
4752 };
4753
4754 static const char* ReduceQStr[] = {
4755 "", "", "*", "*?", "??", "+ and ??", "+? and ?"
4756 };
4757
4758 static int
4759 set_quantifier(Node* qnode, Node* target, int group, ScanEnv* env)
4760 {
4761 QtfrNode* qn;
4762
4763 qn = NQTFR(qnode);
4764 if (qn->lower == 1 && qn->upper == 1) {
4765 return 1;
4766 }
4767
4768 switch (NTYPE(target)) {
4769 case NT_STR:
4770 if (! group) {
4771 StrNode* sn = NSTR(target);
4772 if (str_node_can_be_split(sn, env->enc)) {
4773 Node* n = str_node_split_last_char(sn, env->enc);
4774 if (IS_NOT_NULL(n)) {
4775 qn->target = n;
4776 return 2;
4777 }
4778 }
4779 }
4780 break;
4781
4782 case NT_QTFR:
4783 { /* check redundant double repeat. */
4784 /* verbose warn (?:.?)? etc... but not warn (.?)? etc... */
4785 QtfrNode* qnt = NQTFR(target);
4786 int nestq_num = popular_quantifier_num(qn);
4787 int targetq_num = popular_quantifier_num(qnt);
4788
4789 #ifdef USE_WARNING_REDUNDANT_NESTED_REPEAT_OPERATOR
4790 if (!IS_QUANTIFIER_BY_NUMBER(qn) && !IS_QUANTIFIER_BY_NUMBER(qnt) &&
4791 IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT)) {
4792 UChar buf[WARN_BUFSIZE];
4793
4794 switch(ReduceTypeTable[targetq_num][nestq_num]) {
4795 case RQ_ASIS:
4796 break;
4797
4798 case RQ_DEL:
4799 if (onig_verb_warn != onig_null_warn) {
4800 onig_snprintf_with_pattern(buf, WARN_BUFSIZE, env->enc,
4801 env->pattern, env->pattern_end,
4802 (UChar* )"redundant nested repeat operator");
4803 (*onig_verb_warn)((char* )buf);
4804 }
4805 goto warn_exit;
4806 break;
4807
4808 default:
4809 if (onig_verb_warn != onig_null_warn) {
4810 onig_snprintf_with_pattern(buf, WARN_BUFSIZE, env->enc,
4811 env->pattern, env->pattern_end,
4812 (UChar* )"nested repeat operator %s and %s was replaced with '%s'",
4813 PopularQStr[targetq_num], PopularQStr[nestq_num],
4814 ReduceQStr[ReduceTypeTable[targetq_num][nestq_num]]);
4815 (*onig_verb_warn)((char* )buf);
4816 }
4817 goto warn_exit;
4818 break;
4819 }
4820 }
4821
4822 warn_exit:
4823 #endif
4824 if (targetq_num >= 0) {
4825 if (nestq_num >= 0) {
4826 onig_reduce_nested_quantifier(qnode, target);
4827 goto q_exit;
4828 }
4829 else if (targetq_num == 1 || targetq_num == 2) { /* * or + */
4830 /* (?:a*){n,m}, (?:a+){n,m} => (?:a*){n,n}, (?:a+){n,n} */
4831 if (! IS_REPEAT_INFINITE(qn->upper) && qn->upper > 1 && qn->greedy) {
4832 qn->upper = (qn->lower == 0 ? 1 : qn->lower);
4833 }
4834 }
4835 }
4836 }
4837 break;
4838
4839 default:
4840 break;
4841 }
4842
4843 qn->target = target;
4844 q_exit:
4845 return 0;
4846 }
4847
4848
4849 #ifdef USE_SHARED_CCLASS_TABLE
4850
4851 #define THRESHOLD_RANGE_NUM_FOR_SHARE_CCLASS 8
4852
4853 /* for ctype node hash table */
4854
4855 typedef struct {
4856 OnigEncoding enc;
4857 int not;
4858 int type;
4859 } type_cclass_key;
4860
4861 static int type_cclass_cmp(type_cclass_key* x, type_cclass_key* y)
4862 {
4863 if (x->type != y->type) return 1;
4864 if (x->enc != y->enc) return 1;
4865 if (x->not != y->not) return 1;
4866 return 0;
4867 }
4868
4869 static int type_cclass_hash(type_cclass_key* key)
4870 {
4871 int i, val;
4872 UChar *p;
4873
4874 val = 0;
4875
4876 p = (UChar* )&(key->enc);
4877 for (i = 0; i < (int )sizeof(key->enc); i++) {
4878 val = val * 997 + (int )*p++;
4879 }
4880
4881 p = (UChar* )(&key->type);
4882 for (i = 0; i < (int )sizeof(key->type); i++) {
4883 val = val * 997 + (int )*p++;
4884 }
4885
4886 val += key->not;
4887 return val + (val >> 5);
4888 }
4889
4890 static struct st_hash_type type_type_cclass_hash = {
4891 type_cclass_cmp,
4892 type_cclass_hash,
4893 };
4894
4895 static st_table* OnigTypeCClassTable;
4896
4897
4898 static int
4899 i_free_shared_class(type_cclass_key* key, Node* node, void* arg ARG_UNUSED)
4900 {
4901 if (IS_NOT_NULL(node)) {
4902 CClassNode* cc = NCCLASS(node);
4903 if (IS_NOT_NULL(cc->mbuf)) xfree(cc->mbuf);
4904 xfree(node);
4905 }
4906
4907 if (IS_NOT_NULL(key)) xfree(key);
4908 return ST_DELETE;
4909 }
4910
4911 extern int
4912 onig_free_shared_cclass_table(void)
4913 {
4914 if (IS_NOT_NULL(OnigTypeCClassTable)) {
4915 onig_st_foreach(OnigTypeCClassTable, i_free_shared_class, 0);
4916 onig_st_free_table(OnigTypeCClassTable);
4917 OnigTypeCClassTable = NULL;
4918 }
4919
4920 return 0;
4921 }
4922
4923 #endif /* USE_SHARED_CCLASS_TABLE */
4924
4925
4926 #ifndef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS
4927 static int
4928 clear_not_flag_cclass(CClassNode* cc, OnigEncoding enc)
4929 {
4930 BBuf *tbuf;
4931 int r;
4932
4933 if (IS_NCCLASS_NOT(cc)) {
4934 bitset_invert(cc->bs);
4935
4936 if (! ONIGENC_IS_SINGLEBYTE(enc)) {
4937 r = not_code_range_buf(enc, cc->mbuf, &tbuf);
4938 if (r != 0) return r;
4939
4940 bbuf_free(cc->mbuf);
4941 cc->mbuf = tbuf;
4942 }
4943
4944 NCCLASS_CLEAR_NOT(cc);
4945 }
4946
4947 return 0;
4948 }
4949 #endif /* CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS */
4950
4951 typedef struct {
4952 ScanEnv* env;
4953 CClassNode* cc;
4954 Node* alt_root;
4955 Node** ptail;
4956 } IApplyCaseFoldArg;
4957
4958 static int
4959 i_apply_case_fold(OnigCodePoint from, OnigCodePoint to[],
4960 int to_len, void* arg)
4961 {
4962 IApplyCaseFoldArg* iarg;
4963 ScanEnv* env;
4964 CClassNode* cc;
4965 BitSetRef bs;
4966
4967 iarg = (IApplyCaseFoldArg* )arg;
4968 env = iarg->env;
4969 cc = iarg->cc;
4970 bs = cc->bs;
4971
4972 if (to_len == 1) {
4973 int is_in = onig_is_code_in_cc(env->enc, from, cc);
4974 #ifdef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS
4975 if ((is_in != 0 && !IS_NCCLASS_NOT(cc)) ||
4976 (is_in == 0 && IS_NCCLASS_NOT(cc))) {
4977 if (ONIGENC_MBC_MINLEN(env->enc) > 1 || *to >= SINGLE_BYTE_SIZE) {
4978 add_code_range(&(cc->mbuf), env, *to, *to);
4979 }
4980 else {
4981 BITSET_SET_BIT(bs, *to);
4982 }
4983 }
4984 #else
4985 if (is_in != 0) {
4986 if (ONIGENC_MBC_MINLEN(env->enc) > 1 || *to >= SINGLE_BYTE_SIZE) {
4987 if (IS_NCCLASS_NOT(cc)) clear_not_flag_cclass(cc, env->enc);
4988 add_code_range(&(cc->mbuf), env, *to, *to);
4989 }
4990 else {
4991 if (IS_NCCLASS_NOT(cc)) {
4992 BITSET_CLEAR_BIT(bs, *to);
4993 }
4994 else
4995 BITSET_SET_BIT(bs, *to);
4996 }
4997 }
4998 #endif /* CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS */
4999 }
5000 else {
5001 int r, i, len;
5002 UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];
5003 Node *snode = NULL_NODE;
5004
5005 if (onig_is_code_in_cc(env->enc, from, cc)
5006 #ifdef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS
5007 && !IS_NCCLASS_NOT(cc)
5008 #endif
5009 ) {
5010 for (i = 0; i < to_len; i++) {
5011 len = ONIGENC_CODE_TO_MBC(env->enc, to[i], buf);
5012 if (i == 0) {
5013 snode = onig_node_new_str(buf, buf + len);
5014 CHECK_NULL_RETURN_MEMERR(snode);
5015
5016 /* char-class expanded multi-char only
5017 compare with string folded at match time. */
5018 NSTRING_SET_AMBIG(snode);
5019 }
5020 else {
5021 r = onig_node_str_cat(snode, buf, buf + len);
5022 if (r < 0) {
5023 onig_node_free(snode);
5024 return r;
5025 }
5026 }
5027 }
5028
5029 *(iarg->ptail) = onig_node_new_alt(snode, NULL_NODE);
5030 CHECK_NULL_RETURN_MEMERR(*(iarg->ptail));
5031 iarg->ptail = &(NCDR((*(iarg->ptail))));
5032 }
5033 }
5034
5035 return 0;
5036 }
5037
5038 static int
5039 parse_exp(Node** np, OnigToken* tok, int term,
5040 UChar** src, UChar* end, ScanEnv* env)
5041 {
5042 int r, len, group = 0;
5043 Node* qn;
5044 Node** targetp;
5045
5046 *np = NULL;
5047 if (tok->type == (enum TokenSyms )term)
5048 goto end_of_token;
5049
5050 switch (tok->type) {
5051 case TK_ALT:
5052 case TK_EOT:
5053 end_of_token:
5054 *np = node_new_empty();
5055 return tok->type;
5056 break;
5057
5058 case TK_SUBEXP_OPEN:
5059 r = parse_enclose(np, tok, TK_SUBEXP_CLOSE, src, end, env);
5060 if (r < 0) return r;
5061 if (r == 1) group = 1;
5062 else if (r == 2) { /* option only */
5063 Node* target;
5064 OnigOptionType prev = env->option;
5065
5066 env->option = NENCLOSE(*np)->option;
5067 r = fetch_token(tok, src, end, env);
5068 if (r < 0) return r;
5069 r = parse_subexp(&target, tok, term, src, end, env);
5070 env->option = prev;
5071 if (r < 0) return r;
5072 NENCLOSE(*np)->target = target;
5073 return tok->type;
5074 }
5075 break;
5076
5077 case TK_SUBEXP_CLOSE:
5078 if (! IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_UNMATCHED_CLOSE_SUBEXP))
5079 return ONIGERR_UNMATCHED_CLOSE_PARENTHESIS;
5080
5081 if (tok->escaped) goto tk_raw_byte;
5082 else goto tk_byte;
5083 break;
5084
5085 case TK_STRING:
5086 tk_byte:
5087 {
5088 *np = node_new_str(tok->backp, *src);
5089 CHECK_NULL_RETURN_MEMERR(*np);
5090
5091 while (1) {
5092 r = fetch_token(tok, src, end, env);
5093 if (r < 0) return r;
5094 if (r != TK_STRING) break;
5095
5096 r = onig_node_str_cat(*np, tok->backp, *src);
5097 if (r < 0) return r;
5098 }
5099
5100 string_end:
5101 targetp = np;
5102 goto repeat;
5103 }
5104 break;
5105
5106 case TK_RAW_BYTE:
5107 tk_raw_byte:
5108 {
5109 *np = node_new_str_raw_char((UChar )tok->u.c);
5110 CHECK_NULL_RETURN_MEMERR(*np);
5111 len = 1;
5112 while (1) {
5113 if (len >= ONIGENC_MBC_MINLEN(env->enc)) {
5114 if (len == enclen(env->enc, NSTR(*np)->s)) {
5115 r = fetch_token(tok, src, end, env);
5116 NSTRING_CLEAR_RAW(*np);
5117 goto string_end;
5118 }
5119 }
5120
5121 r = fetch_token(tok, src, end, env);
5122 if (r < 0) return r;
5123 if (r != TK_RAW_BYTE) {
5124 /* Don't use this, it is wrong for little endian encodings. */
5125 #ifdef USE_PAD_TO_SHORT_BYTE_CHAR
5126 int rem;
5127 if (len < ONIGENC_MBC_MINLEN(env->enc)) {
5128 rem = ONIGENC_MBC_MINLEN(env->enc) - len;
5129 (void )node_str_head_pad(NSTR(*np), rem, (UChar )0);
5130 if (len + rem == enclen(env->enc, NSTR(*np)->s)) {
5131 NSTRING_CLEAR_RAW(*np);
5132 goto string_end;
5133 }
5134 }
5135 #endif
5136 return ONIGERR_TOO_SHORT_MULTI_BYTE_STRING;
5137 }
5138
5139 r = node_str_cat_char(*np, (UChar )tok->u.c);
5140 if (r < 0) return r;
5141
5142 len++;
5143 }
5144 }
5145 break;
5146
5147 case TK_CODE_POINT:
5148 {
5149 UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];
5150 int num = ONIGENC_CODE_TO_MBC(env->enc, tok->u.code, buf);
5151 if (num < 0) return num;
5152 #ifdef NUMBERED_CHAR_IS_NOT_CASE_AMBIG
5153 *np = node_new_str_raw(buf, buf + num);
5154 #else
5155 *np = node_new_str(buf, buf + num);
5156 #endif
5157 CHECK_NULL_RETURN_MEMERR(*np);
5158 }
5159 break;
5160
5161 case TK_QUOTE_OPEN:
5162 {
5163 OnigCodePoint end_op[2];
5164 UChar *qstart, *qend, *nextp;
5165
5166 end_op[0] = (OnigCodePoint )MC_ESC(env->syntax);
5167 end_op[1] = (OnigCodePoint )'E';
5168 qstart = *src;
5169 qend = find_str_position(end_op, 2, qstart, end, &nextp, env->enc);
5170 if (IS_NULL(qend)) {
5171 nextp = qend = end;
5172 }
5173 *np = node_new_str(qstart, qend);
5174 CHECK_NULL_RETURN_MEMERR(*np);
5175 *src = nextp;
5176 }
5177 break;
5178
5179 case TK_CHAR_TYPE:
5180 {
5181 switch (tok->u.prop.ctype) {
5182 case ONIGENC_CTYPE_WORD:
5183 *np = node_new_ctype(tok->u.prop.ctype, tok->u.prop.not);
5184 CHECK_NULL_RETURN_MEMERR(*np);
5185 break;
5186
5187 case ONIGENC_CTYPE_SPACE:
5188 case ONIGENC_CTYPE_DIGIT:
5189 case ONIGENC_CTYPE_XDIGIT:
5190 {
5191 CClassNode* cc;
5192
5193 #ifdef USE_SHARED_CCLASS_TABLE
5194 const OnigCodePoint *mbr;
5195 OnigCodePoint sb_out;
5196
5197 r = ONIGENC_GET_CTYPE_CODE_RANGE(env->enc, tok->u.prop.ctype,
5198 &sb_out, &mbr);
5199 if (r == 0 &&
5200 ONIGENC_CODE_RANGE_NUM(mbr)
5201 >= THRESHOLD_RANGE_NUM_FOR_SHARE_CCLASS) {
5202 type_cclass_key key;
5203 type_cclass_key* new_key;
5204
5205 key.enc = env->enc;
5206 key.not = tok->u.prop.not;
5207 key.type = tok->u.prop.ctype;
5208
5209 THREAD_ATOMIC_START;
5210
5211 if (IS_NULL(OnigTypeCClassTable)) {
5212 OnigTypeCClassTable
5213 = onig_st_init_table_with_size(&type_type_cclass_hash, 10);
5214 if (IS_NULL(OnigTypeCClassTable)) {
5215 THREAD_ATOMIC_END;
5216 return ONIGERR_MEMORY;
5217 }
5218 }
5219 else {
5220 if (onig_st_lookup(OnigTypeCClassTable, (st_data_t )&key,
5221 (st_data_t* )np)) {
5222 THREAD_ATOMIC_END;
5223 break;
5224 }
5225 }
5226
5227 *np = node_new_cclass_by_codepoint_range(tok->u.prop.not,
5228 sb_out, mbr);
5229 if (IS_NULL(*np)) {
5230 THREAD_ATOMIC_END;
5231 return ONIGERR_MEMORY;
5232 }
5233
5234 cc = NCCLASS(*np);
5235 NCCLASS_SET_SHARE(cc);
5236 new_key = (type_cclass_key* )xmalloc(sizeof(type_cclass_key));
5237 xmemcpy(new_key, &key, sizeof(type_cclass_key));
5238 onig_st_add_direct(OnigTypeCClassTable, (st_data_t )new_key,
5239 (st_data_t )*np);
5240
5241 THREAD_ATOMIC_END;
5242 }
5243 else {
5244 #endif
5245 *np = node_new_cclass();
5246 CHECK_NULL_RETURN_MEMERR(*np);
5247 cc = NCCLASS(*np);
5248 add_ctype_to_cc(cc, tok->u.prop.ctype, 0, env);
5249 if (tok->u.prop.not != 0) NCCLASS_SET_NOT(cc);
5250 #ifdef USE_SHARED_CCLASS_TABLE
5251 }
5252 #endif
5253 }
5254 break;
5255
5256 default:
5257 return ONIGERR_PARSER_BUG;
5258 break;
5259 }
5260 }
5261 break;
5262
5263 case TK_CHAR_PROPERTY:
5264 r = parse_char_property(np, tok, src, end, env);
5265 if (r != 0) return r;
5266 break;
5267
5268 case TK_CC_OPEN:
5269 {
5270 CClassNode* cc;
5271
5272 r = parse_char_class(np, tok, src, end, env);
5273 if (r != 0) return r;
5274
5275 cc = NCCLASS(*np);
5276 if (IS_IGNORECASE(env->option)) {
5277 IApplyCaseFoldArg iarg;
5278
5279 iarg.env = env;
5280 iarg.cc = cc;
5281 iarg.alt_root = NULL_NODE;
5282 iarg.ptail = &(iarg.alt_root);
5283
5284 r = ONIGENC_APPLY_ALL_CASE_FOLD(env->enc, env->case_fold_flag,
5285 i_apply_case_fold, &iarg);
5286 if (r != 0) {
5287 onig_node_free(iarg.alt_root);
5288 return r;
5289 }
5290 if (IS_NOT_NULL(iarg.alt_root)) {
5291 Node* work = onig_node_new_alt(*np, iarg.alt_root);
5292 if (IS_NULL(work)) {
5293 onig_node_free(iarg.alt_root);
5294 return ONIGERR_MEMORY;
5295 }
5296 *np = work;
5297 }
5298 }
5299 }
5300 break;
5301
5302 case TK_ANYCHAR:
5303 *np = node_new_anychar();
5304 CHECK_NULL_RETURN_MEMERR(*np);
5305 break;
5306
5307 case TK_ANYCHAR_ANYTIME:
5308 *np = node_new_anychar();
5309 CHECK_NULL_RETURN_MEMERR(*np);
5310 qn = node_new_quantifier(0, REPEAT_INFINITE, 0);
5311 CHECK_NULL_RETURN_MEMERR(qn);
5312 NQTFR(qn)->target = *np;
5313 *np = qn;
5314 break;
5315
5316 case TK_BACKREF:
5317 len = tok->u.backref.num;
5318 *np = node_new_backref(len,
5319 (len > 1 ? tok->u.backref.refs : &(tok->u.backref.ref1)),
5320 tok->u.backref.by_name,
5321 #ifdef USE_BACKREF_WITH_LEVEL
5322 tok->u.backref.exist_level,
5323 tok->u.backref.level,
5324 #endif
5325 env);
5326 CHECK_NULL_RETURN_MEMERR(*np);
5327 break;
5328
5329 #ifdef USE_SUBEXP_CALL
5330 case TK_CALL:
5331 {
5332 int gnum = tok->u.call.gnum;
5333
5334 if (gnum < 0) {
5335 gnum = BACKREF_REL_TO_ABS(gnum, env);
5336 if (gnum <= 0)
5337 return ONIGERR_INVALID_BACKREF;
5338 }
5339 *np = node_new_call(tok->u.call.name, tok->u.call.name_end, gnum);
5340 CHECK_NULL_RETURN_MEMERR(*np);
5341 env->num_call++;
5342 }
5343 break;
5344 #endif
5345
5346 case TK_ANCHOR:
5347 *np = onig_node_new_anchor(tok->u.anchor);
5348 break;
5349
5350 case TK_OP_REPEAT:
5351 case TK_INTERVAL:
5352 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS)) {
5353 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS))
5354 return ONIGERR_TARGET_OF_REPEAT_OPERATOR_NOT_SPECIFIED;
5355 else
5356 *np = node_new_empty();
5357 }
5358 else {
5359 goto tk_byte;
5360 }
5361 break;
5362
5363 default:
5364 return ONIGERR_PARSER_BUG;
5365 break;
5366 }
5367
5368 {
5369 targetp = np;
5370
5371 re_entry:
5372 r = fetch_token(tok, src, end, env);
5373 if (r < 0) return r;
5374
5375 repeat:
5376 if (r == TK_OP_REPEAT || r == TK_INTERVAL) {
5377 if (is_invalid_quantifier_target(*targetp))
5378 return ONIGERR_TARGET_OF_REPEAT_OPERATOR_INVALID;
5379
5380 qn = node_new_quantifier(tok->u.repeat.lower, tok->u.repeat.upper,
5381 (r == TK_INTERVAL ? 1 : 0));
5382 CHECK_NULL_RETURN_MEMERR(qn);
5383 NQTFR(qn)->greedy = tok->u.repeat.greedy;
5384 r = set_quantifier(qn, *targetp, group, env);
5385 if (r < 0) {
5386 onig_node_free(qn);
5387 return r;
5388 }
5389
5390 if (tok->u.repeat.possessive != 0) {
5391 Node* en;
5392 en = node_new_enclose(ENCLOSE_STOP_BACKTRACK);
5393 if (IS_NULL(en)) {
5394 onig_node_free(qn);
5395 return ONIGERR_MEMORY;
5396 }
5397 NENCLOSE(en)->target = qn;
5398 qn = en;
5399 }
5400
5401 if (r == 0) {
5402 *targetp = qn;
5403 }
5404 else if (r == 1) {
5405 onig_node_free(qn);
5406 }
5407 else if (r == 2) { /* split case: /abc+/ */
5408 Node *tmp;
5409
5410 *targetp = node_new_list(*targetp, NULL);
5411 if (IS_NULL(*targetp)) {
5412 onig_node_free(qn);
5413 return ONIGERR_MEMORY;
5414 }
5415 tmp = NCDR(*targetp) = node_new_list(qn, NULL);
5416 if (IS_NULL(tmp)) {
5417 onig_node_free(qn);
5418 return ONIGERR_MEMORY;
5419 }
5420 targetp = &(NCAR(tmp));
5421 }
5422 goto re_entry;
5423 }
5424 }
5425
5426 return r;
5427 }
5428
5429 static int
5430 parse_branch(Node** top, OnigToken* tok, int term,
5431 UChar** src, UChar* end, ScanEnv* env)
5432 {
5433 int r;
5434 Node *node, **headp;
5435
5436 *top = NULL;
5437 r = parse_exp(&node, tok, term, src, end, env);
5438 if (r < 0) return r;
5439
5440 if (r == TK_EOT || r == term || r == TK_ALT) {
5441 *top = node;
5442 }
5443 else {
5444 *top = node_new_list(node, NULL);
5445 headp = &(NCDR(*top));
5446 while (r != TK_EOT && r != term && r != TK_ALT) {
5447 r = parse_exp(&node, tok, term, src, end, env);
5448 if (r < 0) return r;
5449
5450 if (NTYPE(node) == NT_LIST) {
5451 *headp = node;
5452 while (IS_NOT_NULL(NCDR(node))) node = NCDR(node);
5453 headp = &(NCDR(node));
5454 }
5455 else {
5456 *headp = node_new_list(node, NULL);
5457 headp = &(NCDR(*headp));
5458 }
5459 }
5460 }
5461
5462 return r;
5463 }
5464
5465 /* term_tok: TK_EOT or TK_SUBEXP_CLOSE */
5466 static int
5467 parse_subexp(Node** top, OnigToken* tok, int term,
5468 UChar** src, UChar* end, ScanEnv* env)
5469 {
5470 int r;
5471 Node *node, **headp;
5472
5473 *top = NULL;
5474 r = parse_branch(&node, tok, term, src, end, env);
5475 if (r < 0) {
5476 onig_node_free(node);
5477 return r;
5478 }
5479
5480 if (r == term) {
5481 *top = node;
5482 }
5483 else if (r == TK_ALT) {
5484 *top = onig_node_new_alt(node, NULL);
5485 headp = &(NCDR(*top));
5486 while (r == TK_ALT) {
5487 r = fetch_token(tok, src, end, env);
5488 if (r < 0) return r;
5489 r = parse_branch(&node, tok, term, src, end, env);
5490 if (r < 0) return r;
5491
5492 *headp = onig_node_new_alt(node, NULL);
5493 headp = &(NCDR(*headp));
5494 }
5495
5496 if (tok->type != (enum TokenSyms )term)
5497 goto err;
5498 }
5499 else {
5500 err:
5501 if (term == TK_SUBEXP_CLOSE)
5502 return ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS;
5503 else
5504 return ONIGERR_PARSER_BUG;
5505 }
5506
5507 return r;
5508 }
5509
5510 static int
5511 parse_regexp(Node** top, UChar** src, UChar* end, ScanEnv* env)
5512 {
5513 int r;
5514 OnigToken tok;
5515
5516 r = fetch_token(&tok, src, end, env);
5517 if (r < 0) return r;
5518 r = parse_subexp(top, &tok, TK_EOT, src, end, env);
5519 if (r < 0) return r;
5520 return 0;
5521 }
5522
5523 extern int
5524 onig_parse_make_tree(Node** root, const UChar* pattern, const UChar* end,
5525 regex_t* reg, ScanEnv* env)
5526 {
5527 int r;
5528 UChar* p;
5529
5530 #ifdef USE_NAMED_GROUP
5531 names_clear(reg);
5532 #endif
5533
5534 scan_env_clear(env);
5535 env->option = reg->options;
5536 env->case_fold_flag = reg->case_fold_flag;
5537 env->enc = reg->enc;
5538 env->syntax = reg->syntax;
5539 env->pattern = (UChar* )pattern;
5540 env->pattern_end = (UChar* )end;
5541 env->reg = reg;
5542
5543 *root = NULL;
5544 p = (UChar* )pattern;
5545 r = parse_regexp(root, &p, (UChar* )end, env);
5546 reg->num_mem = env->num_mem;
5547 return r;
5548 }
5549
5550 extern void
5551 onig_scan_env_set_error_string(ScanEnv* env, int ecode ARG_UNUSED,
5552 UChar* arg, UChar* arg_end)
5553 {
5554 env->error = arg;
5555 env->error_end = arg_end;
5556 }
EFI Development Kit II mirror for PVE